i have a experiment in which i need to fix R and N, so that Fpfd would change continuously, Ref in sweep (1 MHz to 200Mhz) nd RF IN (200MHz to 4 Ghz ) what should be the optimum value of R and N to do so ?
Is there a particular part you have decided to use?
I'm not exactly sure what your doing, but you need to choose values of R and N that will always give you a frequency which the PFD can handle. I'm assuming your prescaler value, P, must be fixed too.
Let's say you chose the ADF4106. The maximum PFD frequency is 104 MHz.
If P = 32, N = 2 and R = 2, then you will meet all the PFD requirements for you input frequency ranges. However, the N and the R signals at the PFD will be different frequencies!
These numbers will give you signals that the PFD can handle, but I'm not sure what you're doing after that!
The ADF4007 may be of interest to you. It has a hardware controlled N and R dividers: http://www.analog.com/en/rfif-components/pll-synthesizersvcos/adf4007/products/product.html
but if i choose that value of R and N so PFD would be the same, like REf in
1 MHz and R=5 therefore Pfd=200KHz and iF We have RF IN = 200 MHz , let
N=1000 , there fore we can have Fpfd=200KHz. now i will sweep both the
Input frequency so that R and N =constant, and also having the same pfd
instantaneously as it would sweep from 200Khz to 4 Mhz. i can do that can't
On Fri, Jul 19, 2013 at 11:03 PM, rbrennan <
I think that should work but I've never seen that implementation before. The main requirement is to meet the maximum PFD frequency and the maximum prescaler output frequencies.
What are you connecting to the charge pump output? If you have a loop filter, the filter dynamics would be constantly changing as the PFD frequency changes.
actually i have an experiment in which i have to check which of the two
signals ( rf in and ref in ) is higher , so to do that i only need to see
the out put at charge pump pin of the chip i.e 2 pin. so i don't need a
charge pump and a VCO.
On Mon, Jul 22, 2013 at 11:14 PM, rbrennan <
It sounds like Shrey is using the PLL as a relative frequency detector.
In a simple case, suppose R and N are equal. Then if the RefIn is larger than the RFIn the charge-pump will pump in one direction, and if RFIn is larger than RefIn then the charge-pump will pump the other way. The specific direction associated with each case is given by PD polarity. So a circuit can tell if RFIn>RefIn or RFIn<RefIn. By setting R and N to particular values multiples can be added to this, so it can tell if RFIn/100 is larger or smaller than RefIn/3. Shrey, is that what you're looking for?
One problem with this, as Rob Brennan mentions, is that the frequency limits on the N-counter and R-counter are different. It may be easier to do this kind of comparison using the N-counter for everything. Suppose that a fixed RefIn and PFD are used. Firstly, one signal is applied to RFIn and a set of N values are tried. For some the charge-pump will pump up and for others it will pump down. Then the other signal is applied to RFIn and the N values are tried again. The advantage of this is that both input frequencies can be anywhere in the RFIn frequency range of the PLL. Two PLLs could be used for this if necessary, one for each frequency.
On some Analog PLLs it's possible to get the N-counter output pulses from Muxout, that could be useful too.
ya i'm looking at the first case you mentioned having ref in =5Mhz to
100Mhz and RF in =200 Mhz to 4 Ghz
for that i have fixed R=5 and N=200 stepsize= 1 Mhz to 20 Mhz. I wanted to
know if the above specs could be used in my experiment as a relative
On Tue, Jul 23, 2013 at 10:56 PM, RobThorpe <
You can do that, though you may encounter a few problems. Keep the following in mind....
Your proposed setup will detect if RefIn/5 > RFIn/200. When the result of this comparison is small the charge-pump will give out very narrow pulses. It may be difficult to detect those pulses. So, it may be useful to filter the charge-pump output.
Lock-detect is only meant to be used for frequency synthesizers, it won't provide any useful information here.
On the parts with internal VCOs such as the ADF4350 and ADF4351 the "autocal" process runs before each lock to pick the band and VCO core. It requires that the PFD isn't changed after locking. So, you can't use those parts for this application.
Retrieving data ...