Possibility of Tube substitution trough MOSFETs

[tiefbassuebertr]

Can I replace in generall power tubes (6550, EL34, 6C33C) trough high voltage mosfets arround Uds 1000V?
If yes, what value of resistance I must choise for the transformator windings (ultralinear, PPP) and what value of resistance I must choise for the source resistor to create the internal tube resistance (OTL)?
Because the value of source resistance is much higher compare to typical solid state amps, there are the possibility to use many parallel devices.
Has anyone this one already tested?
Thank you very much for your advices.

[Wavebourn]

MOSFETs have huge input capacitances, thermal dependencies, so no direct replacement is possible, but you can build a "tube" that has pentode characteristics, gm of amperes per volt, and couple of hundred watt of dissipation. Here is my example (obsoleted already, so I post it here):

http://wavebourn.com/forum/download.php?id=137&f=7

[kenpeter]

This is a big subject: And the only easy answer I know revolves
around faking the internal feedback mechanism of the Triode.
Sometimes the best way to "fake" it is with a small triode as
both the input buffer for high impedance and the reference
model for Mu.

If you were to place the base to emitter junction of a bipolar
transistor in series with your reference Triode's cathode, you
have a Triodlington. Current is multiplied roughly by Beta, but
plate voltage is still controlled almost entirely by the internal
feedback mechanism (Mu) of the tube. If plate and collector
are strapped together, the transistor becomes slave to this
feedback.

You can do the same with a MOSFET, but need to add one
resistor across the Gate to Drain.

Now, if you really wanna fake a Pentode, beam power type,
for UL or whatever... I don't know if enough voltage feedback
gets through a normal pentode's plate resistance to abuse a
small one as a feedback reference. I got a feeling that straight
up Pentodlington based on a Pentode wouldn't work..

Might just maybe could fake your way there with a Cascode
of two Triodlington type emulators???

[Wavebourn]

Peter;
it sounds nice theoretically, but practically triode with BJT or MOSFET is not practical due to huge resulting Miller capacitance. Also, a simple resistor between gate and source does not work because of thermal dependence and much higher forward transconductance of MOSFET VS tube. That's why I had to use a pentode and sophisticated thermal compensation through gyrator.

[tiefbassuebertr]

Thanks for your first comments

Originally posted by Wavebourn
MOSFETs have huge input capacitances, thermal dependencies, so no direct replacement is possible, but you can build a "tube" that has pentode characteristics, gm of amperes per volt, and couple of hundred watt of dissipation. Here is my example (obsoleted already, so I post it here):

http://wavebourn.com/forum/download.php?id=137&f=7

The high capacity (additional non-linear) was for me the main reason MOSFETs as replacement for tubes to use. However, only in a CSPP topology (Circlotron) and only as a source follower (gain less than 1). The whole thing should be placed outside a NFB loop (very important to prevent sawtooth waveforms and TIM).

In a simulation I get very good THD results (see PDF) - If I would have the time, I would build such a power amp device - I'm almost sure that it is possible also in real life excellent sonic results. Maybe another of the members perhaps have time to build such a part - but it should not be a beginner - 2x 400 VDC are very danger.

In the first step, I recommend a voltage from 2x50V to use. The power output amounts now only 1/32, but is enough for checking the sound quality.

At bottom the results of my idea - but please note, that by this simulation all capacitors and resistors so as the voltage source (power supply) and input source (signal generator) are idealized (i. e. no internal resistance by the power supply). In additional there are no lead respective wire inductance and spread capacitance through PCB layout.
Only the effects of transistor spice models so as the pure circuit there are to see.

Idea for CSPP 2x400V, idle power: 13x80mA = 832W, input=18V, output=10V
1) schematic
2) frequency response
3) THD 10 KHz (lin) third harmonic 0,73mV
4) THD 1 MHz (lin) third harmonic 2 mV
5) THD 1 MHz (log)
6) THD 10 KHz (log)

[Wavebourn]

I would call it Super-cyclotron. Good luck!

[kenpeter]

Originally posted by Wavebourn
Peter;
it sounds nice theoretically, but practically triode with BJT or MOSFET is not practical due to huge resulting Miller capacitance. Also, a simple resistor between gate and source does not work because of thermal dependence and much higher forward transconductance of MOSFET VS tube. That's why I had to use a pentode and sophisticated thermal compensation through gyrator.

Where in either of these circuits is there an increase in Miller?
There is definately a huge increase in transconductance. And
if transconductance is the limiting factor, thermals will make
a big difference. But if Mu is the limiting factor...

Matters little the thermals or linearity of either BJT or MOSFET.
The tube is watching the voltage of both parallel anodes, and
applying the rule of Mu. Neither type of sand likely to run away
or be non-linear when locked down by 1/(Mu-1).

The Pentode model would be a cascode of Triodlings, possibly
with different mu needed for screen and G1. That requirement
alone might make the circuit impractical...

[tiefbassuebertr]

Originally posted by Wavebourn
I would call it Super-cyclotron. Good luck!

regarded the parallel devices - this approach is also to find by "Tim de Paravicini's" V20 (V 20) - go to
http://www.stereophile.com/integratedamps/1099ear/
and
http://www.envogue-24.de/Neue%20Seit.../V20vollv.html
for more information
The best tube amplifier that I ever heard. Especially for tweeter driving in bi- resp. multi-amping systems.
In opposite to most other companies use EAR even in this case preamplifier tubes. Translated to solid state amp technology means this a lot parallel small signal devices as 2N3904 or 2N2222 for creating a discrete multi emitter power device (e.g. cluster of 30x30 devices, perhaps in surface mounted design (SMD).

[Wavebourn]

Peter;

Tube is watching, but applies nothing when loaded on an output transformer, compared to what BJT and MOSFET do with temperature.

However, if you intend to use your tubes as cathode followers, you may forget of both Miller and thermal stability, of course. Otherwise it would not be practical because of Miller capacitance, and would not live long due to thermal dependence of mostly Vbe and Vgs on temperature.

Peter, I built a working prototype of such an amp based on the picture I've inserted in my earlier post. Push - pull, 200W nominal output power.

[analog_sa]

Originally posted by tiefbassuebertr

The best tube amplifier that I ever heard.

Really? I always thought this amp was meant as a joke. Fwiw i always try to avoid parallel devices everywhere: preamps, power amps, solid state or tube. Every time i've tried paralleling i have ended up regretting it.