How-to measure low current? What current sense amp do you suggest?

[Lucast85]

Hello,

I need to measure the leakage current of an embedded board under test (DUT).
The DUT is powered by a custom voltage generator board that supply a voltage in the range [0-5] V. The voltage generator must measure both current and voltage, like a SMU. I'm looking for advice about the current sensing.
The voltage is sensed at the load and there is a feedback to compensate voltage drops along cables and other components; thus I could not worry about voltage drop across the current sense resistor (shunt resistor).
I expect a leakage current to measure of a few micro-ampere (I think in the range [1-7] µA).
Those are some desired specifications to design a current sense circuit:

• High-side current sensing; common-mode voltage expected in the range [0-5] V;
• Leakage current to measure range: [0-1] mA (unidirectional);
• Accuracy: ±2 µA;
• Resolution: 200 nA;
• Resolution of ADC: 16 bit in range [0-3.6] V;

I found some current sense amplifier ICs () to design the circuit but it seems they've some bad tricks 8-O like:

• No rail-to-rail output;
• high voltage offset;
• high bias current;

that don't allow to reach the goal. Follow the 2 images of LTspice simulation graph and circuit and an attachment with LTspice simulation files.


However I can disconnect the DUT and compensate for Voffset, Ibias, and other stuffs (reading ADC's value and clear it from the measure via software) before to measure the DUT leakage current.
Is this the right procedure?
Have you got some advice to reach the goal?
Thanks in advance and excuse me for my bad english ...

 

[FvM]

If you are targetting to high performance measurement, you better get rid of arbitrary specifications like 5V single supply.

 

[Lucast85]

...you better get rid of arbitrary specifications like 5V single supply.

Ok, no problem, I can provide also a dual supply to current sense amplifier. Note that the Vcm=5 V in the schematic is the common mode voltage of the sense resistor; it can vary in range [0, 5] V. The voltage supply of the current sense amp. is +3.6 V.

 

[schmitt trigger]

The MAX9934 datasheet indicates an input bias of 0.1 to 100 nA....in my experience, board leakage can be far larger than that, if one is not careful with the layout..
Why do you say the device has high bias current?

 

[Lucast85]

The MAX9934 datasheet indicates an input bias of 0.1 to 100 nA....in my experience, board leakage can be far larger than that, if one is not careful with the layout..
Why do you say the device has high bias current?

You're right, the MAX9934 has optimal input bias current characteristics, but only if common mode voltage are minor than Vcc-0.2 V. Instead, input bias current when Vcm is greather than Vcc-0.2V, is in range [35-60] uA. Probably, the major use of my custom SMU is with common mode voltage at 3-4 V, and this can lead to have an "high" Ibias on Rs-.
I think to "auto-calibrate" the SMU disconnecting the DUT and reading the current-sense-amplifier's output (ideally this output must be equal to zero but it will not be). Finally I will clear (by MCU's firmware) the offset voltage due to Ibias and Voffset at input of current sense amp. This could be a straightforward way?

 

[schmitt trigger]

Looks like a good idea!