Hi,
I’m trying to use some Opal Kelly boards (mainly XEM7310), to control an analog board and read digitized data from it.
However, in my signal readout I notice some spectral lines that are possibly due to digital noise coming from the OK. I’m wondering if anybody has already had a problem similar to mine and solved it somehow, either via HW or via FW.
So far I’ve tried adding a copper shield to cover the OK’s components closest to the analog board and that seems to have mitigated the problem, but it didn’t solve it completely.
I’ve read for example in this post that another user got noise injected via the GND, so if this is the case for me as well is there any recommended way to handle this kind of crosstalk.
Any suggestion on the topic of digital noise reduction is much appreciated.
In noise-sensitive applications, it’s commonly required to shield either the noise source or the sensitive components (and sometimes both). It sounds like you’re on the right track by shielding the Opal Kelly board. You might see an additional improvement if you’re able to enclose more of the OK board within your copper shield structure. Make sure you have good connection from the shield to the PCB ground. This helps to keep the radiating circuitry completed surrounded by the Faraday cage structure provided by the shield and the ground planes within the PCB.
We have a white paper about EMC testing that might be useful for you. Section 3 lists potential contributors to digital noise floor and gives some ideas about how to mitigate them.
Understanding the frequencies of your noise spurs might help to narrow down the specific source(s). For example, if they seem related to the clock and data rates of the board-to-board signals, you might be able to make an improvement by reducing the drive strength at the digital signal source.