THVD2410: A-B Spikes

Hi Shaofeng,

There could be multiple things that could cause this. Since it's periodic on each edge transition, we will ignore noise as a source

1) PCB layout, if one of the traces from the IC to the connector is longer than the other, the A-B signal could become skewed/dis-proportionate and result in what you see.

Note: the above is also true if the receiving board's traces are not the same length

2) The device itself could be switching/outputting with a slight difference in timing (A turns on slightly slower than B or vice versa). The difference results in a skewed A-B waveform.

3) The cable you are using could have slightly different lengths (even on twisted pairs) (probably not likely since you state it looks the same on 10m and 100m)

4) reflections could be occurring (probably not likely since you state it looks the same on 10m and 100m)

5) If you have some filtering caps on the A line to GND and the B line to ground and they are not the exact same value then this could happen (for example a 200pF cap on both A and B but with 20% tolerance, changing them to 1% tolerance would help fix this).

My guess is it's likely a combo of #1 and #2. You can try to fix this by putting a differential cap (a cap between A and B) of around 200pF to try to smooth this out. 

I think the waveform looks pretty clean though so you can ignore it if you want. If you want to clean it up because you want to lower the EMI then add a differential cap there.

-Bobby