mosfet forward bias/ Hartke HA3500

[Alex R]

I wonder if anyone can lighten my darkness. Why might the 2SD2155/2SB1429 power transistors in a Hartke HA3500 (which transistors I have replaced after I found a couple shorted) be overheating, with a forward bias of 0.5v and the bias adjuster as low as I can get it?

Further to this, I emailed Hartke's distributors about this conundrum and they wrote back that "the bias should be <10mV". I am so far from understanding this that rather than enter into a discussion I wrote back and ordered a new power amp board, a thing I hate to do, because why be a fixer of things if you're just going to bolt on prefabricated modules? (because you don't know how to fix them, I know, I know). Had a happy customer, but I feel bad about it and I want to know what the problem was.

I'd attach a schematic but the forum wants it so small that it becomes unreadable. But if you'd like it, post your email and I'll send it.

Alex

[Enzo]

Actually the forum usually has a click to enlarge feature, or you could post it on photobucket or someplace and link to it.

I have the HArtke HA3000 and the HA4000, and I also have the Hartke 3500, and the 3500MOS, but I don't know the HA3500.

What are the Q#s for the three or four positive side output transistotrs on your drawing? And how many are there, three or four?

There will generally be that .5v junction drop. That doesn't tell you how much current is flowing through the transistor. That base/emitter junction acts like a diode. A diode has about the same drop on it whether conducting 1ma or an amp. More or less anyway.

In any case, the part numbers you describe are not MOSFETs, they are plain old bipolars.

You can adjust the bias coldest, then fix the amp, then adjust it to where it wants to be. if the coldest setting is too hot, you have not fixed the amp yet, it is not a matter of adjustment.

If the outputs were shorted, then check the ballast resistors at each emitter for opens. Replace the driver transistors, yes even if they check good. There is a resistor running emitter to emitter between the plus and minus side drivers. Or you could view it as from the + side output bases to the - side output bases. Same same. is it open?

Is the bias transistor OK? Does adjusting the bias actually change the current draw? Monitor the amps entire current draw with an ammeter in the mains circuit. Do that and I will tell you the secret way to bias these amps, once fixed.

[Alex R]

Hi Enzo, and many thanks for this. The power amp has 4 x 2sd2155 as Q313 Q315 Q317 and Q317, and 4 x 2sb1429 as Q314 Q316 Q318 and Q320.

Mm, no, not mosfets are they? - bit embarrassing that esp since I managed to buy a whole new set without realising... Must be mixing it up with a similar Hartke I think I once saw that did have mosfets. Or doing something even stoopider.

I checked the emitter resistors and they were ok, and replaced the drivers too. Quite a few of the power transistors had failed so I reckoned best change the lot in case of stress to non-failed parts.

It had overheated in the first place, not shorted output I think, so clearly there's an unfixed root cause going on here.

The bias transistor is ok, and the bias does adjust both globally and in balance between the sides (it has a trimmer for each).

I will go and check the emitter-to-emitter resistors - R322 and R333 on my drawing I think? (can't find the forum click-to-enlarge button - here is the schem up on my site).

http://www.stevesamps.co.uk/Ha3500.pdf

I will also check the total current draw. It dimly lights my 60w bulb in the mains supply but I will measure it properly. Then I will get back to you and hassle the life out of you for the secret biasing method.

It would be very cool to send the replacement boards back to the supplier especially as I bought two by mistake... I am not feeling as clever as I normally feel about this whole affair.

With humble thanks,
Alex R

[Alex R]

update... I checked those two resistors and they're fine - 12K ish I recall. Rechecked the emitter resistors - all around 0.47ohms. Rechecked the driver transistors - all nice diode effects.

Measured the mains current draw (from 240v AC mains in the UK) - with no speaker connected it started at 1.3 amps and rose as the heat went up - at 1.8 amps I was just thinking enough was probably enough when the 2A fuse on my DMM blew. The amp has a fan permanently on labelled 24v and 0.09A. Seems to be running fine.

Bloody thing. I'm sure I've forgotten something stupid. Do not be afraid of suggesting all kinds of senior moment scenarios.

Alex R.

[Ptron]

I hesitate to jump in since I'm not the expert here, but I'm pretty sure the resistor Enzo was refering to is R321, the 150ohm 2W. R322 and R333 are in a current limiting protection circuit.

[Alex R]

Ah, thanks, of course it is R321. Now if only Enzo had written something nice and clear and unambiguous like this:

"There is a resistor running emitter to emitter between the plus and minus side drivers"

...then I would have been onto it straight away... I will check it when I'm back at the bench. Thanks again Ptron for pointing that out.

Alex R

[Enzo]

Jump right in Ptron, I'll split the pay with you.

Yes, I didn't want to say R321 until I knew we were all on the same drawing. The HA3500 seems to use the same power module as the 3500B.

Q311,312 are limiters. Resistors R323, 332 sample the voltage developed across the output xstr emitter resistor. If the voltage there gets high enough - meaning the current through it is high enough - it will turn on the base of Q311,312. Those transistors when on will shunt the driver bases, thus limiting the current the output stage can conduct.

If R324 had opened, THEN I might expect to see R323 and Q311 burnt up, since they would become the current path to the output buss. But if R324 is OK< then so will the little stuff there be.

An amp like this is class B. The plus side or the minus side conducts, and they hand off right at zero on the output bus. if the output xstrs are running hot, it means both sides are conducting at the same time. If both are full on at the same time, then they act as a short between + and - rails and fuses blow. But if they are both just on a little, they merely fight each other and get hot. Kinda like driving your car with the parking brake partly engaged.

Look at the outputs. Their bases are one junction drop away from the output bus. That bus SHOULD sit at zero DC. So those output bases sit at about .5. (+ and - mirror each other) That means R321 SHOULD have about 1 volt across it. Now back a stage, the drivers bases are one junction above their emitters, and the emitters are connected to the output cases, so each driver base is about 1v above the output bus. So the driver bases should be avour 2 volts apart: +1v to -1v = 2v.

Q308 is the bias xstr. It's job is to maintain that relationship. If that space (the 2 volts) between + and - driver bases grows, then each will be turned on harder, and so turn on the outputs harder. It runs hot. On the other hand of the space shrrinks, then the drivers are turned off and nobody conducts.

We could even short those bases together. Or the output bases for that matter. Then we could apply a signal right to the bases. In that case, a sine wave will do here, the signal rises from zero, but until it reaches the junction voltage, the output device will not turn on. When teh signal finally reaches the mafic half a volt, THEN the output device turns on and conducts current to the speaker. So while the signal was continuous, the amp output had that pause. And of course the same thing happens on negative.

This of course is classic crossover distortion. We slice the bottom half a volt of signal out on each polarity. So we want to allow that space between the bases.
We want just enough voltage at each base that the slightest increase from the signal will in fact increase the current through the xstr. Class B.

Let's say Q308 can adjust from a dead short to open. A dead short maximizes that cross over distortion,but the amp would run its coolest. AS we open Q308 up, as the voltage across it grows, so too does the crossover notch shrink. At the point where the crossover notch JUST disappears, the two sides are right at the edge of conduction as they should be. If we continue to increase the voltage across Q308, then the two output sides start to conduct simultaneously, and the amp runs hotter and hotter. And if Q308 were open, then both sides would have their bases puled to the nearest power rail and all the outputs would conduct hard.

The factory adjust is across the marked test points.

But here is the cheap trick. Once the thing works OK with the coldest setting - admittedly with the xover distortion, then you turn the bias up until it is right. Monitor the power draw through the mains. NO signal present. AS you turn the bias to warmer settings, the mains draw will stay steady at whatever it started at. But at one point, all of a sudden the mains draw will start to climb with your turning the adjustment. The point of climb onset is of course the point at which both output polarities start to conduct together. Set the bias exactly at the point it starts to increase, then back off just a hair. Your outputs are now properly adjusted right at the edge of conduction.

Normally this point coincides with the disappearance of the crossover notch, but due to various things, not always exactly. No one will hear a little crossover distortion, so if it has to have a little to idle cool, so be it.

[Alex R]

Thank you Enzo for your generous input of time and energy. I have read your post with much interest and learned something about how ss amps limit current flow, something I admit I was hazy about before. I'll have another look at the amp in the light of it, learn some more and get back to the forum with the results in a day or two (one or two other issues to deal with first). Still determined to send that new board back to the mfr!

[Ptron]

Don't mean to hijack the thread, but can you tell me, Enzo, what R321 does? I don't quite see why it needs to be there.

[Enzo]

Ensures a current path through the drivers. A more common layout might have separate emitter resistors for each driver, and they meet at the output bus. On this one they bypass that bus and the two resistors now become one

[Ptron]

Oh. I was thinking the drivers could drive the OTs darlington style. Maybe this is not ideal or doesn't work for some reason I'm not seeing.

[Enzo]

What about when the outputs are not turned on, the drivers still need a current path. They have to be conducting before the outputs start. Darlington xstrs often include resistors internally.

[Alex R]

...well I checked all the key components you mentioned, Enzo, and found them ok. I had replaced Q308 with an MPSA42 (twisted the leads around to match the pins) in my initial attempt to shotgun the thing... erm... could that be the problem? he asked with trepidation. Is this an unusually type-sensitive position?

Anyhow I put the ammeter on the mains supply again, turned SVR302 to its coolest and the current draw did settle. I could turn it up a bit as you suggested before the current rise began, and turn back again to settle it down again. Still idles pretty hot, though, drawing 7-800mA without the fan and the heatsink hot to the touch. Across the test points I got about 400mV. Is that where they expected there to be the 10mV they quoted to me as the bias figure?

Revealing my ignorance more recklessly than ever... could you tell me what SVR301 adjusts, and how?

With many thanks,
Alex R

[Alex R]

... oh, and steady 1.26v across R321.

[Enzo]

Well, yeah, your outputs are conducting way too hard. Yes, that is where the 10mv was supposed to be.

Did we measure R321?

Are the voltage drops across D303,304 about like all the other diodes?

I wouldn't think Q308 cared much what was there, but I stock the C3200s in this shop anyway, so I never tried anything else.

What voltage is across Q308? We want to minimize that. You could short across one of those diodes - D303 for example - and that would increase current through Q308, and that would tug the driver bases closer together.

Have you fully explored the prior stage to the drivers? Q304-307? Note Q306 wired as a diode and will show C-B short. Check R304, 314.

The other pot sets the DC offset for the amp. Power on, no load, check for DC at the output terminals. Set that pot for zero VDC.