On IRF540/9540

[Hans Olofsson]

I am working on an hybrid vacuum tube (pcc88) amp/Mosfet output, and tested it for some time- working ok and no sign of distortion or any misbehaviour. However -shortly after some time-the irf540/9540 shorted, blowing the fuse in a flash. The rails are 73 Volts and the 540/9540 are rated +-100V.
Anyone having any records about these mosfets not making the voltage? I assume no design errror as the amp sounded nice prior to shorting. Is it possible that the PS voltage somehow can spike over the rated mosfet ratings? And should I replace them with 640/9640?

[Stuart Easson]

Without a schematic it's hard to know if it's relevant or not, but: if the rails are +/-73v and the mosfets are used in a typical complementary pair, they will see up to 146v on peaks.

With 100v parts working at 146v there's a good chance they will fail in the fullness of time...

If the 640/9640 are rated at greater than 146v there's a good chance they'll live longer if the other parameters are OK.

Rail voltages can vary as the mains fluctuates, allowing a 10% margin might be prudent, so perhaps 200v parts would be good.

HTH

Stuart

[Workhorse]

Change the Mosfet to 640/9640 as they are rated at 200V!!!

[unclejed613]

i always use the formula Ert*2=Vceo
where Ert is total rail to rail voltage, Vceo is the C-E voltage rating of the transistor. multiplying by 2 gives a very healthy safety margin. if devices aren't available that meet the demands of the formula try for 20-50% safety margin. if still no dice, settle for no less than 10%. sounds like there was a bit of an error choosing outputs for your amp, should have been AT LEAST Ert+10%.

[ilimzn]

73V from rail to rail (+-36.5V) or +-73V? If the latter, it is a wonder it worked at all so far. When there is full rail voltage at the output (clipping), the oposing transistor sees 2x single rail voltage. Have you forgotten that?
Also, what happens when the tube stage overdrives the MOSFET output (it can likely generate a larger voltage swing than the full rail-to-rail voltage), or the tube is cold? have you remembered to limit the Vgs of the MOSFETs, so that on clipping it does not become over the maximum 20V?

[nitrate]

Hi,

Don't know if this helps but i've been using IRF640's in my classD creations and i've had nothing but trouble with them. They have always had a tendency to blow and go short accross all pins just for a laugh, Usally at power up. I'm convinced its a fault of the fets as the design seems to under run the fets by a long way. Can't imagine any reason why they should fail so often.... there dosnt seem to be any transients or over voltages to cause failures. Strange...........

Mad.P

[AndrewT]

Hi Hans,
do come back and confirm if your FETs are operating on a single rail supply (like most tubes) or on a dual polarity supply (like most SS power amps).
You'll gather that there is a significant difference from the content of the previous posters' replies.

[Hans Olofsson]

Thank you for all valuable input.
The single supply rail is +48V fullwave rectified voltage, giving mesured unloaded ~73V. The schematics is from EW 04 march issue by author Jeff Macaulay (sorry for the poor photo from the magazine). In his design the rail is +86V, and he is using 640/9640. Mine is estimated loaded voltage ~70V and I assumed that 540:s would be sufficient.
The cold tube will make a slow turnon, and as the pc88 heats up, the bias would bring on the Fets to the idle current adjusted earlier (~60-70mA)- that seemed to work ok. The saftey margin to Vs 20V should also be secured by the setting of the idle current prior to testing.
By the way- the to Fets (+100 -100) in my opinion would stand 200V in series, so the voltage margin should be even better. You say that the rail voltage could spike over the nominal voltage at power-up, how can that be? The maximum voltage available at all times cannot be more than 73V?
I always considered Fets to be rugged devices as they do not suffer by thermal runaway, but I might have to reconsider.
Regards Hans.

[Hans Olofsson]

Sorry, forgot to provide the schematics.
Hans.

[Stuart Easson]

Hans,

The mains voltage is subject to variations, I see +/-5% at my house all the time, my computers' UPS tells me it has 'protected' me from considerably more. With a normal, unregulated transformer based PSU, output will change in the same proportion as the input.

With respect to the output transistors being in series and therefore only seeing only half the rail voltage: This condition is only true for the idle condition. As soon as the output voltage moves one or the other of the transistors is dropping less of the voltage, the other is therefore seeing more. The limiting condition is at clipping where one transistor is completely on, seeing close to 0v, the other nearly off, subject to the entire rail voltage. Unfortunately, unlike power, you can't average device voltage, the transistor can (and will) 'break' in one cycle of audio.

All that having been said, your rail voltages are well within the rating of the FETs. Perhaps the next thing to check is the gate voltage, anything above ~20v will kill the FET very quickly. You might consider zeners to limit the voltage.

HTH

Stuart