Strange uCurrent reading anomaly

[nathancrum]

I have an odd uCurrent reading anomaly I'm very curious if anyone can figure out.

I recently bought a uCurrent for work and was measuring the current draw of an Analog Devices ADXL362 accelerometer.  In the sleep/wakeup mode I was getting an expected ~250nA draw.  Active mode, the nA range is maxed so I switch to uA and I get -116uA?? 

First I thought maybe I had it connected backwards?  Double check connections - nope - connections are right.
Then maybe battery is dying?  2.9V - nope.  Changed out a fresh 3.0V CR2032 anyway.  No difference.
If I measure with the DMM directly - I get ~5uA (expected draw).
If I measure voltage on the current terminals of the uCurrent I get 0.04mV on 10ohm (uA range) and 48.35mV on 10kohm (nA range).  Those both agree with ~5uA current draw just fine.

So I thought maybe my uCurrent was busted.  So I measured current on a 100k resistor with a variable DC supply from 0.1 to 1.0V and nope - all measurements are perfectly fine.  I got 1.0 to 10uA no problems.

I put it back on the ADXL362 and I get -100uA again.   

If I add a 0.47uF cap on the power rails the output on the uCurrent drops to -65uA.
If I add a 10uF cap on the power rails the output on the uCurrent drops to around -1uA.

So - I assume there is some HF black magic going on here, but it is well beyond me.  My scope is being used on another test setup at the moment, but I will probe it for more details when I get a chance.

If anyone can think of a reason for why I get such bizarre readings using the uCurrent and the ADXL I'd love to hear it.  The uCurrent appears ok, but something very strange is happening when connected to the ADXL and beyond the apparent affect of some kind of HF oscillations I have no idea why.  Perhaps this would make a good EEVblog episode - I'm sure there's a trap for young players in there somewhere.   

[Rerouter]

Woudl this device be hooked up to something with a slightly higher supply voltage? if so the input protection diodes could be leaking that beck into the supply your measuring, (e.g. using a 3.3V device with a 5V micro?)

[EEVblog]

Sounds like it has to be a very specific issue with your setup. Likely impossible to troubleshoot unless you are actually there.

[Jeroen3]

How many possible power sources are electrically connected to the circuit under test? Even when it is only ground.