EMRFD Message Archive 5279

Message Date From Subject
5279 2010-09-28 09:13:13 Tim Spice model with realistic stray inductances for MOSFET PA
I've been working on refining Spice (LTSPice in particular) models for the MOSFET PA's I homebrew.

I've come pretty close at reproducing the waveforms I see with a an actual scope.

I think you can see them if you go to the "Photos" section of this group and click on "N3QE Spice Simulations", then look at the MOSFET Amp schematic and the MOSFET Amp waveforms.

This is a 30 meter push-pull class C amp driven by about 250mw and producing circa 20W out from 12V Vcc. I used IXTU12N06 Mosfets, which are kinda lower-R_on IRF510's. I simulate the IXTU12N06 in LTSpice with an IRF510 crudely hacked up for lower on resistance with about the same stray capacitances as an actual IRF510.

The waveforms represent:

GREEN: Output after the low-pass filter

DARK BLUE: Waveforms right before the low pass filter

RED: Drain voltage (you see it swinging from 0V to 24V with ringing)

LIGHT BLUE: The "bypassed Vcc" at the centertapped transformer that feeds the MOSFET drains.

You will notice that the "bypassed Vcc" in fact, shows a lot of ringing +/- a few volts around 12V. The modeled 10nH or so stray capacitacnes associated with the bypass capacitors seems to capture this about right. According to the specs I read a "bare" MLCC chip capacitor has a few nF of stray capacitance, and a "bare" leaded capacitor has more than that, and I find that 10nH is the right ballpark for reproducing the actual waveforms I see on the bench.

Similarly for the ringing on the drains... when the MOSFET is turned on it pretty effectively damps the ringing with its low R_on but when it turns off I see the voltage ring like the bezeesus. Actually the model doesn't accurately predict the upper spikes I see but it's not too bad. Note that I model this with stray inductances circa 10nH on the drain, the source, and between the source and the output transformer.

And because this is all push pull, even though the strays don't exactly match symmetrically, they largely get canceled out in the final output waveforms.

Right now I'm semi-dissatisfied with the ringing on the "bypassed Vcc" and will see what I can do to (in real life) get it down. I'm not sure it'll make any actual difference in performance though.

So that's the experiment from last night :-). My conclusion is that most of the ringing I see on the scope doesn't have much to do with deficiencies in the transformers but can be mostly explained just with stray inductances between components.

Potentially a well-laid out SMD board, instead of dead bug construction could, I suppose, reduce some of these strays by a factor of two. Will try that in real life over the next week maybe. I'm not sure I could realistically claim that the stray inductances ever get much below the 5nH level - reading MOSFET data sheets these are the strays that happen with the MOSFET package itself.

I don't show the gate waveform on the MOSFET in these pics but it has a lot less ringing than the drain waveforms. I am actually rather satisfied about that, I had somehow expected to see a lot of ringing there. I was worried about gate drive waveforms but in this case they actually seem rather straightforwardly clean.

Tim N3QE
5284 2010-09-28 12:10:29 WA0ITP Re: Spice model with realistic stray inductances for MOSFET PA
Hi Tim,

Could you point us to a datasheet for the IXTU12N06 ?
Google doesnt find one.

Thank you.
I love this radio stuff !
72, 73 Terry, WAƘITP

----- Original Message -----
5291 2010-09-28 15:38:01 Tim Re: Spice model with realistic stray inductances for MOSFET PA

My model for in LTSpice was simply an IRF510 with the R_on lowered by an order of magnitude.

I'm gonna be playing around with paralleled 74HC transceiver chips (with 3V on the ground and 9V