Discharge Switches (was: Re: J105 )

[Ian Fritz]

Rene Schmitz wrote:
 
> Don't overlook the leakage effects on the lower end of the scale.
> It would gain nothing if you find a FET with one halve the R(on) and double
> the leakage.

True...but... The J105 "typical" leakage is 10 pA vs. 5 pA for the
2N4859 and 2N4391. Twice as bad indeed, but still much less than for a
MOSFET (e.g., 30 nA for the BS107, 1 muA for the 2N7000, etc.) 

> And two more thoughts on the subject of discharge switches:
> I found an interesting idea in Horowitz & Hill, they show how to make a
> high precision integrator, as switch they use two Fets in series (though
> they use MOSFETS).One has the larger portion of the voltage across it, and
> the other a very small Vds. The middle of this arrangement is grounded via
> a 1M Resistor, so the leakage current of the one with the high voltage
> flows to ground. Kind of a guarding technique. Now what I think is this can
> be done similarly with JFETS (then we would have essentially only gate
> leakage contribution which is 1-10 pA).
> One can use a smaller cap value to bring the speed up at high frequencies.

There are similar (identical?) circuits in Jung's "IC Op-Amp Cookbook".
These came originally from National and Analog Devices. I haven't been
able to find the originals. I would be *very* interested in learning how
to use these techniques with JFETs.

> And the second thing, charge injection from the gate-channel capacitance
> can probably eliminated by adding the inverted discharge signal to the signal
> thru a tiny cap.

Yes, Terry mentioned this possibility also. I'm not aware of anyone
trying it yet. Again, I would be most interested in seeing a circuit if
you work on this.

  Ian