Questions about shunt resistor choosing


#1

Hi Colin,
I am trying to do power analysis on my target, which works on 3.3v. Without resistor, the average current while running is about 140mA.

I used a 10Ohm resistor as a shunt resistor with 5V power supply. I connected the shunt resistor between 5V
and target VCC pin, and measure the voltage change between VCC PIN and GND. (decoupling caps have been removed). However oscilloscope did not give a reasonable power consumption line.

Is there any thing wrong in my setup? Any guidance on shunt resistor choosing?

Thank you very much for any help!


#2

Colin has much more experience here than me, but my first though is why use a 5V supply? I understand that with a 10 ohm shunt the target will get ~3.6V on average, but it will get a much higher voltage when its current draw is lower.

Why not stay with 3.3V and use a much smaller shunt resistor?


#3

Hi,
Thanks for the advice. I tried a smaller shunt resistor (1 Ohm). and then I could stay at 3.3v to drive my target.
However I still could not get a reasonable power trace from this setup, even I remove most of the decoupling cap of power source. Could you help me more on this?


#4

Do you have a good trigger? It’s hard to say more based on what you’re reporting. What do you mean by “reasonable power trace”?


#5

I haven’t got a trigger right now. What I am doing is starting the scope, connecting power to my target and watching power traces.
by “reasonable power trace”, I mean with the shunt resister, the power traces did not show a big difference against to measuring the power source only, It just looks like random noise, without any clues of power consumption.
I also connected my target to chipwhisperer, it showed almost the ripple wave of power source, no obvious consumption peak found.


#6

Hi, of course there’s more than one way to do this but it can be harder to know what you’re looking at, or find what you’re looking for, without a trigger. What’s your target? Noise will always be present to some degree.
If you have a trigger and can capture the same identical operation many times, then you can average those traces to average out the noise and better see what you’re interested in.
Regards,
Jean-Pierre