Need advice on an MK4800 Class AB power amp

[Crystaleyes]

Hi people.
I have this Mark Audio MK4800 amp where the Left channel output gets scorchingly hot and just stays that way, even when the input signal is removed. It's almost like a semiconductor breakdown has occurred.

Originally I thought that it might be only incorrect biasing, and in fact the base voltage to the output transistors (both NPN and PNP) is just short of double the cool running, Right channel, however there are no trimpots to make any bias adjustments and looking at the schematic I'm not sure where a modification could be made to resolve any imbalance?

Amplifiers are not my forté, but when they do (rarely) come my way, I normally find any damaged components and get the unit working again. In fact this unit 'is' working, only with seriously imbalanced outputs.

I've attached the full set of schematics. The relevant channel page is 4: PWR-A.

The overheating starts at T 14A (2SA1837) and T 15A (2SC4793) then affects every other transistor from there up to T 21A. These two transistors are perhaps the root of the problem as the Emitters on the good channel are balanced at around 0.4v whereas (before overheating) the Emitters on the Left channel are around 0.55v and 0.3v. During the overheating they are both stuck between 1.2 and 1.5 volts.

No other components are getting hot, not really even warm. The Left channel won't heat up until it receives a signal past a certain level. It can be used at a low volume for seemingly hours without overheating.

Whilst I swap out the 2SA1837 and 2SC4793 from the good channel...

...Any amplifier techs out there who can help out?

[CapLeaker]

I said I’m my post number 3 that this is not an output transistor problem. Since you’re plan off attack changed by taking things off the board, we’ll keep doing that. That is what I do as well for certain reasons. Do it by stage’s for example all output transistors, all bias transistors, etc. Also strategic placed resistors to isolate a circuit etc. Could be anything even from a short in a component or bad component to an open trace.

[mon2]

Hi. Not an expert on this but we have fixed a number of similar amps. Recall something like this schematic in the past 6 months. The fault was with the end high / low side transistors.

Remove all power. Meter in resistance mode. Measure the resistance from the collector (high side transistor bank) to the coil that is driving the speaker output. What is the resistance?

Repeat the same test but this time from the collector of the south (low) side transistor bank and the speaker again. Anywhere you have a low resistance, that bank is a suspect and at least one of the transistor is defective. Flux and remove the transistor(s) till the resistance is higher. Respectively, you can test the removed transistor by using diode mode of the meter and watching one of the many YT videos on this topic. If I recall correctly, we had to replace one of the transistors to fix the amp. Do not power up with a missing transistor but use this method to locate the bad boy(s).

Curious on your progress so continue to post here.

[CapLeaker]

I think there is nothing wrong with the output transistors. The problem is in the biasing circuit as the transistors are oscillating and getting hot a Eff. Bummer is there is no trimmer pot. Since you got one working channel use it as comparison.

[Crystaleyes]

Thanks for the suggestions fellas.

mon2

I've not encountered that Collector to coil test before but shall definitely remember and use it.

As for the readings, there's some strangeness going on...

The 2SC5198 NPN's all consistently measure around 700kΩ.
The 2SA1941 PNP's, on both channels measure considerably lower. The first measurements (taken just before I started this reply) were around 200kΩ. Twenty minutes later it was 70k, then a few minutes later 10kΩ, on the last reading five minutes ago it is back around 70kΩ on all 2SA1941's. Both channels.

How does that work?

All of these transistors test fine in diode mode on my calibrated Fluke 289, in fact two of them were replaced yesterday.

This has now raised the question in my head as to 'why', if both channel PNP's are reading low in regards to the coil circuit, does the right hand channel hardly get hot, if at all?
I must admit that it seems that the cooler channel could have more power, but we'll cross that hurdle when we get to it.

Gonna leave the amp overnight and take more readings in the morning as it was running for hours today as I tested components and settings.


CapLeaker

Until those coil tests I was in agreement with you regarding the bias and am definitely not ruling that out.

The question is, "Where in the circuit could the output stage bias be tweaked?"

This amp is Brazilian and even in Portuguese, technical documentation for this amp is almost non-existent. It seems to me that they might be using diodes, resistors and capacitors to bias different stages.
Regardless, I don't want to bias the amp too far back down the line.



EDIT;
I'm now wondering if a previous tech has crossed a couple of wires over?

[CapLeaker]

Correct. Usually you don’t find 3 diodes in series.

[Crystaleyes]

Just did some more Collector/Coil testing and now the readings are stable.

The PNP's all measure 0.50 MΩ
The NPN's all measure 0.57 MΩ

None of those seem low so back to somehow adjusting the bias.

As I mentioned the problem does display symptoms of semi-conductor breakdown but I don't have a curve tracer, so maybe I'll just replace certain diodes and see?

[Crystaleyes]

EDIT;
Rewind.. It's not cooling down

I'm fast losing patience with this amp. The left channel just keeps heating up and I have a life to get on with.

Swapped over a couple of the diodes - no difference.
Increased the Left channel R30A and R31A resistors form 470Ω to 620Ω but those transistors are still way too hot to touch. In fact when shining a light through the pcb it can be seen that T14A and T15A have scorched the board. They have no heat sink and I don't believe it's a design fault.

Considering the owner doesn't want to pay a penny, after 15 hours over three days, I've had enough of this shitty amp.


Thanks again for the suggestions and support.

[mon2]

Had a similar day with a macbook repair that we suspect was stolen and was locked. Client stayed on the premises for hours (most of the afternoon) and then decided, never mind, he will consider to buy a new unit. Our (spineless) fee = 'take it bro with a smile - no charge'. What a waste of a day!!

My suggestion is to consider using an external powered PC speaker with a 0.1uf cap at the tip of the input cable to this speaker. The cap will be used as a probe for this defective amp. The owner will likely scrap it so you probably could try some other ideas or not.

14A and T15A have scorched the board.

That is scary. I would review the full part # on the part -> google the type of transistor they are -> review how to test the transistor out of circuit with a multimeter. Watch a few YT videos on this topic to confirm if the part is good or bad.

If good then the output of these parts is shorted and forcing these parts to be over loaded. You may even find typical schematics with the same transistors which you can use as a reference for study on websites / YT. We did for a mix of our audio equipment.

[CapLeaker]

Check out T10A and R26A. If you change that resistor it should set the bias. Anything off spec, well… bias is going to be off as well.

[Crystaleyes]

So had another look. And testing every transistor, diode and resistor in circuit revealed nothing obviously out of the ordinary.

mon2

Those transistors are a BD 139 and BD 140

Can you give more details on that active speaker and capacitor trick? As in are you saying connect one side of the speaker to the amp output and then use the cap to probe through the circuit?

As for the owner, there is no way he will scrap it as it still plays even if incorrectly.

CapLeaker

Both those test fine. The reisistor is 100.02 Ω.

I'll change the value tomorrow and see how it goes.

[CapLeaker]

maybe the problem is further left on the schematic?

[Crystaleyes]

Would make sense.

I'm thinking of swapping from the good channel, all the remaining semi-conductor devices until something presents itself.

Failing that, I'll call it a day. It's not been a complete waste of time as I have learnt more than I knew before..

[CapLeaker]

check out the transistor pair T1A and T2A you need to take them out

[Crystaleyes]

Ok, right...

had a break due to lack of time, however..,

CapLeaker

removed and tested T1A and T2A as well as T4A and T5A but all is in order.

mon2

Went back to your original suggestion and between Collector and coil on all the NPN's it was reading between 5 -7 kΩ, (whereas the PNP's are still above 600kΩ). So, one by one removed each of the 2SC5198's and even the 1N5406's but that made no difference.
The reading actually rises if I keep the probes connected (don't tell me it is a capacitor problem...?) anyway it is clearly not an output transistor problem.


The headfuck is that the readings keep changing

[CapLeaker]

I said I’m my post number 3 that this is not an output transistor problem. Since you’re plan off attack changed by taking things off the board, we’ll keep doing that. That is what I do as well for certain reasons. Do it by stage’s for example all output transistors, all bias transistors, etc. Also strategic placed resistors to isolate a circuit etc. Could be anything even from a short in a component or bad component to an open trace.

[Crystaleyes]

Don't want to speak too soon, however after resoldering the components mentioned above as well as reflowing various resistors and transistors, I ran the amp for well over an hour today and lo and behold, it no longer overheats!
Ok, I don't have 800W speakers nor a preamp but it certainly seems that I can no longer reproduce that problem. The amp was run at full volume, and/or ran with the input signal overdriven and nothing could get the overheating problem to reappear.
The amp has a few years on it and clearly had previous repairs, so perhaps there was a dry joint hidden in there somewhere? The traces side of the pcb is not easy to access but it's a good lesson for me to not be lazy and to just remove each and every suspect component.

NOW, the amp will go back to it's owner


Sincere thanks for the patient support