===============================================================================
Hagerman Technology WebLog 2007.  Comments?
===============================================================================

Dec 28
------

Did a bunch of work today, kinda forget what it was.  The highlight of my day
just happened.  Been listening to LPs again (yea!), going through a bunch of
old stuff that's been boxed up for decades.  Wanna find me the good tracks I
can use for shows.  Did I really buy some of this stuff?  The Billy Joel 
sounded horrible.  Yuk.  No, I'm not talking performance, I mean the sonics I
can rip off the LP with my Dynavector.  Tatto You was ok, better than most.  A
lot of stuff I have is just so-so.  The Jackson Brown sorta dead and unlively,
but it was nice to play it anyway.  Then I find this Brothers In Arms.  Shoot,
I don't even remember buying it.  The cover and sleeve are near mint.  I check
the stamper, SH11 both sides.  Hmmm.  Drop the needle, and ... very quiet.
Huh?  Then pow!  Holy crap, I can't believe how solid and dynamic this pressing
is!  It is easily as good as anything in my library, 200g re-issues and MoFis
included.  Wide open sonics, superb transients and punch.  Oh yeah, baby, this
is the way it is supposed to be.  Ok, so I got me one stellar piece of vinyl.
I hope I find some more.  Gonna run through the entire collection and grade.

So I run through a whole bunch of crud.  Why is it most of my old collection
has a dull lifeless sound?  It it because there is a stale layer of beer on
them?  Or did my Dual 1228 rip them up?  The LPs from a certain era of my life
were not treated too well.  Who would have guessed I end up here.  Anyway, I
get to some new stuff.  So I put on one of those 45s from Classic, a Jacinta
track.  Wholy sheet!  It does get better.  These 45s are unbelievable!  Quiet,
smooth.  Wonderful sound.  I have to get more of these.

So I put the Dire Straights back on, just to make sure I wasn't hallucinating.
Yep, got the whole body goose bumps again.  Incredible musical buildup at the
start of that MTV song (money for nothin).

Dec 27
------

Already thinking about next generation FRYKLEANER.  The FRYBABY is nice, but
the form factor for the FRYGOLD seems to be unpopular.  Either that or the
burn-in craze has greatly diminished.  So I thought of what the next model
should be.  I think the VACUTRACE chassis is perfect.  So the smaller one is
even better.  Rugged and portable.  Easily stored.  Can make with very nice
finished panel.  I think more output power is worth a try.  Have this idea to
use an LM1875 audio power opamp to drive a tranny.  Center-tapped secondaries
will then boost voltage and drive differentially, with no local connection to
ground.  It'll have two controls, one for amplitude (MC, MM, LINE, and CABLE).
The high power CABLE setting should be able to deliver roughly 5x the power
over a FRYBABY.  I think 10Vpp between conductors.  Other control is a mode
switch (OFF, VOLTAGE, CURRENT), so you don't have to play games with various
cable hookup methods.  Let a switch determine voltage or current mode.  One
drawback is that I can't use a wall wart, as it won't fit inside the closed
cover.  But a power cord will!  So will have internal ac supply.  Has XLR
outputs, phono levels (of course).  And the new feature will be an outlet for
burning in power cords.  A very versatile design.  Small.  Way more power.  A
good set of specs for a new generation.  Ar ar.

Meanwhile, ordered a bunch of production parts.  Trying to build this whole
line prior to launch.  So I think I will delay website until I can get actual
photos of the finished items.  Looking at everything to see where I can 
possibly upgrade or make common with another machine.  Sent out RFQ for
production chassis parts too.  Might as well get batch #1 in house.  It'll help
speed up prototyping, like the DA-10.  

Since I re-did the Cinemag SU-10 yesterday, thinking if I really like their
iron, then I can use it in PL-10.  Or at least in a redesign of the COMPRESSOR.
No reason that machine can't be redone in the image of the new VACUTRACE or new
FRYKLEANER.  It'll be a real standout machine with that form factor.  Exactly
what you need to break into a new market.  

Looked into remote controls.  I'm thinking all these fancy plastic moulds look
cool, but are not heavy enough.  I like the solid, weighty remotes.  And this
one will have panache.  Sadly, I limit myself to no custom milling of billets.
Maybe later.  For now, I'll copy the PI-10 and SU-10 concept.  Black painted
aluminum casting for bottom, red anodized panel on top.  Will look ok, not
spectacular.  Will feel pretty good, though.  I fired up the new band saw and
cut out a few wood blocks to check out various sizes for palm fit and feel for
knob control.  I think I'll do the 2" x 4" x 1".  Should be a really good
visual match with the other components.  

Dec 26
------

I redesigned the SU-10 (step-up tranny) to look just like the PICCOLO (PI-10).
That other chassis, however nice, was just adding too much cost.  So I'll make
the same top panel, same size as the PICCOLO.  In fact, I'm upgrading RCAs on
the PI-10 too.  This is cool.  I will now have two MC step-up devices, roughly
the same cost, looking alike, but of course one active the other passive.  I'll
have LO-MED-HI loading switch to kick in resistors on the secondary.  So I'm
going to try LO-MED-HI gain switch on the primary.  Not that I really like to
have switch contacts on phono signals.  It's worth a try, if I can get the 
right metallic junctions.  My Cinemags came in already.  Just need to order a 
panel and wire them up.

Working on an offset-nuller for the PICCOLO.  I neglected to realize that any
dc offset on the output could be a problem for some SS phonostages (maybe I 
work too much in tubes?).  The simple fix is to add 10 ohms in series with R10.
But I want to take that farther.  So something like a pot in parallel with R11
would be better.  Take a 2.21k from the source of Q3B to the wiper of a 10k
pot, which then ties one leg to VEE.  This gives an adjustment range from 217
ohms to 200 ohms.  Should be very easy to dial in a zero.  No need for those
pesky 10-turn pots, either.  I think a simple hot glue of the pot right under
Q3 (oither side of board) is ok.  However, the battery holder standoffs will
have to be much taller.  At least 1/4".

Ok, finally got the VACUTRACE manual updated.  Machine back on the website and
ready to go.  I have this first unit that has minor cosmetic defects, that I 
can give a discount for.  

VACUTRACE new!

Dec 24
------

Catching up with production.  Even today I shipped out 5 items.  Gotta go fill
some stockings now...

Dec 21
------

Ok, I am alive again.  Somewhat.  Taking a while here to recover.  So instead
of doing the things I should have been, I layed out the DA-10.  Yeah, it wasn't
next on the schedule.  But I was thinking.  I really need this for myself.  No
more CHIMEs in the house.  And I had already figured out the new style power
supply design.  So I churned out a schematic and did a proto layout.  Should be
an easy build.  Just like an HA-10 with HAGDAC onboard.

So I was thinking, what can I do to improve on HAGDAC?  What can be upgraded?
Well, there wasn't anything I could find.  That's a first.  No better part to
use.  Only possibility are the output connectors.  The one change I will make
so it fits the DA-10 is to change the pullup resistors on the comparator
outputs from 4.7k to 1.5k.  That way they can directly drive the LEDs.  I don't
need any extra stuff on the DA-10 board.  Front panel is blank, with 4 LEDS.
Low voltage, high voltage, USB ok, and PLL lock.  If they are all on then sound
should come out.  Oh wait, if I could, I would change the linear regulators on
the HAGDAC to the discrete type I came up with for the HAGCLOCK.  They are very
nice and high performing.  But can't because their dropout voltage doesn't work
in this case.

Anyway, the layout came out great.  Routed easier than I expected.  Suprising,
since I did an unusual arrangement for the chokes.  I ran the B+ down the
middle and the heaters on either side.  But instead of a single tranny for 
heaters, I used two of them.  It is a psuedo center-tapped connection.  Both
are full wave but outputs cross the board to feed both sections of chokes.  I 
get both +6V and -6V heater supplies at 0.5 amp.  This requires 11 mH of choke,
so I run two 155B 6mH chokes in series on each side.  At 25% of capacity their
actual henries will be higher.  Shux, that takes six chunks of iron to make a
relatively low power (6W) +/-6V supply.  Believe me, this is the most expensive
way to do it.  It is also the only one that will give clean power factor and no
switching spikes, inherent voltage regulation, soft start, and zero feedback.
Add 40,000uF of capacitor to each side and voltage ripple should be a mere
11mV.  

DA-10 layout

As you can see, with 3 chunks of iron for B+, this board will weigh about the
same as a CASTANET, with 9 chokes and trannys.  All super clean passively
regulated supplies.  The outputs are 7788 fed by a simple no feedback CCS at 
15mA.  Hence, the B+ is identical to CASTANET.  Tubes wired in triode mode for
lots of gain.  Should deliver about 2Vrms full scale at 1.5k output impedance.
That's awesome, and I did it without using the cathode follower.  So that's
improvement #2 over the CHIME.  Even with the same HAGDAC, this should sound a
little better.  Only thing is, you can't roll tubes as easily.  Not so many
brands of this type.  My one concern is induced hum.  I have all the chokes
wired out of phase to magnetically buck each other, and cancel far field.  I
also put them as far away and at the 45 degree angle to the chokes on the
HAGDAC.  Even so, I might get a magnetic coupling beween supply and IV filter
chokes.  We shall see.  I maximized spacing.  Nothing is for free.  To get that
truly organic music, you have to pull out the stops.  

Oh yeah, I also had to not route the outputs all the way to back.  I decided to
just use coax wiring.  It's not a big deal.  Just takes a little longer to wire
up.

Here's the final draft of schematic.  It looks hideously simple.  Exactly!
That's the whole point.  Yet it belies the hidden sophistication.  Much, much 
more going on in there than meets the eye.  This is the culmination of 10 years
of tube audio design (plus 15 more on non-audio).  This should be the best DAC 
I can design from a topology standpoint.  

DA-10 schematic

Dec 18
------

Oh man, just got the worst food poisoning ever.  Today I am drained.  Not gonna
work.  But I will get KOs order in the mail.

Dec 17
------

Spent more time listening last night.  Finally.  This rig is really starting to
open up.  I am getting really good micro details and the ability to hear deep
into a recording.  Mostly, I realize that the separation between instruments
and voices has taken a big step forward.  This is exactly what I heard at
Thom's with his X-1S.  Each sound within the mix remains distinct and does not
mix or get congested together.  Oh, and I'm an idiot.  That wasn't Aja, it was
Gaucho.  Duh.  This time I played the entire Crime of the Century.  It was this
album that got me hooked on hifi in 1976 at some shop in Minneapolis.  Might
still have a lack of overpowering slam.  I think there is still more room for
that extra punch.  The upgraded capacitors in the CO-10 really help with the
tonality and detail, adding no smear.  The notes have no overhang.  Very fast
transients and timing is dead on.  Had to lower VTA a little for improved
tracking.  Bass falls off a little in amplitude, but the articulation stays.
The good news is that this is the first time I've ever been able track this LP
properly.  It now made it through.  I don't know if that is due to the VTAF or
the 20XL.  I also have the new rack with spikes.  And new wiring.

Oh, as far as the VTAF, I grossly overestimated the slop in the bearing.  Yes,
it has some lateral shift and can tilt, but it is not as big as I thought.
Carefully measuring with a dial calipers I got 0.992 for the threads and 0.998
for the sleeve.  That leaves a gap of 0.006.  Exactly what Pete said it should
be.

Spent half the day just going through minro design details of the new product
line.  Trying very hard to give each piece a common look, such that they all
appear to have been designed at the same time and for the chassis style, even
though they all come from widely different backgrounds.  Shifting connectors,
lettering, etc. 1/4" this way, then that way.  Make sure all of the lettering
is common in approach.  It is working.  The lineup of TR-10, CO-10, and SU-10
all look about the same when side by side.  FL-10 (FLUGELHORN) is a little bit
of an oddball, in that logo placement doesn't line up perfectly with the HA-10
and other of the new machines.  You do the best you can.  PI-10 is the old
small box chassis, another oddball.  Also, the wife made me go back to yellow
lettering.  Anyway, put everything on a shelf and they are all HAGLABS.  All
red faces, yellow lettering, same logo, black covers.  Common ac plugs,
switches, connectors, yellow LEDs, etc.  

Even the VACUTRACE now looks similar.  Even has the HAL logo, but I will
initially start selling it again from HAGTECH.  

Dec 16
------

Did a quick check on SNR for the new CORNET.  Hmmm, not so great.  Hooked it up
to listen and mostly hiss with no input, but hum when shorted.  Aha!  The new
power tranny is in a nasty orientation.  I forgot the original was aimed for a
perfect magnetic null along the line of tubes (which is also why they were in a
line).  Ay, the new tranny spews mag field right across the board and inputs.
Fortunately, they are pretty far away, so a little re-wiring of the input
section is all it took.  So now, with a minor EQ tweak and the re-wiring, the
gain is 45dB and SNR 75dBA referenced to 5mV input.  That would be 0.9uV of
input referred noise.  Pretty good for tubes.

Oh yeah, I also tweaked my CASTANET by putting 0.22uF ppfmx caps in parallel 
with the 3.3uF output caps.  My system is now the 20XL on a P3 with VTAF into a
PI-10 / CO-10 / HA-10 and Sennheiser HD600.  Thinking I should rename the new
CORNET to CO-10.  Then also the PI-10 and TR-10 (TRUMPET).  It sorta makes 
sense.  Yes, I am having trouble naming things.  

Anyway, I added 1/4" nylon standoffs to my 20XL, such that the VTA can now be
adjusted both positive and negative.  Not sure this is a good thing sonically.
Nylon is a bit damped.  Methinks I lose some of the rigid coupling to the arm.
It also puts the stylus much lower that the vertical bearing axis, thus causing
velocity modulation on warps.

RB300 standoffs

Finally, I am back to listening.  Off the bat this is quite good.  Way better
than the setup I had in the lab, which was lifeless and thin.  First I did the
test record and adjusted antiskate.  Was able to track all of the tracking
tests.  The +14dB antiskate test was too much, however.  +16dB really booched
it.  Ok, it ain't perfect.  On some familiar records I tried adjusting VTA.
Really expected to hear change in volume, or tone, or vocals or treble.  What I
got was mostly a change in bass articulation.  High it got wooly.  Low and it
was thin (yes, it was thin in the lab!).  Middle wasn't too hard to spot.
Really didn't have to change it much between records.  So I dig into my library
and pulled out some LPs I've never heard.  The Heifetz on LSC was ok, but 
lacking in excitement.  I must not be there yet.  Also, overall I am missing 
that last bit of slam and dynamic control.  Tried an old Doc Severinsen.  Nice
tunes and great playing, but zero bottom end.  Like it needed a different EQ.
Then I tried my copy of Beatles 63-66.  Oh my.  Lifeless, boring.  Vocals were
receeded (could be polarity?).  Wondering how far off I am.  Getting worried.
Threw on Aja, and voila!  It all came back.  Solid, thumping bass with control,
decent dynamics (not perfect yet), and wide open sound.  And this was an
original.  Maybe I don't need the re-issue?  So good LPs sound really good.
The bad ones sound bad.  Overall, the noise floor is fine, zero fatigue, easy
to listen to without tiring at all.  It sounds good both loud and soft.  I
think I'm going to play all of AJA now.  See ya later.

Dec 14
------

Buying tons of parts today.  That's just the way it is.  Making good progress
on parts lists and pricing.  Then I was thinking about the cute little chassis
I bought to match the other HAL stuff.  The PICCOLO didn't fit, so it stays the
same, mechanically.  But how about a step-up tranny?  Ok.  So I am buying some
of the Cinemag trannys that so many people like.  I'll wire them up a little
differently, because I don't want switches or jumpers on the primaries.  Then,
in order for them to mate better with some carts, I'll add a loading switch on
the scondary.  Right now the tranny is for 50k loading.  This limits you to
certain turns ratios.  To get the super low impedance carts you run 1:36 gain.
Maybe bandwidth starts to suffer.  So instead, I'm going to try variable
loading on the secondary, in parallel with the 47k of a phonostage.  So two
gains, 19dB and 25dB (separate RCA jacks).  Three loading settings, LO, MED, HI
(47k).  This should result in some useful combos.  I'll build one and try it
out myself.  Besides, I don't have a step-up of my own right now.  

And so I got the CORNET-HAL up and running last night.  Tweaked on resistor
values to bring it into proper bias and operating points.  Runs perfect!  I do
need to make some standoff and board position adjustments, but otherwise this
thing is ready for production.  Listened to it.  Super quiet.  The noise floor
when used with the PICCOLO was superb.  I had to crank it up way past a loud
listening level for it to be heard.  But even more important was how the groove
noise itself during play was remarkably subdued.  I'm not sure why this is,
but it is a great sonic feature of both the CORNET and TRUMPET.  Quiet during
play.  The coloring is wrong on this unit, so I will keep it for myself.  This
will now be my main phono for listening.  Combined with the HA-10 (yeah, I'm
keeping unit #1) it makes for a great way to test out the new SU-10 (step-up)
proto.  Yeah baby, starting to get close to production.  I'll have to put in
a huge order to Hammond one of these days.  Ok, now working on resistor
upgrades to these phono stages.

Compiled a new parts list for VACUTRACE.  Each part price had to be updated, as
I was working with numbers from seven years ago.  This design sure can stand 
the test of time!  Anyway, the pricing comes out to $1499 from HAGTECH, even
though it has the HAL logo.  That's the OEM direct pricing.  Two boxes, one is
portable VACUTRACE (close the lid with carrying handle) and the other is a
smaller carrying case for the adapter cards.  Price for the SU-10 is high, at
$1500 retail.  More than double a PICCOLO-HAL!  But hey, that's what it costs.
I don't fudge the numbers.  It costs what it costs.  (Note, if I sold the new
VACUTRACE through dealers it would be $3000 msrp.)  

Dec 13
------

A check at my web stats.  I get about 2000 page hits per day (1000 visitors).
Roughly 1000 hits per week on my blog!  Hey, that must be some sort of
milestone?  Most people come from google.  Then asylum, audiocircle, diyaudio,
audiogon, in that order.  And many more.  Hey, I even got a few referrals from
slimdevices.com!  Most of you come from comcast.net.  49% are using internet
explorer, 76% run xp, 9% mac, and only 6% vista.  One of you is still using
win95.  

For a long time I've been wondering about my selection for microprocessor to be
used in the PL-10.  I love the PIC line, used them before, and rave about the
Harvard achitecture.  They are easy to program, and now have the option of
using C.  To me, they are a perfect bridge between code and hardware,
especially analog hardware.  Hard to explain.  Anyway, the problem is that they
have a gazillion choices.  What to pick that won't go out of production?  Or
that is in stock.  Or popular.  I finally narrowed it down to a 16F687, based
on my requirements, mostly pinout and functions.  Seems perfect.  But how long
before it is discontinued?  Well, allay my fears.  An article in Nuts&Volts
covered this subject, and the author (rabid user) picked the 16F690 and one of
his prime choices.  Turns out it is exactly the same part but with larger
program memory storage!  And then, double checking on the microchip.com site,
I see the 690 as a featured item.  Bingo.  I just upgrade to the 690 and my
choice is safe.  Not to get a development kit.  

Here's some shots of the new CORNET-HAL.  Small, very solid unit.  This chassis
wraps around the board very tightly, making it one of the smallest tube phonos
to hit the market.  I'll be painting the covers black.  The silver panels was
a possible alternate choice for color schemes.  Will likely need some resistor
changes in the power supply, as this tranny has a bit more B+ voltage than the
original.

CORNET-HAL front
CORNET-HAL back
CORNET-HAL guts

All wired up and ready to go.  Meanwhile, I fired up the VACUTRACE, but for 
some reason the heaters didn't kick in.  Maybe a problem with the board, or I
didn't get the connector fit just right.  Making mechanical changes to washers
and standoffs to recreate the perfect fit (this panel is much thicker than the
last one.  Should have this done tomorrow.  Anyway, it looks something like
this:

VACUTRACE (new)
VACUTRACE guts

Now I need to work out pricing.  Should I include a small tandem case to carry
the adapter cards?  Or is that too much?  Do I make the jumper card standard?

One thing I did notice, in a dimly lit room, the yellow lettering smokes!  It
comes alive and all of a sudden looks better than the other alternatives.  My
preference for the raw engraving is a bit harder to read.  It also has a mirror
effect when it catches the light just right.  Different.  But maybe too out of
the box?  I need to make these machines please you, not me.  I have to admit,
the yellow does grow on you.  The raw silver is stunning at first, but then
perhaps the illusion wears off a little.  I still like it, but not sure which
is best for HAGLABS.  They are both out of the ordinary.  

Dec 12
------

Built the CASTANET, took photos, finished the manual.  Started shipping them.
Ok, so I ran out of fuse clips, but I still got a bunch out the door.  They
look great and build easy.  Except for the part about the phone jack mounting,
which requires a bit of balance.  Anyway, finally.  I was really expecting to
be here a month ago.  

Catching up on all production, trying to get machines there by xmas!

Meanwhile, I can get back to finishing the new VACUTRACE, CORNET, and TRUMPET
protos.  All of which are 80% to 90% complete.  Also have most parts in for a
new ARCHIVER, a project that was cancelled, to be replaced by the FLUGELHORN.
It still helps me debug fit for the OEM version of ARCHIVER (black with silver
knobs & lettering).  Once these are done, I should be ready for production of
the HAL units (spending even more money on inventory).  And when I find some
spare time, I'll do the relayout of PA-10.

Dec 11
------

Spending a lot of time putting together accurate parts lists for the new
products.  It's the only way to know how to price things.  At the moment, the
expected retail prices (with dealer markup) are:

HA-10         $2100
PICCOLO-HAL   $600
CORNET-HAL    $2150
TRUMPET-HAL   $3400
FLUGELHORN    $2150

Suprising how the three machines ended up costing almost the same.  I'll likely
offer a 30% discount on all machines until I get some dealers signed up.  So
far, all of these products are starting to look killer.  Especially the CORNET-
HAL.  It will be very compact, and with all of the part upgrades (especially
caps) it is a steal.  That's $1505 for a tricked-out CORNET in fancy chassis,
with introductory discount.  The FLUGELHORN is also turning into a nice unit. 
I realized that I didn't purchase a gazillion ARCHIVER boards, so I am not
stuck having to chop them up.  In fact, I can re-design the board to be shorter
and easier to wire up.  The board then only costs $33 a piece, even in low
quantities.  The original ARHIVERs will then be used for a special OEM project.

So much building going on.  I'll be happy to get some of these operational.  At
least to the point where I can take photos and start up the website.  The whole
HAGLABS movement is finally converging upon a solution.  Been working on it a
long time.  Wrote the original business plan 1.5 years ago.  If you go back in
the blog you'll see I spent a lot of time talking and fretting about it.  Never
had any real forward momentum until I started the CASTANET.  It gave me the
new chassis style and a low-volume ramp-up solution that I could actually
afford.  So this is very exciting for me.  Almost ready to exit the tunnel.
Years in the making.

Wow, I just realized what a bargain the TRUMPET-HAL is gonna be.  It didn't
cost me too much to make all these chassis changes, which I thought would be a
problem.  Tons of work putting one of these together, though.  Not bad really,
a TRUMPET-PICCOLO combo for $4k.  That is a LOT of machine.

Ok, CASTANET boards came in!  Yippee.  Now to pick up oarts at the post office
and I can built it up.  Take the photos.  Finish the manual.  I expect to ship
all pre-ordered CASTANET kits by the end of the week.  Just in time.

Customer emailed me saying Jack Elliano at Electra-Print made him some custom
headphone output trannys to replace the 119DA he was using.  Said they sounded
great.  Could be a possible upgrade?  I didn't see anything at Jack's site
about them, but maybe somebody can tweak him.  Imagine if it had the same
footprint!  So I browsed his site a little.  Checked out the DRD (direct
reactance drive), as I didn't know what it was.  Wow, brilliant!  I love the
concept.  Not the only answer, but extremely clever.

Electra-print

I got some new panels in.  HAL red but with no ink for the lettering.  They
look absolutely STUNNING!  Gorgeous!  This is the breakthough I was hoping for.
Superb appearance, this is going to work great with black painted top and 
bottom covers.  Dang!  I just showed it to my wife.  She doesn't like it. 
Said "go back to yellow".  Shoot.  I liked this better.  It is more me.

Dec 10
------

After I got all production out of the way (folks don't wanna wait for their
machines prior to xmas), I got back to the P3.  I read the VTAF website and
apparently the loose fit is on purpose.  Personally, I don't like it, as the
slop gets taken up during play, with the needle pulling the bearing to one
side.  This will slightly affect setup.  So I changed it.  Applied several
coats of nail polish to the inside of the sleeve.  That took up most of the
slack.  It is still loose, but much less play.  No idea if this is better or
not.  Then I'm thinking, hmmm, the VTAF adds 1/4" under the arm at its lowest
setting.  If the arm is already set for average playback with average cart,
then we're off automatically by 1/4".  Well, with cart installed, and my 20XL
is on the tall side, the arm still would not make level.  The instructions say
to put shims under the back of the cart to tip it backwards.  Is that a good
thing?  Instead, I added some fiber washers as spacers.  Now the cart hangs
down below the headshell on these small standoffs.  Ok, still not level, but
pretty darn close.  Certainly cannot do negative VTA.  Maybe I need a thicker
mat.  

Since the machine was apart, I also went ahead with other changes.  Added the
matching counterweight.  Very nice looking.  But I was not able to unscrew the
original stem off the RB300.  And no way was I going to wrench it.  So had to 
put the counterweight on it as is.  Diameter is a little off, but with the set
screw I didn't see a problem.  It snugs in place.  Original just used o-ring
squeeze, which adds some damping & isolation.  Ok, now I put on the dust cover
and shit, it hits the counterweight!  The VTAF is so tall I can no longer close
the lid.  Fortunately, I have me a CNC machine a few feet away.  So I ground 
off the top of the counterwieght.  It now hangs low and with slightly less
mass.  But hey, I can now close the lid.  And it looks sort of cool, too.

VTAF installation

Lastly, I made a new wiring harness.  Decided to stick with coax for now.  Made
it silver plated copper & teflon.  Two coax twised together with two black
ground wires.  Arm ground is now separate from signal.  Man, the wire inside 
the Rega arm base must be AWG infinity.  That's the skinniest wire I ever
worked with.  And short!  I thought for sure I would be sending this arm out
to get rewired.  By some miracle, I got it, no shorts or breaks.  I did,
however, have to use hot glue to insulate the bare solder connections.  Not my
favorite dielectric.  Also changed the clips.  Cable not burned in.  Just go
with it.  Finally, table is back together, ready to tune.

As a side note, all of these upgrades that people do, isn't it just cheaper to
buy a better table in the first place?  I think so.  Maybe next time I just get
a VPI Scout.  Save me some aggravation.

Fired it up on a RIPPER (other stages not ready) with PICCOLO front end.  Best
it seemed to run at 26dB.  Anyway, the sound was already waaay better than what
I had.  Not sure exactly which of the changes made the difference, but it was
back to music, full range.  Messed for awhile.  I was hoping to find that
magical VTA null everybody talks about.  Nope.  Never heard it.  Gets a little
sloppy and wooly when high, not too bad when low.  Really, not the huge changes
I was expecting.  Put on the test LPs, ran a VTA test.  I still couldn't come
up with an optimum.  Reset anti-skate according to the high level tracks (at
least that was easy).  Also noticed that I got best tracking performance with
VTA set to minimum.  Yeah, maybe the arm is still too high?  Once I got the
protractor and balance tweaked yet again, it seemed to balance out a bit.
Music was pretty good, but not up to what I expected.  Bottom end is still a
little on the wild side.  Maybe my ears are tin.  Why can't I hear that optimal
setting?  Is it because I am listening through headphones?

The jury is still out for me.  Honestly, I was hoping I could do a rave review
here.  But considering the challenges of installation, this is not for
everyone.  You have to drill a clean hole.  There is an immediate 1/4" deficit.
The bearing has too much slop for me (perhaps on purpose)?  The parts are
nicely made, I'm just not convinced yet this is the best way to go for a P3.  
Much more appropriate for a custom table or arm board.

Dec 09
------

Finally got the last of the CHIME GOLD machines up on auction.  Just in time
for Christmas.

Audiogon

Started modifying my Rega P3 for the VTAF.  Whoah, it says here you gotta drill
a 1.25" hole.  Ok, I guess this is a one-way street, once you commit that's it.
Went to store and bought a 1.125" bit.  Damn!  Back to store and got the 1.25"
bit.  Plinth drill very nicely on my press even though it started as a 1" hole.
Came out reaaly nice.  Only $5 for the bit.  Now I find the VTAF has the arm
sit in the bushing via gravity, nothing else holding it in place.  Hmmm.  Hey,
maybe it sounds better that way.  I'm a little concerned with the bearing slop,
it has about 0.01" inch play and slop, can chatter.  Also re-wiring the arm,
but that's a whole new can of worms itself.  I have to make a new "plug" for
strain relief.  I'm leaving the wire inside the arm in place for now.  And
separating the ground lead from the signal cables.  Interestingly, I may have
accidentally discovered a Rega secret.  The plinth has hidden cavities!  Top
is solid MDF, covered with veneer.  Bottom has several internal cavities, 
covered up by the veneer so you don't know they are there.  Maybe this is
resonance control?  Just run your fingernail along the bottom surface and you
can hear exactly where the cutouts are.

Dec 08
------

Was looking at my FLUGELHORN proto, flipped the rear panel upside down in order
to get RCAs on the proper side.  When I realized the wood pieces I have to
temporarily hold things together ain't such a bad idea.  For the PA-20, which 
has to be taller, I could use wodden side panels!  That would be both cheap and
very nice looking.  Better than a red strip.  Of course, this has the problem 
of maintaining electrical contact betweer all of the chassis pieces.  Real
important.  So maybe meet half way, use the normal extrusions to hold things
together (and PCB), then add a wood trim piece for visual appeal.  Yeah, I like
it.  Can't wait to get time to put into that 20W amp.

Dec 07
------

So I counted the number of names in my customer list.  Those who have paid for
something.  1739 different customers.  Is that good or not?  Seems like a
pretty big number to me.  A lot of them repeat buyers.  Hey, it's a good start.

Everybody wants hi-res downloads cheap, right?  Gave a try at this musicgiants
website.  Fine, made account, put in some money, downloaded some songs.  Did 
the DRM thing.  Hey, they play fine on WMP.  But I don't want WMP, I want to 
play in foobar with my asio stuff.  Ok, I want bit perfect playback, none of
that kernel resampling crap.  Well, the website is nice, easy to use, the songs
downloaded handily.  Fast, too!  Took me a LONG time to find out where they
went, as my XP machine registry is quite a bit messed up.  Had to use WMP to
find them.  Made copies, put into MyMusic.  They will NOT play in foobar.  Nor
can I convert them into FLAC.  Am I stuck in WMP?  Maybe I can burn to CD and
then rip & convert?  If these hi-res files only play on WMP, then what's the
point?  I'll keep you posted on progress.  Right now I need help.

Boy, can it rain here or what.  At times we can get an inch per minute.  You
gotta see it to believe it.

Best way to envision a new product is to build a proto.  So I put together a
FLUGELHORN today.  Took my panel design and kludged together what it might look
like.  Seems like its a fit.  Just run wires from front to back.  Oops, I just
noticed I put the RCAs on the wrong side.  They should swap places with ac
input.  I could probably lose the gain trim.  But it looks nicer this way.  And
I like to use it as a volume control.  Trying out some different coupling caps
(blue).  Will do an A/B with a normal ARCHIVER when I get a chance.

FLUGELHORN proto front
FLUGELHORN proto back

Got my copy og HIFI+.  First couple of pages I see the new phono stage by
Emille.  Wow, looks a lot like the new TRUMPET in size and configuration (yet
their circuit is curiously closer to a CORNET).  Nice looking boxes.  

Emille phono

Oh yeah, I noticed the DRM and other stuff on the musicgiants site might make 
it so you can only playback on the computer that did the downloading.  Ay.  My
music servers are separate from the machine I download to.  Is that going to be
a problem?  Solved it!  I burned all the songs I bought to CD.  From there I 
can convert to flac and run on my music servers.  Yippee!  And the CD ends up
being a nice backup, to boot.

And now this paper slide rule falls out of HIFI+.  System diagnostic.  Shit!  
It says my stuff is WAY less than low quality.  Crap.  I don't even make it to
the bottom of the chart.  Maybe I should just give up.

Dec 06
------

Six of the trees in my yard blew down the other day.  The soil is not deep here
(hey, we live on the side of a volcano) and the trees adapt to the trade winds.
So they all grow up leaning west.  Great, now a storm wind of 75mph blows east.
I have most of them propped up with ropes.  They look ok, but confidence is not
very high.  

Updated the PA-10 schematic.  Lots of changes.  New driver tube, new CCS, etc.
Just to give you some idea of the debugging process, here's a copy of the
original schematics marked up with lab notes.

PA-10 original schematic

And here is the revised version.  I still need to update reference designators.

PA-10 schematic rev b

I agonize over the thought of trying to re-wire the layout.  It's gonna take a
ton of work.  Super tight fit.  Guaranteed I will run into size problems.  The
big 3.3uF film caps I added are huge.  Well, each part is key to the design.  I
don't add stuff if it ain't needed.  I would much rather try and deal with 1uF
caps instead.  Also, I am raising the voltage on the electrolytics to 450V, in
the case it gets powered up without tubes.  No load condition swings the power
supply up to almost 500V.  I need to withstand the surge.  CCS have changed.  
The output CCS are now separated from driver stage (layout consideration).  Not
perfect, as they are self-biased from their own output (3mA).  They should stay
reasonably well matched if the transistor betas are close.  Added the LEDs to 
indicate if each tube is operational.  Anyway, it is still the Hagerman self-
balanced output stage.  No adjustment required.  I'm thinking you could plug in
a 300B or 45 into one of the spots and it would still run.  Balanced!  It runs
unbalanced at 1/2 power with only one output tube.  Not recommended.  With
these changes, I think the BW will go up a little, noise will go down.
Actually, all specs should measure superb except for HF response.  That will be
on the low side, yet wonderfully behaved.  

Ok, did some more photoshop on the self portrait.  I kinda like this version
the best.

Psycho

Sometimes I am just out of control.  I mean, my designs can really go in
circles, taking wildly different directions.  What happens is that I come upon
a working solution, but later realize I am not happy with it.  Something just 
not right.  Such is the case with the ARCHIVER.  It's a beautiful stage with
great sound, super capability.  I put a TON of work into that design.  But so
far it doesn't sell.  Maybe it is the chassis?  Assembled RIPPERs are no longer
selling either, indicating the market for USB recording has been lost to
competitive offerings.  Ok, so my idea was to remake the ARCHIVER into a RED
HAGLABS product, add a BUGLE power supply, better chassis.  Well, my solution
(presently being fabricated) is ok.  Chassis is plastic (very nice looking) and
big.  I got the entire board in there.  Chassis is like 18" x 4" x 8".  Most of
it hollow.  Maybe it is ok.  However, every so often I stop to ponder.  And
tonight I realized that most of an ARCHIVER will fit into the new red chassis,
same one as HA-10 (CASTANET).  It's a far better chassis, is a perfect match 
for the other components, and ends up being cheaper overall!  The tradeoff is
that I lose the dual inputs, no mono switch.  I replace the wall wart with a
real ac power supply.  More voltage, more headroom.  The rest of the controls
stay.  It all fits rather nicely.  A compact ARCHIVER.  All I have to do is 
slice the board into three pieces.  The panels look something like this:

RED ARCHER panels

I don't know what to really call it.  So far it is the ARCHER.  Maybe it should
be the FLUGELHORN?  That would actually make sense.  All of a sudden this 
becomes no longer an oddity, but a way cool little unit.  It fits the new RED
lineup: PICCOLO - FLUGELHORN - CORNET - TRUMPET.  That gets the business
started.  Excellent quality phono stuff.  Then I add the HA-10, etc.

Dec 05
------

Got my TT stand spiked last night.  Now I just need time to upgrade the arm
with my new VTAF and wiring.  Anyway, got a RED PICCOLO built, tried out some
different caps.  As soon as I get a RED CORNET or RED TRUMPET built I can do
some A/B with a stock unit to see if they are better or not.  Still, running it
into a RIPPER phono was nice.  Even with the cartridge not aligned quite right
the sound is more pleasant than digital.  Can't wait to get one of the better
phonostages hooked up!  I'll have much more stuff to listen to.  Listening via
the HA-10 is an absolute joy.

Did carefully priced parts lists, HAGLABS style.  I am trying out a new markup
formula, not the typical 3x thing.  It is much more complex and I tried to take
into account a lot of variables.  The idea is to get best dealer pricing
possible and still maintain profit and efficiency.  No point underpricing a
product, as it just puts you out of business.  A good example is the CASTANET!
Things ended up costing more than I expected.  Look at the numbers.  I sell a
kit/2 for $99.  Then subract out the PCB ($31), hardware parts ($7), mailing
envelope ($1), and postage ($3).  That leaves $57.  That has to pay for the
roughly 20 minutes average order processing time and customer support.  Or at
$50/hour (salary & benefits) about $17.  Thus leaving $40 net.  Now given that
I've been working on development since July, I estimate I have maybe 160 hours
into this machine.  That's roughly $8k unpaid salary.  And roughly $2k in part
and prototype expenditures.  So to get ROI (return on investment) of $10k that
was already invested, it takes sales of 250 half-kits just to break even.
Yeah, that's how business works.  I haven't even sold that many CORNET2 yet.
The key to profitability is the assembled unit.  So don't think I am ripping
you off when I change the CASTANET pricing to $129 (that's my old 3x pricing
markup).

Meanwhile, I almost finished a RED VACUTRACE.  Missing a few parts, so again I
have to wait.  It looks beautiful.  So much nicer than the old chassis.  My
goal for December is to get this thing back on sale.  That, plus the CASTANET.
Then in January I can do the HAGLABS website and start with an HA-10, CORNET,
TRUMPET, ARCHIVER, PICCOLO (all red).  The CLARION (SE-10) was just a
distraction.  No point to it considering the superiority of the PA-10 (yeah,
back from PP-10) architecture.  Then I put PA-20 and DA-10 on back burner so I
can get some time on the PL-10, which is supposed to be the bread winner.  The
PA-10 and PA-20 (4x6080) may end up being some of the best amps on the planet,
but I don't think I'll sell hardly any of them.  They are just too advanced for
the marketplace.  So it makes sense for me to delay the PA-20.  The PA-10s I
need for myself and RMAF08.  

And so I need a new photo for my website.  Was thinking of something very
dramatic, like the bold side-lighted portraits you see of violinists and such.
Or one of those stark images in vanity fair or whatever.  Well, I don't think I
quite got there.  Here's my first try.  An attempt at boldness, art, and at the
same time look intelligent, confident, reliable.  I thought maybe the few days
of stubble would help (it does Mr. Pitt).  Yeah, yeah, I know, don't quit my
day job.

Self portrait #1
Self portrait #2

I finally got around to checking this online rag out.  Not bad, very
interesting format.  Seems like it builds itself in real-time, streaming ads
from a separate database.  Sort of like google ads.  After a very short time it
got extremely annoying, as one of the same damn four ads showed up every page.
Honestly, I couldn't even finish the issue.

Playbackmag

Dec 03
------

Haven't forgottent the PP-10.  Still working on it.  Tried out the cap to CT
yesterday.  Bad move connecting it to middle node between cathodes, as that
really messed up the waveforms deep into clipping.  Connect it straight to B+
was the answer.  Clips wonderfully, just like no connection, but you get two
benefits.  One, the low impedance forces extra drive into class A2 operation
for extra headroom, two you get an improvement in PSRR.  It's a clever
connection I found in the WE 92A schematic.  The idea I believe is that any
B+ ripple or wander gets fed to all triode grids.  The plate, grid, and cathode
all see the same common mode shift, hence hiding the supply ripple.  Thus, any
low frequency changes in line voltage (due to refrigerators, etc.) are somewhat
filtered out.  The WE chose a high impedance connection, thus keeping operation
out of A2.  I went low impedance, same as Lynn Olsen, to push harder into A2.
In the audio application, we want the extra peak power headroom.  Today I might
try out the 6ep5 as a driver.  Also, I'm going to find a way to put an LED in
series with the CCS on each tube, so you know if tube is operating properly or
not.  A nice feature for something like the PP-20.

Oh my, I just saw this on asylum.  Maybe I should make a product like this.  A
wall of gear preset and ready to go.  This is stunning!

My kind of tube amp

I measured a 6E5P tetrode strapped in triod mode.  It makes a fine driver.
Almost as much gain as the 7788, but with more balls.  I think I'll try it out
in the PP-10.  Highly recommended by Romy.  This has to be the best, most
linear indirectly heated triode I've ever seen.

6EP5 triode curves

Ok, so I wire up one of these tubes into the PP-10.  This amp now looks like
frankenstein.  This RF tube sits on wires.  Is it gonna oscillate or smoke?  
Fire it up.  Hey, not bad!  Whoah, holy moly.  I'm getting way more signal
swing.  Oh my, it is now pumping out 8 watts easily before it starts to clip,
and then it pushes well past clipping to deliver at least 12W.  Sheesh, all
because I changed the driver tube?  Ok, I did a re-bias and re-centering of the
stage.  But still.  It's pushing way more power into the output tubes.  So I
hooked up the spectrum analyzer.  Yup, harmonics all look the same.  Distortion
down from 0.5% to 0.4%.  This is it.  This is the new driver tube for both
PP-10 and PP-20.  In the PP-20 I'll use the 30H plate choke which will increase
the swing even more (mu = 2 on the 6080).  

Finally did a first draft of a CASTANET parts list.  Posted online.  

CASTANET manual

Just hooked up the frankenstein PP-10 up to a speaker.  Ran one channel out of
a CHIME GOLD.  Oops, no volume control.  No noise!  I couldn't believe it.  All
of the changes and playing around with that amp paid off.  The noise floor was
virtually non-existent.  Last time I listened it had a buzz and hum.  Now
almost nothing.  Sound was a little honky, but I don't know if that was because
it was only one channel, the speaker, or the amp.  I hate not having both
channels.  It always sounds like crap until the other channel kicks in.  Maybe
if I had a mono source?  Anyway, a good match for the CHIME now, as I get
nearly full power without clipping.  I can safely say this will be loud enough
for my average sized room with the vandys.  

Then I switched over to the CASTANET.  Still can't believe the bass I get out
of these HD600s.  Unreal.  Did some A/B against a new pair of interconnects I
made.  Very hard to distinguish between the two.  Thought for a minute the
original Bolder's had better vocal presence, but my new ones seemed just a tad
louder or larger, I'm not sure.  Can't say which is better.  The good news is 
that my new design wasn't awful.  I'll make more and see what happens.  They
are simple stranded coax silver over copper with teflon insulation.  Military
overstock.  Small diameter, gold RCAs with brass knuckles.  Nothing fancy.  

Dec 02
------

Shux, now I see that even the Lamm SE amp uses feedback.  Crikey!  That must 
be how Mike at Quicksilver gets all of those great numbers.  Shiese.  It's like
racing against someone on EPO.  I just ain't gonna do it folks.  No way am I
gonna add that feedback.  Another comparison is Lynn Olsen's Karna amp.  300B
PP delivering roughly the same power, but with more headroom into clipping.  
same B+ power level.  Probably lower distortion, but at the cost of 300B and 45
NOS pairs.  Four 6080 is probably about $50.  Or 1/10th the cost.

Raining like hell here.  So I may as well get some more R&D in.  Trying to see
what sort of chassis I can fit into.  Thought perhaps putting the CASTANET
chassis on its side.  Have two 6080 and one 5AR4 sticking out each side, the
6e5P out front.  All tubes in horizontal position.  Looks funky, but maybe not
practical.  Then a flash of an idea, how about making the chassis extrusions
larger?  Right now there is a silver stripe along each side of about 1/8"
height.  It separates top and bottom cover.  What if I replace these with
taller ones?  I change the side panel from 1/8" to 1" or 2".  Now front and
rear panels are much larger, say 6" x 9".  Theoretically this would be cheaper
than two chassis with an umbilical.  It gives me room to make a CASTANET
chassis fit all 7 tubes on top, the two power trannys on back.  All on one
circuit board.  The extra height means I can squeeze the output tranny inside
with a vertical orientation.  It would look just like the other chassis,
expecially the new CORNET.  Very compact and heavy.  So I did a quick study of
costs, making custom extrusions.  Entered rough dimensions into emachineshop
program.  Whoah, it cam back over $60 a pop!  No bargain.  Maybe there is a way
to combine two existing ones?  Might be able to use same trick to squeeze a
PL-10 from two boxes into one.  

Extrusion example

Aye, I got it.  The two extrusions side by side look funny, but if I add
another piece of black chassis in between, well then I get two silver stripes
and it looks perfectly normal.  Funny, the extra aluminum panels they supply
almost fit perfectly.  Just need to cut them down to size and paint.  Or maybe
I could do a red anodized stripe?  Gives me some easy options here to increase
chassis height.  So obvious.  And I can do any height.  I could even convert 
the new TRUMPET back to one box.  Would be less work, but visually maybe not
as cool.  For the phono stages, I like having two boxes, as it gives me big
spacing between power magnetics and input circuitry.

Dec 01
------

Spent all day doing yard work.  So tonight (we didn't go out) I put my brain in
gear.  Came up with a solution for the PP-20.  Ok, maybe I should call it
something else, even though it is sort of a PP-10 on steroids.  For years,
people have been asking me to do a high powered tube amp.  Everyone wants one.
Probably would be my best selling kit.  However, now that I am into the math, I
can't see any way to do this cheaply.  Ok, there is the classic solution of
power pentodes and gobs of feedback.  There is also OTL and gobs of feedback.
Oh my, it seems the ONLY way to crank out high power at low output impedance is
with feedback.  Me can't do that.  I design for sonics, not numbers.  And so
that is how I ended up down the road of four 6080 in push/pull/parallel.  With 
all that tube and all that dissipation I can still only get out a modest 20W.
That will have to be enough.  On the other hand, it will sound like way more,
and be refined and musical as hell.  It won't have feedback.

Looking at available parts, I see my B+ current is rather high.  Voltage is
very low at +250V, but current is 430mA or so.  That's just more than what most
trannys and chokes will handle.  So hey, do it in parallel!  It has that
synergy thing since the output tubes are also paralleled.  I chose a 273CZ
because it has the nice horizontal mount (like the other power trannys in the
HAGLABS lineup), delivers 325Vx2, has 5V at 3 amps for a rectifier, and 6.3V
at 5 amps, perfect for two power tubes!  I'll have to add a small filament
tranny for the driver tube, which will probably be a 6E5P at 25mA.  Anyway,
the trick here is to run two 273CZ power trannys, each with their own 5AR4
rectifier and 159S 4H choke into 800uF (or maybe 1mF) capacitance.  This gives
almost a perfect 250V output at 430mA.  Only 0.15Vrms ripple.  I'm hoping that
is good enough for PP operation with no hum, given some tube mismatch.  If not,
I add another pair of 4H chokes and another bank of big caps.  Hey, 8 of those
100uF panasonic EB caps in parallel gives you some pretty low ESR and ESL.
The driver will be CCS SE into the interstage tranny / phase splitter.  Four
output tubes means eight triodes, each with their own 50mA CCS.  The choice for
output tranny is the 1650H driven by the UL taps.  That's 264 ohms per side.  I
might get a little less than 20W output now, but impedance with be close to 2
ohms.  Hmmm, this is almost a cross between a CYMBAL and a PP-10.  Low voltage,
high current operation.  Very low turns ratio (only 6:1!).  I think it will add
up to some superb performance.  Wow, that's only 6 chunks of big iron.  Now to
find a chassis.  Alternately, I could run a pair of 6C33 in exactly the same 
design?

Just realized that driving via UL taps will bring the primary inductance waaaay
down, potentially destroying the LF performance.  I might have to buy one of
these trannys and experiment.  The Hafler will come in handy again.

No wonder everyone uses feedback.

If I lay all the parts in a line, I get a narrow 10" wide chassis.  By 24" to
30" long!  This amp would be a good candidate for hand-wiring, except for the
CCS.  Hey, maybe I need to make me a CCS module?  I suppose I could jam this 
onto an 10 x 17 chassis.  One monoblock.  Re-did the math for the 1650H, looks
like about 15W into 8 ohms.  A giant SS amp can deliver 200W with 40Vrms.  I 
got 14Vrms (40Vpp) at 2 ohms, or about 1/3 the output voltage.  Is that enough?
That's all I can get out of a proper design.  It'll sound like 50W, though.  

Aha!  I goofed.  Forgot that I only calculated 1/2 the output stage for
impedance.  So in reality I should be getting 1.1 ohm out.  Whoah, that's
pretty decent.  This bumps power up to 19W prior to clipping.  Ok, so let's
call it 20W.  And PP-20 is now a very good name.  But is this right?  I checked
out a really (awesomely) good commercial amp, the Lamm ML1.1.  It's a beauty,
and the measurements show it.  Very close to what I am doing, but with one
difference.  My idle is 225W (from ac line), the ML1.1 is 200W.  Both have Ro
of about 1 ohm.  Mine cranks out 20W, the ML1.1 80W.  What!?  Yep, it has to
be feedback.  This is indicated in the square wave stability test loaded with
2uF in parallel with 8 ohms.  Such would cause an RC bandwidth rolloff in my
amp, but with feedback you get peaking.  Only feedback can do this!  Also,
looking at the distortion vs. power graphs you see a discontinuity and abrupt
jump at clipping.  Mine will be a curve similar to the CYMBAL, gently rising
upwards as power increases.  Ok, so maybe my amp will sound like 80W?

CYMBAL distortion

Nov 30
------

And so I realize last night that the PP-10 architecture can be extended to much
higher power.  Yeah, the article in TAS mentioning the Magnepans need gobs of
power.  What to do?  Ok, so 20Vrms into 8 ohms is 50W.  That should be enough?
That's 30Vp, and normally I get 100Vp out of a 2A3 or 6H30.  That leads me to a
tranny ratio of 4:1.  Not very much.  128 ohms on primary?  What is going to
drive that?  I realize all high powered tube amps run boatloads of feedback.  I
can't do that.  That means tubes with a lot more voltage swing and/or lower
output impedance.  All I can think of is the 6C33.  Is that why everyone uses
it?  Not the most linear of tubes, but many claim it sounds great.  Superb
construction.  What else could I use?  A 6080?  6C19?  I need to think about 
this.  Could be a great project once I get the TEN series out the door.  The
good news is that the PP-10 topology works with many tubes in parallel.  I
imagine a big chassis with maybe 10 output triodes.  If they were 6080 or
similar dual type then a tube could be pulled safely while still operating.
Only the power level would go down.  It would still work.  

Ok, did a quick study here.  Using a 6080 dual triode, they run low voltage and
have ~300 ohm output.  If we take a 5k output tranny and use the ultralinear 
taps, they're at 40%, or 0.4.  That's then a 5:1 turns ratio (0.16 * 5k), or
200 ohms on each side.  Thus, to have 4 ohm output drive on 8 ohm tap we need
100 ohms plate, or three 6080 in parallel.  One 7788 should be able to drive
all three tubes no problem.  At low voltage we use plate choke instead of CCS.
I love it.  All iron!  About 400mA B+ at 150V.  Lots of heater current.  We
can only get about 100Vp output from tubes, so that gives roughly 15Vrms at
output.  A 3.5dB power loss (4 ohms into 8) means we have 10Vrms into 8 ohms
for barely 12W output power.  Hmmm.  This is a LOT of work for not a whole
lotta power.  With 6 tubes in parallel we get 2 ohms out, and 18 watts.  The
tubes are damn cheap!  Hmmm.  If I squeeze 120Vp out of these tubes, then six
in parallel get 23W.  Ok, so we clip at 25W, pushing the design.  That's with a
lot of output tubes.  About 50Vpp into 8 ohms.  Right now I get 20Vpp with the
2A3s in the PP-10.  Yeah, this is worth looking at.

Did some compromise, in order to make this reasonable and a practical
implementation (real world parts).  Four 6080 output tubes, rebiased to 170V at
-80V and 50mA, can deliver 20W into 8 ohms.  It does not stay in class A when
delivering the large peaks.  That's 50Vpp output.  100W of B+.  70W of heater.
Seven chunks of heavy iron.  Nine CCS.  No biasing necessary.  Zero feedback.

Nov 28
------

Well, I've stopped about 75% of my email spam, just by dropping the *.hagtech
mailbox.  Not sure if I am still under control of worm/virus yet.  I think my
firewall is helping.  

More cool amplifier tests.  I connected headphones straight to the output.
This is a great way to hear the noise and microphonics.  It has some buzz and
hum.  We knew that.  But it was nice to hear no change between R = 10 and open.
The 7788 has a very unusual "ping" to it.  Not the normal microphonic ping or
thump you get from a preamp tube.  More of a long lasting shimmer.  The tube
dampers didn't seem to help.  I guess this is due to frame-grid construction?
Anyway, I then hooked up my Hafler power amp as a driver stage, bypassing the
7788.  It drove the interstage tranny.  As expected, the 2nd harmonic went way
down.  In fact, by 14dB, enough to cut distortion by 5x.  

PP-10 distortion using Hafler driver

This is looking all wrong, though.  You want the harmonics in a nice decending
fashion.  Here the 2nd is on par with 3rd (and with R = 100 it is), which does
not lead to good sound.  Again, I see the exact same 6dB change in 3rd between
R = 10 and R = 100.  And the spurs on the fundamental rising at R = 10.  Still
wonder about those spurs.  Is it the 2nd from the driver mixing with the 2nd
from the output?  What makes the 3rd come up so much?  Why is no 4th visible?

Nov 27
------

Some more PP-10 testing.  Seems the trends don't vary much with signal level.  
I am still getting a 6dB increase in 3rd harmonic at R > 100 versus R < 10.  
Did discover that my wein bridge light bulb oscillator has a bit of 5th
harmonic built-in.  Changed to the EMU-0404 output dac as a source.  Now all I
can see are 1st, 2nd, 3rd.  Nothing else shows up.  I can't run the power level
too high or it clips the ADC input.  Maybe I should build a resistor pad.  So
at about 4Vpp output I have roughly 0.5% distortion.  

PP-10 distortion spectrum at 1kHz

It took me awhile to figure out the best operating conditions for the FFT
software.  Always had it set to maximum 65k sample points.  Hey, go for the hi-
res, right?  Nah, you can't see anything but spikes.  The above plot is with 
only 4k points!  Doing so is like adjusting the bandpass filter.  This setting
is good for 1kHz sinewaves.  I also ended up using the blackmann-harris window.
It seemed to give the most useful information.  Most significantly, I can now
see 60Hz or 120Hz spurs riding on the fundamental!  I am getting some mixing 
going on here (multiplication).  That would be IMD?  At R > 100 the spurs drop
way down, but the 3rd comes up.  I suppose that just means I am losing some of
the inherent power supply rejection of the differential stage.  Maybe that is 
where the compromise comes in?  I dunno yet, but my first impression is to drop
the 3rd.  Might have to do A/B listen.  

Ok, so I tried to make a distortion versus frequency plot.  Changed FFT to 65k
for all readings.

PP-10 distortion vs. frequency

It remains solidly at half-percent across the band, then rising a bit in the
treble.  I was suprised it was so good in the bass region.  Usually this is
where things start to fall apart.  Ok, so now distortion versus power.

PP-10 distortion vs. power

Looks like about 0.1% minimum distortion before the noise floor kicks in (which
I'm sure I can make quieter).  Then we get a hair over 1% at 1W.  That's not
bad, but I think I can do better.  Wonder how much of this is driver stage.
Maybe I need to make me a scope probe buffer/attenuator?  That way I can
measure distortion at any point in the circuit.  

Maybe I need that 1uF from between the cathode caps back to the CT of the grid
driver tranny?  Need an input loop retun path for paralle mode.  Don't think I
can jam it into this frankenstein of a circuit.  Too many changes already.
Also realized I don't have the electrolytics bypassed.  One thing of note, the
power level on that 100 ohm resistor is really low.  I was getting about 1mA
through it.  Very low power dissipation.  Theory has it that only the imbalance
signal passes throught it, balanced are cancelled.  So that would mean I have 
perhaps a 2% imbalance between my 2A3?  That's a really good match.  Seems I
need to do a relayout here and try out this massive batch of changes before I
can tweak more on distortion.  

Ok, so I made me a probe buffer out of a BUGLE board.  A 1M resistor and 5pF
added gave almost a perfect probe calibration (preset to match my Tek 7704).  
The 0dB buffer then feeds the EMU-0404 ADC.  The 10x probe allows me to then
look at large signal swings.  Oops, forgot to ac couple it.  Anyway, looking at
the grid drive I get almost the same level of distortion.  Perhaps this just
means most of the 2nd order stuff is from driver, not output stage.  Hey, I
suppose this ain't too bad, 50Vpp at 0.5% with no feedback through a tranny.

PP-10 driver distortion

Hmmm.  I wonder if the driver stage is part of the harmonic balancer solution.

Nov 26
------

Finally done with all of my painting and house repairs!  The 5-color scheme was
more than enough work.  I must have gone through a mile and a half of masking
tape.  Here's a shot looking through the kitchen.  You can see the cranberry
entry at far end, the orange swoop of a soffit that passes through three rooms, 
the green, the moccasin (tan), and the white trim.

House paint

Spent all day with production.  Nothing else to report on.  Oh wait, this just
in.  Rigged up the PP-10 for distortion tests with the PC.  I wanted to map out
the changes as the harmonic balancer resistor value was varied.  Except when I
set it to zero the CCS blew.  Sheet.  Took me an hour or two to straighten that
out with new parts (didn't have any more TIP50 on hand).  My guess is that I
gave it a transient by slipping with an alligator clip.  Anyway, after that was
sorted out I had my full power back and low distortion.  But things just
weren't changing as they should.  Here's what I got at about 1 watt:

R     %     2nd   3rd
----  ----  ----  ----
10    0.49  -49   -75
220   0.51  -49   -69
330   0.51        -69
1k    0.52        -68
220k  0.53  -48   -67

The overall distortion was dominated by 2nd harmonic at roughly 1/2 percent.  I
am wondering if it is all due to the driver stage?  The deal here is, in fully
differential mode (220k), the output stage should cancel all even orders.  It
gave me a 1dB change.  In the wrong direction!  In parallel single-ended mode
(10 ohms) it gave a 6dB drop in the 3rd.  There wasn't much of anything else on
the screen.  Would have saved a jpg but my battery ran out.  Does this make
sense?  I was supposed to find a null somewhere in the middle.  Or so that was
my theory.  Should have been around 250 ohms.  Hmmm.  Maybe I should test at
higher or lower power.  The WE92 schematic has a resistor value of 85 ohms.  Or
maybe it was 65 ohms, I can't make it out.  That was with 300A tubes, so I
figured maybe double to 120, which was my original design value.  What is
missing is the 1uF to grid tranny center tap.  Me thought that was just for
ripple rejection.  Maybe all the magic happens between zero and 100 ohms.

Nov 24
------

Have to keep on schedule here.  So many projects, must keep the goal of RMAF08
in sight.  Mostly, I have to have a PL-10 up and running by that time.  That is
by far the most difficult and time consuming design.  These other ones are
mostly slam dunk.  Yeah, I put a lot of time into the HA-10 and PP-10 so far.
The others build on that work.  For example, I came up with a new design for
DA-10 based on the B+ supply of HA-10, the CCS and driver stage of PP-10, and
the filament supply from PL-10.  Only I had to change a few things.  My trick
this time is to power the filaments separately.  Left channel on +6.3V and 
right channel on -6.3V.  And to be a TEN series means any rectification has to
be done with choke filter.  Ay ay ay, that is so hard for filaments.  In this
case I make +/-6.3V supplies for filaments and HAGDAC.  However, it is not
straightforward and the values are awkward, such as only 0.5A current per leg.
The problem is available chokes.  That will fit.  Again I end up using the 155B
0.006H choke, but two in series for 12mH.  Two per side means 4 chokes just for
heaters!  Also using a split transformer to maintain layout symmetry (and stray
field cancellation).  I end up with 6 pieces of iron just for heaters.  And the
means 9 total, just for a tube DAC.  Probably $140 wrapped up in 9 chunks of
iron, giving a world-class power supply.  I think I came up with an arrangement
where the HAGDAC fits in the middle, a 7788 on ech side for output.  Leave off
the fancy controls and IR.  Maybe a couple of LEDs, one for USB, one for lock.
Plug and play.  No knobs or buttons.  Make it as simple as I can.

Ok, so now I have: HA10 - TRUMPET - CORNET - ARCHIVER - PICCOLO - PP10 - DA10,
all to be completed and shipping before I even start the PL10.  That should
leave me maybe 5 months to do the proto.  Maybe this is too big of a bite to 
chew.  The idea is to get the simple stuff out the door first, set up a base
upon which to build.  Get some revenue flowing.  Meanwhile I still have to get
this new VACUTRACE done.  That should be easy, as it's just a matter of
soldering what I already have together.  I have to put the CLARION / SE10 on 
hold for now, make sure I can get this other stuff done first, while still
keeping up with current production.

Nov 23
------

Hooked up a real CHIME last night to play for the CASTANET.  Well, once again I
am floored at the differences.  The RIPPER was fine, but by comparison is
really in-your-face.  The CHIME was soooo relaxed and smooth.  But that's not
suprising.  What really hit me was the bass!  Oh my.  Thunderous bottom end.
It's like the entire EQ was shifted.  The top end almost seemed rolled off
(back to my original impression of the HD600) because the bass was so huge.
Whoah, this I can live with.  

My 156 came in.  That's the interstage tranny alternative to the 124B used in
the PP-10 (formerly PA-10).  A quick run of bench testing gave it a similar
27kHz self resonance.  Bandwidth, even damped, went up to 50k or so.  I forget
exaclty what the 124B did out of circuit.  Anyway, I'll give it a shot.  Seems
to me nothing will change.  Also got the new filament trannys.  

Staring at the TRUMPET and PICCOLO red chassis.  That yellow lettering just
isn't doing it for me.  The wife still says it looks ok.  White would be ok.  
The issue isn't really the black covers (which are way better than the beige),
but the aluminum strip down the sides.  I have exposed metallic silver color
highlights all over the chassis.  Extruded aluminum and stainless steel.  So
why not make silver the accent?  Maybe the engrave lettering doesn't need any
ink at all?  Silver-red-black.  That's not bad.  And I'm backing up a little,
thinking maybe I should offer a more traditional option.  Since these machines
are likely built to order, I may as well offer a choice in color.  But only
two.  The silver (clear anodize) faceplate of that ARCHER I built was
beautiful.  It would work here.  Silver face, stripe, black covers.  Very
classy and traditional.  So I would have choices of either red or silver.  

I still need to order panels for the new CORNET and the new ARCHIVER, so heck,
it's a perfect opportunity to try out the new styles.  Eventually, I will 
converge on a solution.  That's just the way R&D is for me.  Go in circles,
make a lot of mistakes.  Fix it now while you can.  

Nov 22
------

I am now set up to take pre-orders for the CASTANET half-kit.  Posted schematic
online (final version).  I noticed that the color of the LEDs on the proto
looked incredible in a dim room.  I think the 2mA or 3mA I run them at is just
right.  A deep amber color, not too far off from the glow of the 6H30.  

CASTANET schematic

Had me a bit of a scare yesterday.  Was listening to the CASTANET in the lab
hooked up to my modified cd25.  Playing an old REM CD (I like some of the
songs).  Anyway, I noticed I had to keep turning the volume down.  Oh crap.  I
could listen loud only for a short while.  By the end of the CD my ears were
hurting.  Oh crap.  Is this right?  The sound is vibrant and punchy, why am I
getting fatigued?  There must be an edginess or harshness!  Oh crap.  Did not
post this finding to the blog.

And so last night I brought the CASTANET inside and hooked it up to a RIPPER
used as a USB DAC.  All I had at the moment.  Played an MP3.  Ugh.  It was
awful.  I could hear every last detail of imperfection.  Every encoding
disruption was blatantly apparent.  Gads.  Then started to play my usual test
tracks.  Hey, not bad.  What a big difference going from MP3 to FLAC.  All of a
sudden it was music.  Hey, this is pretty good!  This is coming from a RIPPER?
Most significantly, I found myself cranking it up.  There was no more ear bleed.
In fact, I could pound my ears with ungodly SPL (anyone who listens at full
volume is mental).  No harshness.  No fatigue.  Any volume worked just fine.
Full, beautiful tonality and pace.  Very neutral and uncolored.  You can't get
this from speakers.  I now understand the head-fi crowd.  Easy to hear
everything on the track.  The resolution was brutal.  Nothing gets hidden or
smoothed out.  No sweetness added.  Clearly, the CASTANET is revealing.  If
there is any flaw in the recording, it will be apparent.  Sorry to say, bad
recordings are not made more musical with this amplifier.  If anything, the
flaws are exposed.  I was sort of hoping the CASTANET would make marginal
recordings more listenable.  Sad to say that is not the case.  One last
interesting point.  It was late, I'm slumped in a comfy chair.  The tunes are
blasting in my head yet I could fall asleep.  How is that possible?  Tonight I
try a CHIME instead of the RIPPER.

Now I wonder if my cd25 sux or was it just that REM CD.

Nov 21
------

Ugh, fighting some sort of email virus.  Shoot, seems hagtech.com is now
spitting out a new spam every 20 seconds or so.  Something about a canadian
pharmacy, I think.  Been trying to clean things up, see if I can stop it.
Meanwhile, limiting my email addresses to "jim" only.  Used to be anything@
hagtech.com would get to me.  I had LOTS of spam.  Now, maybe I won't.  Got the
firewall kicked in to higher levels, etc.  My guess is the problem is related 
to some disney interactive online game the kids were playing.  It's not puzzle
pirates.  This computer is slowly dying.  Hard drive don't always boot, it has
years of crap all through the registry, which was also completely screwed by
registry mechanic.  Everything is hanging on by a thread.  Any day now it might
lock up permanently.  Fortunately, all data is backed up in several places.

Here's a shot of the new red TRUMPET.  I'm no longer enamored with the yellow
lettering.  In fact, I wonder about no ink at all.  Just engraving, bare
aluminum.  I have aluminum and stainless steel hardware accents all over, so
maybe it would be a more natural look.  Fewer colors.  The black painted
chassis covers changed everything visually.  I guess I need to experiment a
little more until I find the most promising combinations.  

Red TRUMPET proto

The power supply box is incredibly heavy.  Very solid feel.  The rubber feet I
chose are working great, much better grip and cushioning than the typical vinyl
bumper.  However, there is a vibration in the amplifier.  When I bounce it
there is a rattle.  I think the circuit board within the chassis rails?  If I
press on the board the rattle goes away.  So I need to come up with something.
My first thought is to run a nylon standoff from the centers of the tube
sockets down to the bottom cover.  The #4 nuts are already in place.  Would
this be a good connection or bad?  Might be a solution for othe chassis with
rattles and shipping damage considerations.  I guess I can live with a few
screw heads in the bottom cover.  Mechanical integrity and vibration control is
far more important.  Meanwhile, the CASTANET box seems just fine with the
bumpers inside.  No rattle.  

Nov 20
------

Finished building the CASTANET.  Worked out all issues, just need to implement
them into the layout.  Suprisingly, the numbers came out really good.  Must 
have been due to the component and circuit changes.  I realized that 20Hz boost
of 2dB we had before might have been caused by the power supply.  Yeah, the
supply impedance (Ls and Cs) were ringing in opposition to the signal giving an
artificial boost in level.  Well, I had doubled the B+ capacitance, and yup,
now the +2dB is at 12Hz, but only with the outputs unloaded.  I don't need 
those extra 100 ohm damping resistors then, as I can live with this.  On the
Grados there is no boost.  The Sennheiser bandwidth is thus 8Hz to 355kHz, and
Grado from 12Hz to 267kHz.  That's much wider than it was before.  Output 
impedance is still 9 ohms (LO).  Noise and hum went way down.  Broadband was
0.4mVrms.  A-weighted it shot down to 12uV, or -98dB referred to 1V output.  
I'm happy with that.  Forgot to measure gain and distortion.  Will do that 
after it burns in for a few hours.

CASTANET proto front
CASTANET proto back

Worked out a few minor issues.  Like how to add mechanical support in center of
board.  It can get too much droop.  My first thought was to use nylon standoffs
connected to bottom cover.  But when measuring the gap, I realized the space
from tops of trannys & chokes to cover was almost exactly 1/8".  Doh!  I have
all these 1/8" rubber bumpers.  Bingo.  Just put a bumper on the top of each
tranny.  It worked perfect.  I get the extra structural robustness needed to
survive shipping, plus it has inherent self-damping.  And I don't have to drill
holes in the cover.  Also, the alternate headphone jack I selected needed a
thicker panel.  Must stick with other version.  Sure glad I found this out now!

Oops, I took the earlier bandwidth measurements improperly loaded.  I ran it
open and with Grados.  Well shux, they must not stay at 32 ohms in the ultra-
sonic region.  Ok, so let's do this again at 32 ohm load.  Actually, I ran a
64 ohm load on the HI output (same as 32 ohm on LO).  Ok, now things look
normal.  True BW is 12Hz to about 75kHz.  Distortion mainly 2nd order being
0.4% at 300mV output.  Gain is 2.6dB.  Maximum output level is 4.0Vrms, or you
could say 3Vrms maximum input prior to clipping (which is really soft and
benign).  Here is the square wave response.  That's awfully clean for a tube 
amp and no feedback.  No squiggles or ringing.  

CASTANET at 10kHz

Wow, this really proves the "sum of its parts" theory.  The CASTANET sings.
Way better than it should.  Good topology and holistic synergy go a long way
towards making a great design.  And this machine has it.  This is world-class
performance, and I did it using commercial-grade transformers, chokes, and
caps.  Ok, the 3.3uF films are $10 a pop, but they aren't your V-caps or any
other uberphile part.  It's not about how many over-the-top parts you can stuff
into a design, it's about how you can get them to play together.  And in this
case, the whole is indeed far greater than the sum of the parts.   

Nov 19
------

Oh my, I did a quick look at the mechanicals, and suprise!  The CLARION circuit
will indeed fit on the HAGLABS chassis.  I had figured there was no way enough 
room.  Guess my original 11x17 layout was pretty sparse.  Pretty tight, and it
will weigh a ton, but having the power tranny hang off the back is what did the
trick.  Ok, so now we have PP-10 and SE-10, the latter also to be offered as a
CLARION half-kit board.  I have some changes to make.  Perhaps do a CCS on the
driver?  All the iron and tubes are the same.  

I got most of these TRUMPET pieces sort of together.  And I just am not liking
that yellow.  Maybe I don't have cojones?  To me, the white looks much more
natural.  Something fluorescent in the combination yellow/red.  Must rethink
this.  The VACUTRACE looks wierd too.  The HAGTECH machines will definately go
white.  It's the HAGLABS units that I fret over.  Big difference once I put the
glossy black covers in place.  

Ok, CASTANET fired up ok.  Sounds wonderful right out of the box.  Must do
final tweaks, but so far everything looks spot on.  Every dc voltage is within
a volt.  And this is even after trying out a new tube.  I marvel at how much
sound this thing can put out.  And only the one tube.  Hum is all gone.  My 
tricks worked.  Painted the covers almond again.  But now I'm wondering if the
CASTANET panels should just be au natural.  More contrast with the knob.  HA-10
of course, will be red.  The big CD caps are working out.  Also did a re-wiring
of the primary back to cathode rather than ground.  Will do measurement and 
listening comparisons tomorrow.  I think I am close enough to just do the last
few changes for production, then get the boards ordered this week.  Yeah, I
work on a lot of holidays.  That means I should have boards by 12/10.  

CASTANET proto #2

Nov 18
------

I woke up in the middle of the night, thinking hey, why not put the PA-10 power
transformer on the back panel, just like the new CORNET?  Well duh.  A genius
would have figured this out up front.  It took me several months.  Looked into
it and seems to be a very good fit.  This will save me a lot of time messing 
with the top covers.  And, the amp might look nicer without a visible tranny.
Nothing but tubes.  I needed to make some other changes inside, like rotate the
output tranny 90 degress, move things around a bit.  Plus a ton of changes I've
worked out over the past few weeks.  Also think maybe I change the name to
PP-10, which means push pull instead of power amp.  That way I can later add a
SE-10 if I feel like it.  Would LOVE to put a CLARION circuit into one of these
chassis, except I don't think there is enough room.  

Oh, I also painted the CASTANET covers black.  Shoot!  I was supposed to NOT do
that.  This was supposed to be an example of a DIY-built machine.  Oh well.
Maybe I find me some beige paint and paint it back the way it was.  Ugh.

Nov 16
------

Yesterday I did the paint job on the new TRUMPET chassis.  Whoah!  When I
checked on them this morning, I was stunned at how beautiful they came out.  
What was once a drab beige hammertone (textured) finish is transformed.  The
glossy black paint over the texture is gorgeous!  I can't believe what a
difference this makes.  It went from a piece of test equipment into a hi-end
shiny chassis overnight.  This is such a relief!  Now, if I can only find some
time to get me a real compressor and paint gun.

Discovered that painting is a bit of an art.  Realized I didn't get enough
color on the edges, went back and touched them up.  Well, duh, it seems that
this screwed up the nice even gloss across the top.  Waited a few hours, nope,
didn't go away.  At a certain distance the spray is a little too fine and
sparse, not covering properly.  Seems you need to saturate the entire area with
a wet coat in order for it to dry with even patterned gloss.  Ok, so I threw on
another coat.  

Last night I spent 3 hours in a flurry of part ordering.  Six different 
companies, had all websites on at once, so if mouser was out of stock, I would
find something at digikey, etc.  Agonizing to try and keep it all straight.
Anyway, after I was all done and hit the six submit buttons, I realized I
forgot to order more HAGUSB boxes.  Crap!  Oh well.  I have enough to last
another week.  

Nov 15
------

Thinking maybe one of those LED light pipes might work well for the IR remote
control receiver window.  Better make sure it can pass such a long wavelength,
though.  Forgot this window needs to match up visually with the rest of the
decor.  Focusing hard on WAF.  Yup, that's going to be one of my key selling
points.  Wife Acceptance Factor.  Small, cute, unobtrusive components.  Stuff
that doesn't take up the whole room.  Stack them nicely, if you want.  If you
bring your wife to the store, she'll say, "that one is fine."  Something in me
rebels against the gigantic sculptures that some companies put out.  Unwieldly
contraptions that all compete for the top shelf.  Which is where the damn
turntable is supposed to be!  

Ok, tried some series damping on the PA-10 interstage.  Ended up with pretty
much the same result, but more gain.  Oddly, discovered the tranny is not
symmetrical in it's HF performance, as one secondary loaded with 22k does all
the work, 22k in the other has minimal impact.  So somthing is up with this
tranny.  Is my sample ok?  Are the secondaries wound bifilar?  Seems perhaps
not.  Regardless, when I have it tuned in for maximal performance, the 
waveforms look perfect, just bandlimited to 23kHz.  I'll have to listen some
more to see if the treble is impacted.  Loss of air?  Meanwhile, I order the
156 interstage tranny tonight.  Can try it next week.  Once I get that nailed
down I can do the harmonic cancellation testing.

One more HAGDAC to build and then I can get back to finishing up the HA-10
proto.  More stuff in paint shop.  Oh yeah, built me a new wooden shelf last 
night for LP storage.

Trying to figure out where to put an LED on the fron of the red CORNET.

Nov 13
------

And now I discover my inventory is perilously close to running out of BUGLEs
and POWER SUPPLYs.  Oh my, more stuff to order.  

Getting some chassis parts ready for painting this afternoon, but figured I
better drill the footer holes first.  Need template.  Hey how about this CORNET
board?  Just off a little bit, so I use something else.  Hmmm, isn't it
interesting how the CORNET classic board fits perfectly inside the new red
TRUMPET chassis?  Doh!  Then it hits me.  Shoot, this is an easy one.  I
realize the tranny can be mounted to rear panel, everything fits.  Wow, this
would make for an even smaller CORNET.  And so it was born.  Why not?  I can
offer the RED CORNET in the upgraded chassis with upgraded caps and stuff from
HAGLABS.  Too bad I can't come up with this stuff myself, it has to happen by
accident.  Sort of makes the HAGLABS launch a bit more phono-centric.  Should
have me the RED ARCHIVER, RED TRUMPET, RED CORNET, and RED PICCOLO.  Plus the
HA-10.  The LC-10 and PA-10 will come later, although they both have working
prototypes at the moment.  I decided to keep the VACUTRACE, FRYBABY, and UFO
with HAGTECH.  

Meanwhile I am so busy just filling orders, I can't believe I get anything else
done.  It takes me the first 5-6 hours each day to keep up with production.
After that I move one to the new designs.  

Discovered during that 7788 triode curve test that the VACUTRACE uses too much
plate voltage modulation for measuring gp.  I'm going to look into reducing it
to 25V.  The line on the scope gets a little bent, leading to a slight error.

Ok, back to working this PA-10 board.  Changed the driver tranny primary 
connection to cathode.  Hey, might as well do as much as I can from the 92A
schematic.  It's a good topology.  I also got to look at the effects of the
soft ground connection to the driver center tap.  This is very unusual.  It
lets the common mode float.  What happens is that in grid current saturation, 
the coil gain compensates by pushing the other grid harder.  Funny looking
waveforms in saturation, but the result is symmetrical clipping.  Overdriving
this amplifier is not a bad thing.  It handles it really well and in a very 
nice fashion.  Soft clipping with no harshness.  I still have to add in the
1uF cap to cathodes, see what that does (I think a clever way to add power
supply rejection).  Then I have this idea for positive feedback neutralization,
using the 40% taps coupled through 10pF micas back to opposite grid.  In theory
this can greatly reduce the input capacitance.  The problem for me, is that my
bandwidth limitation is all with that nickel core interstage tranny.  Or is it?
I have yet to try series damping instead of parallel.  Also can try the 156
interstage tranny, which is on order with the L2 filament trannys (have to find
the better balanced).  Oh yeah, and then the harmonic cancellation.  I'd also 
like to do listening tests on each version as I progress.  Last night I rebuilt
my speaker cables.  Ripped them up and converter to a simple twisted pair
teflon over silver.  Mostly I just wanted to remove that plastic jacket braid.
Sure, it looked nice, but that's not exactly what I want as a dielectric.  

Finally, I am having to raise prices a little.  This fall in the dollar has
really started to add up.  My parts costs are up at least 25% across the board
over a couple years ago.  Especially with items like the PCM1704, I'm now 
paying double what I used to.  

Nov 12
------

Got me a better sample of 7788.  New curves look like this.  I now get 1.5k ohm
output impedance, which hits spec.  

Improved 7788 curves

Thinking I'm not going to move VACUTRACE over the HAGLABS.  Maybe such would 
cause confusion.  I should not have any direct sales at that website.  My
desire to launch with lots of product made me do it.  After a weekend of house
painting, I can think more clearly.  KISS.  Just launch the TRUMPET and HA-10
stuff to start.  The VACUTRACE does not need to be red.  Forgot to order
ARCHIVER panels.  Oops.  And now I am running out of BUGLE and POWER SUPPLY
boards.  So much to reorder!  Sales are good, with a big increase in assembled
machines (HAGUSB, PICCOLO, BUGLE).  

Meanwhile, building up my new listening room.  Bought this $160 cutting board
cart for conversion into a TT stand.  Very thick hardwood, has really inert
sounding quality.  Much better than my other racks.  No, I don't have a sandbox
or anything fancy.  I left off the drawer and a middle shelf, only bolted
together what was needed, added glue to the mix.  Once I cone it, should be a
pretty decent, heavy, and inert platform.  Someday I can put a Serac or similar
on it.

Nov 10
------

Sat down and did some thinking last night.  Try to plan my next moves on this
PA-10.  So far, it's actually moving along quite well.  I have a complete
working amp that sounds decent.  Me want more.  So I'm going to try some 
alternate parts.  The 124B is causing me the bandwidth limitations.  So maybe
I try one of the other versions of this coil to see if they work better.  Was
hoping this would be a good piece, due to it's high nickle core content.  Will
also try the L2 heater trannys.  I used them in the CLARION without any hum
imbalance.  Maybe they will work better than the M2 style.  I'll pay the extra
$5 if they do.  

Interesting item, I accidentally discovered that the circuit operated just fine
one one tube.  At about 1/2 the output power.  Output tranny was singing like a
bird, though.  One thing I didn't check was how well the common mode balance
correction works.  That's the series 47k with the center tap on secondaries.
Normally this would just get connected to ground.  However, I believe there is
a certain benefit to not doing such.  You see, when driving into clipping, one
grid enters into positive grid current.  When this happens, the common mode of 
the driver coil shifts to put more signal on the opposite tube.  Somewhat of a
self-compensation.  Forgot to look more carefilly at this mode of operation,
however, I did notice the drive waveform does not saturate and clip like the
output does.  Maybe it is working.  I will try it both ways.  You see, I am
drawing inspiration from the WE92A amplifier of 1935 and comments by Lynn
Olsen.  This extra resistor is found there.  Also the unusual harmonic balancer
mode of operation.  My topology is similar, but with the additional advantage
of CCS loading, and capacitively coupled cathodes, allowing for unmatched tubes
to accomplish the same thing.  Sort of taking it to the next level.

Lynn's comments on WE


Nov 09
------

Got a lot of panels in.  The CASTANET (DIY version) and TRUMPET.  Hmmm, they
both have fit issues.  I did not leave enough room for the insulating washer on
the RCA jacks.  Slight interference in both machines.  Also thinking I want to
add a standoff from board to bottom chassis panel in order to help with the
heavy tranny weights during shipping.  Holding the boards along the sides may
not be enough, looks like there is a little droop.  Drop the box and board
could crack?  A nylon standoff should do the trick, stay insulated, yet still
offer a little vibration damping softness.  I find nylon works great for such
microphonics control.  Perfect as washers under the chokes, too.  I have same
problem with PA-10.  Too much board droop.

Some thinking on the 7788 curves.  Why are they off?  Then it hit me, there is
a little bit of cathode degeneration in the VACUTRACE.  Not an issue with most
tubes, but it shows up with this uber-tube (high mu, low rp).  Doh!  So I have
an idea to reduce the degeneration and make up for it with more gain in the
processing stages.  Will try it out.  Perhaps I can get more correct curves to
match the data sheet.  Right now it shows the rp too high.

More work on the PA-10.  My first try to reduce the noise was putting the bells
back on the output tranny.  It's a good idea regardless, and they fit, even
inside the chassis.  But that didn't remove the hum!  Shoot.  It is 60Hz, not
120Hz.  After a lot of playing, finally discovered that the heater trannys are
not balanced perfectly.  They both measured about 1.237Vac and 1.221Vac
relative to center tap.  Crap!  I'm feeding the 60Hz signal right into the
tubes.  Aha, but I noticed they are connected to the primary side with the
same polarity.  So I switched on, thus cancelling the effect in the output
stage.  Son of a gun, it worked.  That 16mV differential noise on cathode was
multiplied by a mu of 4, divided by a turns ratio of 12.  It measured 8mV on
the 8 ohm tap.  Now it is under 2mV.  Hooked up to the vandy again and could
not hear it.  Just a faint buzz.  Putting on my A-weighted meter, I now have 
85dBA SNR ref 1W.  That's actually pretty friggin good.  Great!  I didn't have
this trouble with the trannys I used on the CLARION.  It shows up with the
166M2.  Hmmm, maybe that's why it is cheaper?

So now work on bandwidth.  The driver stage is flat out to who knows what.  At
250kHz it hadn't done down a hair.  Everything is lost in the interstage
tranny.  That makes sense, as the CYMBAL pushed 95kHz through the output
tranny.  Yep, it all comes down to the 124B.  I have no dc on it (parafeed), so
there is not much else I can do.  Driving it hard with a 1.5k ohm output from 
the 7788.  The 2A3 grid capacitance ain't that bad.  Yeah, it just falls off,
with some ringing and glitchies to boot.  So I tuned the secondary loading,
found out that 22k ohms per side gave a PERFECTLY tuned square wave response.
Absolutely beautiful waveform (at 8 ohm tap).  It just happens to be at half
the bandwidth I desire.  Perfectly behaved rolloff, but the -3dB point is at
22kHz.  Well, that's perfect for CDs, anyway.  A little soft for vinyl.  My 
ears probably couldn't tell the difference.  Sensitivity dropped to 0.6Vrms
for full power.  I should try series damping.  That might bring BW back up.  If
its a resonance with grid capacitance, anyway.

Oh yeah, also redesigned the current sinks for the 2A3.  Back to a 12V zener
circuit, one without feedback (which I think is important).  Still have to 
work out the right value for harmonic cancellation.  Thinking that resistor
needs to be 250 ohms.  My earlier guess was 120.  I'll measure the spectrums at
1W for several cases.  It should lead me to a null.  If it works per theory, 
well then I am stuck with the interstage tranny, because this way cool topology
needs it.  Relayout will be a bitch.  So much changes.  

Anyway, the weekend is here and I gotta get back to painting the house.

Nov 08
------

Finally got me a backup of my CNC files on floppy.  That lab PC crashes and
dies about once a year.  I'm on my 4th machine now, I think.  Anyway, the files
are of G-code programs that tell the CNC machine how to move and cut.  That 
logo stencil I did last week looks like this.

HAGLABS logo machine code

Cleaned up my old Hafler amp.  Repainted the top.  Then painted on a HAGLABS
logo.  Oops, I didn't tape the newspaper down so I got some overspray.  Doh!
Messed up that paint job.  Anyway, will put this amp back into service for
PA-10 comparisons.

I think I'll hold off on converting the assembled FRYKLEANER stuff over to
HAGLABS.  Too many other things on the plate right now.  It can wait.  I did,
however, buy a sample red translucent case for the FRYBABY.  Main goals right 
now are to get the CASTANET kit out the door and HAGLABS website launched.  
That is, with actual product.  Six of them.  People are again asking for
TRUMPETs and VACUTRACEs.  

Huh.  This was wierd.  In a great way.  Got the Hafler working, gave it a day
of burn-in.  Ok, let's try this out.  Hooked up the vandys, CHIME, etc.  Dang,
one channel is all noise.  Must be some lizard shit on the transistors.  Hmmm.
So I ran one channel, did a whole bunch of A/B with the same channel on PA-10.
Hey, they both sound like crap.  That is, the thinness is all due to the awful
listening room conditions.  The big SS amp didn't sound any better.  A little
more punch and better bass control, but otherwise tonality was about the same.
Ok, run one amp per channel, stereo.  I have to admit, it was suprising how 
well they imaged.  Seriously, I had a pretty good soundstage.  But here's the
weird part.  I could hear the left speaker, but not the right.  I could stare
right at the right one and not hear it.  All of the sound was imaged between
the speakers, yet it was easy to precisely pinpoint the left one.  Well, I'm
hoping that is because the right was powered by the PA-10, not due to odd room
effects.  All listening was near field.  I moved from five to six feet away.
Slightly up and down.  Yeah, still the imaging is better slightly above axis.
I can't figure that one out.  Once I got used to the room, I could hear the
full range of music.  Nice to have a bass line.  Some tracks like Jack Johnston
"upside down" from curious george really likes to have a bottom end to give a
pace and rythmic flow.  It really sets the foundation for all of that
percussion layered on top.  And so, I thought maybe I had really messed up with
this PA-10.  But it was mostly room.  And the vandy came out fine.  Maybe a tad
bright on top, or perhaps that was the Hafler.  Even with its obvious technical
flaws, the PA-10 ended up being transparent and musical.  The fact that I could
NOT hear that speaker was key - that is, pinpoint its location.  Oh yeah, the
other good news is that I got the polarity correct!

Nov 07
------

I need to make a new listening room.  A second one.  Heck, my 1st is still all
torn apart because of the construction and painting I did.  But I think it is
time to make more room in the shop for parts.  So I'm building a new area for
testing.  Convenient to have it right where the vandys are sitting.  Terrible
acoustics, but it's safe from dogs and kids.  My Pete Riggle VTAF just came in,
so I'll put that on the Rega and move it into the house.  I need a much better
stand.  Acutally, I might already have one.  Just need to spike it.  Anyway,
expanding.  Hey, the more I think about it, our master bedroom should have
quite good acoutsics.  It is well broken up and has a built-in bass trap.  I
should try that out.  Cement under the wood floor.  Should be perfect for a
turntable.  Hmmm, now I don't know where to put things.

I am so stupid.  Just checked out the acoustics in the bedroom.  They're f*ing
excellent!  Why did I never think of this before.  It has wonderful sonic
features, bass trap in back (closet full of hanging clothes), behind the
speakers I can open windows in the corners, walls are all broken up with odd
features, ceiling is vaulted.  Shoot, it's incredibly dead, non-echo.  There is
a perfect spot to the side for a table and rack.  All I need to add is a small
rug in front of the drivers.  I'll have to lounge on bed for listening, but how
bad can that be?  This should evolve nicely.  Even the Maggys should eventually
fit.  Screw the living room.

Tried out some different fancy yellow LEDs.  Various currents, how they look in
the red panel.  I like the #67-1126 from digikey run at 5mA.  It has 1.88V drop
across it.  Was hoping for less current, but this is what it looks best at.

Designed me new ARCHIVER panels today.  These are made to fit into a proper
chassis, one with top and bottom.  Awesome chassis I found for $41.  Panels
cost a fortune, but hey, this is hi-end.  Needs to look the part.  I'm adding a
BUGLE power supply to it, see if that helps anything.  Looking into possible
capacitor upgrades.  One thing I finally decided to do was to remove the XLR
outputs and just go with RCA.  Suppose I can do XLR on a custom order basis.
In fact, I think I'll revert the new TRUMPET to that too.  Make it RCA only,
customizable to XLR.  I like that better than trying to have both.

Ok, hooked up the PA-10 to a vandy.  Mono only, put on a CHIME.  Well, I guess
I am so far underwhelmed.  Very midrangy, recessed.  Not too dynamic.  How much
of this is due to one channel?  Anyway, seems to have much less bottom end than
the CYMBAL did (why the frick did I sell that thing?).  Maybe I am realizing
the CYMBAL was way better than I thought.  It seemed to have it all.  Except a
lot of power.  I should have kept it, at least as a reference.  Anyway, I got 
some more things to try on this amp.  I can clean up my Hafler to run the other
channel.  Meanwhile, moved the TT inside.  No idea when I can get to the VTAF
upgrade.

Nov 06
------

Got it!  After futzing all morning changing around caps, making new CCS, I
finally stumbled upon the problem.  Doh!  The pin 1 of the right side 2A3 went
into the socket too deep and shorted to the grid trace.  Stupid.  That's why
the grids always shot up positive.  Anyway, a simple insulator did the trick.
Now everything is working quite well.  I am wondering how much I can back track
now to restore the original circuit.  The cap change stays, as the booch caused
that brilliant insight.  I should try listening to it.

The PA-10, not suprisingly, ends up being an awful lot like the CYMBAL.  Here's
what I'm getting at the moment.  6W output power prior to clipping, 0.5% disto
at 1W, 0.5V input sensitivity, 45mA on each tube at 238V, -47V bias, 17Hz to
27kHz @ -3dB.  I'll see what I can do to improve on distortion and bandwidth.
There is a slight but very well behaved overshoot in the step response.  

Ok, a little tweaking.  Still has a bit of audible hum and buzz.  Output
impedance measures in at 4 ohms.  Higher than I had hoped.  Hmmm.  We take 500
ohm output on anode, 12.5:1 turns ratio, that's 3.2 ohms.  Two tubes in
parallel should cut that in half.  I need to shift the current up another 10mA
or so.  Changed coupling cap to 2.2uF and bass response is now -3dB at 9Hz.
The top end is having some trouble, due to the interstage tranny.  I am getting
a 1.5dB peak at 16kHz, then 0dB at 20kHz, to -3dB at 27kHz.  Fast rolloff.  
This appears to be transformer related, as the driver pushes the grids no
problem.  I wonder if I had the wrong tranny leads hooked up?  I'm only using
one of the secondaries for 8 ohm connection.  They recommend both.

Nope, that didn't make any difference.  Hmmm.  I wonder why the bandwidth of
CYMBAL is so much better.  Even the CLARION was way better than this.  I do
lose a bit from the interstage, but something is dragging down the output.  Is
it the higher output plate impedance?  Did I even install a Hammond 1615?  And
so I got to thinking about the really wierd way they wire up these trannys.
They give these connection diagrams.

4 ohm
8 ohm
16 ohm

And so I thought about how they do this.  The answer is that it ain't really an
8 ohm connection, but 9 ohms!  Seems to me they have two primaries, probably
wound in bifilar, and six identical secondaries, which they then internally
strap with some in series and some in parallel.  It looks like this.

Hammond 1615

I figure the turns ratio is roughly 35:35:1:1:1:1:1:1.  Probably a very good
construction architecture.  Two of the secondaries are wired in parallel for a
1 ohm output.  Then there are two other sets of two secondaries in series, each
giving a 4 ohm tap.  Put these in parallel and you still have a 4 ohm tap.  Put
the two 4 ohms taps in series and you get 16.  Put all of them in series and 
you get 25 ohm.  Here's the kicker, doing a 4 ohm in series with the 1 ohm
gives you 9 ohms, not 8.  Hey, close enough!  The 9 ohm wiring recommendation 
is meant to even out the winding resistance, basically 3 secondaries in 
parallel with the other three.  It does make sense.  This odd configuration
gives you 1, 4, 9, 16, and 25 ohm options for a 5k primary.  See the pattern?
That's one-squared, two-squared, three-squared, etc.  

Ok, that gives me something more to think about.  I also need to put back in 
the common mode hum cancellation trick (also from western electric), and then
tune the parallel vs. differential ratio for distortion cancellation.  Up the
current, see what is killing the bandwidth.  The CYMBAL had a similar issue
when I connected a 1608, which made me change to the 1615.

Nov 05
------

That amplifier for maggys?  It need to have more power.  I want 50W or so.  The
good news is that can be designed for maximum power transfer, have an output
impedance of 4 ohms.  Nice and high.  Design the output stage thusly, build it
with two channels and a line-level xover inside.  A bi-wire amp.  Heck, why not
make it specific for the maggy 1.6?  There are sheetloads of those panels out
there.  It is a market unto itself.  This might have to be another HAGTECH kit?
The big issue in front of me is, what tube?  I need a kick-ass triode that is
linear.  And that ain't the 6C33.  

Oops, discovered I left that 7788 connected as tetrode, not pentode.  Pin 8 was
floating.  Here are the corrected curves.

7788 pentode
7788 triode
7788 triode reference plot

My best guess at values is rp = 2.5k, mu = 50, gm = 20mA/V.  This is at bias of
Va = 200V, Vg = -3.0V, Ik = 20mA.

Man, I am having some trouble getting this PA-10 to fire up.  Can't get the 
CCS loading to turn on.  Makes no sense, unless the transistors are blown.
Which is possible after what I've done so far.  The design is really cool, but 
I made the mistake of concentrating on steady-state conditions.  Doh!  I forgot
to consider power up transients.  Just noticed that the RC time constants in 
various places cause some very bad things to happen.  Crank!  I reverse bias
some electrolytics, which then push the 2A3 way into cutoff.  Oh crap.  I've
made a mess.  Sitting at the scope & DVM didn't do it.  Sat there with
schematic in hand.  Just thinking.  Wha, no, yes, huh?  Eureka.  Came up with a
major change that solves a number of issues.  Wow, big.  A total redo of the 
output stage caps.  I need only two of them!  This solves power up, and makes
the current loops even smaller.  And I am starting to hate the CCS design.  So
for now I am just going to rebuild the stage and use 820 ohm pull-down
resistors instead of CCS.  I was doing too many things at once.  Need to get a
basic version up and running first.  Then I start adding CCS and feedforward
stuff.  Wow, this is way cool.  I may not even need the CCS.

Nov 04
------

The serpac cases came in.  Wow, these are beautiful.  This will make for an 
awesome new VACUTRACE.  Also got in parts to finish the CASTANET board.  This
is really starting to shine.  It has attained that rare holistic synergy.  Very
much in tune with itself, everything fitting and playing together like a glove.
The board with parts alone is 7.5 pounds.  That's without any chassis!  Been
upgrading parts here and there.  Big 2" long 3.3uF caps from Cornel-Dubilier
(super polyprops), silver plated fuse clamps, ceramic fuse, high grade low
impedance electrolytics, etc.  All without busting the budget.  Just waiting
for the panels to come in.  Then I think I can write manual, order production
boards.  

Always on the lookout for various tricks.  Bought some pickup truck bed liner
spray today from the auto parts store.  I want to try this is a damping
material, see if it is any better than the cork/rubber adhesive strips.  

Trying to get the PA-10 fired up.  Looks like maybe the lower CCS aren't
working or the 2A3 has problems.  Much interconnectedness, I wonder if I made
a latch condition.  The 2A3 are supposed to heat up first, followed by B+,
which then gives some voltage for the CCS to work into, thus allowing the high
voltage LED to turn on.  I get B+, the driver stage biases correctly.  Dunno
yet.  So I took the time to make measurements of the 7788 driver tube.  Made me
an adapter card for the VACUTRACE.  Measured me two older (1964?) Amperex "PQ".
In tetrode mode got a very noticeable kink.

7788 tetrode

I forget if this was 0.5V or 1V per grid step.  Not sure where zero is.  H axis
is 50V/div plate, V axis is 12.5mA/div.  Screen was approximately 200V.  The 
looping in the curves is due to the screen supply droop in the VACUTRACE.  It 
isn't really there.  Ok, so now triode mode.  I did some static measurements at
about 180V plate/screen @20mA.  Transconductance was off the scale, plate
resistance was 2.8k ohms (somebody said it was about 1/2 this?).  

7788 triode

Not bad, but that kink is still there!  In fact, it's huge.  But from -1V grid
onwards there seems to be no effect.  Quite linear, actually.  I was really
worried when I first saw this, but I think I am ok, at least in the region
where I plan to operate it.  I would estimate mu to be about 50.  So yeah, that
makes this a bit of a supertube.  High current, high gain, low output impedance.
Here is the same tube but zoomed into 5mA/div.

7788 triode zoom

In my opinion, this is awesome linearity.  Visually way better than most tubes.
Hey, my earlier estimate at -2.5V bias at 180V for 20mA was right on the mark.

Nov 03
------

Had this idea to use the new VACUTRACE box for the COMPRESSOR too.  It would be
an intersting solution.  Fixes a number of issues with the design, plus it
makes it very unique in the marketplace.  A rugged portable mic pre chassis,
not the typical rack mount stuff.  I think this would be much more versatile,
and you could make the mic cable runs shorter.  Something tells me all that 
capacitance sitting on a mic output is similar to what happens to an MM cart.
UA makes a portable mic pre.  With handle!  I'm not that far out.  Anyway, just
thinking that since I have this fabulous circuit, I could offer the machine in
a new format.  It costs more, but I'll just have to charge whatever it turns
out to be.  Besides, I really want one of these things for my own use.  Yeah,
after we get the TEN series out there, it's gonna be home recording time.  I
think this will give me a whole new battery of listening tests.  That, and I 
get me a pair of Maggys.  Oh yeah, the COMPRESSOR is a great reason to try out
the cinemag trannys.

Nov 02
------

This is starting to look like an advertisement for Hammond.

Hammond chokes & trannys

It's actually a CASTANET.  Note how I moved the power supply chokes to minimize
hum induction.  Clearly there was some stray magnetic coupling between the
first choke (lot of volt-seconds) and the parafeed chokes.  No way to stop it
except via distance.  So I made fields symmetrical about the channels, then 
moved the first choke as far back as possible, then oriented it such that the
corners lined up.  That is, a 45 degree angle on the bobbins.  You see, the 
bobbins are square, not round.  This means the windings are for the most part
square, with flat sides.  The external field is a cross product vector, and
perpendicular with the wire direction.  In other words, the external field is
strongest out the flat sides, and minimal at the corners.  Bingo!  Line the
bobbins up for minimal coupling at the greatest distance.  When I tested this
out on proto #1 it worked just fine.  

I got just tons of parts on order.  The HAGUSBs keep rolling, constantly buying
parts for them.  Have expensive panels for just about everything else coming in
like the new TRUMPET and VACUTRACE.  New CASTANET panels.  Ten new projects at
once.  That's a bit insane, except they have a lot in common, and I needed to 
set up the architectures in sync so that the product lines all match visually,
electrically, and mechanically.  New chassis for ARCHIVER due in too.  Yeah,
this sort of R&D spending can sink you.  But not to worry, I just secured an
additional $20k in financing.  That should cover the prototype expenditures.

Meanwhile, I have to start signing up some dealers or a distrubutor for the
initial HAGLABS stuff.  Am I doing the cart before the horse?  I dunno.  Need
to get someone selling the red TRUMPETs and red PICCOLOs for me.  The HA-10 is
also right around the corner.  My goal is to launch Jan 2008.  The PA-10 might
not be too far behind (which reminds me I gotta debug that board), with the
rest coming online at RMAF.  Red.  Everything is red.  In fact I just bought
some silver wire today covered in red teflon.

This weekend I am setting up another workbench in the shop (expanding around 
the corner).  This will be for assembly only.  It will be for the hired help.
Also might set up me paint shop.  

One issue I am having trouble with right now is the thread-forming in the
chassi extrusions.  I have been doing it by hand.  What a pain!  It leaves too
much shards.  I would like to automate this process.  And find a way to clean
up the material afterwards.

Ok, fired up the PA-10 proto.  Finally.  Thinking this is gonna be a tough one
to debug.  Powered up with no tubes.  No smoke.  Not sure what else I can do, 
so plugged in all the tubes and went for it.  Still, no smoke or pops or bangs.
Heaters came up, but the current sources weren't.  I need to make a prop to
hold the cover off so I can probe inside.  Better check polarity of the zeners
and transistors.  Here's what it looks like.

PA-10 proto

Looks reasonably simple from the outside.  Inside, though, it is packed full of
stuff.  Components both sides.  Here is the bottom.

PA-10 guts

And I changed my mind again.  Decided I just didn't like the looks of the
toilet seats.  So off they came.  Instead I tried to mount the spikes directly
to the base of the cabinets using some 1/4-20 inserts I bought.  Well, except 
that these cones are not 1/4-20.  Nor 10-32.  Sheet, they're M6!  Great, I 
can't find metric inserts anywhere, and now I have these big holes drilled.  So
to just save time, I hard glued those cones in place.  This ends up giving them
a better (harder) coupling to the box.  Screw the threads, I say.  Ar ar.  This
time I reversed the order.  The toilets and original stands had two feet in
front, one in back.  It now has two feet in back, one in front.  Just made more
sense to me.

Nov 01
------

Had this idea, mark boxes or maybe products with the HAGLABS logo in paint.  
All I needed was a stencil.  The logo was specifically designed with this in
mind.  Hey, I have a CNC.  Shux, I just write a quick program and cut my own.
Took two tries, but I got it nailed.  Actually came out nearly perfect.  Made
it two inches tall.  Can't wait to try it out.  But on what?

Logo stencil

Realizing there is something unusual about the maggy xover.  Didn't hit me
right away, but normally crossovers are designed with the -3dB points set the
same.  Here we have one at 500 the other at 1600.  How is that going to work?
It just means the drivers have a response of there own.  Noticed a comment
somewhere that talked about the rising response at 1k.  Well, there it is.
Also, one of the changes from a 1.5 to a 1.6 was lowering the crossover point!
Ok, so perhaps my idea to raise it by changing the LP section is bad.  Maybe
that is exactly what they used to have?  I dunno.  But seems to me the only to
start answering some of these questions is by measuring an actual panel with
crossover removed.  I wonder if Stu has some.

Oct 31
------

I'm thinking now I should not have sold my CYMBALs.  Probably should have kept
that amplifier around for a reference.  Cost me about $2k in parts to built it,
sold for less than $1k.  Beautiful machine.  I hate to see it go.  

Here's a photo of the new VACUTRACE chassis.  I don't know why I didn't think 
of this earlier.  They also have a smaller one that can be used for holding the
adapters cards.  Handy.  Not only is this way more rugged (and cool), but it
makes the machine far more portable.  All cables, cords, cards, etc. fit into
the two carrying cases.  Now that I think of it, the FRYKLEANER should have 
been designed for such a box.

New VACUTRACE chassis

Worked out a faceplate.  The mechanics are very similar to the original, but
with the controls slightly closer together, the OUTPUT knob moved, and bringing
the BNCs and power cord stuff to the top.  Added some handles.  This will look
fabulous.  Ordering parts.  

New VACUTRACE panel

Meanwhile, I am re-doing the ARCHIVER in red, giving it a real chassis.  Right
now it is open-frame style, made for professional use in studio racks.  But I
think this will work for home hi-fi too.  And so I am designing it to fit into
a 2U chassis.  This will also shift over to the HAGLABS brand.  I likely may 
leave off the 16-bit digitizer - that seems to have missed the market.  So now
I can have both XLR and RCA outputs.  This will be just like the red TRUMPET!
A good match, these two.  Even more interesting is that there is an audio guru
out there who likes it enough to consider OEM-ing it.  That is, build a version
to their specifications (part selection, colors, etc.) for resale.  I'm ok with
that.  Never expected this to be a huge seller, but it definately is a unique
machine.  

Lots happening here at HAGTECH / HAGLABS.  I realize it may seem very confusing
and changing day to day, but there is an underlying plan.  Basically, it is to
create the HAGLABS brand and sell upscale assembled units.  At the same time,
reduce HAGTECH to half-kit only status, plus maybe a few of the lower cost
assembled machines like the HAGUSB.  One brand sold via dealers, the other is
factory direct.  Both approaches work, I am merely trying to re-organize them 
as separate entities.  Hopefully, I can develop the technologies and
innovations for HAGLABS, and then trickle it down into suitable kits for
HAGTECH.  That is the plan.  And right now absolutely everything is in play.

Ok, every once in awhile I take a break from the grind.  Looking up more stuff
on the maggies.  Trying to figure out what to do for a crossover.  The stock
version has 2 pole low-pass and one pole high-pass.  Tweeter inverted.  Hmmm.
Looking at the acoustic measurements, one can see that both respond to -40dB 
per decade, or two pole response.

Maggy 1.6 response

Now there is mention of an acoustic crossover.  I'm thinking what is happening
here is that the tweeter dimensionally has the equivalent of a one pole rolloff
(like a horn).  Heck, there is only so much the small surface area can do.  So
the combined electrical and acoustic networks gives you a nicely designed flat
curve.  The problems are impedance peak and goofed up transient response.  Ok,
so we can fix the impedance issue by using an active crossover and bi-amping.
We can also easily match levels.  The issue now becomes what can be done to 
make the acoustic response perfect.  I think a natural first start is to mimic
the crossover as is.  The possible improvement comes with solving for time and
phase alignment (no inverted tweeter).  This article does a great job
explaining a couple of the more common choices:

Crossover article

From it I am guessing that the stock system mimics the LR-2nd order variation.
A good compromise with it's simplicity.  I can't help but wonder what sort of
sonic improvements can be had by fixing that inversion.  You know me, I'm a
transient response fanatic.  The question is, how to do it simply without
making a mess of things.  The good news is that I can build dummy networks into
the PA-10, such that they can be easily modified for bi-wiring, without having
to use an external xover box.

Back again.  I did some quick sims to get a feel for what the maggies were
doing.  The stock filter values have the HF corner at 1.6kHz and the LF corner
at 500Hz.  Approximately.  Trying to replicate that above measured response, I
added an extra CR high pass at 1/2 the HF corner, or 800Hz.  This sort of gives
me what I see in the graph.  

Stock maggy xover in SPICE

As you can see, there is a big dip in the midband (blue).  Naturally my
inclination is to see what more I could get out of the woofer.  That is, move
the corner up.  Keeping damping the same, I rough out values of 1.5mH and 10uF.
Ok, now crossover is at 1kHz, but the dip is WAY less.  Now look again at the
measured plot.  There is a broad resonance at 1k by the tweeter.  Looks to me
like there is a lot more than what meets the eye.  That's why I shifted the 
woofer up, it now has a small dip centered about this 1kHz peak.  Ideally they
should tame each other.  However, who knows what might actually happen.

Shifting LF corner up

Anyway, this would be my first test.  Try a 1.5H with a 10uF on the low pass.

Oct 30
------

Reading new issue of TAV.  Yeah, that might be what I am looking for.  The 
Magnepan 1.6!  That my be the right speaker for my living room.  But I had 
always heard panels were so hard to drive.  So a little research.  Plenty of
info out there, seems to be a fanatical crowd (a good sign).  The question
becomes, how to drive them?  That would mean a new amplifier?  I suppose the
4 ohm output of the PA-10 will do ok, at least at low levels.  Quality and
clarity without volume.  Looks like the answer is to bi-amp, use a line level
crossover, put the drivers back in phase.  

Ok, so maybe I am a madman.  Once again shifting in the wind.  After a lot of
hand wringing, I decided not to do a VACUTRACE kit.  It's just not worth the
hassle.  You have no idea the debug support that would be required.  It's a 
really complicated circuit.  Not for audio DIY.  And so I looked into making it
in frontpanelexpress rack mount form.  Something maybe a little nicer?  The
problem was one of chassis for the backside.  And then I realized, shoot, the
dang chassis I use for the TEN series might just fit.  It doesn't, but is soooo
close, I had to see what could be jiggled.  Sure enough, the trick is to mount
the tranny to the panel and not the board.  Then move the OUTPUT control off
too the side.  Shux, I am really suprised this might work.  Not only do I end
up with a better chassis, but I can make it match the other HAGLABS equipment.
Stay tuned.  I'll build one up to see.  Honestly, this would be a great boon to
getting HAGLABS off the ground.  I'll have TRUMPET, VACUTRACE, PICCOLO (red),
and the HA-10 on which to launch.  The UFO can also move over, get the upscale
brand name.  Hmmm, the FRYBABY belongs there too, as it is already a dealer
product.  Ok, that is 6 machines for the grand opening.  Also thinking of doing
an upgraded ARCHIVER in red, with both XLR and RCA outputs, no USB ADC, and a
nicer internal ac power supply.  It really is a great sounding unit and belongs
in the HAGLABS class.  Seriously, the $1000 assembled machines do NOT sell at
HAGTECH.  Too much stigma from the kits.  Don't know what to do about pricing
yet.  The VACUTRACE likely will not have the normal dealer markup, expect about
$1500 for the new version.  

Looking at paint tools online.  Dang, looks like I can get a compressor and 
paint gun for under $200.

Oops.  In my zeal to work out the vertical fit issues I neglected to see the 
obvious, where a 10 inch board doesn't fit too well in a 9.3" chassis space.
Doh!  Crap.  I was feeling lucky.  There was a portacab version, but it had 18"
depth when I only needed about 5".  Must ponder some more.

Didn't sleep well, but came up with an improvement to the rack.  Change to some
round dowels, like a wooden closet pole, use them for legs.  The round shape is
nice for spiking.  Anyway, as I lay there thinking about bi-amping Maggies, I 
realize that is four monoblocks.  The old style rack is too tall.  How about a
side by side approach?  Two legs in the middle, bolt to chassis.  One leg on
each outside, on rear of chassis.  This is more of a triangular topology, and
has the interesting visual effect of leaving the front panels hanging out in
space.  No front legs on the outside.  Might look cool.  Anyway, it's an
alternative connection to the standard 4 poster.  Same parts.

Ha!  I knew it would come to me.  Actually, this was an idea from several years
ago.  Use the serpac #R520 case!  Duh.  Talk about test equipment, this is a
perfect fit.  And cheaper, too.  Use the #R120 smaller case to carry the cables
and adapter cards.  Curve tracer to go.  Plenty enough room, no games.  I can
make this look real nice.

Oct 29
------

And so I thought, hey, just get an older Mac Mini on eBay.  They go for $300 to
$400.  Thinking that should be my new transport.  No keyboard or monitor.  Just
run it as quiet as possible.  Well, except for that damn switching power
supply.  Have to give that its own circuit.  Anyway, turns out I can do some
simple control via the USB audio connection.  Like play/pause, skip, stop.  So
a small box, stick in your CD.  Control it via the DA-10 remote.  The mac mini
also comes with it's own remote, so maybe it's not such a deal.  

Should get the HA-10 boards tomorrow (rev 2).  Maybe by the end of the week I
can have a fully functional pre-production quality headphone amp.  Good enough
for web photos and to prove out final board changes.  Then get them production
boards on order next week.  That puts me on schedule for a Dec 01 CASTANET
launch.  

You know, all this speaker work gets me to thinking about turntables.  I know
it is an area I should probably avoid, but something inside me really wants to
build a turntable.  Gads.  It would be something simple.  My brain is leading 
me to think of an MDF brick.  Stack several layers of MDF and then machine it.
All one big plinth as I just can't get into the difficulty of setting up
multiple pieces, like a tonearm base you have to position manually.  Maybe rim
drive.  Maybe direct drive.  Dunno about ac or dc, but the box would definately
have custom HAGTECH electronics.  Guaranteed spots for more than one arm.  It
would be assymetrical to lower boundary resonances.  Even the platter MDF.  All
painted shiny.  Seems to me the right coat of paint would act as a decent vinyl
interface impedance.  You don't need felt or rubber or teflon or PVC, just the
right paint and thickness.  It would run at 78rpm.  And 16 too.  Er, I better
stop thinking about this and get back to the TEN series.  Who do I think I am,
Tim D.P.?

Oct 28
------

Tweeters came in just in time.  Now I can fire these babys up and listen using
my CYMBALs, which sell tomorrow on Agon.  But I noticed the tweeters need to be
modified to fit.  That must be the -46 version.  The bottom part of the rim is
chopped off.  Well heck, how many of us have a $2000 CNC machine sitting
around?  Turned out to be an e