C
colin
Hi,
Maybe this is of interest i dont know if anything like it exists but i
couldnt find anything, ... I thought i would try and see if i could improve
precision phase detection in my RF PLL by reducing the RF output of a mixer
based phase detector at its zero phase point.
I did this by feeding a-b into one port so at in phase its RF output wil be
zero and a+b + 90' into the other port.
Although its doubtfull this would actual improve overal noise perforance, I
didnt calculate the actual response, I just tried it but i was surprised to
find that the phase response had a definate sawtooth shape.
I then realised that the direction of the sawtooth would change depending on
wich signal was higher in frequency, and that feeding this through a
capacitor into antiparallel diodes meant that i get an average of +/-50mv
signal polarity coresponding to relative frequencies wich is basicaly a
frequency detector.
This means that the benefits of a type II PFD (zero phase offset, low op
ripple, and wide lock range) can be acheived with out the use of very high
speed flip flops and their asociated noise spikes and potential dead band.
I since tried to calculate the actual phase response curve using trig of
(a-b).j(a+b) but couldnt duplicate anything like it, until i realised that
the LO input must be overdiven and the A-B must have some slight inbalance.
Colin =^.^=
Maybe this is of interest i dont know if anything like it exists but i
couldnt find anything, ... I thought i would try and see if i could improve
precision phase detection in my RF PLL by reducing the RF output of a mixer
based phase detector at its zero phase point.
I did this by feeding a-b into one port so at in phase its RF output wil be
zero and a+b + 90' into the other port.
Although its doubtfull this would actual improve overal noise perforance, I
didnt calculate the actual response, I just tried it but i was surprised to
find that the phase response had a definate sawtooth shape.
I then realised that the direction of the sawtooth would change depending on
wich signal was higher in frequency, and that feeding this through a
capacitor into antiparallel diodes meant that i get an average of +/-50mv
signal polarity coresponding to relative frequencies wich is basicaly a
frequency detector.
This means that the benefits of a type II PFD (zero phase offset, low op
ripple, and wide lock range) can be acheived with out the use of very high
speed flip flops and their asociated noise spikes and potential dead band.
I since tried to calculate the actual phase response curve using trig of
(a-b).j(a+b) but couldnt duplicate anything like it, until i realised that
the LO input must be overdiven and the A-B must have some slight inbalance.
Colin =^.^=
