We have an intermittent loss of lock problem when using the ADF4151 PLL. For about 3 % of step frequency changes, the VCO tuning voltage ramps up quickly well past the correct VCO voltage, flattens out and after around 20us it ramps down and locks to the correct frequency after about 40us (we need to lock within around 20us). Using the Mux output we have established that the N divider output disappears for around 10us immediately following the end of sending the frequency change request (LE change from low to high). This happens for small frequency steps as well as large ones, is not affected by the charge pump current and is not affected by the use of fast lock. We are using it in Fractional mode (DB8 in reg 2 set to 0). The problem is greatly reduced (less than 0.5% of frequency steps) if we set the Anti-Backlash Pulse Width to 3ns (DB22 in reg 3 set to 1) even though we are using it in Fractional N mode.
The picture, Scope image 1.jpg below, shows the problem. The VCO voltage is at the top in yellow, the N divider output from the Mux is in the middle in blue and the LE input signal is at the bottom in pink.
The frequency is stepped by a small amount at the end of each LE low to high input. Most work fine with just a small deviation in the VCO tuning voltage and a constant output from the N divider but when there is a problem the N divider output disappears and the VCO voltage shoots up rapidly.
Scope image 2.jpg below illustrates how the N divider output is lost at the moment the LE line is changed from low to high.
Scope image 3.jpg below shows that the loss of N divider output is for around 12us.
This problem happens on many synthesizers we have tried (at least 10).
Please can you help: Why does the N divider output disappear intermittently? Why does setting the Anti-Backlash Pulse Width to 3us improve it?
Many thanks in advance,