Does AC coupling the 8 Gbps CML output of a ADCMP572 SiGe comparator allow you to terminate the end of it's transmission line to a different voltage, e.g. 1.2V, than the Vcco of the ADCMP572 (3.3V)?
Hello, I am Anthony Turvey, the design engineer for ADCMP572.
It's been over ten years since I worked on this design, but I did manage to resurrect the design and simulation environment for it. Looking at the output stage schematic I don't see any reason why you could not AC couple this device to an alternative 1.2V termination voltage. I think it will work just fine. Simulations also suggest that it works fine.
The output stage architecture is not very tolerant of a common mode voltage too far from the original design goal. So any transient event (like power up or power down) will result in a temporary non optimal working condition. Optimal performance will be achieved when the coupling caps have reached charge equilibrium and the internal common mode voltage has settled. With your example of 1.2V I wouldn't expect too much trouble.
So as with any AC coupled system there will be a tradeoff between coupling capacitor value and lower bound of the duty cycle. As I mentioned, the output stage was not designed with the intention of supporting common mode voltages too far away from VCCO, if even for a transient time period. Please keep this in mind as you select coupling caps versus the lower frequency components of your signal.
The other thing I'd watch out for is some power supply sequencing issue where your termination potential can push or pull displacements back through the Q and Qb pins as the coupling caps charge or discharge at power up or power down. Those transients will find their way back through the comparator supplies through the ESD diodes. It's not necessarily a problem, and latchup is not a concern with this SiGe process because it's fully dielectrically isolated. But again, just something to be mindful of as I did not simulate for those transient scenarios.
I would encourage you to give this a try. Without actually getting on the bench and trying it myself, I see no reason why it would not work just as you expect. I think it could be a great solution to level translate the outputs to an alternative termination potential.
We are using the ADCMP572 now and are seeing that it imparts about a 800mV DC offset to the serial link connected to it’s input. This is breaking the DisplayPort 5.4Gbps link that usually swings about 600mV around 0V.
Can we AC couple the ADCMP572’s input, say 0.1uF? If so, do we need a DC offset at it's input as well?
Retrieving data ...