Use µCurrent Gold with wide current range 50nA - 20mA

[G7exxh]

I am working on a very low power device and got the µCurrent Gold current adapter to do some precise measurements. The device uses a low power system timer (Texas Instruments TPL5110) to power the main circuit with the MCU and other components.

The device is powered with a CR2032 3V battery. While idle, the device takes only around 50nA of current, but if activated a LED will consume around 20mA current for 40ms.

I read there is no overload protection in the adapter. Does this mean, if I use the adapter with the 1mV/1nA setting, it will get damaged if the LED will draw 20mA?

[Gyro]

Hi, welcome to the forum. 

You are correct about the overload protection but that wouldn't be a problem in this particular case because your CR2032 power source doesn't exceed the supply voltage of the opamp in the uCurrent.

Your problem comes from the burden voltage on the 1mV/nA range. I believe this uses a 10k shunt resistor, meaning that on the 20mA peaks the entire supply voltage would be dropped across the shunt resistor and your device wouldn't work. (20mA * 10k = 200V).

This is a problem with such low drain devices with such widely varying current ratios. A few alternatives I can think of:

1. Change your software to disable the peaks so that you can just measure the quiescent current (undesirable as it changes the operating conditions).

2. Use a much lower current shunt resistor and accept that you need a much higher resolution meter to measure the voltage drop across it (inaccurate).

3. Try to integrate the current draw so that you can read the average current consumption, which is the figure you're probably trying to obtain anyway (for battery life calculation)....

You can probably do this by buffering the supply to your device with a large value capacitor. You would need to pick a capacitor big enough that its voltage wouldn't drop significantly during the 20mA current pulses. To be large enough it will need to be an electrolytic in practice, which will have its own leakage current. So:

- Connect the uCurrent between the CR2032 and the Capacitor
- Wait for the capacitor to charge and it's leakage current to settle. Depending how long it is since the capacitor was last used, it could take a considerable time for the leakage current to settle. There probably isn't a sensible way to speed this up, but it will hopefully end up stable in the 10s of uA region.
- Then connect your device across the capacitor, being very careful not to interrupt the supply to the capacitor or short it.
- The difference between the capacitor leakage current and your new measured current level should be average current draw of your device.
- To be on the safe side, disconnect your device again and check that the leakage current of the capacitor again to make sure it hasn't changed due to the current pulses drawn by your device.

That's as near to a practical measurement as I can think of. . Ideally (to be realistic as possible) you want to size the capacitor so that the voltage drops by about the same amount as the CR2032 does under the same current peaks, (check on a scope). So some experementation would be needed.


P.S. You might be able to get away with a low voltage Tantalum capacitor, in which case its leakage current and settling time would be much lower.

[G7exxh]

Hi Chris,

Thank you very much for the detailed response!

You are right, I need this precise measurements to calculate the battery life of the device. I especially would like to find any sort of current leaks in the device to fix them of add them to the calculation.

Your approach sounds very interesting, I will try this out.

My main concern was to destroy the adapter, if accidentally the LED would draw 20mA, but you are right in my setup this actually can not happen. If the solution with the capacitor does not work, I will simple measure the individual stages of power consumption separately - then just create a timing "diagram" using the mA mode and an oscillator and calculate an approximate power consumption, using the timing and the precise measurements.

[Gyro]

You're most welcome.

If the solution with the capacitor does not work, I will simple measure the individual stages of power consumption separately - then just create a timing "diagram" using the mA mode and an oscillator and calculate an approximate power consumption, using the timing and the precise measurements.

That sounds like a very sensible alternative. It would be interesting (to others too) to see how well the two approaches match (ie. how well the capacitor method performs). A significant difference might indicate that your device isn't quite doing what you expected of course.