Hi,

Is there a way to simulate the PLL response specification of AD9548?

I'm trying to find out the best filter configuration for my application. As the bandwidth of the loop filter in AD9548 is low, it took me a long time to try a group of settings on the eval board each time, and it's difficult to record and plot the test result.

I tried to build a simulation model, but I couldn't figure out the model of the TDC and PFD because it's not mentioned in the datasheet how this module is implemented. Do you have any solution to this problem? Or is there a simple system function that describes the response specification of the DPLL?

Dear Eggcar,

> Thank you so much for replying. I'm interested in time domain response currently,

> cause I need to know how long it takes the loop to get stable enough for an 1-pps

> disciplined atomic clock hold-over application.

I'm a little confused. Are you trying to estimate the lock time, or determine the stability in holdover? The stability in holdover depends only on the system clock input stability. Stability in equals stability out.

The lock time isn't easy to determine, partly because there is no universal definition of "lock."

Before going too much further, I'd like to make you aware that we have another 1pps capable part, the AD9545, which is newer, and has a few more features. However, we plan on keeping both the AD9548 and AD9545 in production.

A common lock time estimate is 20/(loop BW). When using the AD9548, one can speed up this process by first using a profile with a 50 mHz loop BW. After than locks, you can enter holdover, switch to 10 mHz loop BW, let it lock, etc. Also, the lock time is also a function of the damping of the loop.

> I was trying to build a transfer function and put it into Matlab to simulate, but I didn't know how

> to abstract some block in the loop. Could I find the transfer function in the excel file?

> If so, that would be great help.

I'll ask the designer, but once again, if it's lock time you're after, the above estimates are a good place to start.

-Paul Kern