# Passive Normally-Closed Switch For Audio Applications

I am working on a small, audio device with aux input and output (both headphone-level) that has some interesting requirements:

1. In passive mode (dead battery) audio needs to go from input to output
2. In active mode audio will normally go from input to output
3. In active mode the MCU can block the audio from the input, and pass in audio from a DAC/amplifier
4. SSRs that I have found are too big for my application. I'm also not sure what effect they could have on audio (quality is important)
5. This is for a prototype, so working with .4mm BGA is somewhat out of the question

Here is what I have so far:

1. This IC is pretty much what I think I'm looking for, however I (personally) would not be able to use it: https://www.fairchildsemi.com/products/analog-mixed-signal/switches/analog-switches/FSA550.html#product-availability
2. The answer from jippie in this question MOSFETs or BJTs for using as switch for audio signals seems like it may work, however I'm concerned about the +6V input - would this need to be there in passive mode? It seems like the answer would be "no" but I'm not confident
3. This is different from most BT headphone circuits, since in headphones, the aux input usually trumps the BT, whereas here it is not the case.

Can anyone recommend either an IC that performs as a normally-closed analog switch w/o power or a circuit/application through which I can achieve my goal?

• How much ON-resistance can you tolerate say 15 ohms? – Andy aka Jan 13 '17 at 18:48
• omron.com/ecb/products/pry/114/g3vm_354c_f.html – Andy aka Jan 13 '17 at 18:56
• This is sometimes done with depletion mode MOSFETs. Depletion-Mode MOSFET: The Forgotten FET. – Nick Alexeev Jan 13 '17 at 18:58
• Thanks @Andyaka. I have little experience with analog design (background in embedded, so mostly follow datasheets/application notes). So I am not sure how much ON-resistance I can tolerate for an audio signal. Do you have any idea if that's a reasonable amount? – skobovm Jan 13 '17 at 19:11
• Thanks @nick-alexeev. I forgot to mention that I have also looked into depletion-mode MOSFETS. Would sending audio signal through a MOSFET degrade the quality in any way? Would I need any supporting circuitry/filtration? – skobovm Jan 13 '17 at 19:12

There are normally-closed SSRs available in DIP-8 packages and smaller. For example, the Toshiba TLP4176G(TP,F) which has a 7mm x 4mm footprint roughly, but there are others.

That particular one has an on-state resistance of 15$\Omega$ typically (25$\Omega$ max). You may have to add a shunt element to improve off-state isolation (the 'off' MOSFETs will have a tiny bit of leakage, and some capacitance) but that is much less of a problem since you will have power available. There are in fact a few SSRs that have both NO and NC elements in one (somewhat larger) package.

For lowest parasitics, a P-Pchannel JFET works very well. A gate to source capacitor, like 470nF and a very high impedance pull-down on both drain and source, like 1Meg. You should AC-couple the source leg through a sizable cap, sized depending upon your impedance on the input side, maybe 220uF just to prevent phase shift and low-frequencies from attenuating.

The gate itself is driven also through a 1Meg; it requires no pull down. Depending upon the gate threshold, you can turn OFF (OPEN) the drain-source signal with a positive voltage (greater than the gate threshold voltage).

Operation: During conductance (gate at 0V or floating) the audio signal will couple through the gate-source cap and thanks to the 1Meg gate resistor,the gate and source will track, meaning the transistor stays in conductance (Vgs = 0).

When you introduce the OPEN gate control voltage, then the gate sees the signal at a DC offset, thus Vgs = Vcontrol + Vpp audio (not to imply that the audio signal is required for operation). HOWEVER, your gate threshold voltage must be greater than 0.5|Vpp| of your audio signal, such that when it goes negative (due to AC coupled source, or drain -- this is a bidirectional switch) it will not pull the gate below the threshold voltage you are using to control it.

Alternatively, you can make your gate control voltage higher, but be careful not to blow the gate, and, for digital control, high-voltage may not be a straightforward output option.

That's a discrete solution. The SSR options are very well covered for RF, and audio frequencies, with pF of parasitics and 1 to 25 Ohms of on resistance (you can choose betwen higher parasitic capacitance and lower on resistance, or vice versa, depending on the application -- RF cares much more about parasitic capacitance).

Look into Panasonic PhotoMOS relays. Their latest TSON pkg device is amazing, but only available in N.O.. But these sport a 40uA to 1mA turn on/off requirement, and can pass 30VAC to 600VAC. Their other PhotoMOS are actual LED photomos devices; the ones in the TSON pkg are not, yet they still call them PhotoMOS. But they are not regular MOSFET switches -- gate control is an oscillator through a diode bridge, cap and regulation circuit. The point being to completely isolate the switch from the input control, but in a clever way that only requires 40uA to generate Vgs.

As far as what series resistance you can handle, that has to do with any RC parasitics and input impedance, as well as driver impedance. If that is not high and low, respectively, then no lower series resistance, even 0 Ohms, will help recover the losses. So normally, your output driving impedance is a few Ohms, and your input impedance is high, like 10's of k Ohms or more. So 25 Ohms in the path is negligible.