PLL and phase detector

Discussion in 'General Electronics Chat' started by sharanbr, Aug 16, 2015.

1. sharanbr Thread Starter Active Member

Apr 13, 2009
76
1
Dear all,

I am trying to understanding PLL concept and I am having some conceptual doubts with respect to the phase detector.
I don't seem to understand how phase difference can indicate frequency divergence of output frequency with respect to reference input.

The PLL theory also talks about instantaneous phase of reference and output frequency. How can instantaneous phase be calculated.
Based on a sinusoidal signal, one can probably get phase value based on amplitude input if one assume circular notation.
But how can this be done, let us say, for example, for a square wave where the amplitude remains same at every instant during the high period & low period ...

Assuming that it is possible to measure the phase difference, how phase difference between reference and output frequency signal help.
One could have 2 signals with same frequency but with phase lag/lead.

Some articles actually talk about measuring whether phase difference is constant or not in order to see frequency difference.
But the circuits shown to measure phase difference just measure phase difference and not comparisons across measurements ...

I would like to get some clarifications on this please ...

Apr 5, 2008
16,954
2,957

Feb 17, 2009
4,166
1,188
4. dl324 Distinguished Member

Mar 30, 2015
4,731
1,026
@Jony130 Thanks for that. I've been meaning to read the PLL section in my old Signetics Linear Applications Manual, but could never get started. This YouTube video was easy to understand and pretty much to the point (unlike most which are a waste of time and bandwidth). I always wondered how they generated the clock ratios on the microprocessor projects I worked on...

5. Papabravo Expert

Feb 24, 2006
11,065
2,152
I don't think it matters if the original phase error is due to a frequency difference or to an actual phase difference. The purpose of the loop is to make a change and see if the error increases or decreases. It may take multiple cycles to achieve phase lock but the nature of the loop dynamics assures that it will eventually happen.

Last edited: Aug 23, 2015
6. sharanbr Thread Starter Active Member

Apr 13, 2009
76
1
I have tried reading these and have not been able to find exact answer to my question.

Let me re-state ... I have two questions ...

1) Most of the PD blocks are shown to measure the phase difference between input and feedback signals.
Now, it is very clear that there could be phase difference between two signals without any frequency difference.
Obviously, if one were to use instantaneous phase difference to measure the frequency variation then it would be erroneous.

Now, some literature talk about actually, taking phase difference and then checking if this phase difference is varying over a certain measurements in order to infer if there is a frequency variation. But this is not explicit and in the implementation, I don't
see this being shown explicitly.

2) My second question is also related to PD measurement again. Again, in some references, it is mentioned that individual phases are measurements and then difference is computed. However, in the actual implementation, I see that difference is what is measured.

My question is really, in case of sine or cosine signals, one can measure phase based on instantaneous amplitude. But this method fails if one were looking at a square signal where amplitude is constant during on/off duty cycles.

I hope someone will answer these questions ...

7. crutschow Expert

Mar 14, 2008
16,178
4,322
1) There are PLLs that can operate with a constant phase difference between the two signals but with no frequency difference (Type 1 phase detector), and there are PLLs that operate where the phase difference is adjusted to zero, steady-state (Type 2 phase detector).

2) You can use the waveform zero crossing points for the phase reference which works for both sine-waves and square-waves.

8. AnalogKid AAC Fanatic!

Aug 1, 2013
5,567
1,575
Without trying to be cutesy, phase matters because it is a *phase* locked loop. The intent of the circuit is to establish an output with both frequency and phase relationships to the input signal. A more simple circuit is a frequency-locked loop, where a specific phase relationship is not needed.

For all signals, especially non-sinusoids, there are places other than the signal peaks to use for frequency and phase measurement. One favorite is the zero-crossings. All periodic waveforms have them, and they are easy to detect. Even if the leading edge of a square wave is fairly slow, looking more like a trapezoid than a square, you can examine the waveform, remove any DC offset, determine the zero-crossing level, and use a comparator to produce a timing signal.

ak