mosfet forward bias/ Hartke HA3500

[Enzo]

Has your FLuke arrived? We need you get you on diode test. it is not done on ohms scale. It sends a small current through at a low voltage and then measures the actual voltage difference across the semiconductor junction.

Not to belabor it, but did you checkR320 and TH302 with them lifted from the circuit?

If you have a good board there, try comparing measurements between them. That won't dirty up the new board.

One thing to check is your Zobel network. Make sure R335 is not open 4.7ohm 2W, just to the right of Q320 on the drawing.

DO you still have 17v across S307 - or whichever diode? Or has that settled down?

Iif it were me, I would run it without a load right up to clip, and scope the output to see if the clipping is symmetrical. Then i would put on a load and again run it up righjt to clip and see if it still acts right, then I would run it at about 2/3 power for an extended period as burn-in. If it passes that, put a fork in it.

[Alex R]

Hey Enzo, thanks for getting back despite my last weirdly garbled post. Must've been late. Makes sense if you put the other bracket in, but let it pass. No I didn't lift the resistor/thermistor parallel pair from the board, but their paralleled resistance at room temp was the same as the new board, so I thought 'ok'. With Steve's extra resistor it's about half the normal value across what are now three components. I'll check the other stuff you mentioned tomorrow before getting it nice & warm in the manner you suggest. The Fluke should arrive tomorrow also. I'll check all the relevant diodes with it & post the results.

Did I say thankyou for all this help & teaching? Probably not, ignorant git.

Thank you -
Alex

[Alex R]

...strikes me that the extra resistor will change the way the thermistor (as it heats) affects the base voltage as heatsink temp rises & falls - & thus change the thermal management you described. So worth one last look around for whatever causes this.

It's an interesting system. More musician-friendly than the Trace Elliot 'bipolar bear' use of a bimetallic switch that turns the amp off when it (apparently inevitably) overheats. Have you dealt with any hot Traces? What d'you do - bigger fan/heatsink, or (gulp) higher temp thermostat?

Let me say that I have had a LOT of help from the new Trace outfit (some of the same people designing), who are really helpful and very ready to admit to design 'issues' in some of their old amps. The guy worked out a mod to deal with the overheating vreg in some Speed Twins & sent it to me.

Having myself also grown up on tube technology and only fairly recently started to work with durned newfangled transistors I find myself sympathetic with your explanations. The muscle analogy helps me to think about current. Voltage is so much easier to deal with - it stays still while you look at it if you see what I mean.

More tomorrow,

Alex

[Enzo]

Thermal coupling to the heatsink is a way to keep currents steady. It is a separate thing from those thermal cutouts that kill the power. Your thermistor serves as a balancing agent in the active circuit. It wants to prevent problems. The latter thing is just the same idea as a circuit breaker - it trips when there is a problem. Prevents fires.

[Steve Conner]

Well, I still maintain that it was changing that transistor out for a MPSA42 that did it. The MPSA42 is a high voltage transistor, so it has less gain than a low voltage part, because of the tradeoffs inherent in transistor design. That means it will need more encouragement to "squeeze" Enzo's pec deck down to a suitable bias voltage.

You're right that the extra resistor shunting the thermistor will reduce its effect. If you're worried, you should try some higher gain replacement transistors, until you find one that works without the shunt resistor.

Either that, or you've put the MPSA42 in the wrong way round.

BTW: The thing I find hardest about solid-state amps is that they rely on buttloads of negative feedback for their operation, and won't function at all without it. If anything breaks the feedback loop, the whole amp falls down, and it's often hard to figure out exactly what broke. With tube amps, negative feedback was bolted on as an afterthought, and they would still function well enough with a broken loop that you could trace the signal through to the bad stage.

IIRC, the cure for any fault in a bipolar Trace Elliott was to fit the redesigned MOSFET power board. ;-)

[Alex R]

R337 is ok.

The 19v across that diode has gone.

I tested it as you suggested and it clips symmetrical little half-moons off the top & bottom. As it heated up I took readings across the power transistor bases and watched the voltage between them go down from over 2v to 0.92v as the temperature rose & turned on the bias transistor. Then I put the top on & left it on 2/3 power for a while and it was blowing just warm air through from the heatsink. Waveform still good, clipping the same when turned up.

So I am putting that fork in it, botch or no botch! I've got a bit backed up with other jobs whilst struggling with this. I'll exercise the Fluke on something else when it arrives, & no doubt have plenty of questions about that.

Many thanks to all - and I remain really interested to know what might have been the problem.

Alex

[Alex R]

Hi Steve, our posts crossed. Thanks for that about the MPSA42 variation in gain from the stock part. I see. Puts my mind at rest. I guess the resistor is ok in there as - like I said in the crossed post, the circuit still pulls down the bias as the heatsink warms up.

And yes, I did have it in the right way round, crossed leads and all. It was Q311 I put in the wrong way!

...& actually I thought Enzo's pecs were pretty cool.

Let's not beat around the bush - you solved my problem! Thank you!

Alex

[Enzo]

I hadn't thought about the gain thing. Perhaps that does require the tweaking of the grid resistors. SO Steve, you are saying someone threw a couple more weights on the machine.

Steve, the SS amp feedback thing can be a hassle, as any one place in the loop fails and it throws the whole loop off. I have at times in the past when really baffled broken the loop. The output bus connects back to the diffy pair at the input - or to the other op amp input - through a largish resistor. Looks like 33k R318 here. Without signal or load, the output bus ought to sit more or less at zero, ground essentially. SO I would take that resistor and lift it from the output bus and ground that end of it. Now the feedback is broken. The amp will drift around and stuff, and you must monitor what the output is doing, but since the output is what we are trying to settle, that is not a problem. Sometimes this can provide insight into where the problem lies since the amp no longer tries to correct itself.

It may not be accurate, but I have assumed the available gain of these things is such that without NFB, they'd act like op amps. REmove the feedback on them and they go nuts.

I thought of this discussion a minute ago when a friend asked me what a concertina PI was. Squeezing the plate and cathode together and apart. Hopefully he knows what a concertina actually is.

I think too you get used to the feedback in a SS amp. If the output fails on one polarity and cannot push the wvaeform its direction, the feedback gets the inoput stages to max their output trying to compensate. SO when say the bottom half of a sine looks OK, but the upper half is WAY taller, then I guess it is trying to overcome the lack of gain in that direction, so I look for a breakdown in the positive signal chain.

BY the way, these Samson and Hartke amps using the 7317 output protect controller IC will occasionally get real sensitive. CLick into protect all the time and at slight provocation. Un those cases it is rarely the IC, it is usually the little lytic caps dried out, like C317,318 here.

[Steve Conner]

Well, replacing the transistor with a lower gain one isn't so much like putting more weights on the machine. The weights (the rest of the circuit that needs squeezed around) are the same, but we replaced the stocky football jock type who was in there with a tall lanky guy. So we needed to change that resistor to give him some steroids. Or something. Dammit, I'm not sure if this analogy is so good after all

You are right, these solid-state circuits are just like op-amps. Remove the feedback and they will freak out, more than likely just sticking to one rail. Except for some audiophool creations designed by people who think NFB is evil.

[Alex R]

Well, unsurprisingly the distributors won't take the board back, so if you'd like to buy this expensive board cheap...

http://cgi.ebay.co.uk/ws/eBayISAPI.d...m=330146106817

!!

[Steve Conner]

Well, I was collecting bits to make a big power amp for years. I've got transistors, heatsinks, transformers, capacitors, everything except... a driver board So I may well bid on it, it should work fine with the parts I have.

[Alex R]

Well, that would be utterly cool. I wonder how many more of my little problems you might solve?

As I wrote on the Ebay page, I am in fact embarrassed by the possession of two boards due to an unaccountable error when ordering - how about making a big fat stereo...?

In case you didn't know, the board is designed to sit on top of a heatsink more or less 65mm across. The power transistors and drivers solder into the slots you can see either side of the board and project beneath it at 90 degrees, straddling the heatsink and bolted to its sides. It's a neat arrangement, till you want to get at the solder side of the board. Pretty adaptable in terms of heatsink width I woulda thought.

A.

[Jazz P Bass]

Hartke HA5500 Schematic

[QUOTE=Steve Conner;26278]Well, I still maintain that it was changing that transistor out for a MPSA42 that did it. The MPSA42 is a high voltage transistor, so it has less gain than a low voltage part, because of the tradeoffs inherent in transistor design. That means it will need more encouragement to "squeeze" Enzo's pec deck down to a suitable bias voltage.

You're right that the extra resistor shunting the thermistor will reduce its effect. If you're worried, you should try some higher gain replacement transistors, until you find one that works without the shunt resistor.

Received a Hartke HA5500 with a blown bias circuit.
After replacing the shorted transistors I could not get the 10mv idle voltage between the output emitters.
The amp was pulling 115 watts from the mains at idle.
I found the repair above & it seems the repairer had the same issue.
I called Samson & got the correct schematic.
Lo & behold the bias transistor was the wrong one. A TO220 no less.
( I replaced what was there )
As above, according to Steve the wrong transistor may be an issue.
I installed the correct part (2SC3200) & Bingo.
10mv & 50 watts from the mains.
Thankyou Steve & Enzo for the advise.

[soundguruman]

I decided to bridge out TH 302, short the 7.5 K resistor
who needs that anyway?
And I decide to bridge out D 304 and replace D 303 with a higher rating,
and I finally got some bias going in there somewhere...
BUT I think one of the big problems is the location of the fan thermistor, it's mounted to the coolest part of the heatsink, and so it slows down when the end of the heatsink, the last transistor is hot. And so I moved that one to the end of the heatsink instead of the cool end. Now the fan speeds up when the last transistor is hot.
And I thinks a 47K resistor across the fan thermistor, to make the fan speed up a bit.
I think the bias in this amp is ##^*?!! and the heatsink is too small. And I think that D 308 is drawn backwards on the schematic.