There are many counterfeit HUF76639 transistors around (usually these are not even logic level) and probably no originals left. I would use IRL540N as a replacement, at least you can easily get a genuine article and they are more suitable than almost anything else available at the moment. STP40NF10L also a possible option.
Cheers
Alex
P.S. - don't forget to put the bias adjustment pot to the minimum bias before replacing the MOSFETs, otherwise you may not have a chance to adjust it, they would just blow if the bias is too high.
Last edited:
I can understand wanting both channels the same, I would be doing that too. Ebay... I hear far to many horror stories regarding semiconductors so best to avoid imo. You would always be wondering.
re "I also spotted a flatulent smd capacitor....."
High value SMD ceramic caps (>=47n approx) tend to fail under power after they've cracked internally due to physical stresses. Often happens if the board is flexed, or the cap is right next to a mounting bolt that crushes the board a bit, possibly even if it's next to a hot component, it's not the heat as such, it's the thermal expansion at work. Seen hundreds fail like this on new production 1st version equipment being burnt in, it's the type of thing that gets sorted on later versions.
High value SMD ceramic caps (>=47n approx) tend to fail under power after they've cracked internally due to physical stresses. Often happens if the board is flexed, or the cap is right next to a mounting bolt that crushes the board a bit, possibly even if it's next to a hot component, it's not the heat as such, it's the thermal expansion at work. Seen hundreds fail like this on new production 1st version equipment being burnt in, it's the type of thing that gets sorted on later versions.
Last edited:
interesting.. ok I could try replacing it anyway, but with what type exactly? C20 on the 5350's schematic is 100nF bypass capacitor on the TL074's power supply. Is the type of dielectric important? I have 100nF x7r.
The basic question remains: could this capacitor be the cause of the oscillation? or let me rephrase the question, looking at the schematic which part of the circuit or component is responsible for stability? What can i do to work on tracing the oscillation's cause? I will replace the mosfets and adjust the bias but what if it oscillates again? Knowing that it oscillated twice, how will i be relaxed enjoying the music instead of sticking my eyes on the amp for smoke?
Its extremely unlikely C20 would cause any major issue even if it were open circuit. The IC is used as a DC servo to maintain a low DC offset at the speaker terminals. The type of cap and value isn't critical. If you have any doubts then just replace it. Even a small electroyitic would be perfect.
If you suspect oscillation somewhere then you need a scope to investigate. There is no one component responsible for the amps stability, its a combination of the whole design. All the caps in the pf range are there for stability reasons.
If you suspect oscillation somewhere then you need a scope to investigate. There is no one component responsible for the amps stability, its a combination of the whole design. All the caps in the pf range are there for stability reasons.
Another thing is that the sophisticated protection circuits (there is a micro controller responsible for monitoring dc offsets, currents and voltages at various points) failed to protect the amplifier. Looking from a non technical customer's point of view, I ve got a ~ 2K dollars amplifier (Destiny 2 costs €2350 in my country) with hi-tech protection systems and I have barely managed to listen music more than 20 hours and most of them where in anxiety of smoke that might appear in sudden. The microcontroller prevents me from throwing away the final stage and replacing it with two holton modules for example (in case the final stage cannot be permanently repaired), which makes the whole amplifier useless.
Last edited:
no of course I can't be sure, but I think (enlighten me please if I am wrong) that DC is easier to detect and protect from, compared to an oscillation and I assumed that the advanced protections would not allow such a simple failure. But everything is possible and I would be happier if this is a DC case if DC is easier to troubleshoot!
Yes, DC offset is easy to protect from. Oscillation, as far as a protection circuit is concerned, could appear as a valid signal and there is no guarantee that oscillation would upset the average DC offset either.
I was thinking more of a problem like thermal runaway where the bias current creeps up and up and up resulting in failure through overheating. Up to the point of failure you would not be aware anything was wrong.
I was thinking more of a problem like thermal runaway where the bias current creeps up and up and up resulting in failure through overheating. Up to the point of failure you would not be aware anything was wrong.
ok but I suppose this must happen quite fast before the heatsink raises temperature where the temperature sensor is, because there is, one on each heatsink. If what you describe could be the case, would two voltmeters permanently connected accross the drain resistor help? Will I see anything in case the problem appears again?
That would certainly show any current increase. Its normal for the current to vary quite a bit until the amp reaches thermal equiibrium... danger signs would be for example playing the amp loud so that it get hot, turning the volume back down and seeing the current either not fall back to the quiescent value or worse for it to just keep increasing.
A bulb is the most recommended safety device of all when working on amps. The downside is that it won't let you play the amp loud because it will severely limit current. For leaving the amp running under test its great.
I just soldered the STP40NF10 and the amp started, meaning that the power led became green. No voltage drop at the bias resistor again. I had the bias pot turned some turns anticlockwise before installing as Alex suggested. I tried turning the pot clockwise with no change at the voltage drop ( = 0 volts). Then the amp went into protection and since then it does't start. The power led gets green for a while but the protection red led comes on and also the AUX2 indicator led comes on red (normally each selected input's led is green). This combination of lit red leds must correspond to an error code but without the service manual its a useless effort I am afraid. I also tried it with and without load in case the microcontroller wants to see a load, with no success.
Measuring the source-drain diode using this method, I get the following results:
stock HUF76639:
off state: 525
on state: 001
chinese HUF76639
off state: 510
on state: 070
STP40NF10
off state: 530
on state: 205
STP40NF10L direct replacement my &#$#. At least with the chinese HUF76639 I could listen music, even if the bias could not be adjusted. Now what.. I can only try IRL540N and then throw it out of the window.
Measuring the source-drain diode using this method, I get the following results:
stock HUF76639:
off state: 525
on state: 001
chinese HUF76639
off state: 510
on state: 070
STP40NF10
off state: 530
on state: 205
STP40NF10L direct replacement my &#$#. At least with the chinese HUF76639 I could listen music, even if the bias could not be adjusted. Now what.. I can only try IRL540N and then throw it out of the window.
Sounds like you are having problems...
Checking FET's on a DVM isn't particularly conclusive because the high input impedance of the gate means that this terminal can literally float and assume any potential giving totally misleading results ... something the test method in your link tries to address by saying "touch all three pins" but fails to say how important it is or what is trying to be achieved.
Where did you get the new FET's from in the end ? Recognised supplier or ebay ?
What I would do now is fit a bulb tester (don't work on the amp without) and take some basic voltage readings. Lets see if anything obvious shows up.
Basic checks... and don't overlook anything here or assume anything.
1) Rails OK ? That includes the supplies to the opamp.
2) What is the DC voltage on LR1 which is the output of the amp before the relay.
Just remind me what you have actually replaced here. There are four "output" FETs per channel but the outer pair seem to be used as some kind of series pass element.
Which have you changed and were all four faulty ? The diagram shows all four to be the same type.
Could you take the two outer ones from the "good" channel and use those to replace the main outputs in the faulty channel and then look at alternatives for the outer devices. Just an idea... don't rush and do anything, and don't attempt anything without a bulb tester.
Checking FET's on a DVM isn't particularly conclusive because the high input impedance of the gate means that this terminal can literally float and assume any potential giving totally misleading results ... something the test method in your link tries to address by saying "touch all three pins" but fails to say how important it is or what is trying to be achieved.
Where did you get the new FET's from in the end ? Recognised supplier or ebay ?
What I would do now is fit a bulb tester (don't work on the amp without) and take some basic voltage readings. Lets see if anything obvious shows up.
Basic checks... and don't overlook anything here or assume anything.
1) Rails OK ? That includes the supplies to the opamp.
2) What is the DC voltage on LR1 which is the output of the amp before the relay.
Just remind me what you have actually replaced here. There are four "output" FETs per channel but the outer pair seem to be used as some kind of series pass element.
Which have you changed and were all four faulty ? The diagram shows all four to be the same type.
Could you take the two outer ones from the "good" channel and use those to replace the main outputs in the faulty channel and then look at alternatives for the outer devices. Just an idea... don't rush and do anything, and don't attempt anything without a bulb tester.
mouser.com
with the chinese mosfets rails were ok. The pre-amp opamp has its own power supply which i did not measure.
I only had less than 5mV on both channels with the chinese mosfets. Now I was trying to measure DC and during the first second I got overload indication on the multimeter and then protection kicks in, but I have to use an analog voltmeter to get a valid indication, the time is too short for the digital one to get a proper reading if there is actually DC.
outer ones, main outputs, which are which ?
I can't thank you enough Mooly
I just put back the chinese HUF76639 to see if the problem is caused by the new STP40NF10. The problem remains.
Putting an analog voltmeter at LR1 showed me 1.5V DC at the output for 1-2 seconds, then the protection kicks in.
The dc voltmeter showed just a small spike on the "good" channel when turning on the amp and immediately returns to zero until the protection kicks in (because of the dc on the "bad" channel. So I do have a DC problem now on the "bad" channel which did not exist until today.
Putting an analog voltmeter at LR1 showed me 1.5V DC at the output for 1-2 seconds, then the protection kicks in.
The dc voltmeter showed just a small spike on the "good" channel when turning on the amp and immediately returns to zero until the protection kicks in (because of the dc on the "bad" channel. So I do have a DC problem now on the "bad" channel which did not exist until today.
This is going to be difficult I suspect. The "outers" are Q17 and Q18. These appear to be used as "switches" and I'll be honest, their function is unusual and a bit obscure. I've never seen anything like this arrangement before., however
The worry with it now is that the fault has changed. You have fitted the other transistors back in place and it doesn't work... and it did before.
Does the protection circuit only operate on the speaker output relay or does it remove the supplies as well ?
If you measure the rails in the faulty state, what do they read ? Are they present or not ?
I've got some ideas for working on this but we need to establish the basics first.
Its worth checking that the print is OK around the replaced transistors... you never know, could there be a break or short that's occurred with all the work done.
The worry with it now is that the fault has changed. You have fitted the other transistors back in place and it doesn't work... and it did before.
Does the protection circuit only operate on the speaker output relay or does it remove the supplies as well ?
If you measure the rails in the faulty state, what do they read ? Are they present or not ?
I've got some ideas for working on this but we need to establish the basics first.
Its worth checking that the print is OK around the replaced transistors... you never know, could there be a break or short that's occurred with all the work done.