Marantz PM-68 Keeps Frying Output Transistors on Right Channel

[alexpriftis85]

Hello everyone! I have a serious issue with my Marantz PM-68.

A friend gave it to me with severe distortion on both channels. Turned out it only had bad solder joints on the driver transistors.

I repaired the bad joints and it was working perfectly for about a week...



After that it fried the output transistors on the right channel including :

• One of the transistors : bf422 (7272 on the schematic)
• 2 x 47ohm resistors (3322, 3324 on the schematic)
• The 150ohm Resistor (3326 on schematic)
• All the Resistors on the bases of the output transistors (4 x 10ohms) // 3328, 3330, 3332, 3334 on the schematic
• One of the output brick resistor with 3 legs : 3w (2x 18ohms)

After i replaced all of that for the first time it fried the same components again after 2 minutes. Even though the voltages at the bases of each transistor where measuring fine:

• i had my +- 56v from my power supply and and all the transistors where receiving the appropriate voltages.
• I had my +- 1.6volts at the biasing circuit (transistor 2sc3419 // 7268 on the schematic) and at the 2x 100ohm resistors (3314, 3316 on the schematic).
• I had my +- 1v at the emitters of bf422, bf423

and so on..

Long story short all the voltages at the bases and emitters of the transistors up to the final transistors where measuring fine and suddenly the magic smoke appeared...

So i replaced all the transistors (look at the attached photo // schematic), electrolytics, diodes (including the zeners) and it worked fine for about 2 minutes again and then poof!! It fried the same components.

The third time i replaced the output transistors i left their bases "in the air" / not connected and was measuring again and again the voltages in the entire circuit even after it was working for 30+ minutes, once i connected the bases it blew after 2 minutes again.

For some reason the output transistors short on all 3 legs taking with them all the components driving them..

A few final notes:

- The left channel is stable and working fine so i don't think it's a power supply issue.
- I was measuring 0mv bias on the channel that's fried even though the voltages where present where they should be, the trimmer was replaced and checked but no matter how i was setting it no change occured.

• The output transistors where all pretty cool when it happened every time, the driver transistors where hot to the touch but so where the other driver transistors on the left (working side)
• The Dc Offset at the output for the 2 minutes that it was on without load // speakers connected was about 10mv, i could not get it to go lower than that after i replaced all the transistors. Also in order to get the Dc offset to get to the appropriate point i had to turn the dc offset trimmer about 90% of the way to the max, while before the first failure occured it was at about 50% of the way
• The third time it fried the outputs it blew the main 15A power supply fuse while before it did not get to that point.
• The transistor differential pair on the left side measures fine and provides the +-53volts i pulled them and checked them

So this has become a very expensive repair for me since i'm at it for about two weeks now and the same fault keeps occuring.. (3 times so far!!)

Based on the schematic can you think of something i did not?
How could the voltages be correct at every stage but still it blows up?
Is it oscillating for some unknown reason?
Could the transistors i'm buying be fake and causing the fault?

I'm really dissapointed and i ran out of ideas..

 

[nigelwright7557]

I suspect most people would first suggest powering it up with a none LED 60 watt mains lamp in series with the transformer primary to decrease max current.

I would check all components on the pcb before powering it up again.
Shorted caps, o/c resistors, faulty transistors etc etc

If the outputs are blowing it suggests both outputs are being turned on at the same time and shorting out the supply.
I would leave out the outputs if possible and get rest of it working first.

 

[alexpriftis85]

Thank you for your reply!!

- i did try it without the output transistors and all the voltages where stable but as soon as the transistors where in place the fault occured again in like 2 minutes


- I don't have a lamp but i have a power supply +-15v max 1.5 amps. Do you think i could test it like that? 15volts is nowhere near the 56 it accepts in normal operation and 1.5 amps is ten times lower. Could that work without frying everything again?

 

[Mooly]

How could the voltages be correct at every stage but still it blows up?

Voltages are not always a reliable indicator, at least not in absolute terms.

Nigel mentions using a DBT (dim bulb tester) and that has to be the first step. I would then make sure the bias preset is set to give minimum bias current (so preset on maximum resistance) and then retest. If all is well see if the bias adjusts correctly. Assuming it does then turn it back down and reset it again when on full mains. Do not overbias and check as the heatsink warms.

Remember that a base/emitter voltage difference of only a few tens of millivolts can make a massive difference to how hard a transistor conducts... transistors are in any case current driven rather than voltage driven.

 

[alexpriftis85]

I've seen the dim bulb test on videos, i will buy one and retest every component and replace again where necessary.

The truth is while all the voltages at the bases and the emitters where in spec (about 460 mv at the bases of the output transistors for example) the bias reading at the output resistors was always zero no matter where i turn the trimmer.

The only noticable difference was that without the output transistors in place when i was turning the bias trimmer the negative and positive voltages where changing in the driver stage.

At one point i had the bias at minimum and the voltages at the bases was about 200mv instead of 500mv at the 2 output 2sa1941 output transistors but was at about 700mv at the bases of the 2sc5198 transistors.

Any ideas what to look for when i get the dim bulb and replace the components?

Also could i test the board temporarily with my bench power supply?

+-15volts 1.5amps output.

 

[Mooly]

Any ideas what to look for when i get the dim bulb and replace the components?

Aim to get it working with zero bias to begin with as I outlined earlier. It will all work and play normally in that state. Make sure the DC offset is zero and that it runs essentially cold.

At one point i had the bias at minimum and the voltages at the bases was about 200mv instead of 500mv at the 2 output 2sa1941 output transistors but was at about 700mv at the bases of the 2sc5198 transistors.

You check the bias methodically. You have six base emitter junctions to overcome, the outputs, drivers and predrivers. That means the bias voltage generator needs to reach approx 0.6v * 6 which is 3.6 volts. That value is not an absolute as the actual b/e voltage will vary for each device.

 

[Extreme_Boky]

I think that due to the nature of the problem, you should invest in a variac and oscilloscope. Then, set the amplifier potentiometer to the max setting, feed the 1 kHz 500mV PP sinewave to a faulty channel, monitor the output with an oscilloscope, and then slowly increase the variac voltage from 0 setting on a variac... You should observe the clipped output signal: you should see the symmetrical clipping, no oscillations, no overshoot/undershoot. But, you'll see the unsymmetrical clipping and most likely oscillations. Stop increasing the variac voltage as soon as you see the issue on the oscilloscope!! This will save your devices from burning again. Post the waveform here and we'll have a look....

To practice, try the above with a good working channel... you'll get the idea of how to fault-find an amplifier with tricky problems, without destroying the semiconductors. The output sinewave will continue to transform from a symmetrically clipped sinewave to an unclipped sinewave, as you increase the variac voltage. At full mains voltage, reduce the tone generator's 1kHz signal amplitude (from 500mV PP) until you get a beautiful-looking output sinewave. This tone-generator level is what's required to get a full uncliped output signal (with the potentiometer at max) - with no load. You can use this (tone generator output level) as a reference for the faulty channel testing.

All of the above is with no load connected (!)


To summarise...

Once you start to see the:

• unsymmetrical clipping
• oscillations
• overshoot/undershoot

... you STOP increasing the variac voltage. Then, you can trace the signal from the input stage towards the output, with an oscilloscope probe, until you detect the issue.... and your semiconductors will be saved.

Good luck.

 

[Extreme_Boky]

I just checked the schematics. Important:

Isolate the AMP PCBs -> Feed the tone generator signal straight to connector 1251, i.e. remove the TONE section from the picture (of course, the potentiometer setting will not play any role any more during testing)

Remove the 1259 connector, i.e. do not allow the speaker relays to close

Lift the junction of resistors (desolder them from the PCB, and then lift the desoldered ends):
3339 and 3341,
and
3340 and 3342
temporarily

Use T012 / T018 as your output test point(s)

The power supply section does not use any fancy series regulators.... so you'll be okay to use the variac for testing.

 

[alexpriftis85]

Aim to get it working with zero bias to begin with as I outlined earlier. It will all work and play normally in that state. Make sure the DC offset is zero and that it runs essentially cold.

I made a Dim Bulb Tester out of spare parts, replaced the fried components and checked the whole right channel components from the input to the output one by one (schematic left to right). The entire schematic essentially!

This time the only faulty components where the output brick resistors with the 3 legs (2 x 0.18ohms 3336, 3338 on schematic) the output transistors, and the 4x 47ohm voltage input resistors (3295, 3296,3297, 3298, ) on both channels. All the other components where reading fine. I also checked for connectivity problems in the board itself just in case any of the board tracks was open and i did not find any.

I ran the amp through the dim bulb tester and even though the power supply was struggling to click the power relay at first ( the relay was clicking on and off and the dim bulb was on and off almost full brightness) until the capacitors where charged propably because of the limited current of the dim bulb and it works stable the sound is clean and it indeed runs cold.

Tested for about an hour in low volume and the bias was set to zero (Maximum resistance on the bias trimmer).

You check the bias methodically. You have six base emitter junctions to overcome, the outputs, drivers and predrivers. That means the bias voltage generator needs to reach approx 0.6v * 6 which is 3.6 volts. That value is not an absolute as the actual b/e voltage will vary for each device.

I did check again on every set of transistors up to the output transistors with no speakers connected and minimum volume.

The base - emitter voltages are indeed varying especially in the negative side compared to the positive side. With the dim bulb connected i was getting about +1.2v b-e voltage at bf422 (7274) and 0.8v b-e voltage at bf423 (7272) which are the first set of driver transistors at the output stage and the same trend continues up to the final output transistors. The negative side b-e voltages are lower than the positive side.

The difference in voltages starts from the bias driver transistor (2sc3419 // 7268 on schematic) and the two transistors before it 2sa1360 // 7266 on schematic and 2sc3423 // 7264 on the schematic:

• On the emitter of the 2sa1360 (7266 on schematic) i'm getting +1.4v relative to ground.
• On the emitter of the 2sc3419 (7268 on schematic) i'm getting -0.8v relative to ground.
• The +56 and -56 are stable and ok from the power supply on both channels and for the faulty channel the circuit leading up to the 2sa1360 (7266 on schematic) and 2sc3419 (7268 on schematic) provides stable voltages // On the dim bulb tester this value is +45v and -45v from the power supply.

Isolate the AMP PCBs -> Feed the tone generator signal straight to connector 1251, i.e. remove the TONE section from the picture (of course, the potentiometer setting will not play any role any more during testing)

Remove the 1259 connector, i.e. do not allow the speaker relays to close

Lift the junction of resistors (desolder them from the PCB, and then lift the desoldered ends):
3339 and 3341,
and
3340 and 3342
temporarily

Use T012 / T018 as your output test point(s)

Unfortunately i do not own an oscilliscope so that i can do that. All the measurements where made with the 1259 connector disconnected and to test the sound i simply reconnected the tone board.

The 3339, 3340, 3341, 3342 are all 1k resistors going back to the feedback circuit as far as i understand. What would disconnecting these resistors tell me about the circuit in its current state with the measurements i mentioned above at T012 T018? Wouldn't disconnecting the feedback circuit cause damage?

A few more more notes and questions:

- The Dim Bulb was not completely off during testing. It was barely on and increasing as i was increasing the volume. For the 1 hour testing i increased the volume up to 40%

- The dc offset started at 150mv and managed to get it down to 20mv with the dc offset trimmer while on the dim bulb tester and while the bias was set to minimum.
I can't get it lower than 20mv (Manual states 10mv max // left channel is almost zero).
Is it possible that the dim bulb tester affects this reading or is it because of the bias has not been set yet?

- Comparing the values to the left / good channel only the positive side seemed close. The negative side seems low compared to the positive and compared to the left / good channel. Is it possible that this happens because i replaced those transistors and did not gain match them? I've already checked resistors, replaced all electrolytic caps and checked diodes in the entire right channel / faulty circuit.

- To be honest i'm scared to take it off the dim bulb tester (even though it ran fine for about an hour on the dim bulb) and then adjust the bias. Since the dc-offset is not ok yet and because of the reading i mentioned earlier on the negative side.

How do you think i should procceed?

 

[Mooly]

I made a Dim Bulb Tester out of spare parts

Good

I ran the amp through the dim bulb tester and even though the power supply was struggling to click the power relay at first ( the relay was clicking on and off and the dim bulb was on and off almost full brightness) until the capacitors where charged propably because of the limited current of the dim bulb and it works stable the sound is clean and it indeed runs cold.

That sounds promising. It might be worth turning the bias on the good channel down as well whilst using the bulb tester.

The bias current adjustment itself should not alter any DC offset, however slightly low rails from using the bulb tester may well cause the offset adjustment to be out of range.

With the bias turned down on both channels the rails should be close to normal values. In that state the DC offset should be very low and more importantly stable. If the bad channel shows random fluctuations and sudden changes in offset then there could well be a problem.

So that needs checking and confirming OK or otherwise at this point.

(Also remember that base/emitter voltages can not be reliably checked by measuring from ground because as the rails vary so to will the apparent base/emitter voltage. Base/emitter voltages are a very precisely defined parameter and can only be measured across the junction, not from ground to the junctions. At this point nothing suggests an immediate problem)

 

[alexpriftis85]

The bias current adjustment itself should not alter any DC offset

The previous time it blew the transistors the bias adjustment on the faulty channel WAS affecting the DC Offset.

Unfortunately i did not correlate this behaviour with the left / good channel.

While on the dimbulb this time i will check both and see what happens to the dc offset when i adjust the bias on both channels and report back to you. I haven't touch it out of fear so the faulty channel is at zero bias and the good channel was left alone.

One more note about the DC Offset vs Bias on the faulty channel:

Again the previous time before it blew the transistors again and before i had the dimbulb i had to go back and forth between the Bias and Dc offset adjustment to get the 10mv max Dc offset i need. It had settled at 8mv i could not get any lower. The only difference now is that i discovered a partially open resistor 0.18ohms at one of the two sets of output transistors. it was going open intemintently so i replaced it (2 x 0.18ohms 3336, 3338 on schematic). But that could have blown afterwards, it's ok now.

If the bad channel shows random fluctuations and sudden changes in offset then there could well be a problem.

Thankfully on the faulty channel the Dc offset is stable with no fluctuations even though it's a bit high.

Also remember that base/emitter voltages can not be reliably checked by measuring from ground because as the rails vary so to will the apparent base/emitter voltage. Base/emitter voltages are a very precisely defined parameter and can only be measured across the junction, not from ground to the junctions. At this point nothing suggests an immediate problem

Yes i made a serious error measuring like that. What confused me is that the schematic shows all three voltages on B-C-E. I will re-measure and report back Base to emitter voltage drops and collector to emitter voltages drops on each individual transistor.

Thank you very much for your precious advice! i really appreciate it!

 

[Mooly]

While on the dimbulb this time i will check both and see what happens to the dc offset when i adjust the bias on both channels and report back to you. I haven't touch it out of fear so the faulty channel is at zero bias and the good channel was left alone.

The bulb introduces a big variable in the supply voltage meaning as you turn the bias up the rails will fall. That's exactly as it should be but it can also mean that some amps show a change in DC offset, not so much due to the bias changing but due to the supply voltage altering in response to the bias current.

Thankfully on the faulty channel the Dc offset is stable with no fluctuations even though it's a bit high.

That's a good sign. If you have replaced any of the front end transistors then you might expect to see the offset voltage shift and possibly out of range of the preset. Some amps can be pretty dependent on the correct original transistors being used although its usually possible to substitute with only minor changes.

Historically -/+100 mv offset was considered acceptable and if you do the calculations you will see even that level of offset causes only 1.25 milliwatts dissipation in an 8 ohm load.

 

[alexpriftis85]

it can also mean that some amps show a change in DC offset, not so much due to the bias changing but due to the supply voltage altering in response to the bias current.

The minute read this line i got goosebumps.. You really showed me what i was doing wrong the whole time with this one...
You made me realize a few things i did COMPLETELY WRONG the last time i fried the outputs (definately my fault).

I'm writing this reply as a reference just in case someone like me makes these mistakes in the future and to prevent this:

For starters i was trying to measure bias on a partially fried resistor (2 x 0.18ohms 3336 on schematic)

Why did i do that you might say? Because i'm stupid that's why! Look at this photo below:

So this is one of the output resistors that connects to the output emitters that had partially failed..
I don't even know how it could do that. I was measuring the correct ohm values on the top leads but at the bottom leads it was open circuit!!!

How i made things worse the last time?

I was going back and forth between the bias adjustment and Dc offset adjustment until i get the desired dc offset value of 10mv instead of 50mv.

By adjusting bias relative to the Dc Offset and voltages to the bases of each transistor relative to ground i managed to get the dc offset to about 10mv

BUT i had increased the bias too much causing the positive side rail which was higher to go lower in response to high bias current and the negative side to increase a bit and i was getting my +- 1.6volts , +- 1 volts and +- 0.5 volts and at the base of each pair up to the outputs compared to ground.

So what i essentially did is force a desired result by altering the current to a point it blew the amp. I didn't even have the time to measure the bias at the output resistor because i was too busy checking the voltages on each part of the circuit and by the time i got to meausre bias it had already blown.

To sum up my mistakes:

• I got too hang up with the base voltages and did not properly understand how bias and dc offset works (i do now)
• instead of measuring base -emitter voltage drops i was measuring base to ground
• When i was reading 0 bias at the faulty resistor at first it should have tipped me off about something going wrong and i completelty ignored it
• i still don't understand why the output transistors where not hot at all since the bias was supposedly very high.

Right now the amp is still working on the dim bulb and i have not done anything yes because of a busy week. i will check the behaviour of bias asjustment while on the dim bulb turn it back down to zero for both channels take it off the dim bulb and re-adjust.

- The manufacturer states 18mv across the points of the output resistors +-3. I will adjust at 15mv just to be on the safe side.

Thanks again for your precious advice! It was like gold to me and by realizing my mistakes i learned some valuable stuff in the proccess i fairly confident i can get it working now!!!

 

[Mooly]

sounds like you are getting there.

When adjusting the bias make the final adjustment when the amp is fully warmed. It will be normal for the bias to wander and drift until the amp reaches some kind of thermal stabilisation.

 

[alexpriftis85]

Ok i did not manage to fix it after all. I'm about to quit on this one as i don't understand what's going on.

Fire it up with the dim bulb today and and as i was trying to measure bias on the ouput resistors the relay clicked off and now the relays don't engage anymore. As if its detecting dc in the ouput.

• I checked for shorts and did not find any.
• i check for dc in the emitters of both channels and i only found 30mv on the emitters of the output transistors relative to ground and 0 on the good channel.
• Now the amp struggles even more to power on with the dim bulb it clicks the power supply relays on/off repeatedly instead of 3-4 times it did before. Which tells me that something is trying to draw excessive current and the dim bulb won't let it.

-I checked the relays themselves with 24vdc power supply and they work. And the rail controlling the relays is working.


I don't know what to do next. No transistors or resistors appear to be fried and when i t does power on with the dim bulb all the voltages are the same as before but no relay click and nondc is present in the output. I think its highly unlikely that the control board for the relays is fried.

What should i check next? Should i desolder everything again and check one by one? I really don't what to do

 

[alexpriftis85]

-Update-
I found a Leaky transistor AFTER the outputs 7288 on schematic. The emitters connect there and +55v. This was not short but it was like i was reading 2 diodes. I had not replaced that an propably went bad due to excessive abuse from previous faults.


Man this thing has got me tired.

 

[alexpriftis85]

-Final Update-
After replacing that transistor and letting it warm up i adjusted bias to 15mv on. On both channels. The trimmers had to be in different positions for each channel to achieve this and i have 30mv dc offset on the repaired channel. It sounds the same on both channels and can't tell the difference. And honestly this got me really tired and frustrated. So i'm gonna leave it as it is. With low bias and some dc offset in the repaired channel.

Thank you all for your precious advice!! Especially you Mooly!

 

[Mooly]

Well I'm pleased you got there in the end 30mv offset is not going to cause any issues at all.

 

[alexpriftis85]

Well I'm pleased you got there in the end 30mv offset is not going to cause any issues at all.

Well things got bad again....

This time it fried the output transistors on the other/good channel (the one i have not touched at all other than lowering the bias and setting dc-offset close to zero and replacing electrolytics with new ones with same voltage and value)

It was performing great for about 4 days and i was listening to music for about 2-3 hours everyday while i was doing some work at a low output volume (about 15%).

The speakers are fine and are 8ohm speakers so i don't think they are causing the issue. Before i got this amp i had another cheap one and was working with those speakers for years.

The bias was set to 15mv while the amp was somewhat warm with no load connected. Maybe i made a mistake there? The manual does not say whether i should set the bias with speakers connected or not (see image below).

The amp was also plugged in directly to the socket without a dim bulb while adjusting

I ordered yet another pair of output transistors and will replace everything that's faulty on the now faulty channel and get new multiturn sealed pots for both bias and dc offset to replace on both channels. Because i cannot think of a reason this would happen on the good channel other than the pots instantly going out of whack and drifting severly. They are really cheap pots and they are 20 years old so i figured i would install some proper ones that will last.

The fault is sudden, severe and the ouput transistors are barely warm when this happens. The same thing had happened to the other channel which i repaired 3 times already but this time it held up and nothing is fried on the repaired side.

Any thoughts? I'm attaching the manual instructions below and the entire page of the output stage if anyone has any ideas:

Here is what the manual says:

I measured bias across the output resistors as stated in the manual with no load:

The Schematic for both channels: Here

 

[Extreme_Boky]

It would help to know which capacitors you replaced, and with what types... The caps you used... are they of a very low ESR type?

Replace the small signal diodes. They can measure perfectly fine but suffer a junction partial failure that causes oscillations.

C 2263 - 2266 can cause oscillations,.. place the original caps back into the circuit.

You'll need a good oscilloscope....