Several suggestions were posted about my initial question about the Sziklai transconductance.
mirlo's suggestion: GMsz = Afet * GMbjt + GMfet in (#4 &) #6: -> What is Afet?
MarkJohnson has a 'datapoint' in #8: GMsz = 207.6mA/V -> How to calculate this datapoint?
knutn states in #9: GMsz = beta * GMfet -> how does the transconductance and the beta relate, just 'as is'?
steveu comes in #23 with: 16mA/V with a circuit where the Rsz is 10k, driving the J-fet next near to pinch off (the most distorting area).
knutn has a RIAA circuit in #24 that performs fine, the K170/556B prooves a good amplifier stage...
jsdyson wonders in #30 about the very low Id setting (40μA) with "well constrained distortion and useful freq response".
This last 'low Id' contraption is already found in the late 70's Sony TA-E86B preamplifier, it's odd 'bass boost' circuit intended for some audience.
The J-fet (Vdsmax= 12V!) runs at some 60μA without problems. I don't use that output actually.
The redux is about the assumption of a high beta, so the current through Rsz is determining the transconductance of the bjt (aka 'voltage driven' instead of 'current driven').
In this case the Sziklai-transconductance boils down to a simple equation, alongside the fet-bjt also valid for a bjt-bjt combo.
Nevertheless is knutn's remark in #9 and steveu's circuit with a 10k still in debate: what if there's no high beta and/or a very high Rsz, pressing the fet into performance jeopardy?
Comments & corrections are welcome.
(I use "S" for "GM"; S^ = Smax)
mirlo's suggestion: GMsz = Afet * GMbjt + GMfet in (#4 &) #6: -> What is Afet?
MarkJohnson has a 'datapoint' in #8: GMsz = 207.6mA/V -> How to calculate this datapoint?
knutn states in #9: GMsz = beta * GMfet -> how does the transconductance and the beta relate, just 'as is'?
steveu comes in #23 with: 16mA/V with a circuit where the Rsz is 10k, driving the J-fet next near to pinch off (the most distorting area).
knutn has a RIAA circuit in #24 that performs fine, the K170/556B prooves a good amplifier stage...
jsdyson wonders in #30 about the very low Id setting (40μA) with "well constrained distortion and useful freq response".
This last 'low Id' contraption is already found in the late 70's Sony TA-E86B preamplifier, it's odd 'bass boost' circuit intended for some audience.
The J-fet (Vdsmax= 12V!) runs at some 60μA without problems. I don't use that output actually.
The redux is about the assumption of a high beta, so the current through Rsz is determining the transconductance of the bjt (aka 'voltage driven' instead of 'current driven').
In this case the Sziklai-transconductance boils down to a simple equation, alongside the fet-bjt also valid for a bjt-bjt combo.
Nevertheless is knutn's remark in #9 and steveu's circuit with a 10k still in debate: what if there's no high beta and/or a very high Rsz, pressing the fet into performance jeopardy?
Comments & corrections are welcome.
(I use "S" for "GM"; S^ = Smax)
Attachments
Thanks for the clean century.
It took me a full month to realize that the actual 'Sziklai Transconductance Amplification Ratio' ("STAR") is in the very second part of the formula: ... * Rsz * (Ic^/Vt), where the latter is the reciprocal of the internal emitter impedance ('re') or the 'thermal imedance'.
So the transconductance incremental secret of the Sziklai is: ... Rsz / re , alike a current mirror with external emitter resistors.
Jfet: Ssz = Sj * Rsz/re
BJT: Ssz = Sq * Rsz/re
I'm from the pre-MOS era, so I guess...
MOS: Ssz = Sm * Rsz/re
Cheerio,
dW [π°]
It took me a full month to realize that the actual 'Sziklai Transconductance Amplification Ratio' ("STAR") is in the very second part of the formula: ... * Rsz * (Ic^/Vt), where the latter is the reciprocal of the internal emitter impedance ('re') or the 'thermal imedance'.
So the transconductance incremental secret of the Sziklai is: ... Rsz / re , alike a current mirror with external emitter resistors.
Jfet: Ssz = Sj * Rsz/re
BJT: Ssz = Sq * Rsz/re
I'm from the pre-MOS era, so I guess...
MOS: Ssz = Sm * Rsz/re
Cheerio,
dW [π°]
Attachments
The problem with the sziklai pair is the phase input to output at high freqencys. With 2 transistors making phase shift there is much more risk for self oscillation. You will have to decompensate the amplifier to a lower frequency cutoff. That means lower feedback and more distortion.
That is the reason you never see the sziklai pairs in the ultra low distortion amplifiers here.
That is the reason you never see the sziklai pairs in the ultra low distortion amplifiers here.
typo's rule...
Thanks for this other viewpoint. Worth another thread.
A last simplification:
Ssz = Rsz/re * Sf (re = Vt/Ic)
or
Ssz = Rsz/rs * Ic/Vt
Cheerio, Mars