Back in August of 2014, my Ampeg SVT-3 Pro head had
developed this strange scratchy sound that didn’t prevent playing to come
through but was audible enough that it was annoying. I reached out to a local
electronics repair shop that is an Ampeg authorized service center to ask for them
to take a look at it, to avoid having to buy a replacement. For a $50 bench
fee, they told me they wanted to reflow the solder on the input jack. Another
$50 later, and I had an amplifier that was still doing the thing I took it to
them to solve. I may be misremembering the conversation some, but my recollection
is that, when I called them to let them know the work didn’t solve the problem,
they told me it’d be another $50 diagnostic fee because the repair they did
wasn’t at fault. I declined that service, partly because spending unforeseeable
quantities of money on diagnosing a piece of equipment that, at the time, was
19 years old and was still available brand new from the company seemed dumb,
but also because I knew they didn’t value my business. I don’t say that to
disparage them, so much as to say that my first-time customer status combined
with a mystery rinky dink problem that didn’t affect a moneymaking enterprise
pretty understandable doesn’t rate with the only factory-authorized service
center on the north side of DC for a lot of brands. Also, my sense of work
ethic resented being given back something that was just as broken as it was
when I took it there. Again, I get that they probably don’t have time to let
something sit powered on for a few minutes to see what it does, but it still
didn’t sit well with me.
It was around that time that I decided to build my first
desktop computer. Electronics components have been compared to LEGO, so I think
that’s where the itch started: selecting parts to work together to produce a
coherent whole has been something I’ve enjoyed my whole life as far as I know,
and by the time I had made it to 2016, I had started to lean more heavily into
diagnostic and repair tasks. This was the year of the processor thefts from
classrooms, and reading part numbers off processors, while not impressive in
any meaningful way, isn’t really all that far removed from identifying ICs and
diodes and so on. I also started to watch a youtube channel that specialized at
the time in tearing down power tools, which started to show me a world of
materials selection versus cost and was especially helpful in formulating
strategies for identifying suspect components for failure based on environment,
location, and device type. It was here that I built my first test component (an
equivalent series resistance meter for measuring capacitor performance) and
bought my first ‘good’ multimeter (a Chinese domestic market Fluke recommended
for ‘hobbyist’ low voltage DC circuits, of the types that I found myself wanting
to fix most often). Then came the Tektronix 317, a vacuum tube-driven
oscilloscope that I knew would need some refresh work. That led to buying a $10
surplus Tektronix 465 from work, and then fixing them both with a Tek 2336 that
I managed to convince myself to sell on after fixing its latches with a 3D
printed part I learned software well enough to draw up the model for myself. At
that point, I got myself a broken Ampeg PF-500 class D amplifier that was sold parts/as-is
that I managed to resurrect by tracing the circuit back from the known-blown
power transistors and identifying failed components, to include the class D
driver IC, before deciding to build my own Fender 5F1 Champ clone amplifier.
That turned into a commission for a second one, and before you knew it, I had
built up enough skills and equipment to be able to turn my attention to this project
that professionals decided not to take seriously. This was good, because in the
intervening two years, the amplifier had also started to completely lose sound
output periodically and intermittently, which seemed to be, in my opinion, “bad.”
I started with the obvious-to-me, by removing the circuit boards
from the chassis and reflowing ALL of the solder joints. I figured “hey, this
is what the professionals said would help!” It did not. I started to read over forum posts
and, of course, the first thing everyone says is “replace the tubes!” That also
did not do anything. Buying tubes is fun, though, so at least there was that. I
started digging a little further, trying various combinations of search terms,
when the first This Is A Major Issue With These Amplifiers came up: poor bias
on the power output transistors due to changes in position or something of the
biasing trim pot, a device that Ampeg’s parent company apparently decided to
cheap out on when selecting parts.
That’s when I found out the original 1995 power mosfets were
so poorly matched that one of them was barely on while one was near the top of
the adjustment range. There was no way to get them to operate together at a bias
level that was unlikely to create crossover distortion at low volume levels.
This was when I learned about a couple of things: transistor biasing, and parts
catalogs and datasheets. I ran through Mouser for output devices and ordered a
testing device that let me attempt to match them at their turn-on voltages. I
did a bunch of charting, selected the eight closest devices, and got to
installing. Now the amp biases correctly! The voltages you’re supposed to
measure to determine current flow are all within two millivolts instead of
spreading over 23. The amp just sort of generally sounded and felt better at
low volumes... and within a few weeks, I had my next cutout, which meant that
my root problem was still with me.
By this time, I had developed a bit more experience with
both high voltage tube circuits owing to my experience building the two Champs,
and with using a schematic to do diagnostics. Using my digital oscilloscope
revealed something I didn’t see with my analog scope, that the power board’s
tube plate voltage circuit had weird periodic sag issues and a lot of noise. I
went ahead and replaced the filter capacitors and the power transistor that
regulates voltage in that part of the circuit, and lo and behold, now my tube
gain knob seems to work the way it did when I started playing this amp back circa
2002. Cool, another tired transistor out and some fresh electrolytics, and I
seem to have further smoothed out the sound. And then a week after that,
another cutout.
At this point, having had a series of repair dramas with a
bass guitar as well, I am close to saying “okay, one last go at this and then
it’s on to class D. It won’t sound the same to me, but I can’t keep going like
this.” I had figured out previously that the cutouts were only happening at low
volume levels (I could shake a stage for an hour, but practice downstairs was a
problem), I knew that it was more likely to happen with the tube drive knob
below 12:00, and I knew it seemed to be more common when the amp and room were
cold, but sometimes it could come up after a stretch of time playing without
any problems. I also knew, because the amp has a line output, that the problem
was at the very least on the very end of the preboard signal chain or somewhere
on the power board.
I’m glad I gave it one more chance.
Deciding that this was going to take some serious elbow
grease to get to the bottom of, I decided I needed to devote actual,
significant sit-down time to this. I set up with a device that let me record
from the speaker output, and found out that, yep, the problem is between preamp
and power amp. Then I found a potential last stop on the pre board and decided
to move my recording device to the preamp out/power amp in loop, which, if that
were fine, would require the fault to be on the power board for certain. It was
fine. So at this point, I take the power board back out and clean the contacts
in those jacks - I was going to replace the jacks outright, but the modern
parts are a slightly different physical package - because they use their
bridging pins to route signal. Cleaning didn’t fix the issue despite THAT being
a common forum suggestion, so onward with technology. This time around, I
decided to scope it before and after every stage from the last 12AX7 tube on
the preamp. No problems at all between points, so I’m at a total loss when not
only does it keep happening, but it seems to have gotten worse. (Aside: I just
realized that I had closed my vent a little bit so that we’d have more HVAC
pressure into the upper floors. This reduction in temperature was almost
certainly the cause of the worsening symptom, not anything internal or what I
was doing.)
So the sound cuts out one more time and, out of some degree
of frustration and a greater degree of lack of options, I do what everyone who
has been working on the same problem for five years does: percussive maintenance.
And what do you know? The sound cut back in.
Oh ho ho! I have an electromechanical issue... I assumed at
this point that it’s not crusty soldering, and so it must be either a failing
component or a failing jack. I have the scope out still, so I hook up to the
output of the last tube and to the inputs of the power mosfets... still signal
in both places, so I move to the outputs of the mosfets... still signal there,
so I look at the schematic and there’s only one more component in the signal chain:
the output protection relay. With scope on mosfet output and the positive
binding post, I start playing and BLAMMO - cutout with signal before the relay
but not after. I also know now that if you have a nonconductive maintenance
implement (read: chopstick), you can tap suspect components with it to see what
happens. I know this because, some months ago, I did this song and dance with
the tubes. So tappy tap tap - sound comes right back in.
This is where understanding how electronics LEGO works comes
back in. Unlike the output mosfets and the voltage control transistor, or the
electrolytic capacitors, the original Ampeg part is totally obsolete and
unavailable anywhere. One parts site that looks like it was last updated in
2002 lists a newer T9S or something part number on the picture next to Ampeg’s
part number, which is also obsolete and unavailble... so now I have to figure
out what the right replacement part is. I pulled a datasheet for the last known
Ampeg part and go digging on Mouser and Digikey and eventually come up with a
suitable replacement, once I learned what a relay having contacts in form-C
actually meant and realized that the reason the original relays are obsolete is
that the contact materials have been changed and they no longer sell open air
relay units, instead preferring sealed devices (aside: I believe the original
relay failed because of corrosion on the moving contact and surface ablation on
the static contact, issues which perhaps the new materials will prevent and
being sealed definitely will prevent).
And so yesterday, I replaced that relay. For the first time
since probably 2013 or 2014 overall, and certainly the first time in a winter
in at least that long, I was able to sit down, play at low volume, hear no
weird distortions, develop no unwanted noise, and experience no random
intermittent signal drops. I don’t know if the relay was the problem all along,
and if I had been able to devote the time and energy to the diagnostics I did
that one long day earlier, I might have been able to do this sooner. But I have
an amp that, at 25 years old, pulls hard again and sounds better than I have
heard it in a long, long time.
All of this is to say that, given the prospect of being
asked to pay for lazy treatment twice, I’ve ended up learning a ton about
repair, diagnostics, parts cannoneering, electronic component failure modes,
and the value of smacking something with an intermittent failure. I’m happy
with the immediate outcome, assuming I make it through the next few weeks
without any more issues at all... but more importantly, I’m proud at how much
I’ve learned in that time, and what I believe I’m capable of fixing now that I
wouldn’t have even thought I could try, let alone succeed at. Which I think the
takeaway here is probably not “if someone asks you to pay $100 to not fix your
widget, buy an electronics repair lab, fix equipment for that lab, fix a
different amplifier, build two other amplifiers, build some other test
equipment, and then get mad at your widget and hit, thus revealing its secret”
necessarily... but man, what a value I feel like I got from that trip to the
shop.
