Yet another N-Channel Amplifier

[tschrama]

Maybe r50 330Ohm, r37 100Ohm are a bit too low?

[MikeB]

hmm, i suggest following changes:
- reduce c10 to ~20pF, and/or add cdom of ~47pF to q1.
- add 1k to the bases of the 2 ccs-bjts. and/or small cap 1-10nF paralles to the diodes
- add Re's to the diffamp.

These are my changes from experiences, not any simus / calcs...
My experience is, that if an amp is slightly oscillating, the feedbackcap might be too big.

BTW, are you always making that nice pcb's just for prototyping ?
Is it handdrawn ?

Mike

[f4bok]

Hi pelle
Q1 has no resistor in its emitter, gain must be enormous in de negative way....The voltage on Q22 collector cant get higher than 0.6 up to the -V rail.....
Do i think right ??
Rgds

[MikeB]

hi f4bok,principally yes, i would use 75ohm Re for q1.
But, the voltage at q22-c can get much higher than 0.6v relative to V-,
it can't get much lower... (Vce-sat), but low enough to close q1.
if Q22 closes a little, voltage jumps very high, due to "constant" current
from diffamp-collector. But the 67uA into Q1 might be to much for
the currentmirror ?

[richie00boy]

Remove the C10 (across the feedback resistor) and instead use a cap of 22pF-100pF across Q1 (VAS transistor) collector and base. This may help with your oscillation.

Nice PCB, much more compact than my effort at a similar design.

[f4bok]

Hi MikeB
At C of Q22, there is the base of Q1 and emitter of Q1 is at -V rail
and base-emitter is a diode (0.6V) it works in commutation mode no ? just with H11 ?
I just feel there is something wrong there...it does not work with Ohm law

rgds

[MikeB]

hi f4bok, that's right. The operational range is only a few millivolts.
If V at Q1b get's much higher than 0.6v, the amp clips. That's normal
for this kind of circuit. And yes, the gain is enormous. Q1 is the part
in this amplifier that defines the VAS. It does the main job of amplifying.
I think the gain from q1 is at least 1:5000.
Don't forget that transistors have an internal resistance, so Vbe
is not exact 0.6v, it increases with the current rising. Without this
resistance, the gain would be infinite without Re. This would be
a nice world if transistors had a constant Vbe !

[f4bok]

Hi mikeB
Right with you, for internal resistance, that's what i called H1.1
I think pelle shoud add this res in Q1 emitter to solve these many problems of instabillity
rgds

[mikeks]

Originally posted by Pelle
Hi!

I would like some comments of my implementation on a N-channel amplifier.

..a rather odd example of all-n-channel design here:

[Lars Clausen]

mikeks: Thankyou for posting the Siliconix design. It's not so crazy at all. Of course to build it you would have to modernize
the parts list a bit ;-)

The good news it has virtually no rail loss, and equal positive and negative slope speeds. For my taste it has too high feedback, but i guess for PA use where the sound quality is not of extreme importance, or as a subwoofer amplifier, this design definitely has it's merits.

As i tell my employees every once in a while: just because something is old, doesn't have to mean it is useless