Workingman 15 thermal discrepancy

[Delta362]

I have a Workingman 15 that came in with complaints of it cutting out after 10-15 minutes. I found the mains fuse was blown. Q5 (2N6474) had been previously replaced by someone with a 2N3834. However, one of the legs on that was cut short and the connection was made up by a buildup of solder. I replaced the thermal grease on the output transistors. During debugging, Q9 had a broken leg. I replaced that with the 2SA1668. I resoldered the output transistors.

Amp now produces a good clean sound. The scope showed no notch to remove when setting the bias. I must have been lucky with the new Q5 matching the hFE of 40 as the one that came out.

Sound is good and strong. However, I put a thermal gun on the output transistors and am noticing that after about 5 minutes Q10 is at least 15°F hotter (130°F) than Q4 (115°F), with the temperature delta continuing to increase. With no signal applied, I measured voltages on Q4 and Q10 as below. Is there anything that seems out of whack to cause the temperature delta?

Q4
B: 0.473
C: 60.01
E: -0.105

Q10
B: -59.41
C: -0.157
E: -59.99

Thanks!

[eschertron]

I'd expect them to be dissipating the same power based on the numbers you posted. What do the temperatures stabilize at when under constant load? What's the rated case temperature of those transistors? 130F I'd not be worried about. 130C...

Did you do anything to Q4 and Q10 during your maintenance? One may have had a better job of bonding to the heatsink than the other. That would be the biggest factor (in my understanding) involved with the dissipation.

[The Dude]

Did you adjust bias? You may just be overbiased with the new parts. You say, "The scope showed no notch to remove when setting the bias". That means it's not under biased, but it doesn't mean it's not overbiased.

[g1]

Agree with The Dude. Unless my calculations are horribly off, the outputs are idling around 1 amp each. (Assuming 0V on output and 0.1R emitter resistors).
Measure DC directly across R29 and R42.

[Delta362]

Finally back to this. I adjusted bias with an 8 ohm load to remove the crossover. Then without a signal applied, but still having the 8 ohm load, I get the following:

Q4
B: 0.279
C: 63.3
E: -0.260

Q10
B: -62.7
C: -0.250
E: -63.3

Things are running cooler now. With no load and no signal, I measure about 4mV across R29 to R42 (Q4-E to Q10-C), per the bias notes. Seems all that is working ok now.

However, my DC offset is roughly a quarter volt negative. Q5 has hFE of 70, Q3 (and Q8) has an hFE of 47, while Q9 has 126. I don't have another 2SC4382 on hand to try to match hFE with Q9, and Digikey has no stock (also noting a temporary constrained supply). I also did not pull and measure Q4 or Q10 for their hFE, nor did I change them out.

Any guidance on what could help reduce that offset, or if it's acceptable? Thanks!

[Delta362]

Any comments on the DC offset and how to potentially reduce it?

[g1]

Maybe try a new set of Q6/Q7 ?

[Delta362]

Looking at what I wrote above, I see that I somehow missed mentioning that I replaced Q6 and Q7 that were perfectly matched. Must have accidently deleted that when I was proofreading and fixing what I wrote.

[Delta362]

What voltage is considered an acceptable DC offset limit for the output?

[The Dude]

I think it's generally accepted as less than 50mV. Measure directly at speaker terminals and not to chassis ground. On some amps, chassis ground is not speaker ground.

[Delta362]

Thanks for that! I typically have really low DC on the speakers when I'm done, but this one is baffling me. The 0V speaker reference is actual chassis ground in this particular amp, but I've been measuring it at the output terminal block for consistency.

Any thoughts on other things to lower the offset since I already have matched Q6/Q7 installed and I'm unable to source any 2SC4382 to try to match my 2SA1668 stash.

If I'm unable to reduce or eliminate the DC, would a capacitor in series with the speaker fuse be an option? If so, got a recommended value?

[Helmholtz]

Are C16 and C17 good?
If leaky, either might cause a DC offset.

[Delta362]

Are C16 and C17 good?
If leaky, either might cause a DC offset.

I'll check those and let you know. Thanks!

[The Dude]

Agree with g1 on Q6&7, but if you are sure they are matched, then that's that. Are R30A&B both within reasonable tolerance? I have run into some amps that just run higher DC offset, while certainly not ideal. What is the offset with load attached? I wouldn't use an output coupling cap.

[Delta362]

I'm confident that Q6 and 7 are matched. R30A/B are both very much on at 22 ohm (not 220 as the schematic mis-shows). With no load, I'm at -0.285 and about 4mV across R29/R42. With a load, I'm not seeing any change to the DC offset.