• WARNING: Tube/Valve amplifiers use potentially LETHAL HIGH VOLTAGES.
    Building, troubleshooting and testing of these amplifiers should only be
    performed by someone who is thoroughly familiar with
    the safety precautions around high voltages.

Infineon IPA50R140CP N-channel mosfet

By that , do you mean it's made effectively greater?

No!

Effective capacitance is Cg-s * (1 -
973de827210af64b64ed102673366f7a.png
)

I.e. the higher is the transconductance, the lower is the capacitance. However, 15 A/V is a large current transconductance. On the grid drive current it will be lower.
 

exeric

Member
2008-03-24 6:23 am
Perhaps you're right. I was thinking that as long as you could keep the Crss above that 40v it could be operated pretty linearly. Someone I respect thinks it is important for that to be linear. But perhaps I was putting to much emphasis on that and ignored the fact that the linearity comes in at a pretty high level, 40 volts, and goes up abruptly below that 40volts.

Do you have an n-channel mosfet that you recommend for a follower in that application?
 

Hearinspace

Member
Paid Member
2008-06-03 5:18 am
No!

Effective capacitance is Cg-s * (1 -
973de827210af64b64ed102673366f7a.png
)

I.e. the higher is the transconductance, the lower is the capacitance. However, 15 A/V is a large current transconductance. On the grid drive current it will be lower.

OK, that's something you taught me a few years ago but I couldn't hear the tone of voice in the first "however" and needed to check.

I still don't really understand the use of "Bootstrap". It seems to mean a few different things. Th word suggests pulling something up, though it often appears in ways that aren't that.
 
OK, that's something you taught me a few years ago but I couldn't hear the tone of voice in the first "however" and needed to check.

I still don't really understand the use of "Bootstrap". It seems to mean a few different things. Th word suggests pulling something up, though it often appears in ways that aren't that.

Yes, source is pulling the "lower leg" of the "capacitor" up, so if the follower was ideal, the resulting capacitance would be zero.
 
Some of the new C7 superjunctions become wildly nonlinear below 25V.
I have a non-audio use for that sort of capacitor (gate strapped OFF).
But you will have to stand a reserve of volts on the drain at all times
if you want to actually use that gate for your intended linear purpose.

I suggest you look at silicon carbide.
 
Last edited:
In a PowerDrive application the plate of the driver tube feeds the gate of a mosfet wired in source follower configuration. The drain of the mosfet follower is bypassed to ground, and can be considered grounded for the AC analysis case. There are the usual three capacitances involved with a mosfet.

The "input capacitance", Ciss is the capacitance from the gate to the source. In a perfect follower the source "follows" (is identical to) the gate, so any capacitance between these terminals wouldn't matter, since the charge in the capacitance would not change. Followers are not perfect, so there is some charge being moved in and out of this capacitance, but it is said to be "bootstrapped" by the device itself, and the Gm of the device is the measure of this ability.

The "output capacitance", Coss is the capacitance from drain to source. The Mosfet is constantly driving this capacitance. If the bypass cap on the drain is good, this capacitance appears like a capacitor to ground, and can form a low pass filter with the on resistance of the mosfet. We use mosfets in the few ohm range, so this is not usually an issue.

The "reverse transfer capacitance" Crss is the gate to drain capacitance, and since the drain is bypassed to ground, and the B+ supply to the driver tube is also bypassed to ground, this capacitance appears as a capacitance directly across the plate load on the driver tube. It forms a low pass pole with the parallel combination of the driver's Rp and the plate load resistance which can be pretty high if a CCS is used. We want this Crss to be as low as possible, but more important to remain constant with the change in applied voltage caused by the applied audio signal.

A voltage varying capacitance in parallel with the plate load can theoretically generate phase modulation, and phase distortion on high frequency audio signals. It is possible to generate PIM (Phase Intermodulation) in this manner, which is notoriously difficult to measure in small quantities, and this feeds the "all silicon in the signal path is evil" debates. I have found that it takes a pretty crappy mosfet to be audible, but I don't have "golden ears."

The Crss of the Infineon part mentioned goes up pretty sharply below 40 volts, so for it to be used in any follower in an audio application the region below 50 volts of so should be avoided under any signal condition.

In any case the current through the mosfet determines it Gm, so keep the current as high as practical and keep enough voltage across the mosfet to avoid the region where Crss is changing a lot.

The actual value of Crss is more important with a wimpy driver on low current, like a 12AX7 on less than 1mA that it is on a low Rp tube with 10 mA or so flowing through it.
 
A voltage varying capacitance in parallel with the plate load can theoretically generate phase modulation, and phase distortion on high frequency audio signals. It is possible to generate PIM (Phase Intermodulation) in this manner, which is notoriously difficult to measure in small quantities, and this feeds the "all silicon in the signal path is evil" debates. I have found that it takes a pretty crappy mosfet to be audible, but I don't have "golden ears."

Won't the same thing happen if the driving stage rp varies? I mean, even if the input capacitance of a stage is totally constant with signal, if rp of the previous stage varies with signal, PIM will occur when tones mix, right?
 
The varying Rp of the driver tube can cause a frequency response (and therefore phase response) that varies with signal level. We can not do much about it other than optimize the parameters that we can control.

I suppose it would be possible to design something that moved in an equal but opposite direction, to cancel this effect, but I'm not sure if the cure would be worse than the disease.
 
The varying Rp of the driver tube can cause a frequency response (and therefore phase response) that varies with signal level. We can not do much about it other than optimize the parameters that we can control.

I suppose it would be possible to design something that moved in an equal but opposite direction, to cancel this effect, but I'm not sure if the cure would be worse than the disease.

Yeah, we just have to be aware that the effect exists and choose a good tube and operating point.
 

exeric

Member
2008-03-24 6:23 am
I've been thinking about this thread a lot ever since I read Tubelab's excellent explanation of the important parameters in choosing a FET for a source follower in a PowerDrive application. I think I may have just lucked out in singleing out the Infineon because there were so many things I realize I didn't know before reading that explanation. I think the Infineon actually might be a good FET to use if the circuit topology fits it.

I realize though that it won't suit for my purposes because I'm try to build a circuit using minimal drain source voltages to minimize power dissipation. The power drive is only going to drive EL84s or 6V6s. So I'm going to be using about +50 and - 50v between source and drain.

I've found one that may work well. http://www.mouser.com/ds/2/408/TK2Q60D-482885.pdf It looks kind of bad on the chart for Crss but that's only because the scaling is so conservative. It looks to be about +- .5 pf from 11 to 100 volts DS. That's great. I just have to make sure that DS voltage never goes below 11volts. Thanks Tubelab for the great explanation.