mosfet forward bias/ Hartke HA3500

[Alex R]

I made some mistakes in my initial attempt to shotgun this problem, which emerged when I started checking those diode voltage drops and found a few surprising ones. Bet you suspected as much, Enzo. Had to leave the amp unfinished this morning but I am documenting the process carefully and promise in the interests of free information and education to reveal all, hoping that this will encourager les autres to do likewise. Forum threads too often end with a piece of advice that leads the enquirer to a place too embarrassing to relate...

Alex

[Alex R]

Oh this amp. Well I thought I had the source of the trouble when I realised I'd read 3189 for 3198 and fitted Q311 (which I replaced with a BC639) the wrong way around (it wasn't burned out, I was lazily shotgunning what I took to be the bias control area). I like it when I can stop thinking so I did that and turned Q311 the right way around.

...But no, the thing still overheats. I would be very grateful for any further guidance anyone has faith enough to give based on the following responses to Enzo's suggestions:

Diode voltage drops:
D303 and D304 0.5v each
D305 0v
D307 19.2v (not a zener I put in by mistake, and working fine. I checked this...)
D308 0.6v

Q304,305,306 (diode wired) and 307 all show normal diode effect

R321 measures 149ohms

R304 and 314 are ok

Q308 has 2.45v between C and E

There is 400mV between the test points, where there should be 10mV

I do want to find this out and not fit the new board not just because I'm stubborn but because I am starting to get these many-watts-in-little-box bass amps in alongside the valve amps I am so familiar with ( I'm a long-term part-timer), they seem to burn out all the time, and I need to learn more about them. So I am very grateful for all help offered and very happy to take whatever readings you want...

Thanks,
Alex

[Enzo]

Note this difference: THEY don't burn out all the time, but when they DO, you see them. And the more you do it the more you will realize that troubleshooting is troubleshooting, regardless what you are servicing.

D305 - ZERO volts? That sticks right out. Look at the drawing - either D305 is shorted or something connected to it is.


D307 - where does that 19v come from? Are you measuring drops across these while power is on? As opposed to measuring their drops cold with your diode test on your meter?

Either way, how does D307 find 19v?

What vvoltage is on either test point WITH RESPECT TO GROUND? SInce D307 is connected to the output bus via some resistors, if teh output bus is riding at DC, then it could show up at the diode.

WHen you turn the bias control, what voltage range appears across Q308? WHat is the minimum voltage there? That voltage you get seems high. ANy chance TH302 there in the Q308 ckt is open of the wires broken off?

If you lift D305,306, it will eliminate the limiters Q311,312. They could be involved, but the amp will run without them for testing.

All 8 ballast resistors 0.47 ohm 5W have the same drop across them?

[Alex R]

OK I confess to observer bias about the burnouts. Speaking of bias.....

I will try all you suggest. Yes I was measuring diode voltage drops with the amp live. I don't have a diode test dmm (until Monday) but will have a look at the drops & reversed resistances anyhow.

BUT before all that - I noticed a funny thing. Here's a lousy photo (I hope):

http://music-electronics-forum.com/i...attach/jpg.gif

It shows D307 and D308, on the very expensive new board I got from Hartke and would like to send back. The two red dots I've added mark connections to one another and to the ground bus. Zero ohms between them and the ground connector. From the schematic they should be the two cathodes, but on the board it's the cathode of D307, but the ANODE of D308 going to ground at the red dots. You can see the markings on the diodes and symbols on the board underneath suggesting this connection. Doesn't that mean D308 is the wrong way around on the board?

I checked the board again. The anode of D307 is connected via R322 to the base of Q311, right enough, but the CATHODE of D308 goes to R333 and the base of Q312.

It's exactly the same situation on the old board that overheats.

Looks to me as if the schematic is right and the boards wrong, but before I try turning D308 around - what's your view?

Alex

[Enzo]

Keep hoping, the link doesn't work. All i get is the little one in your post, and without seeing the copper side, I have no idea where anything goes.

D307,308 should be wired in "opposite" each other, since one clamps to the negative rail and the other clamps to the positive. NO wait, not rail, but negative and positive excursions of the output bus.

My schematic is right, D307 anode to R322 and D308 cathode to R333. If your diodes both have the same end grounded, it is wrong. My schemo seems to agree with your boards.

If those were live voltages, then zero across D305 is OK, but that 17 volts across D307 can only come from the output bus. Are you sure there is no large DC offset there?

BIG by the way: do you still have the little input connector plugged in? It MUST be connected for the amp to work right. DC offsets are one result of it not.

[Alex R]

Ah yes - another false alarm. I've been looking at a later schematic which has them back-to-back, cathodes to ground. Unless I'm misreading it. It would be interesting to know why the difference. I'll post them both. I have the earlier one with the arrangement as on the little pic; this schem is labelled 'old compressor'.

I tried photobucket but it won't accept pdf files. I'll find one that does and post both schems.

I guess you need the input plugged in because the ground bus is split - is that to avoid loops I wonder. Anyhow more later when I get back to the bench. Delays in doing so are caused by my having to look after my wife's shop this week. It's a women's fashion boutique. Imagine my joy.

Alex

[Steve Conner]

You bet Q302 is a "type sensitive position". You can try shunting a resistor of about 10k across R320/TH302, to give the bias adjustment pot more range. Also, if the old Q302 was mounted in thermal contact with the heatsink, the new Q302 must be too. Same goes for TH302.

[Alex R]

Re diode mysteries, here are what I really do hope are links to the two schematics concerned. The first has the back-to-back arrangement for D307 and D308; the second has D308 the other way around. Just for interest and to0 make the schems available. I'll take a look at the amp and then get back in detail to Enzo and Steve - many thanks for all the help.

http://www.gigafiles.co.uk/files/4855/Ha3500.pdf
http://www.gigafiles.co.uk/files/485...Compressor.pdf

Alex

[Steve Conner]

D307 and D308 are part of the overcurrent protection system and shouldn't have anything to do with the problems you are experiencing. Anyway, it seems to me that as far as these diodes are concerned, the schematic in Ha3500.pdf is wrong, and the boards you have are correct.

[Enzo]

D307,308 won't cause this problem, but if there is 17v across one of them, that's a big clue.

How are we coming towards getting a meter with a diode test function? SImply measuring the voltage across something while powered can provide useful data, but taking junction drop readings while not powered tests the parts in other more useful ways.

[Alex R]

- bought a Fluke the first time you mentioned it - been meaning to do this for ages but have grown attached to the nice red display on my old Racal. But hey I can have both! Should have it by Monday. How's that? Will get back to this thread when I get out of the boutique...

Alex

[Alex R]

OK, well never let it be said that I don't know a quick fix when I read about one, so having a spare hour before departing for my final day of selling frocks shoes and handbags to women so my wife can have a holiday, I bridged a 7K5 resistor over TH302/R320 as Steve suggested. Now I could get the 10mV recommended across the test points. Now I could swing the bases of the power transistors between 0.49v and 0.58v, where before the lowest they would go was 0.62v. And now it runs cool and draws a steady 350mA AC or so from the mains at idle with its fan running.

So is it fixed? God I hope so.

Could my substitution of a MPSA42 for Q308 have caused this? We're only talking about 100mV or so difference I guess. Or D304/D305 having weird v drops and supplying too few mV to the base of Q308.

Just to tie up some loose ends, could you tell me if I'm reading this circuit right? The Q308 bias adjustment circuit takes just over a volt through D303/D304, runs it through R320 and a safety thermistor to allow SVR302 to adjust Q308 into and out of conduction. When it conducts, the bases of the drivers Q309 and Q310 are pulled closer together/closer to 0v, so that (along with the signal, the path of which I'm a bit hazy about & would like to understand better) they conduct less of the +/- rail voltages through to the bases of the power transistors.

In order that the drivers can be left conducting enough to amplify the signal with a clean swing, R321 cancels out most of the +/- rail voltages that the drivers conduct.

So am I getting this right? And if I could understand the signal path before the drivers better, and how the 'phase inverter' effect is achieved, I would rest happy.

Many thanks again to Enzo and Steve, and Ptron. Off to sell frocks.

Alex.

[Steve Conner]

That's frockin' brilliant mate! It's not fixed till it's passed the "Back In Black Test" though.

D303, D304, Q308 and the components around them form what's known in the trade as a "rubber diode". It just generates a small DC voltage that can be adjusted by SVR302. The temperature coefficients of the diodes and transistor decrease the voltage somewhat with increasing temperature, and stop the amp going into thermal runaway. D303 and D304 would have been added to give a different overall tempco, or maybe to sense the temperatures of the driver transistors, if they were different to the output stage. TH302 is probably there just to buy some extra insurance against runaway.

To a first approximation, the rubber diode is just a source of temperature dependent DC voltage and has no effect on the AC characteristics of the circuit. You often see it represented by a battery symbol in amplifier design textbooks, SPICE simulations and the like.

PS: If you're looking for a "phase inverter": there isn't one. Since transistors exist in NPN and PNP varieties, you just "invert" the bottom half of the output stage by making it out of PNP devices, and leave the signal uninverted. It's just an ordinary voltage-feedback power amp design, like an op-amp on steroids, same as every other PA or stereo amp since the mid 80s.

[Enzo]

Other than the legs are in the wrong order, the MPSA42 should be fine. I would not call it fixed yet. Slapping a resistor on there merely identified the base current as the problem. Is R320 open? Is TH302 open? FInd out. The last thing you want is an amp that works whilpe you politely exercise it on the bench, but when it gets real good and hot starts to run away.

YOur D304,303 voltages were OK. You never measured the junction of D305 - there was no current flowing through it, so you got zero. If the junction measured zero, then it would be shorted, by definition. It is likely fine.

R321 doesn't cancel anything, it serves as a load for those drivers - it provides a current path for them.

Transistors are all about current, tubes are all about voltage.

I take nothing from STeve's explanation of the Q308 circuit. I don't know the rubber diode, but I believe him. I visualize the circuit in a different manner thus, perhaps it might offer insight.

I see Q308 as trying to pull the two sides together - like one of those Nautilus machines in the gym, where you sit and your arms start at the sides and you have to bring your elbows together in front of you. It does this by conducting current. The more it conducts, the more my elbows are pressing together. If I reduce my force, my elbows will start to spread apart, and if I reduce the current flow, the voltages from one end to the other will start to spread apart.

I visualize it more, but the voltage space between the two halves will always be the same - no matter how positive or negative the signal takes things, those two driver bases will always be that couple volts apart.

SO we want to turn Q308 on juuuuuust right. SO here is my vision: D303,304, R320, TH302, SRV302, and R319 all form a voltage divider. Whatever voltage is across Q308 forms current through that path. And of course each element makes its own voltage drop. SVR302 samples off the voltage needed at the base of Q308. UNless they are mounted to the heatsink or onto a power transistor, D303,304 are not part of thremal tracking, they just serve to push down the max voltage that can appear at the base of Q3080. Kinda like a tube amp with variable bias when we add a resistor under the bias pot so it cannot be set to zero.

Now we look at TH302. It is thermally linked to the heatsink. The warmer it gets, the lower its resistance becomes. Its resistance tracks its temperature inversely. It is in parallel with R320, so their combined resistance goes up and down according to the heat.

The higher the voltage at the base of Q308, the harder it conducts, and the closer together the emotter and collector come - voltage wise. And the lower the base voltage, the less hard it conducts and the more the emitter and collector voltages spread. ANd the hotter the amp would run as you found.

The hotter TH302 gets, the lower its resistance, so the voltage dropped across TH302/R320 decreases, and that means more voltage across SVR302. And THAT means the base voltage at Q308 will be sent higher. Higher base voltage means Q308 conducts hard, whuch pulls the driver bases closer together, which results in the amp running cooler. That is thermal management - the hotter the amp runs, the more the amp tries to run cooler to compensate.

Steve can tell me if I have made any glaring errors in this story.

I grew up on tubes in the 1950s, learned TTL/CMOS logic in the 1970s, I learned relay logic in the 70s too, and kinda filled in transistors after that. Took me a while to shake tubes aside long enough to make transistors intuitive. But here is a way of thinking about them that works for me. The more you pull the base of a transistor towards its collector, the more it conducts. I mean that in an electical sense.

SO in the positive side of this amp, with the output collectors at some + voltage, and emitters at essentially zero, the more I increase to voltage at their base, the more current they conduct.

[Alex R]

OK, here's some more to go on.

I checked D305 with the amp off and the ohmmeter forward and reverse, and it behaves just like all the other diodes.

R320, TH302 and R319 are ok, I checked them again.

I just changed C309 but it didn't make any difference (ie to the now-cool-running circuit with the 7K5 over R320 now making about 3K5 there with the thermistor cool where there used to be about 7K7 through the thermistor and R320 parallelled.

Turning SVR302 now gives across the driver bases from 2.22v to 2.42v.

So, hmm, wonder what was wrong with it?

I'm going to run it full undistorted output into the dummy load for a bit and see how hot it gets.

Alex