ChipWhisperer Lite EXTCLK

I’ve been playing with the ChipWhisperer Lite board and I am not able to figure out how to use the external ADC clock source feature.

It seems that the clock signal is supposed to be fed into an input pin somewhere but I am unable to figure out which one to use.

Hello,

The “HS1” and “HS2” are the clock output and input pins respectively. This isn’t well documented will fix ASAP, as it’s documented on the old CW-Capture page!

So if you feed a 3.3V clock into the HS2 pin (either via the 20-pin header, or the 10-pin single-line header which has HS2 marked on the silkscreen) it should work. You can use the ‘frequency counter’ feature to confirm this.

Regards,

-Colin

It actually looks like HS1 is the clock input. Other than that, the frequency counter is showing the correct value now. The current problem I am facing is, that the DCM is losing lock on EXTCLK. I guess this could be because my DUT is running too slowly (~4MHz)? When the ADC source is set to CLKGEN running on 3.3MHz, it seems to work fine though.

EDIT: It also seems that just connecting the clock induces a huge amount of noise into the measurement, is that normal or just my somewhat flaky setup?

Thanks for your help

Ah damn… not having HW in front of me made it harder to double-check if HS1 or HS2 was correct, apologies!

The DCM has more trouble locking to an external signal if it’s somewhat haphazardly fed in, so it might be a frequency thing. However it could be the signal isn’t very clean due to missing ground or termination issue… can you try adding a series resistor (somewhere in 22 - 100 ohm range) on the clock pin?

The noise might be related to the above, it shouldn’t change things too much. If the clock isn’t reliable it can cause apparent noise due to issues with the ADC/timing margins.

Do you have more details of your setup? Where’s the clock coming from etc.

The resistor fixed the locking issue. I still have quite a lot of noise when measuring with CLKGEN as ADC clock. This noise seems to be present with EXTCLK too.

The clock is supplied by an STM32L162 on its MCO pin:

Sampling with CLKGEN at 16MHz:

Sampling with CLKGEN at 16MHz and MCO connected to the HS1 pin.

Sampling with EXTCLK x4 (16.8MHz):

Anyway, assuming that the noise is just the MCU clock added to the trace, it should not really affect the attacks right (except for the clipping, that could be a problem I guess)?

EDIT: So it seems that the DCM is still loosing it’s lock, altough it is somewhat less frequent.

The extra noise will make things somewhat more difficult (due to worse SNR)… are you routing both a CLK and GND pin? i.e. if you only route the clock as a single wire it will have issues due to lack of return path… even better is a coaxial cable, but just two wires side-by-side (i.e. piece of a ribbon cable, or twisted-pair cable stolen from ethernet cable) will be better!

Connect both the clock and ground on both sides of the system (MCU and CW-Lite board, there is GND pins beside the HS1), I’d still use the series resistor too.