I'm looking at building a folded cascode power amp that uses smallish Fairchild MOSFETs as the cascode transistors, with BJT for everything else. FQP4N20L is the NMOS and FQP3P20 is the PMOS.
Anyone using those as VAS transistors? Or cascodes?
Anyone using those as VAS transistors? Or cascodes?
Hi Russell,
I've got a very positive experience in using IRF610/9610 MOSFETs as VAS - see here >CF-FET V2.0 front-end - going high-tech (SMD)<
Fairchild ones have got input capacitance (Ciss) around twice as much... just something to keep in mind. Otherwise - should be fine. I don't have the model for these ones, so difficult to say what they will show in terms of overall linearity, but if you already have them - worth trying 😉
Cheers,
Valery
Linearity should be fine, since it will have fixed current bias, around 6 mA, maybe 10mA. I'll look at the IRF, but that's a part I used to use over 20 years ago. Surely, a more modern part will have a higher Gm/Ciss figure of merit.
OK, here's what I found:
Qtotal Ciss Coss Crss Gm
IRF610 8.2nC 140p 53p 15p 0.80 @ 2A
FQP4N20L 4.0nC 240p 36p 6p 1.42 @ 0.245A
Of those, Ciss is the least important in a cascode device, because the R in that RC constant is 1/Gm. Coss+Crss load down the drain node.
I've been playing with the small-signal models, and FETs potentially give much higher overall gain when used as cascode devices, compared to BJTs.
The missing small signal parameter is Ro. VA or lambda or just well-drawn curve-traces, I could determine lambda. I want to compare beta*Ro of the BJT to (Gm*Ro_casc)Ro_mirror for the FET case. If the FET answer is a lot higher than the BJT, and all that C doesn't ruin GBW, I could be on to something. I'll see what LTspice thinks.
Thanks for the reply.
--Russell
OK, here's what I found:
Qtotal Ciss Coss Crss Gm
IRF610 8.2nC 140p 53p 15p 0.80 @ 2A
FQP4N20L 4.0nC 240p 36p 6p 1.42 @ 0.245A
Of those, Ciss is the least important in a cascode device, because the R in that RC constant is 1/Gm. Coss+Crss load down the drain node.
I've been playing with the small-signal models, and FETs potentially give much higher overall gain when used as cascode devices, compared to BJTs.
The missing small signal parameter is Ro. VA or lambda or just well-drawn curve-traces, I could determine lambda. I want to compare beta*Ro of the BJT to (Gm*Ro_casc)Ro_mirror for the FET case. If the FET answer is a lot higher than the BJT, and all that C doesn't ruin GBW, I could be on to something. I'll see what LTspice thinks.
Thanks for the reply.
--Russell
A 2N5551, for comparison, has something like 10 pF total C, < 1/20th as much as the FETs. DC gain is no good to me if I don't also have high UGF (GBW). I bet I end up with bipolar, but I'll still do the investigation.
- Status
- Not open for further replies.