weird fet

[John Larkin]

We were experimenting with a Fujitsu PHEMT, a pseudomorphic
heterojunction gaas-type fet. They are poorly characterized for DC
specs (typical microwave part!) so we were making pulsed measurements
to get the DC transfer curves.

They are normally considered to be depletion parts, but it turns out
they enhance nicely, to about 1.4 * Idss with maybe 0.4 volt positive
gate voltage. But the weird part is this: at around +0.5 on the gate,
the drain current goes *way* up, apparently exponentially on gate
voltage.

So, can a PHEMT go into bipolar transistor mode when you forward bias
the gate? Anybody know anything about this?

John

 

[Jeff Stout]

Can you be a little more specific about the part number?

Jeff Stout

 

[gwhite]

We were experimenting with a Fujitsu PHEMT, a pseudomorphic
heterojunction gaas-type fet. They are poorly characterized for DC
specs (typical microwave part!) so we were making pulsed measurements
to get the DC transfer curves.

They are normally considered to be depletion parts, but it turns out
they enhance nicely, to about 1.4 * Idss with maybe 0.4 volt positive
gate voltage. But the weird part is this: at around +0.5 on the gate,
the drain current goes *way* up, apparently exponentially on gate
voltage.

Interesting. But square law is still exponential, so I don't know that much
changed aside from bleeding a bit of gate current off. Why wouldn't it still be
a channel device with square law response? (I mean at very small increases from
the .4 V you call out.)

I've had overdriven MESFET power circuits that had ended up with what I called
gate-leak bias since it was similar to the tube era grid-leak bias. The DC
power would go down with increased drive. Of course that was due to the whole
circuit since the gate bias circuit (in that case) could not source current to
cancel the leak bias.

So, can a PHEMT go into bipolar transistor mode when you forward bias
the gate? Anybody know anything about this?

I don't know. Pengelly includes gate voltages up to +0.6 in his MESFET
ruminations. I haven't looked close, and am in over my head on this one, but on
a quick persuse it seems like he thinks that a slight positive is within the
reasonable swing for large signal operation.

 

[John Larkin]

Can you be a little more specific about the part number?

Jeff Stout

It's a Fujitsu FSU02LG, really a nice part. The pulsed transfer curve
is almost a straight line above 20 mA, and the curves are amazingly
consistant from part to part. I can post it if anybody cares.

If I get time, I'm going to deliberately bang the gate into forward
conduction with a really fast pulse and see what the drain does.
Funny, nearly all the texts and databooks about PHEMTS stop at Vb = 0,
and ignore this very interesting positive-gate region.

Most of these kind of parts are specified for small-signal RF use,
where you sort of pour some RF into the gate and expect more to come
out of the drain. You're lucky if the datasheets tell you how to bias
them to do even that.

John

 

[Ken Smith]

They are normally considered to be depletion parts, but it turns out
they enhance nicely, to about 1.4 * Idss with maybe 0.4 volt positive
gate voltage.

This also happens with other depletion mode parts. A common JFET will
show a higher drain current if the gain is raise by a few tenths of a
volt.

But the weird part is this: at around +0.5 on the gate,
the drain current goes *way* up, apparently exponentially on gate
voltage.

Something very like this also happens on a JFET. If you forward biad the
gate junction, the drain current increases and the gate shows a neg.
resistance character. It works like a unijunction.