One of our major products uses DS90LV019TMTC/NOPB and is showing issues in production. During testing, we are having 20% of our boards show noise with the part. Sometimes it is just on the two differential receive lines, and sometimes on the ground and vcc pins as well. Sometimes attaching unterminated cabling brings the noise. Other times it's just there with the board having no cabling attached to the connector that sends the the differential lines out to another board. The noise sometimes looks like it has a distinct fundamental frequency. One thing to note is there are ferite beads connecting the board's 3.3V rail and ground to the part's vcc and gnd pins. the vcc pin has bypass caps. Often the noise will be on the power and ground pins of the part, but it will not be on the other side of the ferrite beads that are connected to the actual ground and 3.3V. If we remove the ferrite bead, the problem sometimes gets fixed. Shorting the pins to actual 3.3V and Ground will tie the pins to a stable 3.3V and 0V. Other times the noise doesn't stop and the part needs to be replaced.
We did a lot of isolating to isolate the noise to the chip. We've had luck with a different manufacturer's LVDS chip, but didn't know what the issue was and if there was anything we could do to make this chip work. Image of the basic setup below: