The Domino wave is most stable at 5 GSPS, slowly degrades down to 3-2 GSPS, and at 1GSPS gets some significant jitter. This is for internal reasons in the chip and cannot be compensated by the loop filter. It is therefore important to run it as fast as possible if you want to achieve best timing resolution. As a rule of thumb, the jitter at 5 GSPS is about 20-25 ps, and at 1 GSPS it is maybe 150 ps. If you require good timing resolution, you can use the 9th channel to sample a stable reference clock (100 MHz for example) and measure timing relative to that clock. This way you can bring down the resolution to a few ps at 5GSPS and to maybe 40 ps at 1 GSPS.
We have been using DRS4 IC in our design for quite some time and it is giving good performance.
Till now we were using Domino wave frequency as 1 GSPS by use of reference clock to DRS4 and internal PLL of DRS4. Recently we tried to use 4GSPS by modifying the reference clock.
What I have found that DRS4 domino wave is more stable at 4 GSPS as compared to 1 GSPS by doing the timing jitter analysis. I am not sure if it is the property of DRS4 IC to be having more stable domino wave at higher frequency (by design) or it is due to some external effects like PLL loop filter or any other on board parasitic effects.
Please share if anyone has done any study of DRS4 Domino wave stability at different sampling frequencies.