JFET switching?

[ninjaaron]

Hey, I've been thinking about putting some toggle switches on a circuit that change a fair number of connection (not the on/off switch). Getting a huge switch is pretty much out. Too expensive and stupid. I was thinking some of those micro relays would be the way to go (I'm still thinking that, actually), but I was wondering if there was any reason to consider using a JFET switching system, and how I might implement this. The Boss system looks pretty involoved, and I'll just go with the relays if it has to be that complex.

[R.G.]

Any switch, no matter how complicated, can be decomposed into a number of single-pole, single-throw single contacts timed to work either simultaneously or inversely. For instance, a single pole double throw (SPDT) is equal to two single contacts working in opposition and connected together at one point. A double pole double throw switch is equal to four single contacts, connected as two SPDTs in parallel.

True bypass switching can be done with three simple contacts; two working at the same time, and one working inverse and bridging the other two. This is why the Millenium Bypass can work; there is one single contact closure "left over" in a standard DPDT to run the business of turning on the LED.

For audio signal switching,  you can consider one JFET to be equal to one single contact when properly supported by resistors, capacitors, etc. to get the DC voltages and gate switching voltages correct for it to switch without distorting. At a minimum for this you need three caps and three resistors per JFET in isolated switching. If you are ganging the outputs of two JFETs, you can dispense with one cap and one resistor for each ganged junction. One very nice thing about JFETs is that you can physically put the "switch contact" right at the place it's switching without running wires all to one central relay or IC chip to be switched.

CMOS switching can do many of the same things, but it also needs a DC Bias and DC blocking caps on each switch, just like JFETs to avoid clicking and distortion. CMOS does give you the opportunity to use higher levels of switch connections, like the CD4053, which is three SPDTs in a single IC.

Every SS switch will need something in the way of external parts to behave properly. It's just part of the territory.

Relays howerver, have their own warts. They use a huge amount of power when activated unless you use latching relays. They are big, expensive, mechanically fragile, and also have popping mechanisms that SS switching does not.

Overall, the choice between JFET, CMOS, and relay switching is not simple. They each have advantages and disadvantages.

Have you read the materials on switching at geofex.com?

[ninjaaron]

wow, good stuff.

I haven't read the stuff at Geofex about switching. Maybe I should scoot on over there...

edit: by the way, I should probably add that the switches are to switch between different clipping devices (JFET, Si BJT, and none with an on-off-on SPDT toggle switch).

[R.G.]

wow, good stuff.

I haven't read the stuff at Geofex about switching. Maybe I should scoot on over there...

edit: by the way, I should probably add that the switches are to switch between different clipping devices (JFET, Si BJT, and none with an on-off-on SPDT toggle switch).

For selecting different clipping devices, you have two main situations, those being when the clipping devices are in the feedback loop of an opamp and when they are connected to ground (or other fixed voltage) on one side.

In the feedback circuit of an opamp, both ends are connected to the same DC voltage, usually a bias voltage on single-supply circuits, or to 0Vdc (not to ground - that's different) for bipolar power supply devices. In this case, the JFETs are trivially simple. You simply put the JFET in series with the clipping diodes and arrange to drive the gate to the off voltage through a large resistance. Generally you would also use a series reverse biased diode and a slow-down capacitor to make the transition be really click and pop-free. No biasing resistors and capacitors are needed because the circuit naturally sits at those voltages.

In the clipper-to-ground circuit, you would use a large (1M or more) resistor from each side of the JFET to ground and if the driving point in the circuit does not itself already sit at ground, a large capacitor to break the DC path through the JFET and clippers. Then the same advice applies - a diode, resistor, and cap from the gate to the gate-drive logic or switch. Notice that for clippers to ground in circuits which use a single positive voltage for power (like a 9V battery in most effects) that a P-channel JFET lets you drive the JFET gate with a positive voltage to turn it off. If you use an N-channel JFET in this circuit, you will have to drive its gate negative to turn it off, and that will require making a negative voltage to drive it from, which complicates things. Generally you can drive the more-common N-channel devices from half of a 9V battery to ground if the source and drain of the JFET are biased up at half the power supply (+4.5V).

CMOS switches (CD4066 for example) may even be easier, as you only need to bias the switches so the in and out pins of the switch are sitting at the same DC voltage before and after switching and the control pin is driven to ground or power supply to operate the switch. However, CMOS sometimes generates clicks which can be hard to get rid of.

[ninjaaron]

no no... I'm not hooking them up like diodes. I'm hooking them up as amplifiers with the opamp before to overdrive them.

The more I think about this, the trickier it begins to seem. I have a good solution using two DPDT micro relays, but I'm a bit concerned about power requirements.

Really The best thing might just be to skip all this shit and use two DTDP switches.

Then I could have Opamp -> out; Opamp -> BJT -> out; Opamp -> JFET -> out; and Opamp -> BJT -> JFET -> out...

but damn... doing that would require two different bias voltages for the JFET, wouldn't it?

[R.G.]

Post a tentative schemo and I'll critique it for you.