Replacement power transistors for Yamaha G100-212

[Jazz P Bass]

The mica MUST be there or the backplane of the transistor will short out to the chassis.

[Jimmy74]

I did try the linear option, but it won't let me go to page 2, not even using chrome.

the mica washers are new and correct, I only needed to clip the sides a bit to make them fit in the space where the transistors go....it was most likely the thermal paste...as I said it was so watery it leaked thought the holes and onto the other side of the heatsink.

[DrGonz78]

Pardon me if this has been asked already... What type of thermal paste are you using? It would be very unfortunate if you applied a type of thermal paste that is more like a conductive grease. Sounds like the thermal paste is in question in your head already so let's identify exactly what type you are using.

[Jimmy74]

I used a high conductivety,low bleed, high temp silicone fluid paste...the white stuff that you usually find on transistors...only that this time I went for a cheaper brand

[R.G.]

What's confusing the issue is the difference between thermal conductivity and electrical conductivity. You need a paste that's high THERMAL conductivity, low ELECTRICAL conductivity - which is probably what you have, from your description, but your post makes that unclear.

I'm using a different browser now (chrome as opposed to friefox) it's a bit better but I'm still missing some posts.

I use only firefox, and still don't miss posts. I think the browser may not be your issue there.

I still think it was the crappy thermal paste, which looked more like white water, that shorted the transistors

This is very unlikely, but there is a simple way to find out - smear a glob of the thermal paste onto a bit of cardboard, and use your meter to measure the resistance of first the un-smeared cardboard, then the glob of white paste. You'll need to make sure that your meter prongs stick into the glob but do not touch. Most heat sink pastes are very carefully made non-ELECTRICALLY conductive, for obvious reasons. But here's a question - you say it was like "white water". Heat sink goo is more like toothpaste in consistency, not water. Heat sink goo is usually thick enough not to flow out into a flat puddle. It tends to be stiff enough to sit in a glob. Early heat sink goos (circa 1960s, maybe) were just clear silicone oil or grease, and were clear. Today's stuff is a carrier that is loaded with finely powdered mineral stuff, like zinc oxide or ceramic particles.

because the first time I turned it on there was no noise or smoke, just a loud thump when I turned it off...then when I turned it on again that's when I heard the noise and saw the smoke.

I believe what happened is that the first time you turned it on, it worked normal-ish. When you turned it off, the charged-up 2000uF cap in series with the output discharged through the speaker and made the thump. Single power supply amplifiers were notorious for turn on and turn off thumps for this reason, even when they worked right. That's one reason that bipolar power supplies are about all that's used today.

It is possible that the turn off transient that first time, or the turn on transient the second time is when the output died. Dying is more likely at turn on or turn off because things are changing so fast there is extra stress on the parts - like light bulbs flashing and dying when you turn the switch on.

Do you think that all the transistors are shorted or just one of the new ones? I did test them out of the circuit

.
There is a problem inherent in output stages like this. They often have a chain of destruction failure mode. When one part dies, the part before (and after, if there is an after in that circuit) is put under huge stresses by the failure, then the next part fails. This is one reason that the commonest advice for replacing output transistors is to never replace just one of a pair, replace them both with new ones even if only one shows to be blown. It's about a 50-50 chance that the death of one overstressed the other and it may fail soon too.

Back when people got $3 an hour for skilled labor and power transistors cost $8 each, it made sense to spend an hour or two finding just the one failed part. Today, transistors are down to $5 each in much "cheaper" money, but skilled labor is $50 an hour. It's far cheaper to replace them all than to spend hours ohming things out. As an example, Mouser electronics lists the MJ15015G for $5.82 in ones, $4.94 in tens. The MJE15030G and MJE15031G would be a great replacement for the TO-66 parts, and they're $1.50 each at Mouser. So for $15 plus shipping, you can stop worrying about exactly which transistor is dead or might be dead based on one output dying, and also stop worrying that a "good" remaining device may die soon from the overstress.

And which one of the TO66's would more likely to have burnt up?

It depends on exactly what happened in the last hundred microseconds of the output transistor's life. I can make a case for any one of them either dying from the output dying, or being the cause for the output dying. My approach to the repair would be to sidestep that question except for reasons of intellectual curiosity.

If it was me doing this repair, I would spend my time carefully removing and cleaning up the board, testing each R/C/diode on the board to be sure they're good while the power transistors ship. I would also re-melt/re-flow all the solder joints just to be sure, and ohm out the copper traces.

As an aside, one place that all complementary Class AB semiconductor setups are sensitive is the bias diode(s). If that diode opens, something dies, and FAST because this turns on both outputs simultaneously, and they both try to take down the power supply. The power supply wins that one. So test the bias diode to be sure it's not either shorted (which will cause crossover distortion) or open, which will kill output devices and maybe drivers. The bias diode being shorted to the heat sink can kill either output transistor, as can a broken wire to/from the bias diode.

[Jazz P Bass]

You need to keep in mind exactly what the thermal paste does.

It is used to simply fill in any voids or imperfections in the heatsink/ mica interface.

You do not need globs of the stuff.
(I have seen amps slathered in the paste!)

I light smearing is all that you need.

[R.G.]

JPB is absolutely correct. Too much is bad, and in fact makes heat transfer worse. The idea is that neither the transistor nor the mica washer is perfectly smooth, so that there are microscopic hills and valleys in the surface. Only the "peaks" make contact. The valleys are filled with stagnant air, which is a good heat insulator. Filling the valleys - but no more! - replaces the insulating air pockets with goo that conducts heat better than air, but worse than metal to metal contact or metal to mica contact.

I used to use a brush, but now I use a single edged razor blade to smooth on a layer that's barely thick enough to cover the surface. If you do this, you have to be careful that the razor edge does not cut the metal or mica and turn up a burr or metal whisker. Actually, there are plastic single edged razor blades for non-destructive/marring scraping and smoothing, but that's another story.

You want the minimum that will cover the surface. If any significant amount of heat sink goo squeezes out when you tighten down the mounting screws, you had too much.

[Jimmy74]

Well well lots of information to last a life time there....about the paste...yes I have used proper thermal paste before and I know that too much is bad... but this stuff was like cat diarrhea...I could see the more "solid" stripe in the middle of all this white water..and trying to squeeze out more to get more of the solid stuff out but it was just a big sloppy mess...I tried to clean it down with a tissue but it's obvious that it didn't work very well. I should have taken a photo.
As for the diode...do you mean the new one I just mounted that is screwed to the heatsink? I did actually replace all the electrolytics, the burnt resistors, the drivers and the power transistors. I was asking if there was a way to find out which driver would have gone south because I have a spare 2SA483.
As for the replacement substitutes for the drivers, which would substitute which?

[R.G.]

Well well lots of information to last a life time there....about the paste...yes I have used proper thermal paste before and I know that too much is bad... but this stuff was like cat diarrhea...I could see the more "solid" stripe in the middle of all this white water..and trying to squeeze out more to get more of the solid stuff out but it was just a big sloppy mess...I tried to clean it down with a tissue but it's obvious that it didn't work very well. I should have taken a photo.

I understand. It sounds like the solids settled out and the stuff needed mixing. You're right - poor quality thermal paste. I saw thermal goo do this in the 1970s, but modern stuff is much better.

As for the diode...do you mean the new one I just mounted that is screwed to the heatsink?

Yes, that diode. Check to be sure that the path for current in that diode is not open. If anything interrupts the path to/from that diode, or if either diode lead is shorted to chassis, it will burn out one or the other output devices. In fact, I've had good luck with shorting the whole bias network between the bases of the complementary output drivers and outputs when first bringing up an amp with a burned output section. You get ugly crossover distortion with the short in place, but you make sure that the bias network is not killing your new transistors. After the transistors are proven to work, remove the short and get the bias right.

I did actually replace all the electrolytics, the burnt resistors, the drivers and the power transistors. I was asking if there was a way to find out which driver would have gone south because I have a spare 2SA483.

Now I understand better what you were getting at. 2SA devices are PNPs, so the only one it could replace would the one 2SA483. 2SC and 2SD devices are NPNs. So the question becomes whether to replace only one 2SA483 if it happens that the 483 is the only bad driver.

I would not do that. I personally would replace both the PNP and NPN drivers with new, really complementary devices. But you certainly could test just the driver board as I outlined above. Removing the output load and also removing the output transistors leaves just the complementary NPN/PNP drivers, which are capable of amplifying well enough to test them out, and probably in a non-damaging way since they have 150R resistors in series with them. Removing the output transistors and the output load lets you power the driver board up and measure transistor pin voltages in a stable state. Those voltages will tell you if one or more of them is open/shorted, or working correctly.

As for the replacement substitutes for the drivers, which would substitute which?

The MJE15030 and MJE15031 are complementary NPN and PNPs. The 030 is NPN and replaces the NPN 2SC. The 031 is PNP and replaces the 2SA483.

[Jimmy74]

well after a long long time...I'm back on this one...got the current limiter going and fired it up....did not attach the speakers...using 100watt light bulb....closed the curtains and turned it all on.... amp powers on without any burning or smoking...light bulb lights up for a second and then dims down to almost no light at all...so what's the next step? attaching the speakers?

[52 Bill]

Yes, attach the speaker and see if the bulb stays dim. Also see if the amp passes signal and how it sounds.

If all that goes well, then try without the limiter.

[Jimmy74]

well been playing through it with the limiter for about an hour, the amp reacts properly I'm guessing, no strange hisses or buzzes but the only weird thing is that when I turn it off after about 5 or 6 seconds it gives out a short fart sound....what could that be?

[Enzo]

Get rid of the limiter then, the amp is trying to run on starvation voltage in the bulb.

[Jimmy74]

Ok removed the limiter...amp turns on without anything blowing up so that's good but I have noticed that the EQ is noisy specially the treble and it's not scratchy pots but as if there's a faulty resistor or cap. Also the Pre-Set Volume doesn't seem to do anything at all. The bass control is a bit mushy but the reverb and tremolo work like honey!! Lovely sounding amp I must say but I don't like the distortion at all. And that farty sound still pops up a few seconds after turning it off.

thanks