How low a noise do you need? What's the application anyway?
If the current through the switch is so small that there is practically no voltage drop across the switch, you get the thermal noise voltage of the on resistance of the switch and possibly a very small amount of shot noise current from junction leakage currents, such as the leakage of the ESD diodes of the 40xx chips.
If the voltage across the switch is not negligible, you have to take 1/f noise in account, that is, small and slow random variations of the on resistance. No-one ever specifies how much 1/f noise their switches have.
I think so far, OptoMOS solid-state relays with back-to-back MOSFETs (as recommended by SomekPoland, soekris and abraxalito) are the most practical alternative to P-channel JFETs. You can get on resistances in the few ohms range or even less with the corresponding low thermal noise.
Last edited:
Anyway if I would bias to half of supply in my case 4V (8V supply ) it would be too less for a mosfet threshold to switch when input is 2Vpp.
Currently I could not find any other solution as a p-channel JFET as shown in the link:
https://electronics.stackexchange.com/questions/177398/jfet-audio-switch-does-not-work-in-simulation
his working for bipolar input but only psoitive supply and mosfet gate voltage.
I would be interested if there are circuits with othe mosfets types which can do the same because p channel JFETs are rare.
Thanks MarcelvdG for the noise infos.
The switch I need for a E guitar pickups amplifier which needs to be low noise. So the ON reseistance is not that cruicial because the current through the switch is low.
The OptoMOS are intersting but since I have battery powered the LEDs inside the OptoMOS draws too much current.
Currently I could not find any other solution as a p-channel JFET as shown in the link:
https://electronics.stackexchange.com/questions/177398/jfet-audio-switch-does-not-work-in-simulation
his working for bipolar input but only psoitive supply and mosfet gate voltage.
I would be interested if there are circuits with othe mosfets types which can do the same because p channel JFETs are rare.
Thanks MarcelvdG for the noise infos.
The switch I need for a E guitar pickups amplifier which needs to be low noise. So the ON reseistance is not that cruicial because the current through the switch is low.
The OptoMOS are intersting but since I have battery powered the LEDs inside the OptoMOS draws too much current.
Last edited:
If you could find a P-channel depletion MOSFET, it will behave very similarly to a P-channel JFET except that the gate can't go in forward conduction. I'm afraid P-channel depletion MOSFETs are even rarer than P-channel JFETs, though.
Two bootstrapped NMOS switches in antiseries, maybe? (Ordinary enhancement NMOSFETs, small ones because of source-drain and gate-drain capacitance and because of leakage.)
To explain the idea and assumptions:
I assume the DC voltage on both sides of the circuit is kept at 0 somehow.
When the control voltage is also zero, both MOSFETs will then have 0 DC gate-source voltage. Due to the capacitors, AC voltage on the source is also coupled to the gate, so the gate-source voltage remains essentially 0 with an AC signal present. Even though the MOSFETs have essentially 0 gate-source voltage, their drain-bulk diodes can still go into conduction at signal peaks. Using two in anti-series ensures that there is always one in reverse, so you never get a conducting path from input to output.
When the control voltage is 8 V, both MOSFETs are fully on. The trick with the capacitors then slightly reduces distortion by keeping the gate-source voltage constant, but you probably wouldn't have much distortion anyway.
As an undesired side effect, the 3.3 Mohm resistors load your guitar and inject a very small thermal noise current. You could go for 10 Mohm and 33 nF (or 47 nF) if 1.65 Mohm extra load is intolerable.
The MOSFETs have some parasitic capacitance between gate and drain and source and drain (see their datasheets for details). If the load has a very high impedance, high audio frequencies will capacitively couple right through when the switch is off. You may then need an L-switch or a T-switch (switch and switch to ground, or switch and switch to ground and yet another switch). The same holds for P-channel JFET switches, by the way.
Because of the RC network, the switch turns on and off slowly. You know better than I do whether that is a problem or a feature for your application.
Mind you, the BS107 and BS170 require careful handling because they are quite sensitive to ESD. I hope that is largely solved once they have a big capacitor between source and gate.
I assume the DC voltage on both sides of the circuit is kept at 0 somehow.
When the control voltage is also zero, both MOSFETs will then have 0 DC gate-source voltage. Due to the capacitors, AC voltage on the source is also coupled to the gate, so the gate-source voltage remains essentially 0 with an AC signal present. Even though the MOSFETs have essentially 0 gate-source voltage, their drain-bulk diodes can still go into conduction at signal peaks. Using two in anti-series ensures that there is always one in reverse, so you never get a conducting path from input to output.
When the control voltage is 8 V, both MOSFETs are fully on. The trick with the capacitors then slightly reduces distortion by keeping the gate-source voltage constant, but you probably wouldn't have much distortion anyway.
As an undesired side effect, the 3.3 Mohm resistors load your guitar and inject a very small thermal noise current. You could go for 10 Mohm and 33 nF (or 47 nF) if 1.65 Mohm extra load is intolerable.
The MOSFETs have some parasitic capacitance between gate and drain and source and drain (see their datasheets for details). If the load has a very high impedance, high audio frequencies will capacitively couple right through when the switch is off. You may then need an L-switch or a T-switch (switch and switch to ground, or switch and switch to ground and yet another switch). The same holds for P-channel JFET switches, by the way.
Because of the RC network, the switch turns on and off slowly. You know better than I do whether that is a problem or a feature for your application.
Mind you, the BS107 and BS170 require careful handling because they are quite sensitive to ESD. I hope that is largely solved once they have a big capacitor between source and gate.
Finally some application info, thanks. I assume this is inside the guitar itself? The selector logic also residing in the guitar? Output of the switches goes directly into an preamp or is there a (high resistance) volume pot before it? Is it also about series/parallel hookup for humbuckers? Battery switched on with the typical TRS jack arangement where the negative side of the battery is connected to S (GND) when a TS jack is inserted?
Anyway, low noise is not that important as guitar pickups are several kOhms coil resistance and have quite high thermal noise from this alone.
A bit of signal leakage also is not a problem (unless you wanted to use the "all switches off" state as a "kill switch" option -- not really recommended as that would be noisy and prone to interference with the preamp input being basically open circuit).
AC coupling is feasible and therefore 4066 type of switches/muxers would do but requires some thought for the supply. I would guess the preamp input is AC-coupled anyway, running from a single supply.
Solid ESD protection is a must (don't ask me how I know).
@MarcelvdG , nice circuit idea.
The soft turn-on/turn-off could actually be a feature, making the switchovers click-less.
The soft turn-on/turn-off could actually be a feature, making the switchovers click-less.
@MarcelvdG, yes this interesting cicuit. I have seen this but I did not find much about this antiseries mosfet circuit in relation to analog switches.
But this may worth to do some spice simulations to see if even the capacitor can be left out to make it simpler.
@KSTR yes the switches are part of a bit complicated circuit where I want to switch different capacitance on the pickups and want to keep high impedance not to damp the resonance (battery powerd inside the guitar). Also the output of the preamps are switched via anlog switch. So I need a couple of analog switches/mosfets. A SOT23 J175 does not take much space. But the circuit above maybe still feasableand may have better performance.
But this may worth to do some spice simulations to see if even the capacitor can be left out to make it simpler.
@KSTR yes the switches are part of a bit complicated circuit where I want to switch different capacitance on the pickups and want to keep high impedance not to damp the resonance (battery powerd inside the guitar). Also the output of the preamps are switched via anlog switch. So I need a couple of analog switches/mosfets. A SOT23 J175 does not take much space. But the circuit above maybe still feasableand may have better performance.
One weakness of my circuit is that you temporarily get some current flowing from the control input via the large resistors and the capacitors into the signal path at switch on and out of the signal path at switch off. Using very large resistors limits that current.
Just an idea, if the preamp circuit allows for the GND reference tied to positive battery terminal instead of negative you could use N-channel JFETs like PN4391/92 which have similar cutoff like J175
Thanks for all the contributions. I think I will look for the n channel anti-series construction.
Does anyone have goo seuggestions for dual mosfets in one package to save space? But it should be possible solder by hand.
SOT23-6 would be perfect but SOT363/SC70 is also possible but the smallest package I could solder.
I found BSS138PS, FDG6301N, NDC7002N.
Does anyone have goo seuggestions for dual mosfets in one package to save space? But it should be possible solder by hand.
SOT23-6 would be perfect but SOT363/SC70 is also possible but the smallest package I could solder.
I found BSS138PS, FDG6301N, NDC7002N.
there are audio switches than can handle negative voltages without the extra negative rail. for example NX5L2750CGU, also available at other brands. look for "negative swing capability"