Hartke HA1200 Kickback 12 themal runaway issue

[zero cool]

I am working on a Hartke HA1200 that has an odd issue. it arrived with a blown fuse and one bad output device. I replaced the output device. checked the emitter resistors and everything else and all was good. Brought the amp up on a variac and it was completely stable. so much so i didn't thing the amp was actually working until i injected a signal. and yup it works. hooked up a load, ran it up to full power all looks good. shut it down, connect it to a speaker and a bass guitar, flipped it on and played it for a bit. sounded good, clean output, lots of volume, everything is A-OK i think. then after a few minutes. POP blows the fuse and another output transistor.

This time i replace all 4 output devices, power it up. evertything is A-OK, double check Bias and it seems fairly stable at 10mv per the print on the PCB. all is good. idles for 5-10 minutes. then I run it into a load at high volume but well under clipping. heatsinks start to get warm. everything seems to be ok and suddenly the current draw on the variac climbs rapidly from 1 amp to 3-4-5-6 POP before i can even reach the switch it blows the fuse and another output device.

So now I need the schematics and 2 im curious what in the H E double hockey sticks is going on. the transistors are in good contact with the heatsink and it seems to be blowing one PNP device everytime. I need to work on it again today to see if it is the same device location that is the factor.


Ideas? suggestions? random thoughts that pop into your head?



Zc

[Jazz P Bass]

Hartke Kickback

You really should be replacing the drivers, also.
Next instance, after it is up & running, make a couple of voltage measurements. Like the base of the drivers.
I like that: HE double hockey stick.

[Enzo]

Amen, replace the drivers, and make sure the bias transistor is making good thermal contact with the heat sink if it is mounted by it.

And ALWAYS check the resistors. If outputs blow, they can damage the ballast resistors, and the drivers have resistors of their own which can open.


And what does it do without a load, running at high levels for a while?

[jrfrond]

Yes, replace the drivers as well.

Sometimes experience is the best teacher. My experience over the last 30+ years as a tech tells me to never completely trust static semiconductor checks. I also learned a lot working on military electronics in a civilian shop. The QC department would troubleshoot servo amps (similar in design to most audio power amps) and if a single power transistor shorted, the would cut the tops off of ALL of the TO-3 output devices and check them under an electron microscope. What was OFTEN found on transistors that still checked good with static checks is that their junctions often had a minute hole blown in them due to the transient spike caused by sudden current draw by the shorted transistor. These transistors would actually work fine until you started to ask them to deliver current, and would then go into thermal runaway due to the junction hole. You couldn't see this with typical current-sharing checks. What this usually translates to is wholesale replacement of semiconductors in power amps, which means outputs, drivers, bias diodes/transistors and even current-limiting devices.

[Enzo]

The purpose of static tests is they will show you the failed parts. A shorted transistor can never be masked by the circuitry around it. But the static tests NEVER guarantee the part is good.

One trick to sometimes make troubleshooting more intuitive is to make a copy of the schematic, and use a color marker to highlight any parts you find bad. The point of this is that it will often make more evident the path the failure current took. When transistors burn up, excess current flowed through them. it had to come from somewhere, and it had to go somewhere. generally that means a path through other parts. So if this reveals a line of bad parts from V+, through a burnt resistor, through a driver transistor, through another resistor, and on to the other side driver, but the second driver seems to test OK, we can infer from this path of destruction that the part most likely took some excess current itself and should be replaced. I find the color highlighting creates a roadmap that is more suggestive of the failure path than trying to keep a bunch of bad parts in your head.

[zero cool]

I did not have a chance to get back into this one last night but will try again today.

This is an interesting one. I have repaired hundreds of amplifiers and this is the first time I have seen one act like this. All of the transistors are in contact with the heatsink. I have checked and double checked this as it has a goofy mounting system and I have been careful to make sure its mounts correctly and holds the devices flat.

With no load the amp will run at any level even up to gross clipping just fine. it is definitely something that is thermal related. and at the point it happened. the heatsink was not hot! it was barely warm. so im curious now. Im wondering if maybe the silpad insulator sheet is letting the transistor(s) come in contact with the heatsink once it warms up? I might try some mica sheets under the devices and see if that changes anything.


Zc

[zero cool]

welp...It definitely is a thermal runaway problem! I emailed Hartke and they sent me 2 schematics labeled new and old. none of which exactly matches the amp i am working on! But they are close enough. Both sets of schematics show 3 pairs of output devices yet the amp I have only has 2 pairs. and it is obvious that the 3rd pair was never installed from the factory. they didn't install the emitter resistors etc either. but anyway. a review of both sets of schematics shows that the new version they added a thermistor in the bias network!! hmmm the plot thickens. then i noticed that the bias transistor is not mounted on the heatsink! in fact there is no thermal compensation at all!

I also noticed that the stock transistors are supposed to be 2SD718 and 2SB688 devices and whats installed are 2SC3856 and 2SA1492 devices! The client told me it had been repaired previously! Looking at the spec sheets for both devices they look to be close enough. the latter devices simply being higher voltage/current rated devices. but this could also being leading to the aggravated problem?

Before the schematics arrived in my inbox. I did some testing and ran them amp at idle and monitored the neg side bias as this is the side that seems to be blowing every time. I set the bias to 10mv and watched it for about 10min. and while not completely stable. I could tweak it enough to sit reasonable idle +/- .5mv

But if i injected a signal no load. and turned the volume up just enough to get 30mv across the emitter resistor. the current would start to climb. if i backed the volume back down to zero. the bias would not return to zero. it would drop down to say 20mv and slowly start falling.

Out of curiosity I connected a load and ran the amp up until i got 100mv across the emitter resistors and right away the current started climbing quickly. without changing the input level the bias across the neg side emitters climbed to over 200mv, I shut the input down again and the bias kept climbing quickly up to 300mv and i shut the power down and let it cool.

I decided to see how it would react if i moved the bias transistor to the heatsink as it is mounted on the board about 2" away from the sink. I used some small wire and mounted the transistor on the sink and tried again. this time the bias was more stable and had a tendency to climb down not up so that looked promising. tried the same power test and while slower. it still quickly climbed out of control.....So....back to the drawing board.

They only changed a few parts for the new rev. I may try and located the 500 ohm thermister per the schematic and change the few other parts to bring the board up to rev. as well as replace the output devices with what is supposed to be in there and see if i can get it to be a bit more stable....

This has been interesting.


Zc

[Steve Conner]

Here's a tip I picked up from one of Douglas Self's books:

The best place for the bias transistor is right on top of one of the power devices. It's tricky to mount there, but it gives the fastest reacting thermal compensation, because the transistor package gets hotter, and quicker, than the heatsink.

So I'd extend the wires on the bias transistor, and for a plastic packaged (TO-247 etc) power transistor, stick it about where the maker's logo and part number are printed. Glue it on, goop it with a blob of thermal compound, or whatever. Hopefully that will make enough difference to fix the problem.

What value are the emitter resistors, BTW? You can improve thermal stability by just changing them out for a higher value, but they're usually also used as sense resistors for the current limiter, so you have to take that into account.

One other thing... From the schematic, are the output transistors actually emitter followers? Occasionally you get a weird output stage with the emitters connected to the rails and the collectors driving the speaker. On this output stage, the temperatures of the output devices don't matter: the bias transistor needs to be in thermal contact with one of the drivers instead.

[zero cool]

I found the problem! it seems the 2SC3856 and 2SA1492 devices have there leads closer to the devices back plate then the stock transistors. this was causing the transistors to sit at an angle and not in full contact with the heatsink! I replaced the devices with some that had longer leads so that the leads could be bent in a sort of S shape. now the transistors sit completely flush against the heatsink and guess what. no thermal runaway! Imagine that! The previous tech failed to noticed that the devices didn't sit flush.

I also removed the Bias transistor from the board. it was no where near the heatsink and i know this amp will get played hard so i tack on some short leads and used a little silicone to hold the transistor to the heatsink and that bugger is rock solid now!