FETs used as on/off switches to signal path

[akis]

My Fender guitar amp uses FETs as on/off switches in a couple of places. Signal is inserted at the source and taken from the drain. A hefty negative voltage at the gate keeps the FET switched off, and anything at ground level or above keeps it conductive.

I have tried the same using my two available FETs, 2N5459 and the 2N3819, and they both work well, very well I'd say. Except I believe there is a small signal "leak" when the FET is supposed to be switched off. My oscilloscope shows a 10-20mV p2p on an input of say 8 V p2p.

A lot of it is due to the wires, capacitances on the breadboard etc.

But some is down to the FET, or to the idea of using it in place of a mechanical switch. Anyone with experience in the subject could you please suggest something? Maybe I should be using MOSFETs?

[akis]

Just to say, I would expect the drain-source resistance to be many MOhms when the FET is in the "off" state, and may be I would not get better performance with an ordinary (rotary) switch. Certainly, when I did some simple experiments using a switch, there was also a "leak" (interference), at any audio frequency, but I could not pin that down to the switch it self, or to the wires carrying the signal onto the breadboard and the switch...

So maybe it is impossible to "switch off" entirely an input signal, except perhaps by means of earthing the input?

[picowallspeaker]

Using cmos-ttl technique frees you from rotary switches , you can use simple NO or Nc buttons , to keep everything digital.

[currentflow]

Originally posted by akis
Just to say, I would expect the drain-source resistance to be many MOhms when the FET is in the "off" state, and may be I would not get better performance with an ordinary (rotary) switch. Certainly, when I did some simple experiments using a switch, there was also a "leak" (interference), at any audio frequency, but I could not pin that down to the switch it self, or to the wires carrying the signal onto the breadboard and the switch...

So maybe it is impossible to "switch off" entirely an input signal, except perhaps by means of earthing the input?

It is possible that the leakage you experience is also a result of the FET's drain-source capacitance, particularly if it feeds a high input impedance stage. Some, if not many transistor switch designs utilise a FET switch in series with the signal path, followed by a second FET acting as a shunt to signal ground. When the series FET is 'off' the shunt FET is made to switch 'on' thus further attenuating the unwanted signal.

Steve

[akis]

Yes, that's what the Fender does too, it "grounds" the overdrive signal when not needed while allowing the clean signal to pass through. I will try this tonight when I go home, see what difference it makes.

I also noticed that I could cut interference in half by simply exchanging drain and source. I presume this must be an artifact of the FET I am using? (2N5954)

[AndrewT]

I've been told that most jFETS are symmetrical.
You can swap drain and source and they will work equally well.
There are exceptions.

But, I believe they are not good at accepting reverse polarity signals.
Maybe one needs to parallel Pch & Nch for effective isolation and passing duties.

[akis]

I think I will be good especially with the second FET to the ground (here's hoping).

The only thing remaining now is to make sure that no ripple currents, or other interference in the gate could affect the drain to source channel. The Fender amp has a non polarised capacitor at the gate to the ground, presumably for this purpose.

[AndrewT]

Originally posted by akis
The Fender amp has a non polarised capacitor at the gate to the ground, presumably for this purpose.

that cap could be an RC time constant to delay switch on and thus attenuate switch on buzzes/surges etc.

[Pingrs]

Some useful discussion of CMOS and JFET's as switches in Doug Self's "Electronic Analogue Switching" article (Electronics World Jan-Feb 2004)

[picowallspeaker]

yes , that was what I was referring to.But the magazine didn't show the logic needed for turning on and off all the inputs,if I remember ...
Ciao