GK 1001RB got dropped, now random noise comes & goes

[nevetslab]

Our Guitar Dept brought over a GK1001RB that got dropped, with the complaint that it now produces loud hum when turned on, connected to speakers. I was able to restore the output phone jacks, which have the tiny Ground-Protrusions from their Gnd terminal to dig into the chassis, making the circuit ground to chassis ground that way. Never been a big fan of those tiny contacts getting circuit ground that way, but I did restore that. Probably should have fashioned some brass grounding washers on the connector side of the output jacks, and soldering them to the ground terminal. Other than the usual solder joint issues I find on these when I go digging, I didn’t find anything obvious. Went thru the preamp PCB as well, re-soldering any of the pot terminals that even looked dubious. I believe the amp was mounted in a rack mount road case, but had been removed before getting here.

The steady hum and line-synced noise spikes I saw initially aren’t here now since making what repairs I found and putting it back together, then re-biasing. But, what I’m now finding when I first powered it up was wildly-fluctuating random noise in the 60-100mV range, then goes quiet, being around 500uV RMS, 20kHz BW. It now comes and goes, and I have been hearing it thru the speakers, so I can't ignore it. And, it’s only in the LF Power Amp. Not coming from the Preamp (have both LF/HF amps connections unplugged), nor am I seeing it in the HF amp.

This reminds me of a recent 1001RB that had been posted with similar noise problems, which hadn’t yet been resolved, or at least I don’t recall reading it had been found. I searched briefly for it but didn't spot it. Ah, found it. https://music-electronics-forum.com/...ghlight=1001RB

The PCB Revision I have here is 206-0260-B2. My documentation covers Rev A1 thru A6 with all the schematics, parts lists & PCB layouts, and I also have just the schematic for Rev B.

I’m assuming the culprit(s) causing this random fluctuating noise is in the front end of the power amp, ahead of the two-tier bipolar Output Stage. I calculated the currents thru the front end thru voltage gain stage which has the bias circuit in it, then measured those voltages across the collector & emitter resistors at both ends of the power supply rails, calculated the currents, and it’s very similar to the published voltages.

On the front end side of R39/R17, I have +/- 100.3V, with +/- 100.6V at the output stage side. So, nominal current for the front end is 300mV/10 ohm or 30mA. It would be more, since I’m getting more current thru Q14/Q11, and I only looked briefly at the rail voltages. When I measured across R39 (10 ohm), I was reading around 410-420mV or 41-42mA. The currents thru R33, R15,R16, R24 and R31 bear that out.

Res Voltage Current Resistance
R15 2.67V 1.34mA 2.00k
R16 2.60V 1.30mA 2.00k
R24 2.11V 1.06mA 2.00k
R31 2.18V 2.18mA 1.00k
R33 1.70V 35.8mA 47.5R
R21 0.53V 2.65mA 200R
R22 2.03V 1.15mA 2.00k
R20 2.20V 2.20mA 1.00k
R26 1.47V 30.9mA 47.5R

So, the front end is running close to spec…well within tolerances.

Now, what I’m contemplating, though have never done this, is to disconnect the front end from the output stage power supplies by lifting R39 & R17, disconnecting the AC secondary wiring plug from the power supply, and powering the front end from an external regulated dual-tracking supply (a pair of Lambda LQ534 0-120V @ 0-1.7A power supplies, strapped for Master/Slave tracking use, which I had on my bench at BGW Systems decades ago. I’ve never tried this on the GK amps. On other more conventional power amps, disconnecting the driver stages from the voltage gain stages, shorting out the bias circuit with a jumper, and tying those collectors together to the feedback network let the front end work like a high voltage op amp. And even with the feedback opened, have had it stable.

If I lift up R39/R17, along with D12 & D1 to isolate the upper/lower voltage gain stages from the upper bipolar tiers of the GK two-tier output stage, while leaving Q16/Q7 connected, where just their B-E junctions would be passing current between the bias circuit. I think I’d also want to lift R10 up, it being the feedback resistor from the output stage to the inverting side of the front end.

Would this allow me to go digging for the random noise source in the front end?

1001RB-II_POWER_AMP_PN_206-0260.pdf

[dmeek]

You've run into the same problem I had in trying to isolate the source of the noise https://music-electronics-forum.com/...ad.php?t=50083.
I finally gave up on it, but I think your idea of powering only the input stage is a good one. You probably will need to leave R10 connected for DC stability.

Maybe short the collectors of Q14 and Q11 together (or short C19) and connect R10 to that as the output.

[nevetslab]

You've run into the same problem I had in trying to isolate the source of the noise https://music-electronics-forum.com/...ad.php?t=50083.
I finally gave up on it, but I think your idea of powering only the input stage is a good one. You probably will need to leave R10 connected for DC stability.

Maybe short the collectors of Q14 and Q11 together (or short C19) and connect R10 to that as the output.

That.....shorting out the bias circuit to make Q14/Q11's collectors common, and connecting R10 to it is essentially how you make the front end a high voltage op amp. That, unfortunately closes the feedback loop, so, assuming that noise source IS in this front end, it will show up everywhere when it pitches a fit. I could tempt fate by unsoldering and lifting the base leads of the driver xstrs Q16 & Q7, rather than leave them in circuit, where there would be a current path thru their B-E junctions along with the B-E junctions of the lower tier NPN/PNP xstrs. Lifting feedback resistor R10 opens the loop, now relying on the DC stability of the front end circuit to stand erect without latching someplace and going north or south in the process.

I could also selectively power portions of the front end circuits, like applying power between the neg rail where R17 connects, and power the constant-current source for Q2 (voltage between it's emitter resistors and R17 (lifted from the collector of Q3 where the neg voltage comes in). I just saw I hadn't accounted for where Mute 1 & Mute 2 off of R23 & R5 are sourced. Those have to be unmuted so the front end turns on! Found it. The +/- 15V circuits power the MUTE xstr Q20 which enables MUTE 1. The +/- 15V circuit comes from the LM317/LM337 which is powered off the lower +/- 43V supplies. I suppose I could just take and ground R23, which would be enough to turn on Q6/Q5/Q4. Don't need Q1 on for this exercise.

Overnight, I got to thinking about the two 3.6V zeners, and what good noise sources they make (many a noise generator circuit uses zener diodes as the noise source). Though where they're located is AFTER all the gain in the amp. I was also thinking about the original front end diff pair 2SC3381, which I've used over the years as a front end part in designs at BGW Systems....it's only an 80V part, so I'd have to imagine GK using that on these 1001RB amps with +/- 100V rails, they HAD to have been screening the stock parts from Toshiba for the high voltage yield, and rejecting the rest. What's in this amp on the bench is the 2SC2240, which is a 120V part.

I haven't yet stocked up on 2SA1376, 2SA1380, 2SC3478 & 2SC3502 small signal hi voltage parts. And, seeing the forest of 1/8W Metal Film resistors in the front end, only growing them to 1/4W where needed, this noise source could be one of them. I'll have to look in my parts stash that supports the Hydrophone preamp work I was doing years back to see what I have.

[dmeek]

At least if you still have noise with the output stage out of the circuit. it will be narrow the search down. You could short out Q4 and Q5 too
At some point when as much as possible has been eliminated, replacing the remaining parts will be the only way to go, starting with the transistors.
It could be anything including a small arc-over caused by contamination or something with the PCB
I tried replacing Q2A/B which were 2SC3381 with 2SC2240. That wasn't it.

Without the feedback resistor R10 the output will swing to one or the other rail so it won't pass audio. I tried injecting a voltage into Q2B's base
to zero the output but it was too unstable

I returned the unit I was working on to the store that gave it to me. Don't know if they have returned it to their customer yet or
more likely it's sitting in the junk pile. You're inspiring me to get it back and try again - against all economic sense!

[nevetslab]

At least if you still have noise with the output stage out of the circuit. it will be narrow the search down. You could short out Q4 and Q5 too
At some point when as much as possible has been eliminated, replacing the remaining parts will be the only way to go, starting with the transistors.
It could be anything including a small arc-over caused by contamination or something with the PCB
I tried replacing Q2A/B which were 2SC3381 with 2SC2240. That wasn't it.

Without the feedback resistor R10 the output will swing to one or the other rail so it won't pass audio. I tried injecting a voltage into Q2B's base
to zero the output but it was too unstable

I returned the unit I was working on to the store that gave it to me. Don't know if they have returned it to their customer yet or
more likely it's sitting in the junk pile. You're inspiring me to get it back and try again - against all economic sense!

I don't advise shorting out Q4 & Q5, since you need their Vce drops to power Q2/Q2A. The 555 timer IC generates the delay-turn-on, which then turns Q4/Q5 on enough for their large voltage drop, with Q6 & R21 providing the constant current source thru Q4/Q5 to run the front end at a fixed current.

Before I was able to dig deeper, I found out this amp and most of the other GK amps in our inventory belong to GK, so........I won't get the chance (on CenterStaging's dime) to isolate the output stage and open the loop to try and find the source of the noise. Rats. It will get sent up to GK to sort out. I was rather hoping to see if pulling the circuit apart would lead to the source of this problem. Ah, well, there is a 2000RB in the boneyard that doesn't work, so one of these days, I'll try out that technique on GK's circuits. Though that front end is entirely different....double-differential, with cascoded front end stages, more complexity. Another day for later man.

[dmeek]

I don't advise shorting out Q4 & Q5, since you need their Vce drops to power Q2/Q2A. The 555 timer IC generates the delay-turn-on, which then turns Q4/Q5 on enough for their large voltage drop, with Q6 & R21 providing the constant current source thru Q4/Q5 to run the front end at a fixed current.

Agreed

[drewl]

Considering all the weird problems I've encountered with these due to intermittent preamp PC board connections, I think that might be the problem.

[dmeek]

Some time ago I had a keyboard with an intermittent PCB. The problem was a cracked via, completely invisible but easy to find by tracing the signal.
In this case, with everything inside a feedback loop, There is no way to trace.

[nevetslab]

Considering all the weird problems I've encountered with these due to intermittent preamp PC board connections, I think that might be the problem.

In this instance, the preamp board isn't even connected to the power amp board, other than the supply voltages on that multipin cable. The noise issue is entirely on the power amp PCB. That was the first thing I had checked, finding the problem didn't change with the preamp disconnected. Now, if GK simply sends us another 1001RB, but doesn't request this one to get sent back, I'll later get a chance to further chase the problem. After all, the thrill is in the hunt.