I have been wondering if OFDM based system use PLL or DLL as in single carrier system e.g. GSM to do synchronization at all? Currently, a lot of ofdm based systems talk about implementing the frequency offset and time offset compensation in the digital domain. Do they also implement the time and frequency synchronization in hardware?
Does OFDM based system use PLL or DLL for synchronization?
Started by ●February 22, 2011
Reply by ●February 22, 20112011-02-22
On 02/22/2011 11:53 AM, xsong wrote:> I have been wondering if OFDM based system use PLL or DLL as in single > carrier system e.g. GSM to do synchronization at all? Currently, a lot > of ofdm based systems talk about implementing the frequency offset and > time offset compensation in the digital domain. Do they also implement > the time and frequency synchronization in hardware? >What do you want to bet that some do and some don't? There's advantages no matter which way you go, but unless you've got a really unstable LO you should be able to pull the signal in without adjustment in the analog world, and have more than sufficient information to process it digitally without having to play games out where it's expensive. -- Tim Wescott Wescott Design Services http://www.wescottdesign.com Do you need to implement control loops in software? "Applied Control Theory for Embedded Systems" was written for you. See details at http://www.wescottdesign.com/actfes/actfes.html
Reply by ●March 4, 20112011-03-04
From systems point of view, what is the definition of a stable LO? a 100ppm lo? What are some other specializations? I guess my question really boils down to whether we can use PLL for OFDM based system. Typically for using PLL, we have to generate a reference signal at the carrier frequency. However, for OFDM based systems, there is no single frequency. Then what reference signal can we use? On Feb 22, 1:15�pm, Tim Wescott <t...@seemywebsite.com> wrote:> On 02/22/2011 11:53 AM, xsong wrote:> I have been wondering if OFDM based system use PLL or DLL as in single > > carrier system e.g. GSM to do synchronization at all? Currently, a lot > > of ofdm based systems talk about implementing the frequency offset and > > time offset compensation in the digital domain. Do they also implement > > the time and frequency synchronization in hardware? > > What do you want to bet that some do and some don't? > > There's advantages no matter which way you go, but unless you've got a > really unstable LO you should be able to pull the signal in without > adjustment in the analog world, and have more than sufficient > information to process it digitally without having to play games out > where it's expensive. > > -- > > Tim Wescott > Wescott Design Serviceshttp://www.wescottdesign.com > > Do you need to implement control loops in software? > "Applied Control Theory for Embedded Systems" was written for you. > See details athttp://www.wescottdesign.com/actfes/actfes.html
Reply by ●March 4, 20112011-03-04
On 03/03/2011 09:54 PM, xsong wrote: (top posting fixed)> > > On Feb 22, 1:15 pm, Tim Wescott<t...@seemywebsite.com> wrote: >> On 02/22/2011 11:53 AM, xsong wrote:> I have been wondering if OFDM based system use PLL or DLL as in single >>> carrier system e.g. GSM to do synchronization at all? Currently, a lot >>> of ofdm based systems talk about implementing the frequency offset and >>> time offset compensation in the digital domain. Do they also implement >>> the time and frequency synchronization in hardware? >> >> What do you want to bet that some do and some don't? >> >> There's advantages no matter which way you go, but unless you've got a >> really unstable LO you should be able to pull the signal in without >> adjustment in the analog world, and have more than sufficient >> information to process it digitally without having to play games out >> where it's expensive.> > From systems point of view, what is the definition of a stable LO? a > 100ppm lo? What are some other specializations? From a systems point of view, the answer to any question is "that depends". Stable enough that your signal is cleanly passed by any IF filtering ahead of the analog to digital conversion. What is that exactly? Well, it depends on a whole bunch of factors peculiar to your receiver. > I guess my question really boils down to whether we can use PLL for > OFDM based system. Typically for using PLL, we have to generate > a reference signal at the carrier frequency. However, for OFDM based > systems, there is no single frequency. Then what reference signal can > we use? All of the OFDM specifications that I have seen (and I admit to not having looked at a whole bunch of them) have pilot tones -- some channels that are just left on continuously. These can be used to recover the carrier, and that information can be used to adjust phase either before or after the A to D conversion. -- Tim Wescott Wescott Design Services http://www.wescottdesign.com Do you need to implement control loops in software? "Applied Control Theory for Embedded Systems" was written for you. See details at http://www.wescottdesign.com/actfes/actfes.html
Reply by ●March 4, 20112011-03-04
On Mar 4, 12:29�am, Tim Wescott <t...@seemywebsite.com> wrote:> On 03/03/2011 09:54 PM, xsong wrote: > (top posting fixed) > > > > > On Feb 22, 1:15 pm, Tim Wescott<t...@seemywebsite.com> �wrote: > >> On 02/22/2011 11:53 AM, xsong wrote:> �I have been wondering if OFDM based system use PLL or DLL as in single > >>> carrier system e.g. GSM to do synchronization at all? Currently, a lot > >>> of ofdm based systems talk about implementing the frequency offset and > >>> time offset compensation in the digital domain. Do they also implement > >>> the time and frequency synchronization in hardware? > > >> What do you want to bet that some do and some don't? > > >> There's advantages no matter which way you go, but unless you've got a > >> really unstable LO you should be able to pull the signal in without > >> adjustment in the analog world, and have more than sufficient > >> information to process it digitally without having to play games out > >> where it's expensive. > > �> > �> �From systems point of view, what is the definition of �a stable LO? a > �> 100ppm lo? What are some other specializations? > > �From a systems point of view, the answer to any question is "that depends". > > Stable enough that your signal is cleanly passed by any IF filtering > ahead of the analog to digital conversion. �What is that exactly? �Well, > it depends on a whole bunch of factors peculiar to your receiver. > > �> I guess my question really boils down to whether we can use PLL for > �> OFDM based system. Typically for using PLL, we have to generate > �> a reference signal at the carrier frequency. However, for OFDM based > �> systems, there is no single frequency. Then what reference signal can > �> we use? >Thanks Tim for your input. However suppose you have a 100ppm lo, and the carrier frequency is 2.4G HZ , bandwidth is 10MHZ, then the frequency offset can be as large as 240kHZ. If you use the pilots (there are typically many of them scattered around) as reference signal, as the frequency difference between them is much less than 240kHZ, then you would have issue to lock to any of them.> All of the OFDM specifications that I have seen (and I admit to not > having looked at a whole bunch of them) have pilot tones -- some > channels that are just left on continuously. �These can be used to > recover the carrier, and that information can be used to adjust phase > either before or after the A to D conversion. > > -- > > Tim Wescott > Wescott Design Serviceshttp://www.wescottdesign.com > > Do you need to implement control loops in software? > "Applied Control Theory for Embedded Systems" was written for you. > See details athttp://www.wescottdesign.com/actfes/actfes.html
Reply by ●March 4, 20112011-03-04
On Fri, 4 Mar 2011 08:32:21 -0800 (PST), xsong <songxiaofei@gmail.com> wrote:>On Mar 4, 12:29=A0am, Tim Wescott <t...@seemywebsite.com> wrote: >> On 03/03/2011 09:54 PM, xsong wrote: >> (top posting fixed) >> >> >> >> > On Feb 22, 1:15 pm, Tim Wescott<t...@seemywebsite.com> =A0wrote: >> >> On 02/22/2011 11:53 AM, xsong wrote:> =A0I have been wondering if OFDM= > based system use PLL or DLL as in single >> >>> carrier system e.g. GSM to do synchronization at all? Currently, a lo= >t >> >>> of ofdm based systems talk about implementing the frequency offset an= >d >> >>> time offset compensation in the digital domain. Do they also implemen= >t >> >>> the time and frequency synchronization in hardware? >> >> >> What do you want to bet that some do and some don't? >> >> >> There's advantages no matter which way you go, but unless you've got a >> >> really unstable LO you should be able to pull the signal in without >> >> adjustment in the analog world, and have more than sufficient >> >> information to process it digitally without having to play games out >> >> where it's expensive. >> >> =A0> >> =A0> =A0From systems point of view, what is the definition of =A0a stable= > LO? a >> =A0> 100ppm lo? What are some other specializations? >> >> =A0From a systems point of view, the answer to any question is "that depe= >nds". >> >> Stable enough that your signal is cleanly passed by any IF filtering >> ahead of the analog to digital conversion. =A0What is that exactly? =A0We= >ll, >> it depends on a whole bunch of factors peculiar to your receiver. >> >> =A0> I guess my question really boils down to whether we can use PLL for >> =A0> OFDM based system. Typically for using PLL, we have to generate >> =A0> a reference signal at the carrier frequency. However, for OFDM based >> =A0> systems, there is no single frequency. Then what reference signal ca= >n >> =A0> we use? >> >Thanks Tim for your input. > >However suppose you have a 100ppm lo, and the carrier frequency is >2.4G HZ , bandwidth is 10MHZ, then the frequency offset can be as >large as 240kHZ. > If you use the pilots (there are typically many of them scattered >around) as reference signal, as the frequency difference between them >is much less than 240kHZ, then you would have issue to lock to any of >them.There are almost always features in the signal that can be used to accurately determine the frequency offset. Unless the frequency uncertainty is so large that one can lock onto an adjacent signal with the same format, this is a solvable problem. Even then there is often information in the signal to let you know you got the wrong channel. Is this for a standardized signal of some kind? DVB-T? 802.11? Both (and others) have attritbutes for detecting and removing large frequency offsets. There can be such a thing as too large of an offset (like being unable to discern from adjacent channels), in which case a more accurate LO is needed. Eric Jacobsen http://www.ericjacobsen.org http://www.dsprelated.com/blogs-1//Eric_Jacobsen.php
Reply by ●March 4, 20112011-03-04
On Mar 4, 9:06�am, eric.jacob...@ieee.org (Eric Jacobsen) wrote:> On Fri, 4 Mar 2011 08:32:21 -0800 (PST), xsong <songxiao...@gmail.com> > wrote: > > > > >On Mar 4, 12:29=A0am, Tim Wescott <t...@seemywebsite.com> wrote: > >> On 03/03/2011 09:54 PM, xsong wrote: > >> (top posting fixed) > > >> > On Feb 22, 1:15 pm, Tim Wescott<t...@seemywebsite.com> =A0wrote: > >> >> On 02/22/2011 11:53 AM, xsong wrote:> =A0I have been wondering if OFDM= > > based system use PLL or DLL as in single > >> >>> carrier system e.g. GSM to do synchronization at all? Currently, a lo= > >t > >> >>> of ofdm based systems talk about implementing the frequency offset an= > >d > >> >>> time offset compensation in the digital domain. Do they also implemen= > >t > >> >>> the time and frequency synchronization in hardware? > > >> >> What do you want to bet that some do and some don't? > > >> >> There's advantages no matter which way you go, but unless you've got a > >> >> really unstable LO you should be able to pull the signal in without > >> >> adjustment in the analog world, and have more than sufficient > >> >> information to process it digitally without having to play games out > >> >> where it's expensive. > > >> =A0> > >> =A0> =A0From systems point of view, what is the definition of =A0a stable= > > LO? a > >> =A0> 100ppm lo? What are some other specializations? > > >> =A0From a systems point of view, the answer to any question is "that depe= > >nds". > > >> Stable enough that your signal is cleanly passed by any IF filtering > >> ahead of the analog to digital conversion. =A0What is that exactly? =A0We= > >ll, > >> it depends on a whole bunch of factors peculiar to your receiver. > > >> =A0> I guess my question really boils down to whether we can use PLL for > >> =A0> OFDM based system. Typically for using PLL, we have to generate > >> =A0> a reference signal at the carrier frequency. However, for OFDM based > >> =A0> systems, there is no single frequency. Then what reference signal ca= > >n > >> =A0> we use? > > >Thanks Tim for your input. > > >However suppose you have a 100ppm lo, and the carrier frequency is > >2.4G HZ , bandwidth is 10MHZ, then the frequency offset can be as > >large as 240kHZ. > > If you use the pilots (there are typically many of them scattered > >around) as reference signal, as the frequency difference between them > >is much less than 240kHZ, then you would have issue to lock to any of > >them. > > There are almost always features in the signal that can be used to > accurately determine the frequency offset. � Unless the frequency > uncertainty is so large that one can lock onto an adjacent signal with > the same format, this is a solvable problem. � Even then there is > often information in the signal to let you know you got the wrong > channel. > > Is this for a standardized signal of some kind? � DVB-T? � 802.11? > > Both (and others) have attritbutes for detecting and removing large > frequency offsets.Can you elaborate more on 802.11? What attributes are used to remove large frequency offsets?> > There can be such a thing as too large of an offset (like being unable > to discern from adjacent channels), in which case a more accurate LO > is needed. > > Eric Jacobsenhttp://www.ericjacobsen.orghttp://www.dsprelated.com/blogs-1//Eric_Jacobsen.php
Reply by ●March 4, 20112011-03-04
On Fri, 4 Mar 2011 14:23:05 -0800 (PST), xsong <songxiaofei@gmail.com> wrote:>On Mar 4, 9:06=A0am, eric.jacob...@ieee.org (Eric Jacobsen) wrote: >> On Fri, 4 Mar 2011 08:32:21 -0800 (PST), xsong <songxiao...@gmail.com> >> wrote: >> >> >> >> >On Mar 4, 12:29=3DA0am, Tim Wescott <t...@seemywebsite.com> wrote: >> >> On 03/03/2011 09:54 PM, xsong wrote: >> >> (top posting fixed) >> >> >> > On Feb 22, 1:15 pm, Tim Wescott<t...@seemywebsite.com> =3DA0wrote: >> >> >> On 02/22/2011 11:53 AM, xsong wrote:> =3DA0I have been wondering if= > OFDM=3D >> > based system use PLL or DLL as in single >> >> >>> carrier system e.g. GSM to do synchronization at all? Currently, a= > lo=3D >> >t >> >> >>> of ofdm based systems talk about implementing the frequency offset= > an=3D >> >d >> >> >>> time offset compensation in the digital domain. Do they also imple= >men=3D >> >t >> >> >>> the time and frequency synchronization in hardware? >> >> >> >> What do you want to bet that some do and some don't? >> >> >> >> There's advantages no matter which way you go, but unless you've go= >t a >> >> >> really unstable LO you should be able to pull the signal in without >> >> >> adjustment in the analog world, and have more than sufficient >> >> >> information to process it digitally without having to play games ou= >t >> >> >> where it's expensive. >> >> >> =3DA0> >> >> =3DA0> =3DA0From systems point of view, what is the definition of =3DA= >0a stable=3D >> > LO? a >> >> =3DA0> 100ppm lo? What are some other specializations? >> >> >> =3DA0From a systems point of view, the answer to any question is "that= > depe=3D >> >nds". >> >> >> Stable enough that your signal is cleanly passed by any IF filtering >> >> ahead of the analog to digital conversion. =3DA0What is that exactly? = >=3DA0We=3D >> >ll, >> >> it depends on a whole bunch of factors peculiar to your receiver. >> >> >> =3DA0> I guess my question really boils down to whether we can use PLL= > for >> >> =3DA0> OFDM based system. Typically for using PLL, we have to generate >> >> =3DA0> a reference signal at the carrier frequency. However, for OFDM = >based >> >> =3DA0> systems, there is no single frequency. Then what reference sign= >al ca=3D >> >n >> >> =3DA0> we use? >> >> >Thanks Tim for your input. >> >> >However suppose you have a 100ppm lo, and the carrier frequency is >> >2.4G HZ , bandwidth is 10MHZ, then the frequency offset can be as >> >large as 240kHZ. >> > If you use the pilots (there are typically many of them scattered >> >around) as reference signal, as the frequency difference between them >> >is much less than 240kHZ, then you would have issue to lock to any of >> >them. >> >> There are almost always features in the signal that can be used to >> accurately determine the frequency offset. =A0 Unless the frequency >> uncertainty is so large that one can lock onto an adjacent signal with >> the same format, this is a solvable problem. =A0 Even then there is >> often information in the signal to let you know you got the wrong >> channel. >> >> Is this for a standardized signal of some kind? =A0 DVB-T? =A0 802.11? >> >> Both (and others) have attritbutes for detecting and removing large >> frequency offsets. > >Can you elaborate more on 802.11? What attributes are used to remove >large frequency offsets?The preamble is specifically designed to allow course frequency synchronization followed by fine frequency synchronization. The initial portion of the preamble is a repeating sequence in the time domain that allows detection of a significant frequency offset. Eric Jacobsen http://www.ericjacobsen.org http://www.dsprelated.com/blogs-1//Eric_Jacobsen.php
Reply by ●March 4, 20112011-03-04
On 03/04/2011 02:23 PM, xsong wrote:> On Mar 4, 9:06 am, eric.jacob...@ieee.org (Eric Jacobsen) wrote: >> On Fri, 4 Mar 2011 08:32:21 -0800 (PST), xsong<songxiao...@gmail.com> >> wrote: >> >> >> >>> On Mar 4, 12:29=A0am, Tim Wescott<t...@seemywebsite.com> wrote: >>>> On 03/03/2011 09:54 PM, xsong wrote: >>>> (top posting fixed) >> >>>>> On Feb 22, 1:15 pm, Tim Wescott<t...@seemywebsite.com> =A0wrote: >>>>>> On 02/22/2011 11:53 AM, xsong wrote:> =A0I have been wondering if OFDM= >>> based system use PLL or DLL as in single >>>>>>> carrier system e.g. GSM to do synchronization at all? Currently, a lo= >>> t >>>>>>> of ofdm based systems talk about implementing the frequency offset an= >>> d >>>>>>> time offset compensation in the digital domain. Do they also implemen= >>> t >>>>>>> the time and frequency synchronization in hardware? >> >>>>>> What do you want to bet that some do and some don't? >> >>>>>> There's advantages no matter which way you go, but unless you've got a >>>>>> really unstable LO you should be able to pull the signal in without >>>>>> adjustment in the analog world, and have more than sufficient >>>>>> information to process it digitally without having to play games out >>>>>> where it's expensive. >> >>>> =A0> >>>> =A0> =A0From systems point of view, what is the definition of =A0a stable= >>> LO? a >>>> =A0> 100ppm lo? What are some other specializations? >> >>>> =A0From a systems point of view, the answer to any question is "that depe= >>> nds". >> >>>> Stable enough that your signal is cleanly passed by any IF filtering >>>> ahead of the analog to digital conversion. =A0What is that exactly? =A0We= >>> ll, >>>> it depends on a whole bunch of factors peculiar to your receiver. >> >>>> =A0> I guess my question really boils down to whether we can use PLL for >>>> =A0> OFDM based system. Typically for using PLL, we have to generate >>>> =A0> a reference signal at the carrier frequency. However, for OFDM based >>>> =A0> systems, there is no single frequency. Then what reference signal ca= >>> n >>>> =A0> we use? >> >>> Thanks Tim for your input. >> >>> However suppose you have a 100ppm lo, and the carrier frequency is >>> 2.4G HZ , bandwidth is 10MHZ, then the frequency offset can be as >>> large as 240kHZ. >>> If you use the pilots (there are typically many of them scattered >>> around) as reference signal, as the frequency difference between them >>> is much less than 240kHZ, then you would have issue to lock to any of >>> them. >> >> There are almost always features in the signal that can be used to >> accurately determine the frequency offset. Unless the frequency >> uncertainty is so large that one can lock onto an adjacent signal with >> the same format, this is a solvable problem. Even then there is >> often information in the signal to let you know you got the wrong >> channel. >> >> Is this for a standardized signal of some kind? DVB-T? 802.11? >> >> Both (and others) have attritbutes for detecting and removing large >> frequency offsets. > > Can you elaborate more on 802.11? What attributes are used to remove > large frequency offsets?Have you looked at the specification? I would do it with a unique arrangement of tones, i.e. tones spaced in some unambiguous way so that if you saw those spikes in your spectrum you'd know where to put your carrier. Then, if I were designing for systems that may be off by a whole channel, I'd put some sort of supervisory information in the message to say what channel it was on. So if, for instance, you were trying for channel 4, but you locked onto something that claimed to be channel 3, you could say "oops" and move your frequency over by one channel. But it'll be in the spec.>> There can be such a thing as too large of an offset (like being unable >> to discern from adjacent channels), in which case a more accurate LO >> is needed. >> >> Eric Jacobsenhttp://www.ericjacobsen.orghttp://www.dsprelated.com/blogs-1//Eric_Jacobsen.php >-- Tim Wescott Wescott Design Services http://www.wescottdesign.com Do you need to implement control loops in software? "Applied Control Theory for Embedded Systems" was written for you. See details at http://www.wescottdesign.com/actfes/actfes.html
Reply by ●March 5, 20112011-03-05
On Mar 4, 2:34�pm, eric.jacob...@ieee.org (Eric Jacobsen) wrote:> On Fri, 4 Mar 2011 14:23:05 -0800 (PST), xsong <songxiao...@gmail.com> > wrote: > > > > >On Mar 4, 9:06=A0am, eric.jacob...@ieee.org (Eric Jacobsen) wrote: > >> On Fri, 4 Mar 2011 08:32:21 -0800 (PST), xsong <songxiao...@gmail.com> > >> wrote: > > >> >On Mar 4, 12:29=3DA0am, Tim Wescott <t...@seemywebsite.com> wrote: > >> >> On 03/03/2011 09:54 PM, xsong wrote: > >> >> (top posting fixed) > > >> >> > On Feb 22, 1:15 pm, Tim Wescott<t...@seemywebsite.com> =3DA0wrote: > >> >> >> On 02/22/2011 11:53 AM, xsong wrote:> =3DA0I have been wondering if= > >OFDM=3D > >> > based system use PLL or DLL as in single > >> >> >>> carrier system e.g. GSM to do synchronization at all? Currently, a= > > lo=3D > >> >t > >> >> >>> ofofdmbased systems talk about implementing the frequency offset= > > an=3D > >> >d > >> >> >>> time offset compensation in the digital domain. Do they also imple= > >men=3D > >> >t > >> >> >>> the time and frequency synchronization in hardware? > > >> >> >> What do you want to bet that some do and some don't? > > >> >> >> There's advantages no matter which way you go, but unless you've go= > >t a > >> >> >> really unstable LO you should be able to pull the signal in without > >> >> >> adjustment in the analog world, and have more than sufficient > >> >> >> information to process it digitally without having to play games ou= > >t > >> >> >> where it's expensive. > > >> >> =3DA0> > >> >> =3DA0> =3DA0From systems point of view, what is the definition of =3DA= > >0a stable=3D > >> > LO? a > >> >> =3DA0> 100ppm lo? What are some other specializations? > > >> >> =3DA0From a systems point of view, the answer to any question is "that= > > depe=3D > >> >nds". > > >> >> Stable enough that your signal is cleanly passed by any IF filtering > >> >> ahead of the analog to digital conversion. =3DA0What is that exactly? = > >=3DA0We=3D > >> >ll, > >> >> it depends on a whole bunch of factors peculiar to your receiver. > > >> >> =3DA0> I guess my question really boils down to whether we can use PLL= > > for > >> >> =3DA0>OFDMbased system. Typically for using PLL, we have to generate > >> >> =3DA0> a reference signal at the carrier frequency. However, forOFDM= > >based > >> >> =3DA0> systems, there is no single frequency. Then what reference sign= > >al ca=3D > >> >n > >> >> =3DA0> we use? > > >> >Thanks Tim for your input. > > >> >However suppose you have a 100ppm lo, and the carrier frequency is > >> >2.4G HZ , bandwidth is 10MHZ, then the frequency offset can be as > >> >large as 240kHZ. > >> > If you use the pilots (there are typically many of them scattered > >> >around) as reference signal, as the frequency difference between them > >> >is much less than 240kHZ, then you would have issue to lock to any of > >> >them. > > >> There are almost always features in the signal that can be used to > >> accurately determine the frequency offset. =A0 Unless the frequency > >> uncertainty is so large that one can lock onto an adjacent signal with > >> the same format, this is a solvable problem. =A0 Even then there is > >> often information in the signal to let you know you got the wrong > >> channel. > > >> Is this for a standardized signal of some kind? =A0 DVB-T? =A0 802.11? > > >> Both (and others) have attritbutes for detecting and removing large > >> frequency offsets. > > >Can you elaborate more on 802.11? What attributes are used to remove > >large frequency offsets? > > The preamble is specifically designed to allow course frequency > synchronization followed by fine frequency synchronization. �The > initial portion of the preamble is a repeating sequence in the time > domain that allows detection of a significant frequency offset. >Thanks Eric. However ,using the preamble to do synchronization is essentially in the digital domain rather than using PLL in the analog domain. I just wonder if you could do it using PLL for ofdm based system.> Eric Jacobsenhttp://www.ericjacobsen.orghttp://www.dsprelated.com/blogs-1//Eric_Jacobsen.php
