Hi, From the datasheets, it is looks like the only major difference between DCM and PLL is that PLL additionally does jitter filtering. Rest of the features are present in both these macros. So what decides whether one should use a PLL or DCM in FPGA. The following are the common features present in both DCM and PLL: 1) frequency synth 2) deskew 3) frequency div Additionally DCM can also do phase shift. Regards, Sharan
DCM vs PLL
Started by ●April 1, 2009
Reply by ●April 1, 20092009-04-01
On 1 Apr., 16:01, Sharan <sharan.basa...@gmail.com> wrote:> Hi, > > From the datasheets, it is looks like the only major difference > between DCM and PLL is that PLL additionally does jitter filtering. > Rest of the features are present in both these macros. So what decides > whether one should use a PLL or DCM in FPGA. > > The following are the common features present in both DCM and PLL: > 1) frequency synth > 2) deskew > 3) frequency div > > Additionally DCM can also do phase shift.PLL's do that too. DCM's have a higher potential for failures due tu their state-machine behaviour. For Example EMI could disturb some clocks getting into a DCM locking up and you need a systen reset. Analog PLL's are more tolerant about that and especially if you want to have deterministic jitter-fitering you are lost with the DCM. I would always chose a good analog PLL over DCM.
Reply by ●April 1, 20092009-04-01
"Sharan" <sharan.basappa@gmail.com> wrote in message news:6374ea0e-4cdb-474a-9cd6-df68f8252302@v15g2000yqn.googlegroups.com...> Hi, > > From the datasheets, it is looks like the only major difference > between DCM and PLL is that PLL additionally does jitter filtering. > Rest of the features are present in both these macros. So what decides > whether one should use a PLL or DCM in FPGA. > > The following are the common features present in both DCM and PLL: > 1) frequency synth > 2) deskew > 3) frequency div > > Additionally DCM can also do phase shift. > > Regards, > SharanPLL's have a higher potential for failures due tu their non-state-machine behaviour. For Example EMI could disturb some clocks getting into a PLL phase detector and you get a lock loss. Digital DCM's are more tolerant about that and especially if you want to have deterministic operation that is not component value dependent you are lost with the PLL. I would always chose a good DCM over analog PLL.
Reply by ●April 1, 20092009-04-01
"Symon" <symon_brewer@hotmail.com> wrote:> >"Sharan" <sharan.basappa@gmail.com> wrote in message >news:6374ea0e-4cdb-474a-9cd6-df68f8252302@v15g2000yqn.googlegroups.com... >> Hi, >> >> From the datasheets, it is looks like the only major difference >> between DCM and PLL is that PLL additionally does jitter filtering. >> Rest of the features are present in both these macros. So what decides >> whether one should use a PLL or DCM in FPGA. >> >> The following are the common features present in both DCM and PLL: >> 1) frequency synth >> 2) deskew >> 3) frequency div >> >> Additionally DCM can also do phase shift. >> >> Regards, >> Sharan > >PLL's have a higher potential for failures due tu their non-state-machine >behaviour. > >For Example EMI could disturb some clocks getting into a PLL phase >detector and you get a lock loss. > >Digital DCM's are more tolerant about that and especially if you want >to have deterministic operation that is not component value dependent you >are lost with the PLL. > >I would always chose a good DCM over analog PLL.Now I'm confused :-) -- Failure does not prove something is impossible, failure simply indicates you are not using the right tools... "If it doesn't fit, use a bigger hammer!" --------------------------------------------------------------
Reply by ●April 1, 20092009-04-01
Sharan, The DCM requires less area in silicon, thus we can provide more of them. They are digital, and greatly immune to power supply noise (as compared to PLL's). They are a state machine, so a "crummy clock" signal is NOT recommended, as a DCM will have a hard time, where a PLL is commonly used to "clean up" a "crummy clock" in order to make it useful at all. DCM will add totally random, white, broad band jitter, where a PLL will remove high frequency jitter, while adding low frequency jitter (due to its oscillator having very low Q on the silicon). It will also add as jitter any power supply noise present varies the VCO in the PLL. External PLL's may be used with voltage controlled crystal oscillators where the quality of the on-chip PLL oscillator is unacceptable (for example SONET/SDH transmit clock jitter requirements are tough to meet without a very high Q resonantor). PLL and DCM have their uses. We provide a mix of both, and you may precede a DCM with a PLL if you have a "crummy clock" or pass the DCM through the PLL if you wish to remove all nearly all the high frequency jitter components. The DCM also offers precise dynamic phase shifting, which is not so easy to do in a PLL. Use a hammer for nails, and a screwdriver for screws. They are both fasteners. Which is better? Silly question. Austin
Reply by ●April 1, 20092009-04-01
On 1 Apr., 17:36, "Symon" <symon_bre...@hotmail.com> wrote:> PLL's have a higher potential for failures due tu their non-state-machine > behaviour. > > For Example EMI could disturb some clocks getting into a PLL phase > detector and you get a lock loss.Did i miss some braindead engineer's pll design which needs a reset after it got out of lock like the DCM's i know?> Digital DCM's are more tolerant about that and especially if you want > to have deterministic operation that is not component value dependent you > are lost with the PLL.And i also seem to have missed analog PLLs in FPGA's whose process temperature voltage variations violate the specs.> I would always chose a good DCM over analog PLL.- Zitierten Text ausblenden -That's good as long as you can forbid your devices to pick up EMI or any noise on it's clocks and do a preventive restart every minute or so.
Reply by ●April 1, 20092009-04-01
On Apr 1, 10:36=A0am, "Symon" <symon_bre...@hotmail.com> wrote:> "Sharan" <sharan.basa...@gmail.com> wrote in message > > news:6374ea0e-4cdb-474a-9cd6-df68f8252302@v15g2000yqn.googlegroups.com... > > > > > > > Hi, > > > From the datasheets, it is looks like the only major difference > > between DCM and PLL is that PLL additionally does jitter filtering. > > Rest of the features are present in both these macros. So what decides > > whether one should use a PLL or DCM in FPGA. > > > The following are the common features present in both DCM and PLL: > > 1) frequency synth > > 2) deskew > > 3) frequency div > > > Additionally DCM can also do phase shift. > > > Regards, > > Sharan > > PLL's have a higher potential for failures due tu their non-state-machine > behaviour. > > For Example EMI could disturb some clocks getting into a PLL phase > detector and you get a lock loss. > > Digital DCM's are more tolerant about that and especially if you want > to have deterministic operation that is not component value dependent you > are lost with the PLL. > > I would always chose a good DCM over analog PLL.- Hide quoted text - > > - Show quoted text -Haha, It rings the bell about the battle between DLL vs. PLL not long ago. :-) So far, jitter is the only thing I concern about... so who's the winner ?
Reply by ●April 2, 20092009-04-02
<filter001@desinformation.de> wrote in message news:ec6505f4-7051-48e8-b471-6bd5849ded65@q16g2000yqg.googlegroups.com...> On 1 Apr., 16:01, Sharan <sharan.basa...@gmail.com> wrote: > DCM's have a higher potential for failures due tu their state-machine > behaviour. > > For Example EMI could disturb some clocks getting into a DCM locking > up and you need a systen reset.Did i miss some braindead engineer's DCM design which has terrible stability causing it to get out of lock like the PLL's i know?> Analog PLL's are more tolerant about that and especially if you want > to have deterministic jitter-fitering you are lost with the DCM.And i also seem to have missed DCMs in FPGA's whose jitter performance violate the specs.> I would always chose a good analog PLL over DCM.- Zitierten Text > ausblenden -That's good as long as you can forbid your devices to pick up EMI or any noise on it's clocks and constantly check the phase lock.
Reply by ●April 2, 20092009-04-02
On Wed, 1 Apr 2009 13:26:47 -0700 (PDT) halong <ccon67@netscape.net> wrote:> On Apr 1, 10:36=A0am, "Symon" <symon_bre...@hotmail.com> wrote: > > "Sharan" <sharan.basa...@gmail.com> wrote in message > > > > news:6374ea0e-4cdb-474a-9cd6-df68f8252302@v15g2000yqn.googlegroups.com.=..> > > > > > > > > > > > > Hi, > > > > > From the datasheets, it is looks like the only major difference > > > between DCM and PLL is that PLL additionally does jitter > > > filtering. Rest of the features are present in both these macros. > > > So what decides whether one should use a PLL or DCM in FPGA. > > > > > The following are the common features present in both DCM and PLL: > > > 1) frequency synth > > > 2) deskew > > > 3) frequency div > > > > > Additionally DCM can also do phase shift. > > > > > Regards, > > > Sharan > > > > PLL's have a higher potential for failures due tu their > > non-state-machine behaviour. > > > > For Example EMI could disturb some clocks getting into a PLL phase > > detector and you get a lock loss. > > > > Digital DCM's are more tolerant about that and especially if you > > want to have deterministic operation that is not component value > > dependent you are lost with the PLL. > > > > I would always chose a good DCM over analog PLL.- Hide quoted text - > > > > - Show quoted text - >=20 > Haha, It rings the bell about the battle between DLL vs. PLL not long > ago. :-) >=20 > So far, jitter is the only thing I concern about... so who's the > winner ? >=20DCM shows. PLL places. Starting off with a good crystal oscillator in first place wins by a length. --=20 Rob Gaddi, Highland Technology Email address is currently out of order