Getting rid of audio clicks on a push-to-talk switch?

[andrewclark1]

I previously posted a request for a simple electret preamp/buffer that could be powered from 48VDC phantom power (from an audio mixing desk), for a homemade intercom project.

That thread yielded a couple of suggested designs including the emitter follower circuit shown below, into which I have added a normally-open SPDT switch (latching on - off - momentary on) as a push-to-talk function, in the location marked in red:

Is there a way to 'de-click' the switch, or is there a better location for it that will a) cause the output to be muted unless the switch is activated and b) minimise the loud thump of the switch engaging in either of the 'on' positions?

I'm new to transistors (I skipped them and learnt op-amps as a younger man), and could use some advice!

 

[ronsimpson]

The pop/click happens when the DC level of the audio changes by the switch.
The audio is a very small signal at this point but the 12 volt supply is large. When your switch changes state the level moves most of 12 volts.

I drew a blue switch. It will short out the signal from the mic but I think it will not change the DC level.

 

[andrewclark1]

The pop/click happens when the DC level of the audio changes by the switch.
The audio is a very small signal at this point but the 12 volt supply is large. When your switch changes state the level moves most of 12 volts.

I drew a blue switch. It will short out the signal from the mic but I think it will not change the DC level.
View attachment 126061

Thanks for your reply, but the 'logic' of my switch is the opposite from what you've drawn - it is normally open (ie. muted), and only closed when the operator wants to speak.

Can you think of a way to wire it in that allows that logic to be maintained yet minimises any click? Can the DC at the red location be ramped up rather than instantaneously switched? Can the DC be decoupled by adding a capacitor inline? (My gut feeling is no, because the base and emitter need to be DC coupled, but I don't know why I feel that...)

PS. I need it for this weekend, so getting different hardware isn't really an option on that timescale!

 

[rjenkinsgb]

There is no really "ideal" place to use a single normally-open switch in that circuit.

The only point in that circuit I can see that has a chance of working is if you add the switch in the link from R4 & R5 to R6, so the bias to both transistors is removed when the switch is off.

Add a couple of fairly high value resistors (47K?) from the outputs back to the 12V zener to maintain the mic supply and another 100K across base - emitter on each transistor.

There will still be quite a kick in the DC levels, but symmetrically on both outputs so hopefully it will be cancelled out by the balanced input.

 

[andrewclark1]

This weekend's deadline aside, for a future 2nd revision I'd be happy to add a component or two to make something that works really well.

Could the normally-open physical switch be paired with a BJT or FET, so that the normally-closed electronic switch 'shorts' the signals in the location of Ron's blue switch above, and the physical switch then deactivates it to allow the electret signal to be heard?

 

[andrewclark1]

There is no really "ideal" place to use a single normally-open switch in that circuit.

I was beginning to suspect that was the case! I already tried it on the electret's signal and ground lines, but no joy due to the necessary DC there...
I guess that if I used a dual-pole switch, I could interrupt both electret connections between, say, C3 & R4 and C4 & R5? Would that be better?

The only point in that circuit I can see that has a chance of working is if you add the switch in the link from R4 & R5 to R6, so the bias to both transistors is removed when the switch is off.

What's the difference between your suggestion and mine? Instinctively the red location I've marked feels like the Zener is only powered when the switch is on (and the transistors' bases are connected), whereas yours feels like a close variant and I can't intuit the difference...

Add a couple of fairly high value resistors (47K?) from the outputs back to the 12V zener to maintain the mic supply and another 100K across base - emitter on each transistor.

Yes, I understand keeping the Zener connected to the 48VDC coming in from the mixing desk, but what's the significance of the base-emitter transistors?

There will still be quite a kick in the DC levels, but symmetrically on both outputs so hopefully it will be cancelled out by the balanced input.

I was hoping this would be the case, but would the very slight difference in the performance of the two legs of the buffer (due to resistor and capacitor tolerance) cause a less-than ideal balance in the signals to the desk, meaning that the DC click would not be adequately rejected by the desk's input stage? Without getting too perfectionist(!), would it be worth replacing one of the 100k resistors (R4 or R5) with a variable resistor that could be tuned to match the other?

 

[rjenkinsgb]

What's the difference between your suggestion and mine? Instinctively the red location I've marked feels like the Zener is only powered when the switch is on (and the transistors' bases are connected), whereas yours feels like a close variant and I can't intuit the difference...

It keeps all the "unbalanced" circuitry powered (that would give a differential voltage when switching), but removes the bias from the transistors so there is no audio path.

Any imbalance would be due to variations in the time constant of C3 & C4 plus the resistors.

 

[audioguru]

If you suddenly short or allow the sound signal during audio then probably there will be a POP sound.
I worked with a high quality large intercom system (Pamex) that muted and unmuted (but not suddenly) the audio before and after switching.

 

[crutschow]

If ronsimpson's blue NC switch in post #2 still generates a "pop" then add a large capacitor (e.g. 10uF) in series with the switch.
That should short the AC signal but not affect the DC bias.

To keep the capacitor charged to the bias voltage when the switch is open, add a large resistor (e.g. 1 megohm) across the switch.