Transimpedance amp for DC only (bandwidth limited)

[ZappHappy]

I'm using a classic TIA configuration to measure leakage current through coax insulation in the 200nA range.

The circuit works well using a V source and resistor to calibrate but when I connect the long, high capacitance cable the noise is terrible.

Can I bandwidth limit the op amp to say 1-2Hz using a large capacitance across Rf (1 Megohm) or am I having issues due to large cap (>10,000pF) at input ?

I'm not sure if the "noise" is from the coax acting as an antenna or some other op amp stability issue.

 

[frankrose]

Cf should increase OPAmp stability by adding a left half-plane zero to the open-loop gain.
What kind of noise you experienced? Mains hum?

 

[ZappHappy]

Thank for the response..

Yes it's partially mains noise but also a lot of broadband.

In a TIA configuration is there a maximum Cf that may cause problems ? I'm ok with limiting in the .5-1Hz if this is possible.

 

[frankrose]

UGBW is 3.5MHz for this OPAmp. Not so broadband.
I think there is no limit until you don't use such a big device that it will form an antenna.
Be sure your supply is well filtered, and there is no ground noise / ground loop with your measuring device.

 

[ZappHappy]

" Not so broadband." - I guess your right, I'm usually working with precision DC so the noise was "broadband" in my world

The device is battery powered and I have good bypassing on all op-amps and supplies. I will add enough C to limit to <1Hz and see what happens.

Thanks again

 

[Amadeus]

I think the reason is high capacity of the cable and noise propagation of ground loops.
Cable shiled should be driven by a capacitance neutralization circuit. There is a lot of instrumentation amps which use the trick.

 

[ZappHappy]

Thank you for the response..

You may be correct about noise. The cable is fed high voltage at the center pin and the shield is fed directly to the TIA so I don't know how I would "neutralize" the capacitance in the current architecture.

Rolling off the amp really tamed a lot of the noise but may need more as I'm sensing in the 0-200nA range.

I'm also experiencing (possibly) some polarization effects in the dielectric. At higher voltages, when stable the meter is reading opposite the expected reading. I typically see this only when the voltage is reduced.

 

[asdf44]

You can make Cf as large as you want. It only hurts bandwidth but for DC that's not a problem.

But you may also have a problem with RF rectification so I'd look at bulking up the passive input filtering. C in parallel with your diodes and a ferrite in series may also help.

 

[ZappHappy]

Thanks.. I have .1uF (didn't make it on the schem) across now but ferrite is a great idea. Though if the diodes (actually diode connected 2N3904) are rectifying will higher capacitance cause a DC error ? I'll give it a try anyhow..

- - - Updated - - -

You were right on about the rectification asdf44. I removed the diodes and my linearity fell into spec.. Thanks a giga

 

[KlausST]

Hi,

will higher capacitance cause a DC error ?

No.
Only the leakage of the capacitor may cause problems..
Use high quality polystyrene, if possible.

I removed the diodes and my linearity fell into spec..

* The diodes are for "protection" only, not for rectifying..
* use "low leakage" ones.
* they should have no influence on function, nor noise, nor linearity

If there is an influence, I expect a design mistake elsewhere.

I recommend to post a photo of your amplifier circuit, where we can see wiring, power supply connections, all GND connections.

Urgent design requirements:
* GND plane, inelligent star point type GND wiring
* properly HF decoupled supply for the OPAMP
* I recommend to supply the OPAMP from a battery, if you use any other supply you need a high quality one with low and low leakage/ low capacitance to mains
* properly decoupled and properly low pass filtered HV supply with f_cutoff_supply << f_cutoff_of_amplifier

Klaus

 

[asdf44]

Well I didn't think those diodes would be the problem but for precision DC in a noisy environment every PN junction is a possible source for rectification problems. So it's possible.

https://www.analog.com/media/en/technical-documentation/application-notes/AN-671.pdf

 

[KlausST]

Hi,

Well I didn't think those diodes would be the problem but for precision DC in a noisy environment every PN junction is a possible source for rectification problems. So it's possible.

But the voltage across the diods should be only a couple of millivolts.

Klaus

 

[Dominik Przyborowski]

In general input capacitance for TIA is a source of both instability and noises (in 1st approximation noise is linear with input cap).
I would suggest to integrate signal more (by increase Cf) and check behavior by disconnecting line and emulate it by ceramic caps up to 10n, to eliminate antenna hypothesis.