Is the NU and ND signals output by hmc439 DC? How many error voltages in the graph of error voltage versus phase are the NU or ND of the HMC439? Or is it the output of the filter, and how does the filter in the reference circuit in the device manual work?
ad1/ I think these pins are Open collectors as you can see on the "Functional Diagram".
ad2/ I assume it is an output of subtractor (difference between NU and ND), because the output voltage of NU/ND cannot be negative.
If you understand the output of NU and ND in this way, then the manual says that the output is high 5V low 3V peak-to-peak 2V, indicating that the output is an AC. I also measured that the end of the AC is the amount of DC, but the peak-to-peak value of the AC is not much worse than 2V. How can this be understood?
For the loop filter is the DC voltage important. As you can see on the figure "Error Voltage vs. Frequency", the voltage is marked here as Vdc (the input frequencies REF and VCO are identical).
Thank you for your answer, I am still confused? Assume that the NU output DC pulse voltage is 3V~5V, and the ND output DC voltage is 3V (the phase of REF and VCO input frequency is p/2). Then the error voltage is the voltage effective value of NU output minus the DC voltage of ND?
I don't understand exactly, what do you mean with term "DC pulse voltage". Do you mean amplitude of pulse? Or average value of periodical pulse waveform? Or voltage high/low voltage levels of single pulse? Or anything else? Normally PFD works this way: if VCO rising edge comes first (phase or frequency is higher then REF) signals ND and D are active - you can measure pulses there. NU and U are not active, there are no pulses, just constant voltage. And vice versa for F.ref > F.vco. If the phase shift is Pi, then voltage at one output=output with pulses is (3Vpk_min+5Vpk_max)/2=4Vdc and second 5Vdc (constant). 5Vdc-4Vdc=1Vdc. I am not sure about this type of PFD, I described common behavior of PFD.