ampage Tube Amps / Music Electronics	For current discussions, please visit Music Electronics Forum.

ampage archive

Vintage threads from the first ten years

previous: anon R.G.,I'd like to st... -- 1126260902

Re: mosfet in place of triode

9/9/2005 10:56 AM
R.G.	Re: mosfet in place of triode [QUOTE]The zener diode implanted into most mosfets is parallel to drain-source. Just checked the IRF820A data again. AFAIK these parts are commonly used to switch inductive loads. This diode is there to protect drain-source from overvoltage and reverse polarity.[/QUOTE] That's not an extra zener. It's an inherent part of the channel isolation on the chip. Making a power MOSFET without one of those is not practical. It's also called the "body diode". It's purely luck that it happens to be the right place and right orientation to protect the active device from inductive kick. The nature of this diode is sometimes not fast enough for protection in some fast switching circuits and it must be supplemented by external fast response diodes. [QUOTE]Once the SF can't sink current anymore the source voltage becomes positive in relative to the gate voltage and the diode will forward conduct... and clip the previous stage's output. Not so with a CF, not enough voltage for arcing ![/QUOTE] That's true - so the sink speed is the 100K in parallel with the driving 12AX7. The AX7 plate resistance limits this to about 60K or so additional sink under massive overdrive down situations. It's not a big change. [QUOTE]Thus the tonal change will be due to: 1) more parasitic gate drain capacity to drive for the previous stage[/QUOTE] The effective Cgd is reduced by the current feedback in the follower configuration. The driving triode doesn't see Cgd as though it's a discrete cap. [QUOTE]2) MosFet can source FAR more current and won't clip the upper half wave[/QUOTE] Very true. [QUOTE]3) for the negative half wave the previous stage's plate voltage will be clipped by the additional protection diode. In case of a CF the previous stage notices this only due to tiny parasitic grid cathode capacitances.[/QUOTE] This is a factor only if the driving signal is fast enough to cause slew limiting of the capacitances in the following stage with a 100K follower resistor. In any case, the clipping can be removed by reducing the follower resistor to get back to whatever level of slewing is needed to not turn on the reverse zener. If that's desirable, and if it ever in fact happens. I'd have to do some modelling and real world testing to see it really happening. [QUOTE]4) Bootstrap for medium to high frequencies due to gate-source capacity, which is in the range of several 100 pF. Parasitic capacities of ECC83 and the like are far below 10pF.[/QUOTE] As I pointed out, the effective capacitances seen by the driver are significantly smaller than the raw Cgd. [QUOTE]Conclusion: There are differences in case of overdriven followers with low impedance load such as a Marshall tone stack. Remember: you drive 33k with mids on 0 from a stage with 100k large signal output impedance.[/QUOTE] I think I'd say it another way: There is the possibility of some differences if you actually have situations where the follower is overdriven. In such cases, you may have to adjust the source resistor to get a similar response, depending the capacitances in the tone stack. [QUOTE]But in case of the Marshall distorion circuit the additional protection diode is a must and I would perform extensive listening tests before designing a pcb for it....[/QUOTE] So would I. But then I'd always do extensive listening tests before committing a circuit to PCB anyway. Seriosly, lots of folks have used this in many amps, and I have yet to hear anything other than either "no change in tone" or "hey! that sounds better".

9/9/2005 10:56 AM

R.G.

Re: mosfet in place of triode

[QUOTE]The zener diode implanted into most mosfets
is parallel to drain-source. Just checked the
IRF820A data again. AFAIK these parts
are commonly used to switch inductive loads.
This diode is there to protect drain-source
from overvoltage and reverse polarity.[/QUOTE]
That's not an extra zener. It's an inherent part of the channel isolation on the chip. Making a power MOSFET without one of those is not practical. It's also called the "body diode". It's purely luck that it happens to be the right place and right orientation to protect the active device from inductive kick. The nature of this diode is sometimes not fast enough for protection in some fast switching circuits and it must be supplemented by external fast response diodes.

[QUOTE]Once the SF can't sink current anymore the source
voltage becomes positive in relative to the
gate voltage and the diode will forward conduct...
and clip the previous stage's output.
Not so with a CF, not enough voltage for arcing ![/QUOTE]
That's true - so the sink speed is the 100K in parallel with the driving 12AX7. The AX7 plate resistance limits this to about 60K or so additional sink under massive overdrive down situations. It's not a big change.

[QUOTE]Thus the tonal change will be due to:
1) more parasitic gate drain capacity to drive
for the previous stage[/QUOTE]
The effective Cgd is reduced by the current feedback in the follower configuration. The driving triode doesn't see Cgd as though it's a discrete cap.

[QUOTE]2) MosFet can source FAR more current and
won't clip the upper half wave[/QUOTE]
Very true.

[QUOTE]3) for the negative half wave the previous stage's
plate voltage will be clipped by the additional
protection diode. In case of a CF the previous
stage notices this only due to tiny parasitic
grid cathode capacitances.[/QUOTE]
This is a factor only if the driving signal is fast enough to cause slew limiting of the capacitances in the following stage with a 100K follower resistor. In any case, the clipping can be removed by reducing the follower resistor to get back to whatever level of slewing is needed to not turn on the reverse zener. If that's desirable, and if it ever in fact happens. I'd have to do some modelling and real world testing to see it really happening.

[QUOTE]4) Bootstrap for medium to high frequencies
due to gate-source capacity, which is in
the range of several 100 pF.
Parasitic capacities of ECC83 and the like
are far below 10pF.[/QUOTE]
As I pointed out, the effective capacitances seen by the driver are significantly smaller than the raw Cgd.
[QUOTE]Conclusion:
There are differences in case of overdriven
followers with low impedance load such
as a Marshall tone stack.
Remember: you drive 33k with mids on 0 from
a stage with 100k large signal output impedance.[/QUOTE]
I think I'd say it another way: There is the possibility of some differences if you actually have situations where the follower is overdriven. In such cases, you may have to adjust the source resistor to get a similar response, depending the capacitances in the tone stack.

[QUOTE]But in case of the Marshall distorion circuit
the additional protection diode is a must and
I would perform extensive listening tests before
designing a pcb for it....[/QUOTE]
So would I. But then I'd always do extensive listening tests before committing a circuit to PCB anyway.

Seriosly, lots of folks have used this in many amps, and I have yet to hear anything other than either "no change in tone" or "hey! that sounds better".