Setting bias on a Ampeg SVT 4 Pro

[oc disorder]

After replacing all the IRFP240-IRFP9240 in one side plus just about every source and gate resistor, a few drivers ,15v zeners and other misc resistors finally there's no DC on the output (well a few mV)and signal goes in and out (one side slightly more) .
According to the service notes should adjust for approx 25mV across the .47 5w "source resistors".
First question .Do they mean 25mV plus or minus 5mV
Quote
"Bias adjustment with no signal and no load.
Adjust AP101 to obtain an average of 25_ + 5mv across R139,R140,R146,R147,152,153,158,159,164 & 165."
I'm assuming they couldn't line up the + and - over the top of one another in the print process.
I assume therefore anything from 20mv to 30mv is ok but preference should be given to the low side..
meaning if they are all heading towards 30 it's a good idea to back it off slightly.

I'm running the thing through a 200w (240v)light bulb which is about a quarter lit (new test terminology! )which gives me approx 60volts DC on the drains... instead of the aprox 80 volts I should get with no light bulb limiting.

I figured out to put one test probe in-between the heatsinks to where the 0.47 5 watters join (spkr output)
and the other I can clip onto the source of the output devices on the outer side of the heat sink.

My dilemma is this
At 60v aprox some of the fets are conducting about 23mA (the two in the centre) while the others are much lower ,4 to 9 mv.

Measuring on the good side (the one that didn't blow) am getting less variation 8 to 19mv approx.

At this stage neither trim pot has been touched.

Have tried it into a dummy load (8 ohms) and the sine wave looks symetrical but as lamp limited soon as I crank it "she go crazy" well it wants more power supply and relays click off and come back on as I would expect.. makes a good light show !

I'm hesitant to apply full juice as I expected the bias readings across the 0.47 5watters to be more consistant.

Does this mean I have to buy bags of fets to match them into lots of 5 ?

Or should I stop being so fussy and plug it in and "rock on" as I've heard mentioned...?

The amp was only just out of warranty ,is well looked after and just smoked for no reason one night.

any help in this area appreciated.

[MarkusBass]

The MOSFETs should be matched but even with transistors assembled in AMPEG factory you may get such discrepancies. I often see differences like 10-20mV. Their matched MOSFETs are not perfectly matched .
The bias should be set with correct supply voltage. I don't understand the idea with "light bulb". I thought it works only for valve amps. SVT 4Pro is a transistor amp (at least when looking at the output stage). If you remove the bulb, you will get completely different bias. So there is no point of setting it with a bulb.
Differences like 10-25mV are acceptable. If you have higher differences, you should match the transistors.

Mark

[oc disorder]

Thanks for the reply.
No I'm not intending to set the bias while the amp is limited by the bulb.
I use it to get the thing up and running after a big repair and if there is a fault the bulb saves me from a major catastrophe.
I suppose I'll have to suck it and see.
It seems stable enough I'll turn it on with the bias measuring in place.
So if it seems to start climbing too quickly I'll pull the plug.
In future I may buy a few extra o/p devices to try and get them in the same ball park.
I'm hoping with the correct voltages it will even out more although the original side had closer readings which I suppose reflects the fact that they have been matched to some degree.

[Steve Conner]

Just keep an eye on the bias as the thing warms up, and recheck it once the heatsinks have got hot. Then run it a bit with a signal generator and dummy load to get it hotter still, and recheck again.

If the MOSFETs are poorly matched, then there's a danger that the hottest one might run away and blow up, without the others getting the heatsink hot enough to activate the thermal sensing transistor. (Does the SVT4 Pro even have one? If it does, you could relocate it on top of whichever MOSFET has the highest bias reading.)

I never liked Ampeg's use of the IRFPxxx, they're best kept for switchmode power supplies.

[oc disorder]

Well some figures before I sleep on it.

P9240----P240----P240 Orig amp
9.8-------6.0------23
7.8-------1.1-----15.3
7.5-------1.4------5.8
10.1------19.2-----3.5
14.6------18.5-----6.2

Well I guess I can swap a couple around from one amp to the other
and get an improvement.
I'll measure the P9240's amp 2 tomorrow.
The original amp was drawing 31mA when I turned it on so
I backed it down to 23 to measure it.
The repaired side I tried to make the max about 20mA but 19.2 was ok
for comparisons sake.

[MarkusBass]

I'm not sure what kind of help/support you expect. The difference between 1 mA and 19 mA is definitely too high and you need matched MOSFETs. I usually buy 10 MOSFETs to get 4-5 matched.

Mark

[oc disorder]

Well the kind I'm getting.. so I now know from first hand experience
that there can be huge differences in gain from the same supplier .
I'll have to get some more in and pull the others out and start again.
I guess I was just thinking what I could do now until I have more to choose from ie take the 23 and 15.3 from amp two (its a dual mono) and swap
for the 1.1 and 1.4 which are closer to 5.8 and 3.5.. 'spose its idle curiosity
Then again if the gains are different overall ie the negative half and the positive half that would be undesirable as well I guess ?
Thanks for your input

[Steve Conner]

Yes, you can swap them around like that. But I think you'll need some more to make two matched sets. Out of the 10 P-channel devices you have available, you have 4 hot ones and 6 cold ones, so you need to go looking for another hot one.

Matching between the positive and negative sides doesn't matter. The N and P channel FETs have totally different characteristics to each other anyway, but negative feedback makes it all OK. In particular, the DC servo should make the bias currents on both sides add up to the same at all times.

[MarkusBass]

I'm not sure whether the term "gain" is correct regarding MOSFETs.
But the most important is that you have to match transistors within one group (e.g. P240). You don't need to worry how negative side is related to positive side. As far as I remember, SVT 4 Pro has a DC servo which takes care of this problem. You have to make sure that e.g. currents on one side are within let's say 20% (if you remeber 25 mA and +/-5 mA example). Silightly higher margin is also allowed (if you look at transistors assembled in the factory). But not 1 mA and 19 mA.

EDIT: 25 mV on 0.47R corresponds (more or less) to 50 mA. So above example in not that precise.

Mark

[lowell]

I'm about to start repairing one of these too. Poweramp B has fried. Bunch of toasted gate resistors and MOSFETS, as well as bias DC servo transistors too. (I think that that is what that part of the circuit is called) Anywho these MOSFETS mustve gotten quite hot as the thermal insulator strip is practically glued on and it's tearing when I remove the transistors. SO, my question is as of yet, where can I find thermal "stripping" or "tape" like this. It's not pads, but a strip for about 10 MOSFETS. I've searched on Mouser for thermal tape, thermal strip, bulk thermal insulator etc.. and can't seem to locate it.

[Enzo]

So get 10 individual pads.

[lowell]

That was my plan B. If anyone else knows where to get the bulk stuff holla back, be nice to have on hand and be able to cut it up as needed.

[J M Fahey]

I can get the stuff in bulk in Argentina, so by definition it *must* be easier on USA.
This is the 1 Ft wide x 50 Yds roll.

That said, it's incredibly expensive if you buy it by the square yard.
Fact is, an individual pad is stamp size, and, say, a dollar is not much.
But a square yard is like , say, 2000 to 5000 individual pads, so price becomes scary.
That's why it's seldom bought in bulk, unless you are, ahem!! , Ampeg or Fender or somebody their weight.

[imix500]

I've purchased a 12 x 12 sheet of silpad before. Might have been digikey, but it's been a couple years.

[J M Fahey]

Just checked.
There's a 12" x 12" sheet on EBay @ $20. Not bad.
Or $180 the square yard.
Don't want to even imagine the price of the roll I posted