Any hot JFETs other than NE3508?

[Joerg]

John Larkin recently pointed out the NE3508 and NE3509. They are really
nice, steep Vgs versus Id and so on. Even better than the BF862 except
for one major obstacle: At >>50c they are way out of league for many
designs. High drool factor, followed by a depressed look at the BOM budget.

Question: Are there similar hotrod JFETs that are in the 10-20c range?

 

[John Larkin]

John Larkin recently pointed out the NE3508 and NE3509. They are really
nice, steep Vgs versus Id and so on. Even better than the BF862 except
for one major obstacle: At >>50c they are way out of league for many
designs. High drool factor, followed by a depressed look at the BOM budget.

Question: Are there similar hotrod JFETs that are in the 10-20c range?

The BF862 is a rusty old tractor compared to the NEC hj parts. The
hjfets have roughly 10 times the transconductance and 10 times less
capacitance than the jfet, which is a 100:1 gain-bandwidth advantage.
The Rds-on ratio is, like, 20:1.

Also, in my experience, the hj fets are much more repeatable than
jfets. The Idss of the 862 is spec'd at 10-25 mA, which is actually
good for a jfet. Spec-sheet ranges of Idss are 10:1 for some jfets!

In my business, fast time-domain stuff, we don't use jfets. Their
poorly specified dc characteristics, and their low gain, make them
pretty much useless, especially since I can buy 1.8 GHz opamps and 10
GHz mmics nowadays.

John

 

[Joerg]

The BF862 is a rusty old tractor compared to the NEC hj parts. The
hjfets have roughly 10 times the transconductance and 10 times less
capacitance than the jfet, which is a 100:1 gain-bandwidth advantage.
The Rds-on ratio is, like, 20:1.

Also, in my experience, the hj fets are much more repeatable than
jfets. The Idss of the 862 is spec'd at 10-25 mA, which is actually
good for a jfet. Spec-sheet ranges of Idss are 10:1 for some jfets!

In my business, fast time-domain stuff, we don't use jfets. Their
poorly specified dc characteristics, and their low gain, make them
pretty much useless, especially since I can buy 1.8 GHz opamps and 10
GHz mmics nowadays.

I was thinking more about oscillators that must work with really low
supply voltage, a few hundred millivolts. The high gain of the NEC parts
is really beneficial there while the speed would have to be muffled big
time so they remain stable. Pricing is a serious problem though.

 

[John Larkin]

I was thinking more about oscillators that must work with really low
supply voltage, a few hundred millivolts. The high gain of the NEC parts
is really beneficial there while the speed would have to be muffled big
time so they remain stable. Pricing is a serious problem though.

Jfets are good for HF front-ends and for oscillators, ald for
high-impedance instrumentation. The hj fets are noisy at low
frequencies, especially gate current noise. I don't know if they have
the "dispersion" problem that mesfets typically have, namely changes
in Gm at sub-MHz frequencies, or if they have the trapping-state
problems that mesfets have.

I'm trying to do a "pin driver" that will swing from 0.25 to 5 volts
p-p, with +-5 volt programmable offset, and produce an arguably square
wave from dc to 1 GHz. This isn't a very cost-sensitive application.

John

 

[Uwe Bonnes]

In sci.electronics.components John Larkin

The BF862 is a rusty old tractor compared to the NEC hj parts. The
hjfets have roughly 10 times the transconductance and 10 times less
capacitance than the jfet, which is a 100:1 gain-bandwidth advantage.
The Rds-on ratio is, like, 20:1.

However the Gate leckage current isn't specified, with makes it hard to see
if the part is usefull for high impedance work, like charge sensitive
amplifiers.

 

[John Larkin]

In sci.electronics.components John Larkin


However the Gate leckage current isn't specified, with makes it hard to see
if the part is usefull for high impedance work, like charge sensitive
amplifiers.

Oh, the gates leak microamps. But multi-GHz signals don't exist at
high impedances, so it's OK.

Properly tuned, uncooled, the 3509 has a noise figure of 0.4 dB at 2
GHz, about 29 Kelvins. Try that with a jfet!

John