How is the thermal stability of this biasing scheme with a normal N+P-channel MOSFET follower (like M2)? I tested a simplified version of this today with a JFET CCS instead of the precision CCS you suggest and I couldn't get the bias to stabilize. As the output FETs warmed up, bias just kept creeping upwards. I tried incorporating thermal feedback with an NTC, which helped slow the runaway, but did not entirely fix the problem.
it works flawlessly , with all of these arrangements in OS:
NMos + PMos
(small mosfet in front of) NSIT + PMos
(small mosfet in front of cascoded) NJFet + PMos
NMos + (small mosfet in front of) Germanium
NMos + (small mosfet in front of) Big Fat bjt
NMos + (small mosfet in front of) Schaded PMos
(small mosfet in front of) Big Fat N bjt + (small mosfet in front of) Big Fat P bjt
/////////////////////////////////////////////////////////////////////////////////
you can't use JFet CCS there - too much TempCo; try with plain resistor - it'll work tons better than JFet
current is, well, minuscule - you need great stability in temperature domain, and - as I chose set of compromises, also in voltage domain
already said that, for my knowledge ( seen some simplified schematic blocks) - Papa is using resistors for same purpose ....... ZM Chicken for that
NMos + PMos
(small mosfet in front of) NSIT + PMos
(small mosfet in front of cascoded) NJFet + PMos
NMos + (small mosfet in front of) Germanium
NMos + (small mosfet in front of) Big Fat bjt
NMos + (small mosfet in front of) Schaded PMos
(small mosfet in front of) Big Fat N bjt + (small mosfet in front of) Big Fat P bjt
/////////////////////////////////////////////////////////////////////////////////
you can't use JFet CCS there - too much TempCo; try with plain resistor - it'll work tons better than JFet
current is, well, minuscule - you need great stability in temperature domain, and - as I chose set of compromises, also in voltage domain
already said that, for my knowledge ( seen some simplified schematic blocks) - Papa is using resistors for same purpose ....... ZM Chicken for that
one detail more - me being all rage about that small P Channel mosfet for role as level shifter - really not a problem when xformer gain FE stage used - modulation envelope well greater than rails......
search for mosfet in case is for amps having active FE, and I don't want introducing elevated rails for FE, just for sake of "missing 2V"
in fact . missing 2V from voltage envelope from bottom side means 4Vpp missing overall ......
OK, only you know that for sure ...... though , as I said - JFet as CCS is simply not good enough
I was thinking of some other solutions ( except resistor, yeah) but in the end, didn't found anything worth fussing, when ready made solution is available and, well, cheap in overall picture
I forgot to say - it works even with (small mosfet in front of) NSIT + PMos Puck
didn't tried exactly same for two big pucks, have different solution for that
Mr. Pass has written about the linearity of BJT vs linearity of MOSFET. MOSFET
gain increases ~forever. BJT gain increases to a peak and then decreases. This causes different harmonics for BJT vs MOSFET.
@ Itsmee
only that way - see sketches - R+ and R- reasonably symmetric/same values
in short - only if you have two mosfet gates at appropriate ends of R+ and R-, you can shape symmetrical/same behavior in both static and dynamic realms
use anything else than mosfet, expect either predictive or non-predictive control electrode (that being gate or base) current, plus different control voltage (mosfet vs. anything else) ...... so long symmetrical behavior of circuit
only that way - see sketches - R+ and R- reasonably symmetric/same values
in short - only if you have two mosfet gates at appropriate ends of R+ and R-, you can shape symmetrical/same behavior in both static and dynamic realms
use anything else than mosfet, expect either predictive or non-predictive control electrode (that being gate or base) current, plus different control voltage (mosfet vs. anything else) ...... so long symmetrical behavior of circuit
Last edited:
I was writing in "small mosfet" related musings strictly about mosfet used as source follower/level shifter
push enough current through, it is invisible in circuit per se, while doing everything I want it to do
regarding general contemplations about mos vs. bjt and vice versa, that's hardly a topic here
I am using bjt(s) in some of actual amps (OS), deliberately ..... and we'll see what result is going to be
stay tuned ( been listening for few days Ikebana Mono T-Bed-ed blasphemy amp - 2SA/2SC in output ......... it sung ....... after proper stereo test . I'll know was it just fun time, or is it worthy having a proper name) ........
) , so - please - when you see that I have them, be alert
preliminary and still not-so-detailed string of information from my Official Second set of Ears..... I got Goodness Approval
compared with reference (SissySIT) - no drawbacks, same character, little more detailed, Foot-tapping factor not disturbed
good enough for me
Attachments
I did keep the picture for my wife... my desk is quite clear compared to this one...Now I see why ZM was photographing my bench when he was here.
She will never again ask me for wiping this out!!