JFETs as switch in 2-channel tube amp

[muadib25]

Originally Posted by counter culture

J111 & J112 are ideally suited to FET switching. If the source is treated as the input then it should be strapped to the gate using ~22k to ensure that Vgs remains @ 0V in the presence of an input signal. These FETs are apparently symmetrical.

Rather than try to reiterate the entire section from the book, I would suggest reading pages 408-421 of Small Signal Audio Design by Douglas Self which contain a comprehensive treatment of this subject.

Counterculture, thank you! It's added to my reading list, I am sure to cover a lot of things I am missing. As for the FETs I am using here, they are 2N5638, they belong to the "switches and choppers" class, same as J111, J112, J113 etc. Their ON resistance is at around 10-20 Ohm.

On the circuit I have the Drain facing the signal input and Source facing the signal output. But yes, these FETs are symmetrical.

Originally Posted by Mooly

You must drive the FET's via a resistor. The diode leaves the gate floating which is no good. Make sure that -14 volts is within the G-S rating of your chosen FET.

Now I get it!
My misunderstanding here was that from 0V and upwards(+) the JFET's resistance would not change, hence my proposal of using a diode.

So, if I get it right, when the JFET is ON (Vc=0V) the resistor will be conducting the necessary offset to the gate.

When the JFET is OFF (Vc=-13) there will be some of the gate's negative voltage present... Won't I be needing some caps after each JFET for not having DC injected to the signal?

Here is the circuit with the 22k resistors added:

[Wavebourn]

22k is too low resistance. Try something like 2.2 mega ohm or as high as possible, to lower loudness of Thump when switching. This resistor is needed to discharge gate charge, but unfortunately it provides parasitic path for control signal to output. T-configuration as I suggested before would be better in terms of Thumps.

[Mooly]

Lets think about this... and I would recommend you try the design before building for real.

FET 1 and FET 2. Is the feed from V1B AC coupled such that this pont is at zero volts ? If so then both these FET's are tied to ground via the 2M2 and the unmarked resistor. The other side of each FET is AC coupled to the next stage and ground referenced again. So far so good. The gates can be pulled low (to a negative voltage) via the diodes.

FET 5 and 6. Not ground referenced on the output side as drawn ? Where does R19 go ? I'm guessing like the other FET's. That still leaves no ground reference. Add a high value resistor from the output of these FET's to ground.

[Wavebourn]

Mooly, he has potentiometers that provide path to ground, but needs some grid leak resistor. But such rough switching on very sensitive input of guitar amp is something I would never consider.

[Mooly]

Originally Posted by Wavebourn

Mooly, he has potentiometers that provide path to ground, but needs some grid leak resistor. But such rough switching on very sensitive input of guitar amp is something I would never consider.

Yes, but no grid leak. That is what I mean when I say add a high value resistor from output of these FETs to ground.

That is what you mean... yes

[muadib25]

Originally Posted by Wavebourn

22k is too low resistance. Try something like 2.2 mega ohm or as high as possible, to lower loudness of Thump when switching. This resistor is needed to discharge gate charge, but unfortunately it provides parasitic path for control signal to output. T-configuration as I suggested before would be better in terms of Thumps.

I understand... Would a small cap in series with the signal path could isolate the control signal from the output?

FET 1 and FET 2. Is the feed from V1B AC coupled such that this pont is at zero volts ?

Yep!

If so then both these FET's are tied to ground via the 2M2 and the unmarked resistor.

Oops, sorry for that. The unmarked resistor is 2M2 too!

FET 5 and 6. Not ground referenced on the output side as drawn ? Where does R19 go ? I'm guessing like the other FET's. That still leaves no ground reference. Add a high value resistor from the output of these FET's to ground.

Oops again. Yes, R19 is like the other resistors.

Ok, how about this?

[muadib25]

By the way, I got in touch with Kevin O' Connor from London Power in Canada. He told me I should have a split-rail control voltage of both polarities, say +/-15 Volts. This would allow the audio signal to live within the control range window...

[Mooly]

Originally Posted by muadib25

By the way, I got in touch with Kevin O' Connor from London Power in Canada. He told me I should have a split-rail control voltage of both polarities, say +/-15 Volts. This would allow the audio signal to live within the control range window...

If by control volts he means the gate drive voltage then I don't agree. Look back at this,
My MOSFET amplifier designed for music.

It is a similar arrangement to yours. The input is AC coupled and both sides of the FET are ground referenced. As soon as the gate voltage goes above around 0.2 volts (relative to ground) then the FET G-S or G-D junction (depends which way round you have the FET and for these types it really doesn't matter as they are symetrical) will begin to conduct and DC will appear on the output. Just what you don't want. So the positive "on" voltage has to be kept lower than this value. The negative "off" gate voltage has to be more than the peak voltage you are trying to block.

At some point you now have to build the circuit and confirm (or otherwise) that it works as expected .

[muadib25]

Yes, I see now. Perhaps there is something I misunderstood. I will build the circuit and get back to you.

By the way, on your schematic I see you have the gate supplied with +/-9 VDC. Isnt it the same thing Mr. O' Connor suggests?

Again, thank you all one more time for the invaluable help and your patience!

[Mooly]

Originally Posted by muadib25

By the way, on your schematic I see you have the gate supplied with +/-9 VDC. Isnt it the same thing Mr. O' Connor suggests?

Yes, and I did it to highlight what happens when you increase the gate above around 0.2 volts.

The gate channel begins to conduct for real and to pass current. That gives a DC shift on the output which you don't want. Try it