A puzzle about PLL in the locked state

We know that if charge pump has up/down current mismatch, it will create a phase offset in the PFD input when the PLL is in the locked state.

If the PLL is locked then there is no net charge going into the loop filter and then there is no reference spur problem because the current mismatch is compensated by the constant phase error. So the puzzle is why people still observe reference spur although it is compensated by the phase offset.

Welcome to discuss.

In that ideal case, you won't see any reference spurs.

Now, add a small amount of leakage current in the charge pump, and a small amount of leakage current in the loop filter. Now, your loop filter output voltage is decaying, and the mismatched charge pump's have to pump current back into the charge pump. This gives you reference spurs.


Dave

Hello Dave,

Can we just treat the leakage current as one kind of mismatch? In that case we can not see the reference spur.

One possible expalination is that the very small leakage current will have very small net charge feed into the loop filter and the resulted loop filter voltage will cause very small VCO control voltage change. Because the loop gain is not infinite big so there is not enough correction on the phase error caused by this leakage. Is it relative the loopBW or loop gain? I am not sure.

There several things going on here, and they all operate to increase spurious content at the reference frequency. I'm going to assume that you have a loop filter that doesn't provide much attenuation at the reference freuqency, unlikely in practice.

Leakage current first. Even if you have perfectly matched charge pumps, you will see a spur at the reference frequency because that is the frequency that you are pumping current into and out of the loop filter. If you are locked in an ideal system, then you won't see any spurious because there isn't any leakage. If you add leakage current to this case, then you will see some spurious content at the refence frequency.

If you have charge pump mismatch, then the reference spurs will be larger. Again, without leakage currents, you won't see any reference spurs when the pll is locked.

As far as what happens when the PLL drifts off lock, that depends on how small the dead zone around lock is. If the dead zone is large, then the PLL will have lots of room to drift, and I suspect that the spurs will be larger.

Any more than this is going to require numbers instead of handwaving.


Dave

I think even with perfect matching, very small spurs will be there!
This is got from simulations!