How to spec jitter for a digital NCO
in [DSP]
|
Prev: Kenlighten - A social network for knowledge seekers and providers
Next: LMS to remove the unknown sinusoidal signal
From: Tim Wescott on 31 May 2010 15:30 On 05/31/2010 11:30 AM, Steve Pope wrote: > gretzteam<gretzteam(a)n_o_s_p_a_m.yahoo.com> wrote: > >> Hi, >> Say I have a simple x-bit NCO running at 'fin' (100MHz). M and N are >> programmable. The output frequency of the NCO is given by: >> fout = fin*(N/M) >> >> Assuming a perfect input clock, how can I specify the jitter of 'fout'? I >> understand there are different measure of jitter but can't really figure >> out how they work with this system. > > RMS jitter is as follows: > > Form a reference clock signal that is jitter free and of the same > exact frequency and same average phase as the NCO. The RMS value of > the time diffeence of the NCO clock edges relative to the reference > clock, divided by the period, is the RMS jitter as a dimensionless > ratio. > > (If you don't divide it by the period, it's the RMS jitter in > time units.) > > For a given NCO you might be able to compute this analytically > from parameters such as your N and M, if it is important to > do so. I don't think that dimensionless jitter works out right for being phase jitter -- at the least it'd be off by a factor of 2 pi, wouldn't it? I've seen oscillators specified for phase jitter, timing jitter, and frequency jitter. I've seen all three specified as spectra instead of as single numbers, and I've seen the spectra specified both in terms of 'energy' per Hz and in terms of Allan Variance (which has a lot going for it as a measure, and probably isn't popular yet because Mr. Allan is still alive, kicking, and posting his opinions on his web site). -- Tim Wescott Control system and signal processing consulting www.wescottdesign.com
From: Steve Pope on 31 May 2010 15:35 Tim Wescott <tim(a)seemywebsite.now> wrote: >On 05/31/2010 11:30 AM, Steve Pope wrote: >> RMS jitter is as follows: >> Form a reference clock signal that is jitter free and of the same >> exact frequency and same average phase as the NCO. The RMS value of >> the time diffeence of the NCO clock edges relative to the reference >> clock, divided by the period, is the RMS jitter as a dimensionless >> ratio. >> (If you don't divide it by the period, it's the RMS jitter in >> time units.) >> For a given NCO you might be able to compute this analytically >> from parameters such as your N and M, if it is important to >> do so. >I don't think that dimensionless jitter works out right for being phase >jitter -- at the least it'd be off by a factor of 2 pi, wouldn't it? Sure, you need to say whether it's in radians, periods, degrees, minutes, or leagues per earht circumfrance. >I've seen oscillators specified for phase jitter, timing jitter, and >frequency jitter. I've seen all three specified as spectra instead of >as single numbers, and I've seen the spectra specified both in terms of >'energy' per Hz and in terms of Allan Variance (which has a lot going >for it as a measure, and probably isn't popular yet because Mr. Allan is >still alive, kicking, and posting his opinions on his web site). Phase noise I always see as a spectrum, or an integrated number derived from a spectrum, but jitter I usually see as an RMS value. YMMV. Steve
From: Tim Wescott on 31 May 2010 15:41 On 05/31/2010 12:35 PM, Steve Pope wrote: > Tim Wescott<tim(a)seemywebsite.now> wrote: > >> On 05/31/2010 11:30 AM, Steve Pope wrote: > >>> RMS jitter is as follows: > >>> Form a reference clock signal that is jitter free and of the same >>> exact frequency and same average phase as the NCO. The RMS value of >>> the time diffeence of the NCO clock edges relative to the reference >>> clock, divided by the period, is the RMS jitter as a dimensionless >>> ratio. > >>> (If you don't divide it by the period, it's the RMS jitter in >>> time units.) > >>> For a given NCO you might be able to compute this analytically >>> from parameters such as your N and M, if it is important to >>> do so. > >> I don't think that dimensionless jitter works out right for being phase >> jitter -- at the least it'd be off by a factor of 2 pi, wouldn't it? > > Sure, you need to say whether it's in radians, periods, degrees, > minutes, or leagues per earht circumfrance. > >> I've seen oscillators specified for phase jitter, timing jitter, and >> frequency jitter. I've seen all three specified as spectra instead of >> as single numbers, and I've seen the spectra specified both in terms of >> 'energy' per Hz and in terms of Allan Variance (which has a lot going >> for it as a measure, and probably isn't popular yet because Mr. Allan is >> still alive, kicking, and posting his opinions on his web site). > > Phase noise I always see as a spectrum, or an integrated number > derived from a spectrum, but jitter I usually see as an RMS value. > YMMV. Ditto for me on the phase noise, but I've seen "total" phase noise specified. And yes, jitter usually comes as an RMS value, although many of the really good digital designers that I know want to see it specified as peak-peak, because it's not usually the average error that wreaks havoc in a digital circuit, it's those outliers that cause race conditions that do it. -- Tim Wescott Control system and signal processing consulting www.wescottdesign.com
From: gretzteam on 31 May 2010 15:42 >I would love to see the OP's definition of _his_ NCO. I have seen the >term "NCO" to mean a lot of different things, >-- >Tim Wescott >Control system and signal processing consulting >www.wescottdesign.com What I meant by NCO was a B-bit counter, modulo M, incrementing by N at every clock cycle. People use this in a DDS, where the value of the counter is usually truncated and used as the address to a look-up table containing one period of a sine wave. Usually M is set to 2^B in this application. In my application, I use it to generate clocks. The 'overflow' rate of the NCO can be used to clock a separate counter that generates divided clocks. I believe that a generic NCO is simply this B-bit modulo M counter. If my usage of NCO is incorrect, please let me know!
From: Steve Pope on 31 May 2010 15:45
Tim Wescott <tim(a)seemywebsite.now> wrote: >On 05/31/2010 12:35 PM, Steve Pope wrote: >And yes, jitter usually comes as an RMS value, although many of the >really good digital designers that I know want to see it specified as >peak-peak, because it's not usually the average error that wreaks havoc >in a digital circuit, it's those outliers that cause race conditions >that do it. Yes but... isn't there always some very tiny probability of a really large phase excursion? "Peak" quantities can sometimes be stated for real (i.e. "peak to average" of a modulated signal makes sense), but I'm not sure about "peak" jitter. S. |