0
\$\begingroup\$

I have a 5 V 5 A power source which is connected to a simple development board with a processor that draw current dependent on utilization. I use a multimeter that measure 1000 samples / s to measure energy consumption. I measure a voltage drop across a shunt resistor coupled in series with the input power to calculate average power and energy consumption of the development board. My question is : What can we say about error in this case ? Between two samples with a 1 ms interval, how frequent and how much can the power fluctuate in this interval ? Is there any way to find this out ? It would be unfortunate if there are several violent spikes between the samples.

\$\endgroup\$
  • \$\begingroup\$ What you can say about the error is that the magnitude of such spikes in the power drawn is under the control of the PCB designer. They can be reduced to any level desired by increasing the reservoir capacitance and improving the power distribution. Andy's answer is tantamount to doing this just for the measurement process, but it can be a useful improvement to the basic design. \$\endgroup\$ – Brian Drummond Apr 16 '13 at 14:04
  • \$\begingroup\$ hmm, ok, thanks. Is there anything we can say about this board : arndaleboard.org/wiki/downloads/supports/… ? There's schematics of the input power at section 6.1. \$\endgroup\$ – user1511956 Apr 23 '13 at 17:49
1
\$\begingroup\$

In order to overcome these sort of sampling measurement errors you can make the problem "easier" by adding a large electrolytic capacitor on the power feed to the development board after the series measurement resistor. This is an "attempt" to measuring average current taken by the dev board. OK, I say you are measuring "average" but it depends how big your capacitor is in uF terms and how long your "violent spikes" are.

If you have a 1 ohm series resistor and a 100uF capacitor, the time constant of this is 0.1ms so you'll need more capacitance or more resistance. To get reasonably decent measurements go for a time constant of about 10ms - this should suit your multimeter.

As an aside, your multimeter may be perfectly capable of averaging over a 1ms period and giving you the results you think you may not be getting.

\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.