Cmos switch implementation

[Rogy]

Hi All,

In my 30-years old EMT console, the solo function is carried out by interrupting the signal of all other channels (Solo-in-place).

The signal interruption is done using a Cmos switch (Intersil/Harris IH5352) followed by an opamp configured as a buffer (audio to positive in, out fed back to negative in).

I found this config to produce annoying pops when solo'ing, and changing the buffer to an inverting gain of -1 amplifier seems to cure this (I am of course aware of the resulting polarity reverse).

I configured the inverting opamp with a 22K R in and a 22K R feedback, and a parallel feedback compensation cap.

My question: Do I take the opamp's positive in to ground using a 0 Ohm jumper, or using a resistor  that equals 22K//22K? Pros/cons?

Thanks for your insights!

Best Rogy

 

[JohnRoberts]

Hard connected to ground should be OK... 

Eliminating the high impedance buffer may raise distortion slightly, but those are RF transfer gates so low impedance (75 ohm). So tens of K is not much loading. 

Back in the early '80s i used cheap cmos transfer gates inside the NF loop of an opamp and distortion was below my bench measurement capability (for back then).

JR

 

[Rogy]

Thanks for your insights John!

After a couple of hours of soldering, my console now  has a much more click-free solo system!

The solo bus sits at ground potential when no soloes are active and gets pulled to -15V with activated solo.

This transition still triggers deflection of the bus master meters.

I'm wondering if slowing down this transition using eg an RC snubber network would make sense... Any suggestions?

Best Regards,

Rogy

 

[JohnRoberts]

Thanks for your insights John!

After a couple of hours of soldering, my console now  has a much more click-free solo system!

The solo bus sits at ground potential when no soloes are active and gets pulled to -15V with activated solo.

This transition still triggers deflection of the bus master meters.

I'm wondering if slowing down this transition using eg an RC snubber network would make sense... Any suggestions?

Best Regards,

Rogy

I used to use two separate bus lines. one for solo logic and one for the solo audio. Multiplexing them onto one line saves one pin on the header but a 15v step in an audio line is not insignificant.

I can't believe the original design just ignored a 15v click. Perhaps some short term mute that silences the click when it detects the voltage step?

JR

 

[abbey road d enfer]

Thanks for your insights John!

After a couple of hours of soldering, my console now  has a much more click-free solo system!

The solo bus sits at ground potential when no soloes are active and gets pulled to -15V with activated solo. 

I guess you mean the "solo sense" bus, not the audio solo bus.

I'm wondering if slowing down this transition using eg an RC snubber network would make sense... Any suggestions?

You must do some more investigation. Generally, CMOS switches have a bit of logic that takes the control voltage and makes it suitable for the switching elements, and most of the times, these resulting voltages are very abrupt. Slowing down the control voltage would have no effect on these internal voltages.
The charge transfer between the gates of the CMOS and the audio path has always been a concern of CMOS switch designers, and of those who use them. The susceptibility of the receiving end is a major concern; operating with the highest possible level and with the lowest possible impedance is key. Maybe you should try and reduce the resistors to 4.7k? That should reduce clicks by 12dB without really impairing other performance aspects. In this respect, the current SS switches are better by two orders of magnitude than those of 20 years ago.