uCurrent design question

[alexwhittemore]

This one is mostly for Dave but do chime in if you have thoughts: I was thinking about the uCurrent adapter the other day and I realized that I wouldn't have used a shunt resistor setup for the low mA or uA/nA ranges. I would have thought, if we already need a really good super low offset amp anyway, why not set it up in an inverting current sink configuration across a larger valued sense resistor? The only requirement in such a configuration is that the op amp can sink all the current across a sense resistor when at the negative rail, which means the upper end of the mA range is something like 20. So why use a shunt at all when the active curren to voltage circuit gets you zero burden voltage? Especially when really accurate 1k resistors are much more common than .1 resistors?

Schematic in question: http://t.co/gMuWhaH

[tekfan]

It's a great idea but would it work with alternating currents? The great thing About the uCurrent is that you can measure low frequency AC and DC currents. The chopper opamp is just an amplifier and will amplify whatever voltage is across the resistor. You can then set the external multimeter on ACV range and measure the current.

[alexwhittemore]

For an alternating current, amplifying the voltage off a shunt resistor still requires swinging +output to -output at the frequency of the signal. Using my converter requires the same kind of swing. The only difference is that there's more current in and out of the amplifier. The only thing that I can think of being a problem is that if you're actually alternating sinking/sourcing 20mA of AC, the amp will heat up. But that definitely isn't a problem in the nanoamps range.

Also, the leakage currnt at the inputs of the amp is probably on the order of ~10 nanoamps, so how does amplifying even the shunt resistor not totally destroy the measurement?

[alm]

I believe this is what's known as a feedback ammeter. It's described in Keithley's low level measurements book (PDF version is available for free from their website after the usual registration nonsense).

[alexwhittemore]

Actually, it's the second google result for 'feedback ammeter,' here: http://www.keithley.com/data?asset=6169

Maybe that will answer my question as to the design tradeoffs, but the question is still open here as far as I'm concerned.

[EEVblog]

This has been discussed before elsewhere on here I'm sure.
There are a couple of reasons why I used the shunt approach.
I'd used the Maxim chip on other designs, and liked it, had some spare so I just wanted to use them. And I thought some theory on chopper amps would make a good addition to an article. (This was originally an article for Silicon Chip magazine)
I wanted a couple of hundred mA capability too, which ruled out just the opamp approach. Yes, I could have increased the current capability if needed, but just didn't want to go with the feedback approach anyway.
Plus I didn't need nor want pA resolution range.

Sometimes there are other factors that decide what approach is taken in a design, it's not just about what might be the better technical solution.

Dave.

[Zad]

The current measurement in the LCR meter in my blog uses this method of current measurement (integrated into the AD5933) and that uses a transimpedance amp which is specced up to 100kHz, so I can't imagine that AC is a big problem. What can be a problem though is instability resulting in oscillation. This requires a capacitor in the feedback resistor which is bound to compromise HF performance.