[sdiy] MOSFET leakage.

[René Schmitz]

Hi Ian and all,

> Ah, yes.  Same technique should work, though, unless the leakage is so 
> small that the instrumentation current is much larger that what you are 
> measuring.

In that case you wouldn't need to care about it anyway.
And your figure for the residual leakage suggests that cap leakage can 
be neglected.

> No!  Just plugged it all into my whiteboard. :-)
> 
> Plastics are actually very good insulators.  I remember a friend showing 
> me that if you pot the contacts of a coax connector with 5-min epoxy 
> then measure the resistance with an electrometer that the resistance is 
> over 10^14 Ohm.  That's roughly the point where you have to go to 
> teflon, etc.
> 
> The main worry would be conduction along the surface, but it doesn't 
> seem to be an actual problem.

Much like I suspected. I think into the pico ampere range that is 
sufficient, its the fempto region where this could create problems.

  > What I've been looking at  is the speced "drain cutoff current" 
which is
> .5 uA.  So 2 pA is a factor or 250,000 smaller.

Ok, I didn't check and had a value of 1nA in my head.

> Yeah, that's what I'm working on.  Instead of your series R trick, I'm 
> using  pulse shaping to try to tame the overshoot.  Keeping the 
> comparitor from going into deep saturation seems to help a lot.  Also I 
> want to use a device with less capacitance.  The BS170 and 2N7000 units 
> have 60 pF input C.

There really aren't many good small signal MOSFETs, seems most of them 
are switching devices.

> Interesting idea.  I'll need to think about that some more.  You mean 
> generate counter-glitches using a second cap?

Exactly. An old RF trick, to cancel out influences of device 
capacitances. Maybe the cancellation is not 100% perfect, but even 80% 
or 90% would be an improvement.

Cheers,
  René

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