Reduce Amplifier Drift when load is near input impedance

[Hyo X]

I built a voltage amplifier based on the AD625 instrumentation amplifier. This chip has a 1 G-Ohm input impedance.

When the load has a resistance of 100 kOhm, output drift is negligible.
When the load has a resistance of 100 MOhm, output drift is significant and not acceptable.

Are there any post hoc tricks to limit the drift of the output? tricks with ground loops or something? thanks

 

[marcusl]

Why would you want such a high load resistance?

 

[Hyo X]

Thats the lowest resistance of the samples I characterize. It is set by physical limits and we can't reduce it. Other samples have 10 Gohm or higher resistance.

 

[marcusl]

Are you confusing source and load impedance? The load is what the amplifier drives.

 

[Hyo X]

Are you confusing source and load impedance? The load is what the amplifier drives.

I must have my terminology wrong!
The sample resistance must be the source. it is the resistance between the V+ and V- inputs of the amplifier. The load resistance should be negligible as this output voltage goes to a data acquisition system.

How can i reduce the drift of the output of the voltage amplifier when the source resistance is close to the input impedance of the amplfier?

 

[jim hardy]

I assume you meant "source impedance"

a higher z opamp might help
http://www.ti.com/lit/ds/symlink/opa128.pdf

a common trick is "guarding" the input pins so as to minimize leakage into the summing junction
http://www.ti.com/lit/an/snoa664/snoa664.pdf

 

[marcusl]

Without knowing your application in detail, it is difficult to say where this "drift" is coming from--possibly out of the air, given your high source impedance. Instrumentation amplifiers have extensive features to eliminate stray field pickup and common-mode pickup, but they must be implemented properly. Be sure to read Analog Devices's application notes carefully, not just the data sheet. AN-244 will get you started. Shield your source in a Faraday shield; pay close attention to cabling and connection topologies, cable shields, how grounds are routed, and proper power supply practice to avoid ground loops there. That's the best one can offer without sitting in your lab.

 

[Baluncore]

You should consider using an “Electrometer Amplifier” as a unity gain buffer before the AD625 IA.
EAs have an input bias current measured in fA. Consider also the ADA4530-1

As suggested by Jim Hardy, the output should be used to guard the input. The shield or screen on any input cable used should also be driven as a guard. It is often a benefit to lift the input pin from the PCB to eliminate surface contamination currents.