The Official Dual Rail Thread

[Merlinb]

No production amp from a major company that used one as a regulator or variable voltage dropper.
(I discount the boutique stuff like Maven Peal and London Power, who can't be critiqued because they don't publish schematics, and the VVR kits, because I don't think they're engineered to commercial standards.).

London Power does publish the schematics in his books. Also, the soon-to-be-released Marshall AFD100 has a MOSFET for screen voltage control.

[R.G.]

OK, conceded :-P I can also see that the SOA curve shows 0.2A at 600V, the full 160W dissipation. If that really is true (and not just a calculated figure like the APT paper suggests) then this device would be a great choice for a VVR type circuit. I also see they're in stock at Farnell for about $1 each, so maybe I should get some and indulge in some Mythbusting.

I heartily agree. The info about low power linear failures of MOSFETs became available at a time when I was not reading much about power MOSFET reliability. I don't read all the literature all the time, so some things escape me.

It appears from the tech reports that I dig up now that you bring it to my attention that the effect is real; if nothing else, the introduction of "linear rated" MOSFETs by semiconductor companies says that they think it's real enough.

I think some "mythbusting" is in fact in order, at least over this particular application. A test setup for whether the MOSFETs die in the expected case under situations of expected use and abuse would not be that hard to set up and run.

About the only unknown involved in setting this up is the expected value of screen current to be put out. A test rig with HV supply, linear pass MOSFET, and an active load which can apply programmed load profiles including overcurrents to the DUT could be done.

This does leave some questions and issues in my mind. There are obviously some examples of power MOSFETs being used in linear applications successfully. These include audio amplifiers and voltage regulators. There is also the issue of how much is enough - at some dissipation level, a MOSFET would survive. And there is the issue of other and compound devices. An IGBT might be fine - don't know, I'll have to look. Also a compound smaller MOSFET with a bipolar.

More work to do.

[Wilder Amplification]

According to the KT88 data sheets, a KT88 draws 15mA of screen current @ maximum output. Of course both KT88s shouldn't be drawing this at the same time since they are operating out of phase. Also, not sure how much they'll draw at REAL maximum output (clipped to the max/bouncing off the rails) and I'll have to do some measuring once I get a basic working model up and running to know that.

[R.G.]

That helps. From the literature I have read on the issue so far, it appears that the actual dissipation limit and voltage limits to a power MOSFET under linear operation like this would require amount to a voltage limit at low currents. It also appears that this is not specified by the makers of the chips. However, the paper on linear operation of one manufacturer's radiation hardened devices shows much better results than the theoretical "Spirito curve" estimation would predict. It's kind of open as far as I can tell.

However, all the indications are that there is some remaining current capability at a big fraction of the rated BVds. It may be that the issue can get solved by a totem pole of MOSFETs to take the voltage stress on any one MOSFET low enough to not be an issue. Using two or three 800V MOSFETs doubles or triples the price of the devices, but radically cuts the voltage stress to any one of them.

Then there is the linear region of an IGBT. I'll go look at those.

Measurements of real screen current under maximum warp drive would be useful at doing the next bit of work. Thanks for any data you can add.

[R.G.]

IXYS at least has power MOSFETs rated specifically for linear operation.

[defaced]

I know this isn't in the current scope, but would screen current change if the tubes were driven into AB2?

[mooreamps]

I know this isn't in the current scope, but would screen current change if the tubes were driven into AB2?

yes it would !! :} Maybe the friendly neighborhood R.S. ain't so dumb after all......

-g

[Wilder Amplification]

I know this isn't in the current scope, but would screen current change if the tubes were driven into AB2?

Absolutely! The harder a tube is driven, the more plate voltage drop you get. The farther below screen voltage the plate drops, the more electrons are drawn by the screen.

Once you're at clipping and plate voltage is pulled down to the (-) rail (i.e. ground) and stays there, average screen current increases as it has to sustain a constant current draw for a length of time, which causes the infamous "screen glow".

[R.G.]

I think I forgot to mention this before - I do know that in fact linear operation of the IRF820 and IRF830 work well as both source follower voltage regulators and signal followers without going into thermal runaway because both I and other people have done this and it works.

That being said, a lot of the literature I read indicates that the continuing specialization of power MOSFETS for higher transconductance, lower Rds, and better switching seems to be enhancing the thermal runaway problem in general. So maybe the "old" IRF820/830 was more immune by virtue of being older and cruder.

[Steve Conner]

Yes. I've also used older MOSFETs for linear applications without any bother.

But there's always the risk that something labelled "IRF820" nowadays will have a smaller chip inside made on a more modern process.

I've done some work with IGBTs, and I don't think they seem that promising. Some of the IGBT datasheets do actually say, "Not for linear use".

I'm getting psyched up to go and trash some MOSFETs with my 600V, 1kW power supply, does anyone have any requests?

[R.G.]

Yes. I've also used older MOSFETs for linear applications without any bother.

But there's always the risk that something labelled "IRF820" nowadays will have a smaller chip inside made on a more modern process.

That's always a danger. In the effects forums I keep telling people that part numbers and datasheets are merely suggested limits. They may get any part which barely makes it inside the numbers on the datasheet. If there is no number there, no guarantees. And the maker is free to sub in anything they make, by any process, which gets inside the lines.

I've done some work with IGBTs, and I don't think they seem that promising. Some of the IGBT datasheets do actually say, "Not for linear use".

From yet more reading, I agree.

This is a really ironic twist. I was trying to design power supplies back when we had only bipolars, and getting a bipolar to reliably switch rectified-AC-power at 25kHz was a very chancy thing because of the FBSOA and RBSOA. The appearance of power MOSFETs stopped that. Now specialization of the MOSFETs is leading right back to SOA problems, just not exactly switching SOA problems.

I'm getting psyched up to go and trash some MOSFETs with my 600V, 1kW power supply, does anyone have any requests?

Yeah - start with the cheap ones.

A couple of the articles I quoted had tests of some actual devices. For test conditions, I'd like to see 60mA at 300V across drain to source for nominal operation of a screen supply, and 600V/0.001A, for emulation of a current limited screen supply.

Or just make the straw-man screen supply with current limiter, and plot its path as it goes into limiting. Screwing down a thermistor, bipolar, or diode to the DUT for checking temperature would be nice too.

I looked last night - my variable HP tube power supply only goes to 400V.

[mooreamps]

Split-rail power supply...


-g

[R.G.]

Uh, Gary - apropos of what? Not much content on the current discussion there.

[mooreamps]

Uh, Gary - apropos of what? Not much content on the current discussion there.

I do also build a "dual-rail" HV supply. Both both of those rails are of "equal" voltage.

-g

[R.G.]

I do also build a "dual-rail" HV supply. Both both of those rails are of "equal" voltage.

Oh. OK. Good for you.

Do you also wear shoes, drive a car, use a computer? Keep on, then. Those are good for you. Hope that continues to work OK.