>On Sat, 17 Aug 2013 00:57:39 -0500, Vladimir Vassilevsky ><nospam@nowhere.com> wrote: > >>On 8/17/2013 12:38 AM, gyansorova@gmail.com wrote: >>> On Friday, August 16, 2013 2:20:40 PM UTC+12, Vladimir Vassilevsky >>> wrote: >>>> On 8/15/2013 5:57 PM, commsignal wrote: >>>> >>>>> (1) Most of the tutorials discuss the effect of the loop >>>>> bandwidth and the >>>> >>>>> damping factor on the PLL behaviour. >>>> >>>> >>>> >>>> Because PLL of the 2nd order is nice linear problem. >>>> >>>> >>>> >>>>> Is there any reference about the same >>>> >>>>> with respect to the SNR? >>>> >>>> >>>> >>>> If SNR is above 10 dB, PLL behaves like linear filter. Just >>>> integrate >>>> >>>> noise over the bandwidth. >>>> >>>> If SNR is below 10 dB, PLL gets nonlinear and intractable, and you >>>> left >>>> >>>> to simulations and approximations. >>>> >>>> >>>> >>>>> (2) Any reference deriving the tracking ivariance, pull-in >>>>> range, >>>> >>>>> etc.? >>>> >>>> >>>> >>>> It is easy to get ballpark estimate; however it depends on >>>> particulars. >>>> >> >>> >>> Nonsense, I have used a PLL down to 0dB (with great difficulty >>> though). >> >>That is no good idea. >>PLL in nonlinear zone is clearly inefficient method for tracking; and >>even less efficient for acquisition. > >I've never had that much trouble down to the ~0dB range, either. You >just need to know what you're doing. > >>> Most people don't understand how to stabilise the loop. >>> I normally use a lag-lead compensator, most of the textbooks give >>> simple solutions. >> >>Most people have no clue what they doing. > >But some here do. > >> >>VLV > >Eric Jacobsen >Anchor Hill Communications >http://www.anchorhill.com >Thanks everyone. I will definitely go through the text and will come back here. Just being inquisitive, when you say that you have used the PLL down to 0 dB range, what roughly are you doing; using a digital PLL or analog, which part should be concentrated on for operation in that range (is it the phase error detector only)? _____________________________ Posted through www.DSPRelated.com
PLL and SNR
Started by ●August 15, 2013
Reply by ●August 17, 20132013-08-17
Reply by ●August 17, 20132013-08-17
On Sat, 17 Aug 2013 18:35:26 -0500, "commsignal" <58672@dsprelated> wrote:>>On Sat, 17 Aug 2013 00:57:39 -0500, Vladimir Vassilevsky >><nospam@nowhere.com> wrote: >> >>>On 8/17/2013 12:38 AM, gyansorova@gmail.com wrote: >>>> On Friday, August 16, 2013 2:20:40 PM UTC+12, Vladimir Vassilevsky >>>> wrote: >>>>> On 8/15/2013 5:57 PM, commsignal wrote: >>>>> >>>>>> (1) Most of the tutorials discuss the effect of the loop >>>>>> bandwidth and the >>>>> >>>>>> damping factor on the PLL behaviour. >>>>> >>>>> >>>>> >>>>> Because PLL of the 2nd order is nice linear problem. >>>>> >>>>> >>>>> >>>>>> Is there any reference about the same >>>>> >>>>>> with respect to the SNR? >>>>> >>>>> >>>>> >>>>> If SNR is above 10 dB, PLL behaves like linear filter. Just >>>>> integrate >>>>> >>>>> noise over the bandwidth. >>>>> >>>>> If SNR is below 10 dB, PLL gets nonlinear and intractable, and you >>>>> left >>>>> >>>>> to simulations and approximations. >>>>> >>>>> >>>>> >>>>>> (2) Any reference deriving the tracking ivariance, pull-in >>>>>> range, >>>>> >>>>>> etc.? >>>>> >>>>> >>>>> >>>>> It is easy to get ballpark estimate; however it depends on >>>>> particulars. >>>>> >>> >>>> >>>> Nonsense, I have used a PLL down to 0dB (with great difficulty >>>> though). >>> >>>That is no good idea. >>>PLL in nonlinear zone is clearly inefficient method for tracking; and >>>even less efficient for acquisition. >> >>I've never had that much trouble down to the ~0dB range, either. You >>just need to know what you're doing. >> >>>> Most people don't understand how to stabilise the loop. >>>> I normally use a lag-lead compensator, most of the textbooks give >>>> simple solutions. >>> >>>Most people have no clue what they doing. >> >>But some here do. >> >>> >>>VLV >> >>Eric Jacobsen >>Anchor Hill Communications >>http://www.anchorhill.com >> > >Thanks everyone. I will definitely go through the text and will come back >here. Just being inquisitive, when you say that you have used the PLL down >to 0 dB range, what roughly are you doing; using a digital PLL or analog, >which part should be concentrated on for operation in that range (is it the >phase error detector only)?In my case it's usually a comm signal around 0dB using low-rate capacity-approaching FEC (Turbo Codes or LDPCs). The PLLs are all-digital, for both carrier recovery (carrier phase lock) and symbol timing. You do have to pay attention to the detectors, as well as numeric precision, loop BW, etc., etc. You don't have to do anything exotic, just pay attention to details.>_____________________________ >Posted through www.DSPRelated.comEric Jacobsen Anchor Hill Communications http://www.anchorhill.com
Reply by ●August 17, 20132013-08-17
On 8/17/2013 6:45 PM, Eric Jacobsen wrote:> In my case it's usually a comm signal around 0dB using low-rate > capacity-approaching FEC (Turbo Codes or LDPCs). The PLLs are > all-digital, for both carrier recovery (carrier phase lock) and symbol > timing. You do have to pay attention to the detectors, as well as > numeric precision, loop BW, etc., etc. You don't have to do anything > exotic, just pay attention to details.Great job for graduate stupident. A lot of averaging sure cures many problems. In my case, it is usually a comm signal around -15dB SNR using orthogonal functions, at symbol rate of at the order of several bauds. VLV
Reply by ●August 19, 20132013-08-19
On 8/17/13 5:19 PM, Vladimir Vassilevsky wrote:> > Great job for graduate stupident.graduate stupidents are people, too. -- r b-j rbj@audioimagination.com "Imagination is more important than knowledge."
Reply by ●August 20, 20132013-08-20
On Sunday, August 18, 2013 11:35:26 AM UTC+12, commsignal wrote:> >On Sat, 17 Aug 2013 00:57:39 -0500, Vladimir Vassilevsky > > ><nospam@nowhere.com> wrote: > > > > > >>On 8/17/2013 12:38 AM, gyansorova@gmail.com wrote: > > >>> On Friday, August 16, 2013 2:20:40 PM UTC+12, Vladimir Vassilevsky > > >>> wrote: > > >>>> On 8/15/2013 5:57 PM, commsignal wrote: > > >>>> > > >>>>> (1) Most of the tutorials discuss the effect of the loop > > >>>>> bandwidth and the > > >>>> > > >>>>> damping factor on the PLL behaviour. > > >>>> > > >>>> > > >>>> > > >>>> Because PLL of the 2nd order is nice linear problem. > > >>>> > > >>>> > > >>>> > > >>>>> Is there any reference about the same > > >>>> > > >>>>> with respect to the SNR? > > >>>> > > >>>> > > >>>> > > >>>> If SNR is above 10 dB, PLL behaves like linear filter. Just > > >>>> integrate > > >>>> > > >>>> noise over the bandwidth. > > >>>> > > >>>> If SNR is below 10 dB, PLL gets nonlinear and intractable, and you > > >>>> left > > >>>> > > >>>> to simulations and approximations. > > >>>> > > >>>> > > >>>> > > >>>>> (2) Any reference deriving the tracking ivariance, pull-in > > >>>>> range, > > >>>> > > >>>>> etc.? > > >>>> > > >>>> > > >>>> > > >>>> It is easy to get ballpark estimate; however it depends on > > >>>> particulars. > > >>>> > > >> > > >>> > > >>> Nonsense, I have used a PLL down to 0dB (with great difficulty > > >>> though). > > >> > > >>That is no good idea. > > >>PLL in nonlinear zone is clearly inefficient method for tracking; and > > >>even less efficient for acquisition. > > > > > >I've never had that much trouble down to the ~0dB range, either. You > > >just need to know what you're doing. > > > > > >>> Most people don't understand how to stabilise the loop. > > >>> I normally use a lag-lead compensator, most of the textbooks give > > >>> simple solutions. > > >> > > >>Most people have no clue what they doing. > > > > > >But some here do. > > > > > >> > > >>VLV > > > > > >Eric Jacobsen > > >Anchor Hill Communications > > >http://www.anchorhill.com > > > > > > > Thanks everyone. I will definitely go through the text and will come back > > here. Just being inquisitive, when you say that you have used the PLL down > > to 0 dB range, what roughly are you doing; using a digital PLL or analog, > > which part should be concentrated on for operation in that range (is it the > > phase error detector only)? > > > > _____________________________ > > Posted through www.DSPRelated.comI had a low bandwidth PLL locking into a carrier at 455kHz - all analogue. In fact if I remember right we had it working to about -3dB but it needed an Amplitude Locked Loop in front of it and not a hard limiter. a limiter just amplifies the noise.
Reply by ●August 20, 20132013-08-20
gyansorova@gmail.com writes:> [...] > > I had a low bandwidth PLL locking into a carrier at 455kHz - all > analogue. In fact if I remember right we had it working to about -3dB > but it needed an Amplitude Locked Loop in front of it and not a hard > limiter. a limiter just amplifies the noise.Amplitude Locked Loop? You mean an AGC? -- Randy Yates Digital Signal Labs http://www.digitalsignallabs.com
Reply by ●August 22, 20132013-08-22
On Wednesday, August 21, 2013 1:48:34 AM UTC+12, Randy Yates wrote:> gyansorova@gmail.com writes: > > > [...] > > > > > > I had a low bandwidth PLL locking into a carrier at 455kHz - all > > > analogue. In fact if I remember right we had it working to about -3dB > > > but it needed an Amplitude Locked Loop in front of it and not a hard > > > limiter. a limiter just amplifies the noise. > > > > Amplitude Locked Loop? You mean an AGC? > > -- > > Randy Yates > > Digital Signal Labs > > http://www.digitalsignallabs.comNo, an Amplitude locked loop. This is a high gain loop which compresses the envelope like a compander but doesn't do this near the zero crossings. so it has flat envelope and then jumps out of lock when the carrier disappears and freewheels through to when the carrier returns. It has an AGC before the amplitude locked loop which has a high dynamic range, the ALL has a small dynamic range of about 20dB (AGC about 120dB)
Reply by ●August 22, 20132013-08-22
On Thursday, August 22, 2013 12:26:25 AM UTC-4, gyans...@gmail.com wrote:> On Wednesday, August 21, 2013 1:48:34 AM UTC+12, Randy Yates wrote: > gyansorova@gmail.com writes: > > > [...] > > > > > > I had a low bandwidth PLL locking into a carrier at 455kHz - all > > > analogue. In fact if I remember right we had it working to about -3dB > > > but it needed an Amplitude Locked Loop in front of it and not a hard > > > limiter. a limiter just amplifies the noise. > > > > Amplitude Locked Loop? You mean an AGC? > > -- > > Randy Yates > > Digital Signal Labs > > http://www.digitalsignallabs.com No, an Amplitude locked loop. This is a high gain loop which compresses the envelope like a compander but doesn't do this near the zero crossings. so it has flat envelope and then jumps out of lock when the carrier disappears and freewheels through to when the carrier returns. It has an AGC before the amplitude locked loop which has a high dynamic range, the ALL has a small dynamic range of about 20dB (AGC about 120dB)ok this is interesting if the time constant or BW for the ALL is very fast on the same order as the signal BW, such that it tries to follow the noise envelope then its behavior would be the same as a limiter...no? or is the time constant slower? Mark
Reply by ●August 22, 20132013-08-22
On Friday, August 23, 2013 2:11:51 AM UTC+12, mako...@yahoo.com wrote:> On Thursday, August 22, 2013 12:26:25 AM UTC-4, gyans...@gmail.com wrote: > > > On Wednesday, August 21, 2013 1:48:34 AM UTC+12, Randy Yates wrote: > gyansorova@gmail.com writes: > > > [...] > > > > > > I had a low bandwidth PLL locking into a carrier at 455kHz - all > > > analogue. In fact if I remember right we had it working to about -3dB > > > but it needed an Amplitude Locked Loop in front of it and not a hard > > > limiter. a limiter just amplifies the noise. > > > > Amplitude Locked Loop? You mean an AGC? > > -- > > Randy Yates > > Digital Signal Labs > > http://www.digitalsignallabs.com No, an Amplitude locked loop. This is a high gain loop which compresses the envelope like a compander but doesn't do this near the zero crossings. so it has flat envelope and then jumps out of lock when the carrier disappears and freewheels through to when the carrier returns. It has an AGC before the amplitude locked loop which has a high dynamic range, the ALL has a small dynamic range of about 20dB (AGC about 120dB) > > > > ok this is interesting > > > > if the time constant or BW for the ALL is very fast on the same order as the signal BW, such that it tries to follow the noise envelope then its behavior would be the same as a limiter...no? > > > > or is the time constant slower? > > > > > > MarkThe bandwidth of the ALL is fast but it has the ability to drop out of lock in amplitude and back in again. I should point out that for a constant carrier they would be the same, but I was dealing with a carrier that dropped to zero and back again as in fading. When the carrier approaches zero (imagine a double sideband suppressed carrier waveform) the ALL drops out of lock and doesn't amplify the noise.
