Re: gEDA-user: analog switch idea

[al davis <ad106@xxxxxxxxxxxxxxxx>]

On Thursday 03 August 2006 18:38, gene glick wrote:
> I'm looking for some ideas on how to multiplex some
> single-ended audio signals.  The plan is to connect the mux
> to a differential amp, which in turn drives an ADC.
>
> The obvious (at least I think it's obvious) choice is to use
> some variety of analog switch IC. ON-semi has some decent
> ones actually, but they have a THD numbers around 0.01%.  Not
> bad, but not good enough either.  I'd like to get at least an
> order of magnitude better.
>
> I recently saw a "design idea" in EDN where the author used
> an analog switch to change the gain of an op-amp to zero.
>  Cute idea actually. But, he used a really expensive
> low-noise op-amp.  So I guess my next criteria is low cost.
>  How low? not sure.

Many years ago I did some of that ..  As I recall, the "analog 
switch IC's" are really mosfets, series or shunt.  The 
distortion spec was made under certain operating conditions.  
Since you didn't say what you are using, or the circuit, I 
can't judge how to make it better.

This type of switch works best at a virtual ground, so the 
voltage is zero or constant.  Since it is used as a series 
element, the distortion can be minimized by using a larger 
series resistor.  This makes the nonlinear part of the 
resistance a smaller part of the total.  This may increase 
noise, which may or may not be an issue.  You need a pair of 
switches, or a single-pole-double-throw, for each signal. You 
need to consider what happens during switching.  Make sure that 
you never short the op-amp input to ground (which effectively 
removes negative feedback) and you never let the voltage get 
big.  Either of these problems will result in a pop or click 
during switching.

You also need to consider that distortion numbers usually 
increase as the signal gets larger.  Running at a lower signal 
level will usually result in lower distortion.  For the devices 
I used, the distortion was almost all second order, which is 
the least audible type.

Considering all that, by the time all optimization was done, the 
performance impact of the switch IC was essentially zero.  Its 
distortion and noise were masked by other distortion and noise 
in the system, and that was so low that the distortion was 
difficult or impossible to measure, and noise was essentially 
what has predicted by theory and dominated by other stages.

I don't remember what brand or device we used.  I do remember 
that it was mainstream and cheap.


_______________________________________________
geda-user mailing list
geda-user@xxxxxxxxxxxxxx
http://www.seul.org/cgi-bin/mailman/listinfo/geda-user