Hi I have a squarewave subcarrier signal @ 2Mhz modulated with a 250Khz NRZI data signal (no pulse shaping). The data is send in bursts with a 32 microsecond preeamble (subcarrier only). I sample this witha a free-running ADC @ 10Mhz. This will give me 40 samples/symbol so I decimate this with a rate of 4 and BP-filter the result. I would like to demodulate this signal with a digital Costas Loop. Having a square wave, using I&D-filters (matched filter) in the I and Q-arm is a good choice. I do not want to loose SNR in these filters so I must have a perfect symbol synchronization so that I dump the I&D-filters at the correct time, how do I accomplish this? �or should I just use FIR-LP-filters instead? Implementing the I&D-filters; should I output the intermediate values every Ts or should I integrate the whole symbol time and then output the values from these filters. If I take the values at symbol boundaries, it will result in a decimation by ten...and thus I can not use this for symbol sync�I lack the experience of this, how do one do? I have seen that symbol synhronization algorithms like Gardner etc works with 2-4 samples/symbol and have a rather long lock time, sometimes several hundred symbols. For example the Mueller-Mueller detector locks in 15-20 symbols.I would like to lock my Costas Loop and symbol sync in 2-3 symbols�is that possible. I have a squarewave and 10 samples/symbol, do I need an interpolating algorithm like the above mentioned or should I use a digital pll (i.e early-late-gate) to extract the clock?? Do I get less timing jitter using an interpolating algorithm?? I am thinking about implementing this in this order: (1) Demodulate (2) Symbol Sync & clock extraction (3) Decide what was send Is that possible at all in burst-communication, or should one do it reversed; symbol sync, demodulate�. As you see, I have major problems. Please help /Karl
BPSK demodulation and synchronization (square wave subcarrier)
Started by ●February 6, 2006
Reply by ●February 6, 20062006-02-06
Karl wrote:> Hi >=20 > I have a squarewave subcarrier signal @ 2Mhz modulated with a 250Khz NR=ZI> data signal (no pulse shaping). The data is send in bursts with a 32 > microsecond preeamble (subcarrier only). I sample this witha a > free-running ADC @ 10Mhz. This will give me 40 samples/symbol so I > decimate this with a rate of 4 and BP-filter the result.So you do have 10 samples per symbol.>=20 > I would like to demodulate this signal with a digital Costas Loop. Havi=ng> a square wave, using I&D-filters (matched filter) in the I and Q-arm is=a> good choice. I do not want to loose SNR in these filters so I must have=a> perfect symbol synchronization so that I dump the I&D-filters at the > correct time, how do I accomplish this? =85or should I just use > FIR-LP-filters instead?In your case, matched filters are simple boxcars. You can synchronize on = the zero crossing in I channel. Since you have as many as 10 samples per = symbol that will give you enough accuracy to adjust the sampling instant = directly without NCOs and PLLs. Costas loop could be free running; it=20 does hot have to be symbol synchronized. Of course there are some better = methods to skin the cat however it depends if the result is worth=20 complexity.>=20 > Implementing the I&D-filters; should I output the intermediate values > every Ts or should I integrate the whole symbol time and then output th=e> values from these filters.Your filters should run continiously, i.e. 10 times per symbol.> If I take the values at symbol boundaries, it will result in a decimat=ion> by ten...and thus I can not use this for symbol sync=85I lack the exper=ience> of this, how do one do?There are two methods to accomplish anything: do it yourself or hire=20 somebody else.>=20 > I have seen that symbol synhronization algorithms like Gardner etc work=s> with 2-4 samples/symbol and have a rather long lock time, sometimes > several hundred symbols.=20 > For example the Mueller-Mueller detector locks in 15-20 symbols.I would=> like to lock my Costas Loop and symbol sync in 2-3 symbols=85is that > possible.=20Those algorithms are the overkill in your simple case.>=20 > I have a squarewave and 10 samples/symbol, do I need an interpolating > algorithm like the above mentioned or should I use a digital pll (i.e > early-late-gate) to extract the clock?? Do I get less timing jitter usi=ng> an interpolating algorithm??=20Sometimes, extra knowledge only hurts you :-)>=20 > I am thinking about implementing this in this order: > (1) Demodulate > (2) Symbol Sync & clock extraction > (3) Decide what was send > Is that possible at all in burst-communication, or should one do it > reversed; symbol sync, demodulate=85. >=20 > As you see, I have major problems. Please helpNo you don't. There is the only one major problem: where to go for lunch.= Vladimir Vassilevsky DSP and Mixed Signal Design Consultant http://www.abvolt.com
Reply by ●February 6, 20062006-02-06
Reply by ●February 6, 20062006-02-06
Karl wrote:> Hello, > > I forgot to mention that the NRZI data signal is BPSK-modulated. > > /Karl1. Please include pertinent context when replying to posts -- see just about any other reply in this newsgroup for an example. 2. Do you mean that the NRZI data signal itself is BPSK modulated, or do you mean that it modulates the carrier? -- Tim Wescott Wescott Design Services http://www.wescottdesign.com
Reply by ●February 6, 20062006-02-06
Vladimir Vassilevsky wrote:> > > Karl wrote:...>> As you see, I have major problems. Please help > > > No you don't. There is the only one major problem: where to go for lunch.Now here's a man who has his priorities straight! :-) ... Jerry -- Engineering is the art of making what you want from things you can get. ¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯
Reply by ●February 7, 20062006-02-07
Hi First there is a squarewave subcarrier, then this squarewave is modulated with BPSK with NRZI-data. I'll usa a Costas + DTTL!
Reply by ●February 7, 20062006-02-07
Karl wrote:> Hi > > First there is a squarewave subcarrier, then this squarewave is > modulated with BPSK with NRZI-data. I'll usa a Costas + DTTL!I'm curious about the square-wave carrier. A 2 MHz square wave has significant energy at 18 MHz. Some 20% of the power is in the harmonics. What good does it do? Jerry -- Engineering is the art of making what you want from things you can get. �����������������������������������������������������������������������
Reply by ●February 7, 20062006-02-07
Jerry Avins wrote:> Karl wrote: > > Hi > > > > First there is a squarewave subcarrier, then this squarewave is > > modulated with BPSK with NRZI-data. I'll usa a Costas + DTTL! > > I'm curious about the square-wave carrier. A 2 MHz square wave has > significant energy at 18 MHz. Some 20% of the power is in the harmonics. > What good does it do? >Could be that it was easier to implement that way, maybe it is old technology. John
Reply by ●February 7, 20062006-02-07
john wrote:> Jerry Avins wrote: > >>Karl wrote: >> >>>Hi >>> >>>First there is a squarewave subcarrier, then this squarewave is >>>modulated with BPSK with NRZI-data. I'll usa a Costas + DTTL! >> >>I'm curious about the square-wave carrier. A 2 MHz square wave has >>significant energy at 18 MHz. Some 20% of the power is in the harmonics. >>What good does it do? >> > > > Could be that it was easier to implement that way, maybe it is old > technology.One wouldn't provide transmission or storage capacity for a 2 MHz square wave when a 2 MHz sine wave would do, and if no transmission or storage is involved, why encode? I'd like to know, rather than guess. Maybe Karl will tell us. Jerry -- Engineering is the art of making what you want from things you can get. �����������������������������������������������������������������������
Reply by ●February 16, 20062006-02-16
Hi A squarewave subcarrier is common in satellite communication and road tolls....I guess because its very easy to modulate a squarewave with BPSK. One just have to count edges and then change the reflect coefficient. Karl>john wrote: >> Jerry Avins wrote: >> >>>Karl wrote: >>> >>>>Hi >>>> >>>>First there is a squarewave subcarrier, then this squarewave is >>>>modulated with BPSK with NRZI-data. I'll usa a Costas + DTTL! >>> >>>I'm curious about the square-wave carrier. A 2 MHz square wave has >>>significant energy at 18 MHz. Some 20% of the power is in theharmonics.>>>What good does it do? >>> >> >> >> Could be that it was easier to implement that way, maybe it is old >> technology. > >One wouldn't provide transmission or storage capacity for a 2 MHz square>wave when a 2 MHz sine wave would do, and if no transmission or storage >is involved, why encode? I'd like to know, rather than guess. Maybe Karl>will tell us. > >Jerry >-- >Engineering is the art of making what you want from things you can get. >����������������������������������������������������������������������� >