Forums

Question about damping in PLL

Started by fl October 29, 2011
Hi,

I am learning PLL from a web download PLLTutorialISSCC2004.pdf after I
read two books on PLL. I do not understand the following statements
from the web pdf:

...........................
Less ringing and overshoot as Zeta = 1
Severe overdamping --> ringing and overshoot
Ringing at high damping due to low oversampling (large R) � Gardner
limit.

.............
Low damping ---> less period jitter, slower response, more phase error
High damping --> low oversampling (large R) causes oscillation
...................................

The above are from one .pdf document. In the slide like page, it does
not show equation and symbols. I feel there are some conflicting
between the statements.

'Severe overdamping' means Zeta >> 1? Of course I guess so, it should
not be overshoot to me.

What is low damping means? Zeta < 1 ? Then PLL will have slower
response? I doubt it.

Could you explain it to me more? Thanks

On Oct 29, 12:19&#2013266080;pm, fl <rxjw...@gmail.com> wrote:
> Hi, > > I am learning PLL from a web download PLLTutorialISSCC2004.pdf after I > read two books on PLL. I do not understand the following statements > from the web pdf: > > ........................... > Less ringing and overshoot as Zeta = 1 > Severe overdamping --> ringing and overshoot > Ringing at high damping due to low oversampling (large R) &#2013266070; Gardner > limit. > > ............. > Low damping ---> less period jitter, slower response, more phase error > High damping --> low oversampling (large R) causes oscillation > ................................... > > The above are from one .pdf document. In the slide like page, it does > not show equation and symbols. I feel there are some conflicting > between the statements. > > 'Severe overdamping' means Zeta >> 1? Of course I guess so, it should > not be overshoot to me. > > What is low damping means? Zeta < 1 ? Then PLL will have slower > response? I doubt it. > > Could you explain it to me more? Thanks
I am not sure. When I think of an overdamped system, I think of a system that is likely to be stable but have a slow response (for a given bandwidth). When I think of an underdamped system, I think of a system that is more responsive, but can possibly oscillate. This seems to be the opposite of what you stated above.
On Sat, 29 Oct 2011 09:19:46 -0700, fl wrote:

> Hi, > > I am learning PLL from a web download PLLTutorialISSCC2004.pdf after I > read two books on PLL. I do not understand the following statements from > the web pdf: > > ........................... > Less ringing and overshoot as Zeta = 1 Severe overdamping --> ringing > and overshoot Ringing at high damping due to low oversampling (large R) > &ndash; Gardner limit. > > ............. > Low damping ---> less period jitter, slower response, more phase error > High damping --> low oversampling (large R) causes oscillation > ................................... > > The above are from one .pdf document. In the slide like page, it does > not show equation and symbols. I feel there are some conflicting between > the statements. > > 'Severe overdamping' means Zeta >> 1? Of course I guess so, it should > not be overshoot to me. > > What is low damping means? Zeta < 1 ? Then PLL will have slower > response? I doubt it. > > Could you explain it to me more? Thanks
He's not using damping factor the way that it normally is -- it looks like he just pulled some equations from Gardener's book and only half understands them. Using a damping factor makes sense when you have a single pair of resonant poles that dominate the transfer function response. When this is the case, then a low damping factor makes for more ringing, while a high damping factor makes for sluggish response. Both more ringing and sluggishness will increase settling time (and ringing implies peaking), so for that hypothetical 2nd-order system, a damping factor close to 1 (usually you want something between 1 and 0.7) is a good thing. I suspect that with whatever loop he's building, when you turn off your brain and just use the equation then when the "damping factor" (his meaning) goes up, that you have some other pole pair that becomes under damped and leads to peaking. This means, of course, that you don't have a system with a _single_ dominant pole pair, which means that use of the unqualified "damping factor" is a misnomer. -- www.wescottdesign.com
On Oct 30, 9:36=A0am, Tim Wescott <t...@seemywebsite.com> wrote:
> On Sat, 29 Oct 2011 09:19:46 -0700, fl wrote: > > Hi, > > > I am learning PLL from a web download PLLTutorialISSCC2004.pdf after I > > read two books on PLL. I do not understand the following statements fro=
m
> > the web pdf: > > > ........................... > > Less ringing and overshoot as Zeta =3D 1 Severe overdamping --> ringing > > and overshoot Ringing at high damping due to low oversampling (large R) > > =96 Gardner limit. > > > ............. > > Low damping ---> less period jitter, slower response, more phase error > > High damping --> low oversampling (large R) causes oscillation > > ................................... > > > The above are from one .pdf document. In the slide like page, it does > > not show equation and symbols. I feel there are some conflicting betwee=
n
> > the statements. > > > 'Severe overdamping' means Zeta >> 1? Of course I guess so, it should > > not be overshoot to me. > > > What is low damping means? Zeta < 1 ? Then PLL will have slower > > response? I doubt it. > > > Could you explain it to me more? Thanks > > He's not using damping factor the way that it normally is -- it looks > like he just pulled some equations from Gardener's book and only half > understands them. > > Using a damping factor makes sense when you have a single pair of > resonant poles that dominate the transfer function response. =A0When this > is the case, then a low damping factor makes for more ringing, while a > high damping factor makes for sluggish response. =A0Both more ringing and > sluggishness will increase settling time (and ringing implies peaking), > so for that hypothetical 2nd-order system, a damping factor close to 1 > (usually you want something between 1 and 0.7) is a good thing. > > I suspect that with whatever loop he's building, when you turn off your > brain and just use the equation then when the "damping factor" (his > meaning) goes up, that you have some other pole pair that becomes under > damped and leads to peaking. =A0This means, of course, that you don't hav=
e
> a system with a _single_ dominant pole pair, which means that use of the > unqualified "damping factor" is a misnomer. > > --www.wescottdesign.com
You are both forgetting that this is a closed-loop system and therefore you can often get the best of both worlds, fast response and low overshoot. Fast response through high bandwidth and low overshoot by using phase- lead compensation. To use 2nd order time-domain models is poointless. You need to plot an open loop Bode-Plot of the PLL and measure (predict) phase-margin (for stability) and look at the closing unity gain bandwidth (for speed of response).
On Mon, 31 Oct 2011 00:51:41 -0700, HardySpicer wrote:

> On Oct 30, 9:36&nbsp;am, Tim Wescott <t...@seemywebsite.com> wrote: >> On Sat, 29 Oct 2011 09:19:46 -0700, fl wrote: >> > Hi, >> >> > I am learning PLL from a web download PLLTutorialISSCC2004.pdf after >> > I read two books on PLL. I do not understand the following statements >> > from the web pdf: >> >> > ........................... >> > Less ringing and overshoot as Zeta = 1 Severe overdamping --> ringing >> > and overshoot Ringing at high damping due to low oversampling (large >> > R) &ndash; Gardner limit. >> >> > ............. >> > Low damping ---> less period jitter, slower response, more phase >> > error High damping --> low oversampling (large R) causes oscillation >> > ................................... >> >> > The above are from one .pdf document. In the slide like page, it does >> > not show equation and symbols. I feel there are some conflicting >> > between the statements. >> >> > 'Severe overdamping' means Zeta >> 1? Of course I guess so, it should >> > not be overshoot to me. >> >> > What is low damping means? Zeta < 1 ? Then PLL will have slower >> > response? I doubt it. >> >> > Could you explain it to me more? Thanks >> >> He's not using damping factor the way that it normally is -- it looks >> like he just pulled some equations from Gardener's book and only half >> understands them. >> >> Using a damping factor makes sense when you have a single pair of >> resonant poles that dominate the transfer function response. &nbsp;When this >> is the case, then a low damping factor makes for more ringing, while a >> high damping factor makes for sluggish response. &nbsp;Both more ringing and >> sluggishness will increase settling time (and ringing implies peaking), >> so for that hypothetical 2nd-order system, a damping factor close to 1 >> (usually you want something between 1 and 0.7) is a good thing. >> >> I suspect that with whatever loop he's building, when you turn off your >> brain and just use the equation then when the "damping factor" (his >> meaning) goes up, that you have some other pole pair that becomes under >> damped and leads to peaking. &nbsp;This means, of course, that you don't >> have a system with a _single_ dominant pole pair, which means that use >> of the unqualified "damping factor" is a misnomer. >> >> --www.wescottdesign.com > > You are both forgetting that this is a closed-loop system and therefore > you can often get the best of both worlds, fast response and low > overshoot. > Fast response through high bandwidth and low overshoot by using phase- > lead compensation. To use 2nd order time-domain models is poointless. > You need to plot an open loop Bode-Plot of the PLL and measure (predict) > phase-margin (for stability) and look at the closing unity gain > bandwidth (for speed of response).
I'm not sure what you're trying to comment on. When you're analyzing a system the damping factor is a handy concept, and it's a handy target to shoot for regardless of what loop closure frequency you also shoot for. So regardless of what technique you're using to attain bandwidth, you still can't ignore damping, whatever language you may choose to describe it. A damping factor very far away from 1 in either direction generally isn't a good thing, and a damping factor much lower than 1 is often a Very Bad Thing. -- www.wescottdesign.com
On Nov 1, 7:16=A0am, Tim Wescott <t...@seemywebsite.com> wrote:
> On Mon, 31 Oct 2011 00:51:41 -0700, HardySpicer wrote: > > On Oct 30, 9:36=A0am, Tim Wescott <t...@seemywebsite.com> wrote: > >> On Sat, 29 Oct 2011 09:19:46 -0700, fl wrote: > >> > Hi, > > >> > I am learning PLL from a web download PLLTutorialISSCC2004.pdf after > >> > I read two books on PLL. I do not understand the following statement=
s
> >> > from the web pdf: > > >> > ........................... > >> > Less ringing and overshoot as Zeta =3D 1 Severe overdamping --> ring=
ing
> >> > and overshoot Ringing at high damping due to low oversampling (large > >> > R) =96 Gardner limit. > > >> > ............. > >> > Low damping ---> less period jitter, slower response, more phase > >> > error High damping --> low oversampling (large R) causes oscillation > >> > ................................... > > >> > The above are from one .pdf document. In the slide like page, it doe=
s
> >> > not show equation and symbols. I feel there are some conflicting > >> > between the statements. > > >> > 'Severe overdamping' means Zeta >> 1? Of course I guess so, it shoul=
d
> >> > not be overshoot to me. > > >> > What is low damping means? Zeta < 1 ? Then PLL will have slower > >> > response? I doubt it. > > >> > Could you explain it to me more? Thanks > > >> He's not using damping factor the way that it normally is -- it looks > >> like he just pulled some equations from Gardener's book and only half > >> understands them. > > >> Using a damping factor makes sense when you have a single pair of > >> resonant poles that dominate the transfer function response. =A0When t=
his
> >> is the case, then a low damping factor makes for more ringing, while a > >> high damping factor makes for sluggish response. =A0Both more ringing =
and
> >> sluggishness will increase settling time (and ringing implies peaking)=
,
> >> so for that hypothetical 2nd-order system, a damping factor close to 1 > >> (usually you want something between 1 and 0.7) is a good thing. > > >> I suspect that with whatever loop he's building, when you turn off you=
r
> >> brain and just use the equation then when the "damping factor" (his > >> meaning) goes up, that you have some other pole pair that becomes unde=
r
> >> damped and leads to peaking. =A0This means, of course, that you don't > >> have a system with a _single_ dominant pole pair, which means that use > >> of the unqualified "damping factor" is a misnomer. > > >> --www.wescottdesign.com > > > You are both forgetting that this is a closed-loop system and therefore > > you can often get the best of both worlds, fast response and low > > overshoot. > > Fast response through high bandwidth and low overshoot by using phase- > > lead compensation. To use 2nd order time-domain models is poointless. > > You need to plot an open loop Bode-Plot of the PLL and measure (predict=
)
> > phase-margin (for stability) and look at the closing unity gain > > bandwidth (for speed of response). > > I'm not sure what you're trying to comment on. =A0When you're analyzing a > system the damping factor is a handy concept, and it's a handy target to > shoot for regardless of what loop closure frequency you also shoot for. > > So regardless of what technique you're using to attain bandwidth, you > still can't ignore damping, whatever language you may choose to describe > it. > > A damping factor very far away from 1 in either direction generally isn't > a good thing, and a damping factor much lower than 1 is often a Very Bad > Thing. > > --www.wescottdesign.com
Damping factor is a second order system concept. Although you can infer second order dominance, the idea of damping factor is essentially time-domain and 2nd order. Bode plots will work with any order.