Question - Biasing different types of Mosfets

[Samuel Jayaraj]

Hai everybody! Have a question for you experts out there.

Yesterday I completed a 4 channel version of Anthony Holton's N-channel Amplifier. Please don't get confused. This is the one that uses N-type Mosfets only in the output stage and uses an LM317T and IRF610 Mosfet to set the bias.

I used IRFP460 outputs on two boards and IRFP250N on two of the others. In the case of the IRFP460 boards, I was not able to set the bias beyond 180mA (60mA per device at +-70volt rails). Beyond 180mA, turning the pot reduced the bias current gradually. So I left it at 180mA.

In the case of the IRFP250Ns, I set the bias to 240mA (80mA per device at +-70volt rails). The bias could go higher but I set it at this point).

First question; Why this phenomenon? Does it have anything to do with N types verses the plain types or is it some other parameter?

The on load rails were at 69.7 volts and I assume that the off load voltage should have been over 72 volts. The transformer was a 1.1KVA toroidal and total capacitance was 60,000uF per rail.

I connected one channel using 460s and another using 250Ns to Jamo System II speakers and a CD player for source. There was a discernable difference in the character of the sound of each channel. Overall, the sound was magical.

After about 20 minutes, the channel using IRFP250s gave up the ghost. There was a flash and the output devices, Gate and Source resistors charred.

Reasons I could surmise were that the off load voltage was too high and heat sink temperature could have been a bit too high. (Although I should say that there was sufficient heatsink and a fan to cool it).

Any other reasons for the IRFP 460s surviving and the IRFP250Ns burning, apart from the difference in the bias setting?

Comments and analysis welcome for the benefit of all who may be attempting to build this amp.

Thanks in advance.

[Samuel Jayaraj]

Any inputs or insights from anyone as yet?

Thanks again.

[UrSv]

For the bias setting I would guess that this is normal considering the different Vgs characteristics for the two different FETs. IRFP460 has some 3-5 V of Vgs and IRFP250 has some 2-4. Typically that would mean, AFAIK, that the bias circuitry would be able to bias them differently.

As far the smoke I offer no info...

Where did you get the boards?

[mcp]

Hi Sam

Was the heatsink burning hot after it blew? If it was, the thermal controller was either not working properly or not connected properly to the heatsink. The IRF types are like bipolars, they must have thermal tracking.

If the heatsink was cold, chances are high frq oscillation (ie. unstable). Check the output with a scope.

[Kilentra]

In my sym-amps I have had 5-6 failures like you describe, and I think they were all caused by thermal runaway. This amp seems really sensitive to the bias and temperature. The key is to bolt the bias controlling transistor right through with the hottest of the output stage devices. You could consider using a bipolar transistor there instead of an IRF610, since BJTs have higher temperature coefficient (they'll cause a slight overcompensation reducing the bias with increasing temperature). The sym amp has a BD139 there, I did try the IRF610 but the transistors tended to heat up too much over time.

[sam9]

OK. I need some new info.

I haven't done much with MOSFETs but I thought one of the advantages was negative thermal response, hence runnaway is not possible and themal feedback not needed.

Am I misinformed? Regarfing all MOSFETs? Regarding only some? If some- which are which?

Or - is the BJT drivers "upstream" from the MOSFETs that are the problem? that are the culprits?

[Upupa Epops]

It is very nice expression , but I mean, that mistake is somewhere else. If circuit diagram of your amplifier is the same as on Holton's pages, and I mean that probably is, mistake is not in choise of types fets, but the real cause circuit diagram. Because this amplifier have not protection zener diodes in gates "bottom" fets, if goes to the overload ( or clipping, if you can ), voltage on theirs gates in this case will be higher then 20 V ( typical Ugsmax is +/- 20 V ), gate will "break true" and fet is destroyed. If you cannot try it again, connect to the gates of fets zener diodes ( parallel to G and S ) approximately 10 V / 500 mW in series with normal diodes 1N4148, which stoped your problems

[jam]

Did you hook a scope to the output and check for an oscillation?

Regards,
Jam

[mcp]

Originally posted by sam9
I haven't done much with MOSFETs but I thought one of the advantages was negative thermal response, hence runnaway is not possible and themal feedback not needed.

So far, only the lateral mosfets from Hitachi (like 2sk1058/2sj162) do not need thermal control. IRF, IRFP are like bipolars......they are not immune to thermal runaway.

[millwood]

this is from irf540 datasheet.

Is the temp co-efficient positive or negative? Anyone having a similar chart for the Hitachis?