Hey SHH,
1) Those bumps make me think of noise on the lines and they appear periodic to me. Is the customer able to confirm the ground connection is good and their scope probes are good? If they have an inductor (motor or mechanical relay) turning on and off at a high frequency, that could cause the ground to bounce a little and cause spikes.
2) Also the environment of testing could be an issue and the SCL/SDA lines could be picking up EMI through the air. I've also seen power supplies transmit EMI at 60 hz~120Hz range and sometimes the harmonics couple onto the signals as well. You may want to try pulling the device away from any lab PSU and turning surround PSU off.
3) You may want to ask them to watch Vcc and see if it bounces when you see these small bumps during the pull down of SDA/SCL. If that is the case, then adding more capacitance on Vcc could help.
4) Hooking this up to a spectrum analyzer may tell us what frequency these spikes are occurring. If the frequency is higher than our clock frequency we could always filter it.
This short paragraph pdf may also be of interest:
Also, using small series resistors could help dampen these spikes.
Lastly, adding small capacitance on the SDA/SCL bus may help get rid of this if they have room to do so. This isn't a normal suggestion but if the ripple is bad this would help.
I don't see any immediate concern if the device is working properly. The output/inputs are the drain/source pins of FETs so these small ripples won't damage it.
If you run into any problems, let us know.
Thanks,
-Bobby
