Capacitor in place of battery to measure noisy low-power devices

[sebmadgwick]

I need to measure the power required by a battery powered device in sleep mode.  The device uses a DC-DC converter (TPS63000) which makes measurement of the current difficult.

I used a uCurrent connected to a DS4024 scope to display the battery current.  By switching the uCurrent range between uA and mA, I can see that the current is just over 100 uA for a majority of the time but peaks at 250 mA.  It is not possible to obtain meaningful current measurement given this dynamic range.

My solution is to instead power the device from a capacitor and measure how long the capacitor takes to drop by a given voltage.  For example, a 3300 uF capacitor dropping from 4 V to 3 V in 20 seconds is equivalent to a 21k resistive load.

I was wondering if this is a common technique for measuring power consumption of noisy devices?  And if there are any tips, tricks, or traps I should be aware of?

[danadak]

Use a low leakage capacitor.

Also examine C vs V curve for the cap, some exhibit large C change
with applied V.


http://www.cliftonlaboratories.com/capacitor_voltage_change.htm


Regards, Dana.

[BrianHG]

I use this technique smooth out all the bumps when measuring battery powered devices which go In & Out of sleep with HUGE current spikes while drawing in the ua range for most of the time, like pulsing a high current LED once every 2 seconds...

Send power to a first main cap, a series precision 10 ohm resistor on the + between that cap and the + of a second equivalent cap which then powers your device.  All the - should be connected to GND.

Now, on the power source side of the 10 ohm resistor, place a 1k resistor to a 1000uf cap + and the - of that cap to the other side of the 10 ohm resistor.

Now, place your meter on mv and measure the voltage the 1000uf cap.
Every 10mv = 1ma.  If the reading is still too inconsistent raise the 1000uf.
If the reading takes way to long to settle at a particular reading, shrink the 1000uf cap.

Changing the precision 10 ohm to 100ohm means 100mv = 1ma.
(Hope I got my decimal places correct...)

Adding additional series resistors between the PSU and first cap, like 10 or 100 ohm, will also smooth things out even more as well as using something like 470000uf caps as the first and second cap.

[sebmadgwick]

Thanks for the link, danadak.  It confirms my understanding that "electrolytic and tantalum capacitors change capacitance and dissipation factor very little with applied voltage".  Ceramics, etc. would not be appropriate for this application anyway.

I expect that leakage is not a concern for discharge periods of the order of 10s of seconds.  However, it may be beneficial to design for a much longer periods to decrease the required voltage drop and/or increase the accuracy of the measurement.  Leakage could then become a concern.

[Siwastaja]

Tantalum would be optimal from the capacitance stability viewpoint. Leakage is probably OK for 10s of seconds.

[danadak]

Of course T  vs C and leakage effects an issue for precise measurements.

http://www.yageo.com/exep/pages/download/literatures/E-cap_2011.pdf


Regards, Dana.