Voltage Mapping for current source

[nitish111]

I have a current sink which sinks 2 mA DC through a load (10 k -20 k typically) up to 40 V DC. I am using voltage controlled current sink for this purpose(schematics attached). I have to display the voltage across the load for which I am using a differential amplifier(AD8479). I get a pretty good accurate reading for ADC but this is also causing error in current through the load. For 10 k load there is a drop of about 25 uA and  for 20k the drop is about 50 uA. I can reduce the drop by increasing R2 but then I lose the resolution for ADC which I intent to keep same. Is there a way to measure this floating voltage with minimum loss like less than 10 uA for 20 k load?

[Zero999]

Increase both R1 and R2, by a factor of 10.

Can't you reduce the current to 250μA and eliminate R1 and R2?

Alternatively, move R1 and R2 to after the instrumentation amplifier. The values can then be reduced by a factor of 10 and a 1nF capacitor placed across R2, to provide current when the ADC's sample and hold circuit takes a measurement or if the reduction in bandwidth is unacceptable, add a unity gain buffer, between the potential divider and ADC.

Another solution is to drive the 10k to 20k load, with a constant voltage and measure the current.

[SiliconWizard]

Yes you could just increase R1 AND R2 and keep the same ratio.

Just as a side note, I don't think the AD8479 is specified to work at 3.3V single supply. I think it should be 5V min (+/-2.5V min dual supply). So I think you're out of spec.
Additionally, it has a limited operating range (like all diff. amps) meaning that you will get in trouble if your differential input gets too close to the power rails. In your case, it may be a problem when the voltage across the load gets too low (since you're on a single supply). One way of mitigating this is to offset the output by connecting the REF pin to a voltage reference above 300 mV or so (see datasheet: "Because the output can swing to within only about 0.3 V of either rail, an offset must be applied to the output.").

Another option would be to use an instrumentation amp. such as the AD8226 that works down to 2.2 V, and attenuate both inputs (with two voltage dividers to ground), thus limiting the common-mode voltage as well as the differential. You'd still have to connect the REF pin to a voltage ref to offset the ouput and keep it in its operating range for low differential voltages.

[nitish111]

Thanks for the reply. I increased R1 and R2 by factor of 7 coz I had resistors of those values laying around. This helped in reducing current but ADC voltage got saturated to 2.5 V.

I am looking for a way to isolate or mirror voltage in some way so that I can reduce the voltage after isolation without disturbing the load.

[David Hess]

The simple solution if you want to keep the existing circuit is to add a voltage follower powered by the 40 volt supply to drive R2.  Then the error is reduced to the bias current of the voltage follower which will typically be nanoamps to picoamps.

[Zero999]

Why not use the ordinary op-amp differential amplifier configuration? It has the advantage of not needing a part rated to >40V.

The constant current sink needs more than 40V, to put 2mA into a 20k resistor.