Leakage/bias current cancelation techniques

[DragonPetter]

I became curious as to how this is possible after seeing this thread:
https://www.physicsforums.com/showthread.php?t=562370

How are these electrometers able to have bias currents of 1-3 femtoamps? What kind of active leakage cancelation circuits are capable of this? Can this be used on other components (diodes, capacitors, Vds channels, etc.)?

I tried to see if it was possible, so I did a simulation experiment. The simulations are attached below. I reverse biased a diode into a capacitor to collect the charge, and I see that in my simulation that there is a reverse bias of 2.5nA. I try to cancel the leakage current with a KCL at the diode node, and I do this with a current mirror with a leakage reference current using another diode. I had to adjust the biasing just right to eliminate the leakage current. I admit that this circuit is impractical and I just made it to see if the idea would work.

Does anyone have any insight as to how this is done in these electrometers? Am I in the right direction?

 

[Carl Pugh]

Some operational amplifiers have extremely small bias current.
You may want to research the LMC6041IN. It has only 0.002 picoamp bias current typical and is low cost.
0.002 picoamp = 2 femtoamp=0.000,000,000,000,002 amp.
There are operational amplifiers with lower bias current that cost more.

 

[jim hardy]

maybe this will help..


http://www2.cambr.uidaho.edu/papers/dielectrically_isolated.pdf

hmmm apparently the DIFET process make them radiation tolerant, they're talking about 200KRAD devices...

a google on : dielectrically isolated integrated circuit

led to several IEEE articles i can't access...

and TI OPA128 folder has a link to spice model.

but I'm getting in over my head. maybe you'll educate me on this one.

Have fun!

 

[es1]

Looks like they did a little more than use DIFETs, but I am sure that is a big part of it.

http://www.ti.com/lit/an/sboa061/sboa061.pdf
"Replacing the reverse-biased gate-to-substrate isolation diode structure of BlFETs with dielectric isolation removes this large leakage current component which, together with a noise-free cascode circuit, special FET geometry, and advanced wafer processing, allows far higher Difet ® performance compared to BIFETs."

I don't think using a current source to cancel input bias leakage would work at this range as it would need to hold to fA accuracy over voltage and temp. I suspect process would be the only way to go.

 

[DragonPetter]

I'm sorry, but all of those documents including the difet isolation technique are what i'd call passive techniques. Basically they are reducing leakage current by the materials and process rather than canceling existing leakage currents.

From the original thread, I was under the impression that there is some kind of circuitry, signal processing, or mathematical cancelation of the bias current. Is there such a thing, or is it only possible through process and material selection?

I still am not sure how electrometer opamps can have input bias currents within a couple femto amps or less. The most interesting paper I've read is called "Modern electrometer techniques" but it is dated to 1979, and I'm sure there must be more information on it since then.

p.s. that TI appnote was interesting still, and I was surprised that a larger resistance reduces noise.