[sdiy] MAX SPDT swithces

[René Schmitz]

Hi Harry, Aaron, Chad and all,

> You can probably live with resistance if if is constant over applied 
> voltage and
> symmetrical.  

Voltage here is two constant voltage levels. Symmetry can be trimmed if 
deemed necessary by changing one of these voltage levels.
We're trying to make a squarewave to be fed back to an integrator after 
all. This switch is nothing else but a push-pull output stage added to a 
comparator.

Those CMOS transmission gates are used in digital circuits as well, and 
there they switch signals near the rails all the time. As long as the 
load is much larger than the internal resistance they will stay near 
those rails.

But there are also other possibilities for the particular problem:
- Using an inverter instead of that switching arrangement. (Swap + and - 
inputs on the comparator then.)
- Or make a class B output stage out of two BJTs.
- Or make a precision schmitt trigger from a window comparator
and a flip flop.
- Or clamp using opamp ideal diodes.


> If you put the switch at a summing junction of an opamp  
> most or the
> resistance problems are minimized... You can put one in the feedback of 
> an opamp
> and compensate gain errors due to switch resistance...

Even without a switch in the feedback this works remarkably well. The 
switch sees only tiny voltages this way. And you can switch the full 
voltage range, delimited only by the opamps. (You can see an example of 
this on my site under Misc->signal switcher)
But this has all nothing to do with the use of CMOS switches to make a 
symmetrical squarewave coming out of a schmitt trigger.

> Check out Ian Fritz's uses of the analog switches... he has a good 
> handle on that

IIRC, I suggested to use switches for clamping to him. :-)

Cheers,
  René

-- 
uzs159 at uni-bonn.de
http://www.uni-bonn.de/~uzs159