Audio amp step-by-step repair

[Audiorepair]

Get those Fets out.

[xavier60]

Yes, it sounds like a high - DC offset is occurring on TP12,

[okw]

Get those Fets out.

Q6 and Q9 are out. Q9 D-S is bridged.

TP10 (U5, pin 7) is 246mV DC.
TP12 (U5, pin 1) is finally 6mV DC.

Output is 160mV DC, 30V AC (DMM).
Pk-pk is 88V on the scope, Cyc RMS 30.6V. Not sure how to interpret the pk-pk vs. cyc RMS in the audio context.
Touching TP12 only slightly increases pk-pk to 91V.

Put a 2K resistor over Q6 D-S (VTL5C7 datasheet says 5 kΩ @ 0.4 mA and 1.1 kΩ @ 2 mA). This drastically lowered the output AC (30V->1.7V). However output DC is up (160mV->559mV).

New readings:
TP10 (U5, pin 7) is 270mV DC, 0V AC.
TP12 (U5, pin 1) is 2mV DC, 48mV AC.

Any clues?

[Audiorepair]

The Optocoupler seems to be a peak limiter, and will only operate when the output is near +/- 50v.

At all other times it is switched off, so the resistor you should substitute is its off state resistance, maybe for all intents and purposes open circuit.


The scope photos do seem to show some DC offset, but presumably this is without a load, so may be misleading.

Run the tests again with some sort of dummy load, (Don't need full power, just check the dc offset).
it may just be that the only problem is the Fets.


I would first make sure, using a meter on ohms mode with the amp off, that the 16v power supply ground, the 50v power supply ground, the ground on the op-amp, and the ground on the speaker output are all connected together by zero ohms.

Also, the speaker ground return runs through the headphone jack, which is a switched type, so that inserting a jack plug cuts off the speaker.
This switched jack is notoriously bad at failing, so don't put your dummy load on the speaker cable, put it directly on the power amp, and replace the headphone jack later.

[okw]

I'll try that. Measure each grounds points, and see if they differ. If not, then remove optocoupler and put a 1M instead.
I have 5pcs of this (100W 8R): https://www.aliexpress.com/item/1005002749358940.html
I would think one suffice as load?
I always measure output on J23. Is this the point you mean for output?

By the way, I didn't see the test point conditions note in the schematics:
1kHz sine 170mV, vol+level@10, tones@5.
I used 440Hz 721mV, all pots at unity (@5).
It probably won't make much difference, since I adjusted the sine amplitude until the first test point showed the spec 700mVrms. Most TP after this was within spec.

[Audiorepair]

Yes, J23 is your load test point, thus eliminating anything past this from the equation.

One 8 ohm resistor is perfect to test, you don't need to run it full power or for long to see whether it still has DC offset or not.



The Otpo and the fet are in parallel with R23, 47k.
Mute/gain reduction is achieved by reducing this 47k considerably, so I think you can just take the opto out, 1meg isn't going to do anything in parallel with 47k.



I wouldn't worry too much about precise frequencies, levels, etc, you are just finding a fault at the moment, and service manuals do sometimes have incorrect data for the particular revision you have on your bench.

[xavier60]

When dealing with offset problems, it's useful to note the polarity of the voltage. It can indicate whether a stage is causing a problem or is just responding to a problem elsewhere.
Sometimes the act of taking of a DMM reading can temporarily clear a fault. In cases like this, a DSO with a X10 probe should be used.
I sometimes tie a 100K resistor to the end of my DMM probe to reduce circuit disturbances. Not the safest idea.

[okw]

Measured all grounds to be equal. Except opamps (pin 4), but they are connected to -16V if I understand the power table correctly.
Removed opto and the 2k resistor. Q6 and Q9 is still removed, Q9 D-S jumper'ed.
Hooked the 100W 8R on output. Still 500mV DC.

[xavier60]

If the power amplifier stage is working properly, signals including DC offset at TP12 will appear multiplied by 33 at TP11.
Some offset at TP10 could possibly be normal.
Also check for distortion when the output is loaded.

Edit: added "at TP12"

[okw]

With load, but without signal (forgot to connect the cable):
TP12: 0.1mV
TP10: -236mV
TP11: -492mV

With load, with signal (170mVrms 1kHz, all pots@5):
TP12: 0.1mV
TP10:  -294mV
TP11: -515mV
Output: -520mV.

(all DC).

I attached images with the 8R load. load1 are three different zoom (not sure if that helps). load2 is with touching TP12 with my DMM set at VDC (causes offset of 0.5V -probably normal).

I'm clueless at this point.
Could it be caused by some capacitors not filtering the DC? I have not replaced any caps yet.

[xavier60]

Although not a massive amount, there should not be that much offset at the output. "TP12: 0.1mV" rules out an input problem.
Measure pins 5 and 6 on the opamp for now. Use a resistor attached to the DMM probe so as not to disturb it too much.
Im beginning to suspect board leakage. Has there been any yellow/brown glue on it? Any signs of extra flux having been used which could be the wrong type?

[okw]

You mean 100k in series with the positive DMM probe?
U5:
pin 1: 19mV
pin 5: 4.45mV
pin 6: 1.5mV
(all DC).

There is this transparent/light milky colored silicone (acoustic vibration) around some caps and the 10W resistors, but I measure absolutely no ohms in them (Fluke 183). I tried removing some, but it is extremely elastic and hard to remove.

[okw]

I tried with a dummy jack in the headphones to break the connections:
U5, pin 1, 0.0mVDC
Output still 520mVDC
JW51: 80mVDC
JW52: 134mVDC (which
(In my early measurements before removing Q6/Q9/opto, etc, TP6 was 158mV and TP7 127mV.)

I also measured resistance between gnd and headphones L+R (with amp OFF):
TP6: 45Mohm, very slowly rising.
TP7: 3Mohm and slowly rising.

Does this mean something? JW52 is pulled "closer" to ground which results in more DC on this side?

[xavier60]

You mean 100k in series with the positive DMM probe?
U5:
pin 1: 19mV
pin 5: 4.45mV
pin 6: 1.5mV
(all DC).

There is this transparent/light milky colored silicone (acoustic vibration) around some caps and the 10W resistors, but I measure absolutely no ohms in them (Fluke 183). I tried removing some, but it is extremely elastic and hard to remove.

Yes put a 100K in series with the DMM probe if needed. You can monitor the output with your DSO while taking voltage measurement with your DMM. No need to use the resistor if there is little disturbance to the output.
I think you are chasing a moving target. In this post you have Pin 1 which is TP 12 at 19mV which would cause the output to be 19 x 33 = -627mV. The opamp's 4.5mV input offset would alter it a bit.
That adhesive sounds safe.
You need to monitor the DC offset at TP 12 and TP 11 at the same time.

[okw]

Wow, measuring TP11 while touching TP12 makes TP11 drop under the magical ~10mV.
TP12: -0.1mVDC
TP11: 8.3mVDC
What does this tell you? Is it the internal DMM resistance that pulls TP12 down? What bad component does this lead you to?

PS. no DSO reaction when touching only TP11. When touching TP12 it offsets about 500mV up.

[xavier60]

Wow, measuring TP11 while touching TP12 makes TP11 drop under the magical ~10mV.
TP12: -0.1mVDC
TP11: 8.3mVDC
What does this tell you? Is it the internal DMM resistance that pulls TP12 down? What bad component does this lead you to?

PS. no DSO reaction when touching only TP11. When touching TP12 it offsets about 500mV up.

I can't make any sense of that. TP12 is a low impedance point. It shouldn't matter what it's touched with. What is the voltage on TP12 measured through a 100K resistor?

I haven't had much to do with JFETs, Ill leave that to others who know more.

[okw]

Through 100k: 19mVDC and 165mVAC.
Directly:  0.1mVDC and 159mVAC.

[xavier60]

Through 100k: 19mVDC and 165mVAC.
Directly:  0.1mVDC and 159mVAC.

Weird things like that sometimes are caused by something oscillating like U5A. So it could just come good when the circuitry is restored back to original with new JFETs.
Just try one more thing, bridge pin 1 directly to pin 2. and check what effect it has on the output offset.

[okw]

U5 pin1 to 2?

As for Q6/Q9. I can read the datasheet of J112, but I don't know which specs the transistors really need for this amp (many specs of J112 are over-spec-ed for the application, as for all components in all projects, just a few properties is important for each project).

I have these. Would any of these replace Q6 and Q9? (short term and/or long term)?
2N5088
2N3904
2N5457
2N2222
2N3053
BF245B
C548C (old, not sure of condition)
C548 (old, not sure of condition)
F485 (old, not sure of condition)
And for later, what is a better replacement for the J112 (to keep in mind when I order from TME/Mouser/etc./or even cheap from Aliexpress)?

[xavier60]

U5 pin1 to 2?

As for Q6/Q9. I can read the datasheet of J112, but I don't know which specs the transistors really need for this amp (many specs of J112 are over-spec-ed for the application, as for all components in all projects, just a few properties is important for each project).

I have these. Would any of these replace Q6 and Q9? (short term and/or long term)?
2N5088
2N3904
2N5457
2N2222
2N3053
BF245B
C548C (old, not sure of condition)
C548 (old, not sure of condition)
F485 (old, not sure of condition)
And for later, what is a better replacement for the J112 (to keep in mind when I order from TME/Mouser/etc./or even cheap from Aliexpress)?

Yes, pins 1 and 2 of U5. Before doing this, try to observe the possible oscillation. The trick is to load TP12 as little as possible. Set your DSO probe to X10. If you see nothing, try a small capacitor, a few pF in series.
All of those parts appear to be BJTs, ordinary transistors.

[okw]

U5 pin1 to 2?

As for Q6/Q9. I can read the datasheet of J112, but I don't know which specs the transistors really need for this amp (many specs of J112 are over-spec-ed for the application, as for all components in all projects, just a few properties is important for each project).

I have these. Would any of these replace Q6 and Q9? (short term and/or long term)?
2N5088
2N3904
2N5457
2N2222
2N3053
BF245B
C548C (old, not sure of condition)
C548 (old, not sure of condition)
F485 (old, not sure of condition)
And for later, what is a better replacement for the J112 (to keep in mind when I order from TME/Mouser/etc./or even cheap from Aliexpress)?

Yes, pins 1 and 2 of U5. Before doing this, try to observe the possible oscillation. The trick is to load TP12 as little as possible. Set your DSO probe to X10. If you see nothing, try a small capacitor, a few pF in series.
All of those parts appear to be BJTs, ordinary transistors.

So I want to 10x probe pin 1 and watch for oscillations, then bridge 1+2, while continuing to look for oscillations? Or do you mean oscillations on the output (j23)?


What about BF245B? It seems it is a "JFET VHF/UHF Amplifier".
Some discussion have been made here:
https://www.diyaudio.com/forums/parts/183603-bf245b-bf245c-jfet-suitable-audio.html

I'll get the proper components later, just for now.

[Audiorepair]

The Fet's aren't really part of the amplifier. 
They just cut  the signal in the case of a fault (Q9), or mute the amp on power up, using a slowly ramped version of-16v, so it doesn't thump at turn on (Q6), so you can just leave them out of the circuit until you have sorted the DC problem.


At this stage, I would just replace Q1, Q3, and Q4.  That's 3 transistors I know I have in stock, it would take 5 minutes, they don't cost much, and if it still has DC, then I know it is not down to these transistors.
That's a lot of knowledge gained for 5 minutes work and a couple of dollars of parts, as against hours  of pondering and measuring.

Sometimes knowing what the problem ISN'T is extremely valuable, as you can then use this fact to test the rest of the circuit.

[okw]

I'll do that. I have MJE340 and MJE350G. But I don't have MPS-A06. Which in the list above, would you recommend for a replacement?

[Audiorepair]

2n3904.   

It's pin for pin compatible, lower voltage than the MPS, but that shouldn't matter for testing the DC offset.

[Audiorepair]

Thing is, you can have a transistor like these that test perfectly OK, but in circuit they don't work.

This is when you have to resort to just changing them, as you will never find the duff one otherwise.