Hurumph! Grumble, grumble ...
Today was frustrating. Requested several SPICE models for larger FETs, all came back negative. From "Sorry, we don't have one" to "The customers are encouraged to roll their own".
I mean, how do they engineer these things? Doesn't that sort of require major SPICE sessions?
Maybe you are in a dead-end portion of the engineering biz ?:-)
...Jim Thompson
Not really. Right now it's power electronics, not exactly a dead-end market.
Sell the models?
Nobody is buying.
...Jim Thompson
The guys at Modelithics seem to do pretty well selling their models -- at work we have a copy of their "CLR" library, since it's a lot cheaper to buy a copy than try to measure everything ourselves (particularly when it's just, "well, this MIGHT be a useful part... but is it worth half-an-hour to model it?), and we've found the models to be of very good quality.
Unfortunately I suspect that they probably don't have models for the particular transistors Joerg is after. They're more than willing to do so on contract, of course, if there's some money around.
I suspect as electronics becomes further specialized, more and more manufacturers will just contract with a place like Modelithics to produce models rather than developing them in-house -- particularly any manufacturer that isn't doing "cutting edge" products.
---Joel
One of the vendors (IRF) actually wrote back that when there is enough business in my case I might be able to convince a sales guy to get me a model. Oh well, another company has already been financially impacted by this old piece of American sales wisdom because I went to a competitor that had models: "If you don't take care of your customer, somebody else will".
I think it might have to do with product liability. The controls on discrete device manufacture are looser than with I/C's. Thus the reluctance to release models, particularly a collection of models necessary to represent process corners.
So Joerg may be shooting himself in the foot if he relies too heavily on the models provided by his "helpful" vendor.
...Jim Thompson
Simple fix: Put a disclaimer in there. That's what some companies do.
No, I design with plenty of margin. What I need to do is, for example, simulate how deep and for how long I am diving into the SOA and such when switching and sloshing "weird" loads around.
Anyhow, I'm fine. Looks like ON Semi might enjoy some new business while others coulda, shoulda, but didn't.
Proceed with caution. Awhile back I compared 8 models for the 2N7000 FET that everyone likes, and posted the results. Phenomenal variation between vendors.
...Jim Thompson
Postscript...
I think what I'd do is roll my own low level model (like the crap simplified model in the textbooks ;-), by taking data with a curve tracer, then use the capacitances from the data sheet.
NOTE: This will NOT adequately model sub-threshold, or off-state leakage. That requires a high level model and VERY fancy measuring equipment.
...Jim Thompson
I dun that :-)
That's the problem. Leakage isn't but I must spend some time in the linear range with the PMOS device, including around the threshold. Peaks to 100W.
>Have something to add? Share your thoughts — no account required.
Ask the community — no account required