| View previous topic :: View next topic |
| Author |
Message |
MotoDan
Joined: 30 Dec 2011
Posts: 55
|
| Jitter-Free Clock Divider Using NCO |
 Posted: Mon Jan 16, 2023 11:48 am |
 |
|
Hello all,
I'm working with a PIC16F18426 and am planning on using the Fixed Duty Mode (50% square wave) of the NCO to provide a precise, jitter-free output with an external clock as the NCO clock input. The external clock to the NCO will be a TCXO which I need to be able to divide down to various frequencies.
From what I can tell, the only way to get an external clock to the NCO is via the CLC LC1OUT. I plan to route the external clock to an CLC input which will basically just pass-thru to the LC1OUT1 which will clock the NCO.
I could use a separate counter such as a CD4040, but thought I would try doing the division with the NCO.
My main concern is jitter. I have not used the NCO before, but know from past experience that dividers based on PIC timers do not produce a stable output.
Has anyone been able to get a stable output from the NCO? My project can not tolerate any jitter on the divided output.
Thanks! |
|
 |
temtronic
Joined: 01 Jul 2010
Posts: 9097
Location: Greensville,Ontario
|
|
| Posted: Tue Jan 17, 2023 10:00 am |
|
|
Well that just seems 'silly'... can't tie an external clock to the NCO ! You would think THAT would be an option.
I've never used NCO, hmm...'jitter'.. I wonder if when you change the NCO, you might get 'jitter', regardless of the input clock to it ??
It's probably one of those, drag the scope out, cut code and test in the REAL WORLD time
used a LOT of the CD4060 years ago,when I needed a 24Hz clock.....
sigh, I feel old today...thanks.
Jay |
|
|
MotoDan
Joined: 30 Dec 2011
Posts: 55
|
|
| Posted: Tue Jan 17, 2023 12:16 pm |
|
|
Jay,
Apparently some PICs do have an external input to the NCO. I may keep looking to try and find one that does as well as meets my other criteria.
As far a the jitter rewquirement goes, I'm not concerned about any initial jitter that may occur after updating the NCO. I just need for the NCO output to be stable. I'm hoping that the NCO hardware will not be "interfered with" by other PIC operations.
I gues my age is showing as well, The 4060 counter (and others) were the mainstay back in my early days and still seem to be in demand based on DK's in-stock offerings.
I'll report back once I've tested the NCO... |
|
|
dyeatman
Joined: 06 Sep 2003
Posts: 1912
Location: Norman, OK
|
|
|
MotoDan
Joined: 30 Dec 2011
Posts: 55
|
|
| Posted: Tue Jan 17, 2023 1:07 pm |
|
|
| Thanks, I've seen that link and several others. |
|
|
Ttelmah
Joined: 11 Mar 2010
Posts: 19215
|
|
| Posted: Wed Jan 18, 2023 2:50 am |
|
|
An NCO inherently produces jitter. However it is just one clock. A standard
counter will not have this. The PIC data sheet itself tells you this:
| Quote: |
This linear advantage over divide-by-n timers comes at the cost of output
jitter. However, the jitter is always plus or minus one NCO clock period that
occurs periodically, depending on the division remainder. For example, when
there is no division remainder then there is no jitter, whereas a division
remainder of 0.5 will result in a jitter frequency one half of the overflow
frequency.
|
It'll depend on the division you select. So for some divisions there will never
be a remainder, and so no jitter.
If you can chose a multiplier that is a multiple of the divisions you want to
select, voila, no jitter. If not, you will have jitter.
The standard way frequency synthesis is done, is to effectively have an
NCO feeding a PLL. |
|
|
MotoDan
Joined: 30 Dec 2011
Posts: 55
|
|
| Posted: Thu Jan 19, 2023 10:58 am |
|
|
| Well, the results are in. As Ttelmah suggests, there is not discernible jitter when the NCO increment (count) is a binary multiple. Otherwise, the NCO will automatically adjust a separate count (accumulator) to try and average out the desired frequency and thus will cause the NCO's output to wander or jitter. |
|
|
|
FAQ 
Search
Register
Profile 
Log in to check your private messages 
Log in
