Broken MOSFET power amp- MOSFET questions

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.

bobschwenkler

Well-known member
Joined
Dec 8, 2005
Messages
483
Location
Olympia, WA
I'm working on the first MOSFET device I've worked on so far and I have a few questions about power MOSFETs. I think I know what the answers are, but I want to check with people who'd know before I go chasing my tail working to repair this amp.

It's a push pull amp and the power stage has ~+/-85 volt supply rails. The drivers for the power stage are medium power BJTs fed with pretty close to the same supply voltage. The same thing has happened in both channels (at separate points in time), so I'm wondering if there's an underlying cause, but until I get things back to normal I don't feel I can trouble shoot that possibility. The driver BJTs on the + rail went short base-collector, which then would have put the full voltage of their supply rail on the gates of the + side of the power amp and also probably the same on the - side of the amp.

I'm reading really low resistances all directions across the MOSFETs with a multimeter, which is totally wrong, right? Is it possible that this failure mode took out all of the power MOSFETs in this amp in one go? If it's relevant to anyone, the power amp is comprised of IRFP9240 and IRFP240s.
 
bobschwenkler said:
The driver BJTs on the + rail went short base-collector, which then would have put the full voltage of their supply rail on the gates of the + side of the power amp and also probably the same on the - side of the amp.
In my experience it's more likely that the MOSFETs go short circuit first and that sometimes takes out the drivers too. It can be a chicken-and-egg situation though. If the supply voltage (full + to - voltage, not + or - to ground) is too close to the VBE max rating of the drivers, they can blow first. The only way to tell exactly which devices have gone is to measure them out of circuit. If it's a paid repair job it's often cheapest just to replace them all.

Make sure the bias is set correctly for the replacement MOSFETs though. They vary a lot more than BJTs, and contrary to the popular myth, MOSFETs DO suffer from thermal runaway if quiescent current is too high. The temp coefficient of their gate-source threshold voltage is typically negative up to about half of their maximum drain current, so in any practical class A/B design thermal runaway is a distinct possibility.
 
I've never had much success making a linear (non switching) power amp, with vertical (switching) MOSFETs. The old lateral Hitachi mosfets were different animals, and much easier to use in audio amps. 

Good luck... I can only offer general advice, when power amp output devices fail, they often take nearby parts with them. 

JR
 
try to get matched/selected pairs/quads/etc fet's for replacement. the amp will sound much better then ;-)
also the drivers if they are broken of course...
 
This amp's got 2x5 power FETs per channel (2 channels, so 20 MOSFETs). Are there sources for this type of transistor? Or is matching MOSFETs difficult?

I went through the amp today and there's only one FET that isn't shorted! I'm going to order a new set, buying from the (presumably) same batch ought to be more well matched than putting new and old side by side I'd think.
 
> there's only one FET that isn't shorted!

Of course.

Find a rock so heavy that you need five good ropes to lift it. Start lifting. If one rope is nicked and breaks, the other four have a 5/4= 125% OVERload, and they break.

And of course, if only most of one side is blown and the amp sounds like crap, the users turn-it-up in hopes it will get better.

Of course when the P-types turn short, the N-types will try to pull the short, and they will blow.

And yes. The drivers can fail and melt the outputs OR the outputs can fail and the drivers bust a gut trying to move the load directly.

Unless you love puzzles, wholesale replacement is a direct path. If you can find truly "equivalent" replacements.

IMHO, there were a lot of just-good-enuff MOSFET amps. MOSFETs avoid some breakdowns that plague BJTs. But real life eventually finds other ways to stress power parts. A stage-worthy FET amp should be at least as over-built as a BJT amp or a console stand. The console stand must suport the console (normal load), or a producer leaning on it (normal overload), or freak accidents (Bob throws 800-Watt amp across the studio). The range of "impossible" loads and powers at an amp/speaker interface is beyond the comprehension of many smart designers. Who do learn (or don't get called-back). So I'm saying that buying a more-mature amp product might be the wiser path.
 
PRR said:
So I'm saying that buying a more-mature amp product might be the wiser path.

I may sell it off at some point. I bought it broken for almost half its value when functional.

FWIW, it's an Ampeg SVT4-PRO, pretty decent amp. It definitely sounds pretty good. Hopefully it remains working while I have it, these power MOSFETs are not exactly cheap.

I'll actually never be running it near its limit while I've got it. I don't play loud music and it will otherwise be used for recordings.

Edit:Wrong model number changed to correct one.
 
Oh, one other thing. If there are any diodes involved in the output or driver stages, you probably ought to replace them too. A bad (or intermittently bad) diode is often the start of a cascading failure.

Peace,
Paul
 
bobschwenkler said:
This amp's got 2x5 power FETs per channel (2 channels, so 20 MOSFETs). Are there sources for this type of transistor? Or is matching MOSFETs difficult?

I went through the amp today and there's only one FET that isn't shorted! I'm going to order a new set, buying from the (presumably) same batch ought to be more well matched than putting new and old side by side I'd think.

i would change ALL fet's. i got matched ones from borberly audio here in germany for a hafler studio amp some years ago.
http://www.borbelyaudio.com
erno is a specialist with fet-amps
the amp is working for years now and sounds pretty good. i can't remember the type of the fet's...
matching ain't dificult but you have to order a greater batch of transistors.
could be a little more than ordering matched ones from a special suplier  ;-)
 
pstamler said:
Oh, one other thing. If there are any diodes involved in the output or driver stages, you probably ought to replace them too. A bad (or intermittently bad) diode is often the start of a cascading failure.

Peace,
Paul

Is there a normal failure mode for diodes? Is it short? Do you know at all what's happening in a diode that's intermittently bad/
 
The same thing has happened in both channels (at separate points in time), so I'm wondering if there's an underlying cause

What kind of protection does it use? A crowbar of some sort? Can you post a schematic?

Justin
 
Maybe you could use a simple mosfet tester like the following
A 10VDC etc regulated supply with a high enough current rating and a 100ohm power resistor and a pot 10K lin etc and a temp heatsink maybe using a clamp of some type  and insulator and some heatsink compound or thermal pad.

10VDC across 100ohms= about 100ma with a fully on low RDSon power mosfet

N channel
+10VDC to 100 ohm
100 ohm to drain of the mosfet with a temp heatsink.
source to ground

10k lin pot
Top to 10VDC
wiper to gate
Bottom to ground

Set the wiper to maybe 1VDC, 2VDC 3VDC 4VDC and then 9VDC
Measure and calculate and write down the voltage drop across the drain resistor for the drain current at each gate voltage using ohms law or maybe use a current meter
Measure and write down the drain to ground voltage at each gate voltage

You will want a mult-point match look for the same Id and/or drain to ground voltage at the same gate voltage

Things to keep in mind often the higher the power and lower the RDSon the higher the gate capacitance sometime 1000's of pf
So try to match the input cap and RDSon as well as the other parameters.

I used a circuit like that to match mosfets for a multplexed LED driver circuit.  I did not want to use any dropping/balance series resistors so I did a few point gate voltage vs Id match


 
> it's an Ampeg SVT3-PRO

Oh.

Odd power stage. Many comments about unreliable bias.

But "both channels"? Isn't there just one channel? One input, tube preamp, tone/EQ, that wacky big-volt-amp (CF+WCF, WTF?), cap-coupled to 8 Gates.

http://music-electronics-forum.com/t10203/
http://elektrotanya.com/ampeg_svt-3-pro.pdf/download.html

Apparently you should replace the output bias trim-pot. It isn't clear to me how it would fail "smoke and burn", but never trust 5-cent bias pots.

Remove MOSFETs. NO load. The output should servo to zero DC, less than a tenth-volt.

Measure DC from a N Gate to a P Gate. You MUST be able to trim this to less than 8V, and I'd want to have 4V available. The "low current" end of these MOSFETs is about 4V gate-source; if bias is higher the MOSFETs conduct many Amps and fry. Going lower will give rough sound on small outputs (this beast is not made to play small) but is where you want to be for now.

These MOSFETs are $5 parts but are going out of style. DigiKey has stock of 240 but not 9240. Octopart shows a few. Makes you wonder if you should stock-up while you can, or bail-out ASAP.

For initial tests, load just the two sockets with 330 ohm resistors to the "protection" transistors. This should do big signals clean in no-load or in 50 ohms (or 100 ohm 50W in series with a speaker). If not, you are only down $10 instead of $80.

The bias process is not in the info I found. If run "cool" then small (<2V) signals will kink through crossover. If too "warm" it will play great but with some danger of thermal runaway and disaster.
 
> Maybe you could use a simple mosfet tester

At the price of these giant MOSFETs, neither "Ampeg" nor Bob can afford to "match". And it should not be needed. The emitter (whoops) resistors drop ~~2V at full 450W out. MOSFET turn-on from "not-cool" bias is quite smooth. One MOSFET may hog the first Amp, but 16 Amp peaks will share across all four devices.

And it really isn't "supposed" to be "perfect". There is zero audio NFB around the MOSFETs.

IMHO: eight TO3/TO247 devices is not enuff for 450 Watts from a stage-amp, and the "protection" probably isn't.
 
looks like Vishay makes them now
http://www.vishay.com/docs/91210/91210.pdf
http://www.vishay.com/mosfets/list/product-91239/
http://www.vishay.com/mosfets/
 
Yup, it looks like they at least did design the bias pot correctly, when it fails (trimpots typically go open-circuit) the bias voltage drops to very low level.

I never had much success with getting low distortion and good bandwidth with vertical FETs but  for an instrument amp it can make noise.

I have always been suspicious that MOSFETs were used to make a marketing association with the well respected Hafler (Lateral) MOSFET amps. The customer doesn't know that all MOSFETs aren't the same...   

JR
 
Back
Top