Crate GX-212 power amp puzzle

[DGstrat]

Hello folks,

Long post, but I'm trying to give a good picture of what I have measured and replaced so far.

I have a crate gx-212 that suffered a failure in the power amp section. When I received it all power transistors (Q12, Q13, Q15 and Q16) were shorted in one way or another.
One .33 ohm 5W emitter resistor was literally split in half, 3 others were open circuit.

I replaced Q12, Q13, Q15, Q16, all 4 resistors and then I discovered that Q10 and D11 had failed, and replaced them as well.

Testing on the light bulb tester still showed a high current draw. Measuring the base to emitter voltage on the power transistors showed that they were all being driven on by the opamp at IC3.
I also measured a significant voltage level coming from IC5. I replaced both IC3 and IC5, but still had the issue. No amount of adjustment of the bias control (AP1) would affect the output of IC3.
At this point I have removed IC3 so that I can take measurements without the current limiting light bulb.

+/- 40V looks good. +/- 16V looks good.

My suspicion is all around D29. I am seeing 7.6V at the junction with R53, 7.0V at the junction of R56. Now a .6 drop makes sense for a diode, but I would expect the value to be
centered around 0V, not at 7. I am also seeing 14.5 volts at the junction of R53, R54 and IC5 pin 7. IC5 Pins 5 & 6 are both 0V....

Since I have already replaced IC5, I'm scratching my head on this one...

Thanks in advance for any ideas you may have.

Dave
Crate GX-120, GX-212 Schematic.pdf

[Steve Conner]

The drive op-amp shouldn't be able to turn on all the transistors at once, no matter how it swings its output. When it turns on one bank of output transistors, the other side should go off automatically.

I think you made a mistake elsewhere in the circuit (transistor pinout swapped, missing Sil-pad, NPN and PNP mixed up, resistor blown open circuit that you missed?) that has caused one bank of output devices to be jammed on, and the opamp is turning the other bank on full in an attempt to wrench the output back to 0V. Do you read any DC on the speaker output when you power it up with IC3 removed? Does the DC persist when you apply a load? (use a dummy load as DC will roast a speaker in seconds)

[DGstrat]

I agree, but from what I measured it looks like the biasing network is hosed and both halves of the opamp get turned on...

I took the measurements you requested, with IC3 removed, I see 0.024 V on the outputs with no load. With an 8 ohm dummy load I see 0V.
All 4 output transistors are idle (off, base voltage pretty close to 0) with IC3 removed.

thanks,
Dave

[Enzo]

First, are you loading the amp? DO NOT connect a load until you KNOW the amp is not making DC output. Excess draw can come from various places. If you positive and negative side output transistors all turn on at once (loss of bias) tons of current is drawn. If the amp swings over to DC, it ONLY draws excess current when loaded. DIfferent causes. Remember, if you look at all the output transitor bases and find +20v on all of them, only half are turned on by that. +20 won't turn on the PNPs in the negative side.

You have had a bunch of transistors burn out. It is common for resistors to burn open when this happens. An open resistor is JUST as likely to cause excess currents as a shorted transistor, so check them all.

You report the +/-15v are good. Are they BOTH present right at the power pins of IC5? And IC3, note it does not run on +/-15v, it runs on its own 39v from D10. Is that present at the pins?

These op amps are meant to center within their power rails, and at +/-15v, that would mean zero volts. You have +15 leaving IC5. Either it is driven there or missing a voltage. What did this do when IC5 was removed? If the circuit centers without it, then it is all working.

How is R77, 0.1 ohm down by the extra speaker jack? Open?

[DGstrat]

I have not had a load on the amp until I put an 8 ohm dummy load on to answer Steve's question. With IC3 removed there was only 24millivolts of DC on the output and that went to zero with the 8 ohm dummy load.
When IC3 was installed, I did not have any load connected, and I measured around 600 millivolts.

Yes, I have been checking for open resistors. The only ones that I found were the .33 ohm emitter resistors and they have been replaced (and connectivity verified from the emitter of each transistor to the speaker out).

R77 is measuring under 1 ohm (I assume that the accuracy of my meter is not going to show me 0.1 ohms).

IC5 has +16.1 on pin 8, -16.4 on pin 4, 0V on pins 5 and 6, yet 14.5V coming out pin 7... I have replaced it (because the previous one was also showing a high output voltage on pin 7).

IC3 (when removed) has +20.7 on pin 8, -20.5 on pin 4.

The zener D10 tests good with the diode function, I am measuring 40.5 volts across it.

When IC3 was in circuit, Q12 and Q13 had 2.0V on the base (pin 1 of IC3 was 2.5V), Q15 and Q16 had -.7 (pin 7 of IC3 was -0.9V)


Thanks,
Dave

[Enzo]

Your output centers when IC3 is out, fine, but I am unclear, did we try it with IC3 IN and IC 5 OUT? Does that center? The output of IC5 is trying to shove the whole amp that way. And if it does it at idle, I suspect the amp is unable to move the other direction. If the output of IC5 is at +15, it is trying to push the amp output there. If the amp is a little bit negative, the IC5 tries to push it positive, but if it CAN'T move the output more positive, it tries harder and harder until it runs out of room and hits the power rail. IC3 merely follows the orders from IC5, so we'd like to see it absent.


At first we had bright bulb. So with the entire power amp there, no missing ICs, does that bulb glow bright with AN without a load or only with? And either way, when the bulb is bright, is there substantial voltage on the output? I mean a good number of volts, not some number of millivolts.

With IC5 gone, IF there is some voltage on the output, then measure pin 6 where IC5 was. R65 R66 form a 100:1 voltage divider from the output bus. Is the voltage at pin 6 more or less that 1/100 of the voltage on the output? (Checking that the feedback is getting to the IC)

Knowing whether the amp will settle but IC5 is shoving on it or whether IC3 is causing mischief that IC5 is causing is I think central to this effort. That is the point of my IC3 in IC5 out test.

I know you put in a new IC5 and assume it is good, hell I assume it is good too. But we are trying to find the failure path so we need to find if IC5 is involved, even innocently.

[DGstrat]

Ah. Thanks for the insight to the interaction between the two opamps. I think the results below match your analysis.

Sorry, I realized yesterday that I did not answer the question about the output net of IC5 when the chip is absent, so I repeated that experiment.
I also did the combination of IC3 present and IC5 absent (that is the state the amp is in right now).

With both IC3 and IC5 removed, the network of R51, R53, D29, R54 and R56 behaved as expected. The voltage at pin 7 of IC5 (the junction of R53 and R54) was -0.145. the other ends of R53 and R54 were 0.184 and -0.48, respectively, across the diode drop. I also lifted one side of D29 to check it with the diode test on my meter to double check it is OK.

I then installed a new NE5532 at IC3. The readings on its pins are:

1 = 0.125
2 = 0.185
3 = 0.188
4 = -20.26
5 = -0.474
6 = -0.533
7 = -2.49
8 = 20.4



So now the negative half of the 5532 is driving Q15/Q16 to turn on. I am seeing a DC output on the speaker wire that seems to slowly climb (over the course of 15-20 seconds) to around -1.27 volts.
This is with no load connected.

The voltages at the pads of IC5 (which is removed) are:

1 = 0
2 = 0
3 = 0
4 = -16.8
5 = 0
6 = -0.012
7 = -0.159
8 = 16.5

[Steve Conner]

I take back my earlier comments. I didn't notice that the schematic had been posted. In this circuit it is indeed possible for the drive op-amps to turn both sides on at once.

What happens if you short out D29? (If IC5 is removed, connect both ends of D29 to ground.) Pins 1 and 7 of IC3 should both go to 0V and the output transistors should turn off.

Also, do you see any sign of HF oscillations?

[DGstrat]

I shorted both legs of D29 to ground and observed -1.5V on pin of IC3 and +1.2 on pin 7 of IC3. At that point I observed -0.450V on the speaker output. All input pins on IC3 are at 0V.

I put a scope probe on the output and I am seeing about a quarter of a volt of high frequency noise when I do not ground the ends of D29.

[Steve Conner]

In this state with both ends of D29 shorted to ground, does the amp run without excessive current draw? What happens if you short across D29 (instead of grounding the ends) and reintroduce IC5? Does it work properly and pass a signal?

If so I think the bias trim pot AP-1 or one of the resistors associated with it might have failed open.

[DGstrat]

There is no excessive current draw with both ends of D29 shorted to ground.
There is a significant difference in the DC level of the speaker output with and without grounding those signals (-0.5V vs -1.25V).
I'm also puzzled over the non-zero output voltages on the IC3 opamps even though both inputs are 0.

I also suspected AP1 and the other resistors, however I have measured them when IC3 was out of the circuit and saw reasonable values as I adjusted the pot. I will recheck them.

[Steve Conner]

One other thing, have you checked the values of the 1k and 6.8k 1% resistors used to set the IC3 opamp gains? Maybe one of those was overheated by the fault and has drifted high?

The non-zero output voltages are caused by AP1 and its friends applying an offset. They're just more non-zero than we would like. They should be about + and - 1.4V with D29 unshorted, but you have +/-1.4V with it shorted.

To a first approximation, the inputs of an opamp will always measure the same voltage, no matter what the output voltage is, because to a first approximation the gain of an opamp is infinite. Even a tiny difference between the inputs causes a huge change in the output voltage that drives the input difference back to zero.

When troubleshooting I use that as an indication that the opamp and its feedback network are working.

[DGstrat]

I have but I will recheck now. I also did the math and realized where the opamp output voltages are coming from when D29 is shorted to ground. doh.

[DGstrat]

The opamp resistors all appear to be in spec.

I am focusing on the power supply rails now because I suspect that the small difference between the +/- 16V could be throwing the bias network off. I am measuring +16.6 on one rail and -16.95 on the other.
I am also watching the -16 rail climb slowly after power-on, from around -16.5 (where it would be close to the positive rail) eventually to -16.95, as if a cap was charging up slowly (this ramp time appears
to match the ramp that I see on the output of IC3 pin 7. I'm also able to measure ripple on the - voltage rail, but none on the + rail.

Perhaps the original short circuits have stressed the power supply filter caps or the rectifier.

[Steve Conner]

Interesting. Try replacing the zener D26 or just checking its solder joints and PCB traces. These simple zener regulators run hot and often fail.

You might even consider replacing the zener regulators with three terminal ones like the 7815/7915, though the 40V input voltage is a little high for those parts.