More MOSFET Follies

[diagrammatiks]

These are expensive ($7 in 1 off, $3 in 25 off) BUT are what I used on the MOSFETs in the Laser Diode driver I did for the day job.

Normally open, 90 degrees C switch.

Airpax 67F090
http://www.farnell.com/datasheets/29932.pdf

They screw down under the MOSFET mounting bolt.

Cheers,
Ian

RG. I demand LASER DIODE follies.

boo to solid state.

unless it's lazers.

[NateS]

What's the purpose of D2? The gate is already diode clamped by Vf of D1 and Z1.

[R.G.]

D2?

Hmm. Good question. I don't remember at the moment, as it's been a while since I did that and I did it in a very short time. I'll try to think it through.

About all I can dredge up at the moment is that it seems like I put that in at the last minute after some "Oh, yeah!" moment in thinking about it. It seems like it was about protecting the transistor, not the MOSFET under one or another failure mode, possibly rectifier failure or what happens at turn off when the input voltage goes to zero. Just off hand, I can't remember why two diode junctions in D1 and Z1 wasn't good enough. Maybe I was worried about the reverse surge current rating of Z1, which isn't specified for most zeners.

Sorry - I'll see if I can figure out what I meant with that.

[NateS]

This circuit continues to befuddle my weak spice-fu. A few of my observations:
1) I was intending to put a cap on the gate for ripple suppression. I expected it to lower Vgate by half the ripple, but for some reason my simulation doesn't really work without it. It also has a substantial impact on the peak charging current.
2) When used as draw in RG's article, I get much much lower voltage output than expected, UNLESS I put in a filter cap which isn't current limited as input filter. If I add this, the circuit works more or less as expected.

Observe:




vs





After some more messing around with this and related circuits, I'm thinking that D2 (your schem) is in fact the sink path to keep the feeble zeener from bravely sacrificing itself. 60m was based on idle current for 3 12AX7's and 45mA for a pair of EL-34's idling. I don't really believe my sim on the weak output, but I can't seem to get it to tell me anything substantially different. It doesn't appear to be too low of a current limit, because it's not actually limiting more than the first few cycles. The input filter cap (meaning the limit is working on the DC current draw) would suggest otherwise though.

[R.G.]

(a) what happens if you make C1 much, much smaller?
(b) what happens if you put a diode in series with the whole mess into the filter cap to precent back biasing the MOSFET?

I did try this thing in several sims and in an actual circuit and it seemed to work fine. I don't remember using a C1 or a diode.

[NateS]

C1 was intended to slow the gate down, in order to control switching noise when control function is added to the limiter. What I hadn't intended is that for large time constants, it makes an illfated attempt to pretend to be a psedo inductor and fails miserably because it can't store energy and nobody else is either.

Diode AND small series resistor helped, as did a base current limiting resistor. (ngspice seems to be really sensitive here.) Also sprinkling diode clamps and convergence resistors around liberally helps too. But I still can't seem to get more than 400v out of the source when limiting rectifier pulses. It works at DC,and 120Hz isn't exactly fast enough for weird esoteric behaviors to occur.. I have a squirmy baby in my arms and can't really dig further at the moment, but I do have it working (as described) without the gate cap now.

[NateS]

Aha! I think I found my answer, and the results I was getting suggest this is the case. NGspice's mosfet defaults aren't particularly well suited for power mosfets. The model I used is a subcircuit model from onsemi, but it's adapted from some other spice. Even specifying Rd/Rs won't get you close since the magic area numbers that all spice mosfets use which nobody publishes on datasheets are way off for power mosfets.

[Gingertube]

The MOSFET needs to be able to switch at the rectified AC frequency (100 or 120 Hz). The 1M5 + 100nF timeconstant is WAY to big (150ms) to allow this. Ditch C1 entirely, R1 could come down from 1M5 to say 470K, and add a gate stopper (say 220R to 1K) to the MOSFET to supress parasitic oscillation.
Cheers,
Ian

[NateS]

I didn't update the schematic, but yes, it's gone, The gate pullup is now 3x330k, to minimize dissipation when it's shut off, but otherwise, when functioning as a limiter, it shouldn't ever be switching. But I was still seeing Vg - Vs >> expectations, even for modest currents, and that was the fault of the mosfet model.