I’m trying to run the H-Field Probe Demo 1 to make sure that I use the H probe correctly. However, I can’t get the result similar to the demo video “CW Tips: H-Probe Usage for Side Channel Analysis.”
At first, I didn’t change anything in that ipynb file except changing platform to ‘CW308_SAM4S’. With the gain set to be 70 (I also tried to set the gain to be 50), I get the waveform to be something like this:
Also, the CW Husky’s ADC and Glitch red lights were blinking. I tried to print out the error message with “print(scope.errors)“ and I got the following message:
The Husky ADC errors are sticky, so make sure you clear them after you adjust your gain before trying again, otherwise you could be chasing your tail.
Your traces with gain=70 are clearly clipping (max ADC range is +/- 0.5). 50 is much better but may be a bit too high as you move the probe around.
Our video uses an XMEGA target; the power traces from an XMEGA will look very different from a SAM4S, they’re two totally different processors. Don’t rely on what you think traces “should” look like.
You will have to move the probe around a lot to find a good position, unless you’re lucky. Patience is key.
In our demo notebook we suggest using TVLA: this gets you away from “what do I think good traces should look like” and instead focuses purely on “are these traces capturing any useful side-channel leakage”. Because the latter is ultimately what you’re probably interested in.
Thank you for your response! I solved the ADC errors by lowering the gain (also clear the error messages after I adjust the gain).
Also, I got some fine results using TVLA, but I’m still trying to find a better position to achieve higher t-value.
