I'm now trying to use ad9548 to produce low jitter and accurate 10MHz clock. The system clock source is an atomic oscillator, and the reference source is a gps receiver.
The problem is, the datasheet indicates the lowest loop bandwidth of ad9548 is 0.001Hz. But it's not low enough to filter the noise of the gps received 1pps signal. I tried to calculate the filter coefficient to achieve 0.0001Hz bandwidth, and it seems the short term frequency stability(1s - 10000s alan deviation) is improved observably, but I notice that long term frequency accuracy(8 hour average of the frequency measurement data) is drifting along sereral days while locking to the gps 1pps signal.
I'm not sure if ad9548 support loop bandwidth lower than the specific value. What could happen when I set lower bandwidth to it?
I know ad9545 could achieve 0.0001Hz bandwidth, I would try it later.