Adcom GFA-535II problem

I just got an Adcom GFA-535II for free in good cosmetic condition.

F603 was blown so I replaced it.

I did the bias adjustment.

There's -1.79Vdc across the left channel speaker terminals.

There's also -483mVdc across the right speaker terminals.

Is there a part known to fail that would cause this?
Good morning,

There's a large number of possible culprits. I've attached a service manual found on the web.

I assume you've measured supply rails, that they're OK, and there's nothing obviously failed.

This amp features a DC servo, so there's no offset adjustment.

To narrow the search, would you report voltages at the following points, all measured with respect to ground. Please report as signed voltages with best resolution your meter provides.

IC601, all pins noted in the schematic.
Base Q601
Base Q603
Output at TP2 or where convenient.

These are all left channel points. We'll need the same in right at some point.

Be safe! Thanks.


  • gfa535_ii.pdf
    862.4 KB · Views: 93
Oh I see so the problem could be in the DC servo circuit, right or perhaps something else the circuit cannot correct for?

Only thing I don't like about the amp is it has no speaker protection relay, however there are speaker protection relay kits out there that can be connected between the amp and speakers.

I might get one of those kits later and install it in a box.

I'm also tempted to get banana plugs to put on the speaker wires since the speaker terminals can accept them.
It's unlikely to be an issue with the servo circuit. Check pin 6, the output of the servo, it should be around 3-6V if things are working right. There is an intentional imbalance in the tail currents of the LTP, and so the resulting correction by the servo is predictably positive. (It has a single-sided supply) More or less voltage at the servo output means something is out of balance and the servo is struggling to correct.

There's actually not a lot that goes wrong with these amps. About the only persistent problem I see is that Q609 overheats and develops cracked solder joints, or loses Hfe and gets leaky. You'll notice the board is toasty brown around it. Solution is to re-solder and either install a TO-92 heatsink (assuming the transistor is still good) or install a physically larger transistor like KSA1381. This miiiiiight be your DC offset issue.

Once you have it working right, I wouldn't worry about a protection circuit. These amps rarely go DC, due to the low supply voltage and generous SOA in the output stage. Nothing is under much stress in there, except for Q609 and that can be fixed.

If you want even more reliability, upgrade from two pairs of output transistors to three pairs. There are pads on the boards, meant for the GFA-545 which uses the same boards but more power. You'll have to take out the originals and install whole new sets so that they match. NJW1302/1381 is excellent, and if you order them from a legit supplier, they're likely to be from the same batch code and match plenty well.
Thanks for that. Will check Q609 first.

Q609 losing Hfe could explain why both channels have different DC offset voltages on the speaker terminals as both of the transistors would not lose the exact same amount of Hfe.

Would the same issue with Q609 also apply to the GFA-545II?

Also does the amp need to be recapped like most Carver amps made around the same time as this Adcom amp do or are the electrolytic caps still good?

Also I like how relatively simple the amp circuit looks to be. It doesn't seem to me to be overly complicated like a lot of other amps are.
Last edited:
Indeed, same issue with GFA-545! Also an extremely reliable amp.

Sometimes the power supply caps will dry up on these, especially with a lot of power-on time. They used nice quality caps though, usually Rubycons. The other caps on the board are usually fine, but there's so few of them, I just change them for peace of mind. Part numbers in my ADCOM BOMs doc.

Yeah, the design is pretty similar to the elegant simplicity of the original Nelson Pass designed GFA-5x5, but with an AC-coupled input, DC servo, triple-darlington output, and a bias compensation circuit that tracks closer thermally. (I think the poor bias tracking in the MK1 is actually how Adcom implemented it, not NP's mistake.)
  • Like
Reactions: 1 user
Part numbers in my ADCOM BOMs doc.

Thanks for that.

I may leave the caps alone for now and save up for your power supply board for the 535 II along with some of the other mods you list.

I had thought about upgrading the power supply caps to 10,000uF earlier today.

Would you recommend using banana plugs on the speaker wires?

If those two transistors are good, I will definitely get the heatsinks and install on them.
Last edited:
Here's the measurements you asked for.

I wonder if both channels have the same problem, but one isn't as bad as the other yet.

1 measure.png
Thanks. Interesting readings.

In the bad channel, the servo opamp appears to be desperately attempting to correct the amp's offset error, but it's more error that can be corrected.

In the better channel, the pin 6 voltage suggests the servo is within control range, but results are not good.

Do we have a data sheet available for the ADCOM 3A opamp? It looks like it's a bipolar opamp, or is defective. I say this in view of -140mV on pin 3 in the better channel. The design uses large 4.7M bias resistors, so your meter load resistance is affecting measurement. I suggest using the meter to measure between pin 2 and pin 3. You should see only a few mV.

I would suggest a TL081 opamp as a replacement, and install a socket if the amp doesn't already have one.

A new opamp may fix the better channel. There may still be other issues. I believe the bad channel also needs a new opamp, but I'm less optimistic that it's the only problem. But working opamps will simplify troubleshooting.

Good luck!
Last edited:
*Edited cuz I had the Bad and Better channels switched up.

Yeah, your op-amp inputs, pins 2&3, should be the withing a few mV of each other if it's able to satisfy its feedback and DC operating point. The "BAD" channel op-amp looks like it's op-amping as hard as it can, with the inverting input so far negative of the non-inverting, it's doing what you'd expect, sticking to the positive rail. The "BETTER" channel op-amp indeed looks defective, it should be going more positive than that. There's nothing really loading the output but a 1.5M resistor, so if it wants to go positive there should be nothing stopping it.

The ADCOM 3A is actually an LT1006 and there's no good reason to use anything else. The AD820 is also proven and pretty much the same thing but more expensive. Don't pay more for low-offset parts, it only make a couple mV difference at the amp output.

Also, you can isolate this issue by simply lifting one leg of that 1.5M resistor on the output. Then you can see what the amp is doing without the servo confusing things. You should see some offset at the amp's output, no idea how much, but it should be in the same ballpark on both channels, same polarity.
Last edited:
I have to retract my recommendation of the TL081, as I remembered wrong! Its common-mode includes the positive rail but not the negative. This usage relies upon input voltage at negative rail, i.e. ground. Sorry for bad advice.
  • Like
Reactions: 1 users