JFET vs MOSFET

That's the company name I couldn't remember.

The devices used in the Elektor PSU were probably TO220.

Bizarre.

Indeed, and they're even "programmable" -- note the symbol is double-gate: not side-by-side, but stacked. They're uncommitted EEPROM cells!

They even have undocumented programming pins. Likely, the "IC*" pins are brought to +12.5V or something like that, maybe with a serial programming interface as well, to enable post-packaging trim.

You can expect Vgs(th) to drift slowly over time (accelerated by heat), because the embedded gates really are completely isolated, floating metal. SiO2 is an amazingly good insulator, and holds a charge for centuries, at reasonable temperatures.

Tim

A JFET is nothing more than a too-well-made UJT.

I don't recall noting much bulk modulation effect in Si JFETs, which implies they're rather well made indeed -- the channel connections being heavily enough doped that added (minority) charge carriers don't reduce the total D-S "on" resistance appreciably.

Perhaps it's more dramatic in GaAs or GaN, for various reasons. 2DEG has a stupendously high mobility, but it's a very thin layer, and would be used absolutely minimally (read: extremely tiny junction) to get the most benefit in an RF transistor. As a result, even as conductive as the channel is, there's so much (lightly doped?) bulk around it, that it can be modulated rather strongly by emitted charges.

The downside would be IGBT-like performance: the majority carriers shut off quickly, followed by a long "drool" of minority carrier recombination. It's probably still a lot faster than Si, but I do suspect the speed*power product is lower.

Tim

You could use it as a dosimeter for xrays I guess, like the voltage references:

The FET might also make a simple square-law detector, though the temperature coefficient is probably large. You could use a second one to do the square root and make an RMS converter, then the temperature coefficients might cancel out.

Fine, but what good is that? Get real... ;-)

Tim

No, you are not totally screwed up. This group is totally screwed up.

Why so snotty? Didn't get laid last night ?>:-} ...Jim Thompson

I have a suggestion: go away!

Floating-gate voltage reference:

And this:

If you don't like doing something, the first step is to stop doing it.

"Jim Thompson" wrote in message news: snipped-for-privacy@4ax.com...

Jim can't-get-with-the-times Thompson seems to have forgotten his complex math theory. ;-)

Tim

Hardly... lots of Churchill under my belt. ...Jim Thompson

Where did I say I didn't like it?

I'm glad it's only a suggestion.

Probably not much use (except on a chip maybe), but it is something that people sometimes try to do with Schottky diodes, which are perhaps an even worse approximation of a squarer.

Can you think of any uses for a fet with Vth = 0V which couldn't be done just as well with other FETs?

Analog Devices makes some jfet-based voltage references, sort of analogous to the bipolar bandgap concept. ADR420 and such. Very low noise and TCs.