TSE Bias and MOSFET

[cddeluca]

I've been trying to find some further information about the biasing and MOSFET implementation of the TSE and have struck out. Can anyone point me to some discussion or clarify the design basis? Is the bias arrangement to limit current at 150V? The particular values are arbitrary? And with the MOSFET between the bias network and the grid, does that mean that you're essentially biasing the drive voltage from the 5842 rather than pulling the output tube grid down? Very confused.

Also the MOSFET source resistor (bias resistor?); how is that determined? The 'Power Drive Cookbook' has quite a discussion of SE Amp CAD but it doesn't address how to translate the results into the MOSFET implementation.

Any info is greatly appreciated! Thank you.
Charlie

 

[cddeluca]

Some Follow-Up

I think i found two answers on a guitar amp forum

-setting the bias on the gate leg of the mosfet is the same as biasing the tube grid directly minus 3-4 volts since that is the (fixed?) difference between the gate and source of the mosfet

-the bias resistor arrangement seems to be based on negligible power dissipation as to not 'steal' amps from the circuit; micro-amp currents

Has anyone any hints about the mosfet biasing question? THANKS!

 

[cddeluca]

Follow-Up #2

I found most of the rest of the answer here:
http://www.diyaudio.com/forums/tubelab/161524-understanding-tubelab-se.html

It discusses biasing the mosfet to "a reasonable" current within it's linear region. What's linear? Looking at the datasheet for the 2SK3565 Vds is constant for Vgs greater than 8V, which is easily satisfied in the TSE. Rds appears constant for Id < 1A. Since the grid current will be very small in any case, biasing the FET somewhere between 0.01 and 1A would be sufficient, at the expense of dissipation. So it seems that 50mA of bias might be good. With Vdd about 400V and Vb at around -150 then a 20K resistor would bias the mosfet at ~30mA, so I guess that's how it's done.

I would still appreciate anybody's take on this if you're so moved. Thanks!

 

[cddeluca]

I SHOULD have said

I found most of the rest of the answer here:
http://www.diyaudio.com/forums/tubelab/161524-understanding-tubelab-se.html
With Vdd about 400V and Vb at around -150 then a 20K resistor would bias the mosfet at ~30mA, so I guess that's how it's done.

With the grid bias at -70V and the bias supply at -150V the mosfet bias is ~3mA

 

[boywonder]

It discusses biasing the mosfet to "a reasonable" current within it's linear region. What's linear?

I would still appreciate anybody's take on this if you're so moved. Thanks!

I'm no expert here, but I believe the parameter of interest is Crss. IIRC, a low Crss is good, but trading the absolute value for a nice flat curve is worth it. You want to bias the mosfet so that it's on the relatively "flat" part of the Crss curve.

Here is a related thread

http://www.diyaudio.com/forums/tubelab/181369-mosfets-tubelab-se.html

 

[cddeluca]

Thanks, boywonder

I appreciate the link - the Crss thing seems to be minimized for Vds >100 in the 2SK3565 that I've bought. Can't find anything to relate it to bias current, though.