Re: [digitalradio] PSK Modes
Hold those thoughts for the ARRL Prize. But yes there is lots of thought about narrow moodes, 500 Hz and less and wider modes...3.5 KHz and perhaps 5-6 KHz. The ARRL has proposed a 3.5 KHz bandwidth limit for wide bandwidth modes but I think the FCC wants to limit it to less than 3 KHz (maybe only 2.5 KHz) as the widedest bandwidth in the current SSB/SlowScan TV bands. Walt/K5YFW KV9U wrote: Why do some modems use more rectangular waveforms instead of what appears to be the optimum waveform for HF modems? Or are there downsides to raised cosine waveforms? In terms of bandwidth, it seems to me that for most uses, a 500 Hz bandwidth is a wise choice. This seems to be a good tradeoff in width vs potential throughput for keyboard modes and even some higher speed data modes like Pactor 2 can do under better conditions. Also, 500 Hz filters have been commonly available for CW. With more rigs using DSP filters, I admit that it is less of an issue to tailor make it to one particular width. How about a two tone, DPSK scheme with 50 (maybe even 25?), and also 100 and 200 baud rates? Even if you would initially require manual adjustment. Then we also should have a mode that can run in a voice channel, probably 2.4 to at most 2.7 KHz width. How about an 8 tone DBPSK and maybe switchable (manually at first) to higher PSK rates? All of these modes could have similar timing for ARQ. How about 0.5 seconds pause between transmissions and awaiting an ACK or NAK or control signal? With right control signals you could change the length of time for the packet burst. But for starters, maybe just a simple packet size. 73, Rick, KV9U Walt DuBose wrote: One solution suggested to me was that each tone be individually shaped/filtered before transmitting and then each tone have an individual brick wall filter before it is decoded. I believe that there is not going to be one mode or mode configuration that works well on 3-30 MHz...we will probably have several sets of configurations or perhaps even one optimum configuration for each band or set of conditions. The more options you have that you can adjust of the fly or that can be used adaptively the better off the mode will be. Someone (perhaps all) needs to keep technical notes on what modes work best on what band. Also, we need to come to an agreement on what mazimum bandwidth and user throughput we want as well as how robust and how sensitive we want the mode to be. My personal belief is that we go for 500 Hz wandwidth, 400-800 WPM user throughput 99.9% error free and work down below a 0 dB SNR (0 to -5 dB) on a poor CCIR channel. Put your thinking caps on and make your wish list. 73 CLU, Walt/K5YFW KV9U wrote: Some of us did try Chip modes when Nino first came out with them, but they did not seem to perform as well as existing modes. I really implore to our treasured programmers to see if they can come up with some modes that can compete with Pactor modes. Especially some ARQ modes that can work on MS OS. We know from Pactor 2, that a raised cosine shaped pulse is likely a very good basic waveform. Then for the most robust mode, a two tone DBPSK modulation is used and as the conditions improve, the modulation changes to DQPSK and then with further improvements to 8-DPSK and even 16-DPSK for maximum throughput when conditions are very good. This is what enables Pactor 2 to send about 700 bits per second at the peak speed and do it in only a 500 Hz wide span. We know this can be done at the higher speeds under good conditions with sound card modes since SCAMP was even faster than P2, although a much wider signal. The problem with SCAMP was that it had no fallback position. Pactor 3 is runs an occupied bandwidth of about 2.4 kHz, but raw speed is over 2700 bps. Instead of 2 tones, P3 uses up to 18, separated by 120 Hz and modulated at 100 baud DBPSK or DQPSK. SCS has some fairly detailed data on Pactor 3 at: http://www.scs- ptc.com/download /PACTOR-III- Protocol. pdf http://www.scs- ptc.com/download /PACTOR-III- Protocol. pdf http://www.scs-ptc.com/download/PACTOR-III-Protocol.pdf I wish someone could explain why we can not have a sound card mode that is roughly the same as Pactor 2 at least. Even if there was no ARQ at first. And how different is Pactor 3, than what the SSTV hams are using everyday? Aren't they using OFDM with QAM? If you recall what Tom Rink said back in 1995 on the TAPR HF SIG: As mentioned in the introduction, PACTOR-II uses a two-tone DPSK modulation system. Due to the raised cosine pulse shaping, the maximum required bandwidth is only around 450 Hz at minus 50 dB. ASK, which was also tested in the early stage, provided poorer results in weak conditions compared with a higher DPSK modulation, as different amplitude levels are more
Re: [digitalradio] PSK Modes
I did some more research and found out that according to the Wikipedia (which I find to be the most incredible resource on the internet for general encyclopedic information), PSK (Phase Shift Keying) can be considered to be a subset of QAM (Quadrature Amplitude Modulation). That rather surprised me since QAM is using two 90 degrees out of phase carriers that convey data by changing the amplitude of these carriers (tones). PSK is only the changing of the phase but keeping the amplitude constant. One reference considers QAM to be a combination of ASK and PSK, where there is at least two phases and at least 2 amplitudes. And QAM doesn't have to be digital in nature. PAL is a type of analog QAM and perhaps by extension, so are SECAM and the NTSC color TV generation? (For those who are not aware, NTSC is from the organization that came up with the U.S. color TV system that was sort of compatible with black and white TV, the:National Television System Committee. There are some who insist it really means Never The Same Color:) Apparently, one of the main attributes of QAM is the significant savings in bandwidth. Electronic Design pointed out at: http://www.elecdesign.com/Articles/Index.cfm?AD=1ArticleID=2372 That 4-PSK (same as QPSK) having a single amplitude level is actually the same thing as 4QAM. In terms of the coding types, I was thinking that pactor used Viterbi and Clover II used Reed-Solomon. But if Pactor uses both, then this makes it difficult to compare directly, and apparently the combination is better. 73, Rick, KV9U Jose A. Amador wrote: KV9U wrote: If I understand it correctly, the raised cosine pulses tend to be more efficient with power, reduce the crest factor (Pactor 2 is under 1.5), and perhaps make it easier to have a cleaner signal. Raised cosine is, above all, less bandwidth greedy. Just for clarification I have a question: Is QAM modulation a form of ASK? It would seem so to me but I am not sure. Otherwise, what other modulation forms fall into the ASK category? It may be seen as that. Just depend on what abstraction you make to reach that conclusion. FSK is a form of complementary ASK of two carrierswhich is bad is pure ASK because one state is pure signal and another pure garbage (noise, etc). QAM can be seen as an ASK of four phases at a fixed amplitude. Using two quadrature modulators, you create four states keying them with (1,1) (1,-1), (-1, -1) and (-1,1). 1 is the same phase, -1, reversed phase. Combine them and you get a constellation with points every 45 degrees. off the XY axis. Although the SSTV modes are not automatically adaptive, there is a limited choice of number of tones, but for the most part I believe that they have found 16QAM to be about all you can get to work well on many HF circuits, particularly on the lower frequencies. It depends on the signal to noise ratios. There is a video presentation of Doug Smith on Georgia Tech about Digital Voice which is pretty illustrative. Some Googling should find it. It shows the constellations and the effects of noise on it. The more complex the constellation, the less distance there is between constellation points, and so, less leeway for noise before confusing the decoder. I wonder how a 4-QAM mode compares to say a 4-PSK mode when up against the ionosphere? Should be about the same... There must have been a reason that DRM uses QAM instead of PSK? Any thoughts on why? When you need a modulator for 64QAM for the MSC, it is rather easy to create 4QAM with the four extreme points of the 64QAM constellations. In terms of coding, it would be very interesting is to compare two multitone modems, perhaps a 2 tone and an 8 tone (similar to pactor 2 and 3) and have one with R-S and one with Viterbi and see if there is any difference on various circuits. Pactor II and III use both Viterbi decoding and block encoding with interleaving... That is not the test that needs to be done. The difference between P2 and P3, is that P3 stays with the most robust and capable constallation, 4DPSK, and starts deploying carriers using it. The coding tricks are about the same. What I don't know so far is how does it distribute the traffic among the carriers. Jose, CO2JA
Re: [digitalradio] PSK Modes
KV9U wrote: I did some more research and found out that according to the Wikipedia (which I find to be the most incredible resource on the internet for general encyclopedic information), I can agree on that8-) PSK (Phase Shift Keying) can be considered to be a subset of QAM (Quadrature Amplitude Modulation). I told before that it all depends on the abstractions you make. I do not exactly agree on PSK being a subset of QAM. For those who can understand spanish, there is a poem by Calderon that goes like this: En este mundo traidor nada es verdad ni es mentira todo sera del color del cristal con que se mira I used spanish to keep the rhyme. More or less it can be translated as: In this treacherous world, nothing is false or true, it all will be tinted the color of the crystal you look thru. But PSK can be thought as an ASK of two out of phase sine waves. That rather surprised me since QAM is using two 90 degrees out of phase carriers that convey data by changing the amplitude of these carriers (tones). PSK is only the changing of the phase but keeping the amplitude constant. One reference considers QAM to be a combination of ASK and PSK, where there is at least two phases and at least 2 amplitudes. And QAM doesn't have to be digital in nature. No, it NEVER is digital. Whatever modulates a transmitter with digital information is analog... that MAY convey digital info. All those slow envelope or frequency transitions...the use of LINEAR amplifiers is needed for ANALOG signals. Once again, the message, the content, may be digital or analog. PAL is a type of analog QAM and perhaps by extension, so are SECAM and the NTSC color TV generation? PAL and NTSC use QAM with analog messages, two color difference signals (or I/Q in the case of NTSC, a variation of color difference signals) SECAM is not. SECAM is a sequential analog color system that uses FM to transmit color difference (R-Y/B-Y) signals, and that cannot send two components at once, as NTSC and PAL can, using two quadrature displaced sinewaves (some may prefer the mention of a sine wave and a cosine wave) that are linearly independent and can send two messages simultaneously. In practice, it translates into a chroma vector whose phase (referred to the burst phase) conveys hue and its amplitude conveys saturation. Some jokers qualify SECAM as : System Essentially Contrary to the American Method. It is really too complicated in the studio, and yes, it does not suffer the problems of NTSC, but suffers some new worse ones. (For those who are not aware, NTSC is from the organization that came up with the U.S. color TV system that was sort of compatible with black and white TV, the:National Television System Committee. There are some who insist it really means Never The Same Color:) On its beginnings it was nicknamed Never Twice Same Color...now it seems it is not justified. Feedback solid state video amplifiers made great differences with regards to its original performance. The same system with open loop tube circuits with AGC and reactance tube property was really asking for trouble, as it made differential gain and phase distortions inherent to those old circuits. Apparently, one of the main attributes of QAM is the significant savings in bandwidth. Yes, it can save half the bandwitdth, just by itself. Electronic Design pointed out at: http://www.elecdesign.com/Articles/Index.cfm?AD=1ArticleID=2372 http://www.elecdesign.com/Articles/Index.cfm?AD=1ArticleID=2372 That 4-PSK (same as QPSK) having a single amplitude level is actually the same thing as 4QAM. Yes and no. The PSK constellation is on the axes, and the QAM constellation is 45 degrees displaced. In terms of the coding types, I was thinking that pactor used Viterbi and Clover II used Reed-Solomon. But if Pactor uses both, then this makes it difficult to compare directly, and apparently the combination is better. To be true, I cannot assure by heart that the block coding method that pactor 2 uses is RS. I have to reread the fine manual once again. But one of the properties of block codes is that they can detect and CORRECT errors without retransmission. How many bits can be corrected at once, depends on the code used. As far as I understand, Pactor II outperforms Clover. I have never owned Clover equipment, but all I have read so far indicates that. It had a fundamental weakness negotiating speed in worsening propagation conditions, it lost the link without being able to negotiate a speed fallback. Pactor II and III seemingly do very well on that front. Jose, CO2JA -- MSc.Jose Angel Amador Fundora Departamento de Telecomunicaciones Facultad de Ingenieria Electrica, CUJAE Calle 114 #11901 e/ 119 y 127 Marianao 19390, Ciudad de la Habana, Cuba Tel:(53 7) 266-3352 Email: [EMAIL PROTECTED]
Re: [digitalradio] PSK Modes
Please allow me to make some comments based on research done here in San Antonio. As some may know, SouthWest Research Institute is located here in San Antonio and has done work on high baud rate modes with very poor SNRs. Also they are part of the current project that is flying a space craft passt Saturn or Jupitor going to Pluto. Note that the project and space craft were launched after Pluto was demoted. There research has shown that a tone/carrier with more than 8 phases is going to be very hard to detect and that in fact to get it to be error free, the decode time may be as long as using a simpler mode with ARQ. They showed one case where it took 4 seconds to decode 1 second of transmission. There is really great potential in OFDM types of modems if bandwidth is not a concern. However, if we want to keep the bandwidth under 500 Hz, then you are going to be limited to the number of tones you use, the baud rate, gaurd band, etc, etc. There are notes about brick wall filters and of course tone shaper filters, etc. One solution suggested to me was that each tone be individually shaped/filtered before transmitting and then each tone have an individual brick wall filter before it is decoded. I believe that there is not going to be one mode or mode configuration that works well on 3-30 MHz...we will probably have several sets of configurations or perhaps even one optimum configuration for each band or set of conditions. The more options you have that you can adjust of the fly or that can be used adaptively the better off the mode will be. Someone (perhaps all) needs to keep technical notes on what modes work best on what band. Also, we need to come to an agreement on what mazimum bandwidth and user throughput we want as well as how robust and how sensitive we want the mode to be. My personal belief is that we go for 500 Hz wandwidth, 400-800 WPM user throughput 99.9% error free and work down below a 0 dB SNR (0 to -5 dB) on a poor CCIR channel. Put your thinking caps on and make your wish list. 73 CLU, Walt/K5YFW KV9U wrote: Some of us did try Chip modes when Nino first came out with them, but they did not seem to perform as well as existing modes. I really implore to our treasured programmers to see if they can come up with some modes that can compete with Pactor modes. Especially some ARQ modes that can work on MS OS. We know from Pactor 2, that a raised cosine shaped pulse is likely a very good basic waveform. Then for the most robust mode, a two tone DBPSK modulation is used and as the conditions improve, the modulation changes to DQPSK and then with further improvements to 8-DPSK and even 16-DPSK for maximum throughput when conditions are very good. This is what enables Pactor 2 to send about 700 bits per second at the peak speed and do it in only a 500 Hz wide span. We know this can be done at the higher speeds under good conditions with sound card modes since SCAMP was even faster than P2, although a much wider signal. The problem with SCAMP was that it had no fallback position. Pactor 3 is runs an occupied bandwidth of about 2.4 kHz, but raw speed is over 2700 bps. Instead of 2 tones, P3 uses up to 18, separated by 120 Hz and modulated at 100 baud DBPSK or DQPSK. SCS has some fairly detailed data on Pactor 3 at: http://www.scs- ptc.com/download /PACTOR-III- Protocol. pdf http://www.scs-ptc.com/download/PACTOR-III-Protocol.pdf I wish someone could explain why we can not have a sound card mode that is roughly the same as Pactor 2 at least. Even if there was no ARQ at first. And how different is Pactor 3, than what the SSTV hams are using everyday? Aren't they using OFDM with QAM? If you recall what Tom Rink said back in 1995 on the TAPR HF SIG: As mentioned in the introduction, PACTOR-II uses a two-tone DPSK modulation system. Due to the raised cosine pulse shaping, the maximum required bandwidth is only around 450 Hz at minus 50 dB. ASK, which was also tested in the early stage, provided poorer results in weak conditions compared with a higher DPSK modulation, as different amplitude levels are more difficult to distinguish in noisy channels than more phase levels. Additionally, ASK increases the Crest Factor of the signal. For these reasons, it is not used in the final PACTOR-II protocol. Basic information on these items can also be found in the first part of this series. Although not ASK, doesn't QAM employ amplitude changes as part of the modulation scheme? What happens if you use a multitone DPSK? It seems to a non-engineering person like myself, that a lot of what P2 and P3 are made up of are really a series of PSK100 or PSK200 tones (carriers). Isn't Q15X25 a similar modulation scheme? It even runs at 83.33 baud rather than a minimum of 100 baud such as P2. Why did it not work as well as P modes? Or is it because it has no coding such as Reed-Solomon block coding or Viterbi
Re: [digitalradio] PSK Modes
Nino, That was kind of my thought...interesting about the Chip64 decoder...I will have to study the mode more. In my post I did day that it was a 100 mile path but did not stipulate that is was all over land. Also, unless you live in an area where the ground conductivity changes a large amount over 100 miles...as in transmitting from the coast to inland, you won't generally see all that much change as the ground conductivity is gradual...except as the above case and some places in Colorado, California and Germany that I know about. Thus I normally tend to disregard loss due to ground loss/attenuation. Also, you might be thinking that the first hop of three hops ia 33 miles, the second 66 miles and the third 100 miles. Not necesarly so and generally not. The First hop might be anywhere from 10-15 miles to 35 miles, the next hop more or less than 66 miles and the third hop more or less than 100 miles. When the signal hits the ground it spladders as my old PhD in physics Elmer used to say...he maintained that it created another set of groundwaves and that groundwaves from the various hops could mix with the skywave signals and cause even worse signals that you describe Nino. REalize that this was in the 60's and his observations were from the 30's and 40's when little was really known about the ionosphere. Also there are hops between the F1 and F2 layer during the day so in fact you might have 3-6 hops before you receive the signal with only 2 being skywave hops. Very complex. The key to overcoming all this has got to be a way to know exactly which signal is the real signal. There is a system that originally used an atomic clock to track signals and today uses a GPS clock. I can't say much more about that system. (Because I don't know much more about it.) IMHO, the PSK modes have dealt adequately with the noise problem but the problems caused by the ionospheric have yet to be adequately addresses. Thanks for your input Nino. 73 All and CUL, Walt/K5YFW Nino Porcino (IZ8BLY) wrote: Walt/K5YFW wrote: if you may be receiving 1, 2 and 3 hop signals. How does this affect BPSK and QPSK signals from for example PSK31/63/125? the 3 different signals will sum at the receiver, but, having each one a different phase, the sum is destructive with the result that they tend to cancel. If the paths are stable you notice a drop in the signal strength but if paths are unstable (as it is often the case) one signal may win over the others and the phase of the PSK decoder will wander back and forth. The clock recovery is also problematic because of the unstability of the reference. Among the possible solutions to multipath there is the spread spectrum modulation (as in Chip64) where the symbols at the receiver aren't expected at a precise timing, but are decoded in a clockless manner. In Chip64 signal scope you can actually see the signal trace wandering left and rigth due to path hopping or see the ghosted trace of the secondary path. Nino/IZ8BLY
Re: [digitalradio] PSK Modes
If I understand it correctly, the raised cosine pulses tend to be more efficient with power, reduce the crest factor (Pactor 2 is under 1.5), and perhaps make it easier to have a cleaner signal. Just for clarification I have a question: Is QAM modulation a form of ASK? It would seem so to me but I am not sure. Otherwise, what other modulation forms fall into the ASK category? Although the SSTV modes are not automatically adaptive, there is a limited choice of number of tones, but for the most part I believe that they have found 16QAM to be about all you can get to work well on many HF circuits, particularly on the lower frequencies. I wonder how a 4-QAM mode compares to say a 4-PSK mode when up against the ionosphere? There must have been a reason that DRM uses QAM instead of PSK? Any thoughts on why? In terms of coding, it would be very interesting is to compare two multitone modems, perhaps a 2 tone and an 8 tone (similar to pactor 2 and 3) and have one with R-S and one with Viterbi and see if there is any difference on various circuits. Does anyone have information on this already? 73, Rick, KV9U Jose_Angel Amador Fundora wrote: We know from Pactor 2, that a raised cosine shaped pulse is likely a very good basic waveform. That is for saving bandwidth, mostly. It might allow better decoding, as well. Pactor 3 is runs an occupied bandwidth of about 2.4 kHz, but raw speed is over 2700 bps. Instead of 2 tones, P3 uses up to 18, separated by 120 Hz and modulated at 100 baud DBPSK or DQPSK. SCS has some fairly detailed data on Pactor 3 at: http://www.scs-ptc.com/download/PACTOR-III-Protocol.pdf I wish someone could explain why we can not have a sound card mode that is roughly the same as Pactor 2 at least. Even if there was no ARQ at first. I don't know if the least complex of it all is ARQ...most likely, the rest is harder to implement. Yes. A key requirement is having the highest distance between constellation points to have an edge against the noise (or QRM). That's why, in DRM, the FAC uses 4QAM, as it allows to send the reduced but very important info it conveys. But the MSC must use 64QAM, because the amount of data to be sent does not allow otherwise in the least bandwidth. What happens if you use a multitone DPSK? It seems to a non-engineering person like myself, that a lot of what P2 and P3 are made up of are really a series of PSK100 or PSK200 tones (carriers). Isn't Q15X25 a similar modulation scheme? It even runs at 83.33 baud rather than a minimum of 100 baud such as P2. Why did it not work as well as P modes? Or is it because it has no coding such as Reed-Solomon block coding or Viterbi convolutional coding? Certainly...all those tricks add up, and most likely, in a non proportional way...I cannot assure it by heart, but is very likely. One of the gains of the code used in pactor modes is using convolutional encoding with Viterbi decoding. The Viterbi decoder, knowing the history of what has been sent, as the convolutionally coded stream depends on what has been sent previously, makes a soft decode of what is the most likely symbol transmitted. RS coding, after deinterleaving, on the other side, may allow to recover erors WITHOUT retransmission, which may save more bandwidth than what is wasted on the FEC overhead. Also, P2 and P3 avoid the edges of the channel to have the least amplitude and delay differences between carriers. That's why a reduced version of Q15X25 is being more succesful in holding the link. 73, Rick, KV9U 73, Jose, CO2JA __ __ __ __ Correo enviado por ElectroMAIL. Facultad Eléctrica. CUJAE Dominio: electrica.cujae.edu.cu
Re: [digitalradio] PSK Modes
I'm always showing my ignorance, but what is cam. I kind of understand psk and bqpsk as my ARD9800 uses bqpsk for the carriers carrying the voice data and sst and ascii keyboarding function of the modem. Sorry to tie up the group with an elementary question that I should know. 73s, Jack wa5rop - Original Message - From: KV9U [EMAIL PROTECTED] To: digitalradio@yahoogroups.com Sent: Tuesday, January 23, 2007 9:55 AM Subject: Re: [digitalradio] PSK Modes If I understand it correctly, the raised cosine pulses tend to be more efficient with power, reduce the crest factor (Pactor 2 is under 1.5), and perhaps make it easier to have a cleaner signal. Just for clarification I have a question: Is QAM modulation a form of ASK? It would seem so to me but I am not sure. Otherwise, what other modulation forms fall into the ASK category? Although the SSTV modes are not automatically adaptive, there is a limited choice of number of tones, but for the most part I believe that they have found 16QAM to be about all you can get to work well on many HF circuits, particularly on the lower frequencies. I wonder how a 4-QAM mode compares to say a 4-PSK mode when up against the ionosphere? There must have been a reason that DRM uses QAM instead of PSK? Any thoughts on why? In terms of coding, it would be very interesting is to compare two multitone modems, perhaps a 2 tone and an 8 tone (similar to pactor 2 and 3) and have one with R-S and one with Viterbi and see if there is any difference on various circuits. Does anyone have information on this already? 73, Rick, KV9U Jose_Angel Amador Fundora wrote: We know from Pactor 2, that a raised cosine shaped pulse is likely a very good basic waveform. That is for saving bandwidth, mostly. It might allow better decoding, as well. Pactor 3 is runs an occupied bandwidth of about 2.4 kHz, but raw speed is over 2700 bps. Instead of 2 tones, P3 uses up to 18, separated by 120 Hz and modulated at 100 baud DBPSK or DQPSK. SCS has some fairly detailed data on Pactor 3 at: http://www.scs-ptc.com/download/PACTOR-III-Protocol.pdf I wish someone could explain why we can not have a sound card mode that is roughly the same as Pactor 2 at least. Even if there was no ARQ at first. I don't know if the least complex of it all is ARQ...most likely, the rest is harder to implement. Yes. A key requirement is having the highest distance between constellation points to have an edge against the noise (or QRM). That's why, in DRM, the FAC uses 4QAM, as it allows to send the reduced but very important info it conveys. But the MSC must use 64QAM, because the amount of data to be sent does not allow otherwise in the least bandwidth. What happens if you use a multitone DPSK? It seems to a non-engineering person like myself, that a lot of what P2 and P3 are made up of are really a series of PSK100 or PSK200 tones (carriers). Isn't Q15X25 a similar modulation scheme? It even runs at 83.33 baud rather than a minimum of 100 baud such as P2. Why did it not work as well as P modes? Or is it because it has no coding such as Reed-Solomon block coding or Viterbi convolutional coding? Certainly...all those tricks add up, and most likely, in a non proportional way...I cannot assure it by heart, but is very likely. One of the gains of the code used in pactor modes is using convolutional encoding with Viterbi decoding. The Viterbi decoder, knowing the history of what has been sent, as the convolutionally coded stream depends on what has been sent previously, makes a soft decode of what is the most likely symbol transmitted. RS coding, after deinterleaving, on the other side, may allow to recover erors WITHOUT retransmission, which may save more bandwidth than what is wasted on the FEC overhead. Also, P2 and P3 avoid the edges of the channel to have the least amplitude and delay differences between carriers. That's why a reduced version of Q15X25 is being more succesful in holding the link. 73, Rick, KV9U 73, Jose, CO2JA __ __ __ __ Correo enviado por ElectroMAIL. Facultad Eléctrica. CUJAE Dominio: electrica.cujae.edu.cu Announce your digital presence via our DX Cluster telnet://cluster.dynalias.org Our other groups: http://groups.yahoo.com/group/dxlist/ http://groups.yahoo.com/group/themixwgroup http://groups.yahoo.com/group/contesting http://groups.yahoo.com/group/wnyar http://groups.yahoo.com/group/Omnibus97 Yahoo! Groups Links
Re: [digitalradio] PSK Modes
Why do some modems use more rectangular waveforms instead of what appears to be the optimum waveform for HF modems? Or are there downsides to raised cosine waveforms? In terms of bandwidth, it seems to me that for most uses, a 500 Hz bandwidth is a wise choice. This seems to be a good tradeoff in width vs potential throughput for keyboard modes and even some higher speed data modes like Pactor 2 can do under better conditions. Also, 500 Hz filters have been commonly available for CW. With more rigs using DSP filters, I admit that it is less of an issue to tailor make it to one particular width. How about a two tone, DPSK scheme with 50 (maybe even 25?), and also 100 and 200 baud rates? Even if you would initially require manual adjustment. Then we also should have a mode that can run in a voice channel, probably 2.4 to at most 2.7 KHz width. How about an 8 tone DBPSK and maybe switchable (manually at first) to higher PSK rates? All of these modes could have similar timing for ARQ. How about 0.5 seconds pause between transmissions and awaiting an ACK or NAK or control signal? With right control signals you could change the length of time for the packet burst. But for starters, maybe just a simple packet size. 73, Rick, KV9U Walt DuBose wrote: One solution suggested to me was that each tone be individually shaped/filtered before transmitting and then each tone have an individual brick wall filter before it is decoded. I believe that there is not going to be one mode or mode configuration that works well on 3-30 MHz...we will probably have several sets of configurations or perhaps even one optimum configuration for each band or set of conditions. The more options you have that you can adjust of the fly or that can be used adaptively the better off the mode will be. Someone (perhaps all) needs to keep technical notes on what modes work best on what band. Also, we need to come to an agreement on what mazimum bandwidth and user throughput we want as well as how robust and how sensitive we want the mode to be. My personal belief is that we go for 500 Hz wandwidth, 400-800 WPM user throughput 99.9% error free and work down below a 0 dB SNR (0 to -5 dB) on a poor CCIR channel. Put your thinking caps on and make your wish list. 73 CLU, Walt/K5YFW KV9U wrote: Some of us did try Chip modes when Nino first came out with them, but they did not seem to perform as well as existing modes. I really implore to our treasured programmers to see if they can come up with some modes that can compete with Pactor modes. Especially some ARQ modes that can work on MS OS. We know from Pactor 2, that a raised cosine shaped pulse is likely a very good basic waveform. Then for the most robust mode, a two tone DBPSK modulation is used and as the conditions improve, the modulation changes to DQPSK and then with further improvements to 8-DPSK and even 16-DPSK for maximum throughput when conditions are very good. This is what enables Pactor 2 to send about 700 bits per second at the peak speed and do it in only a 500 Hz wide span. We know this can be done at the higher speeds under good conditions with sound card modes since SCAMP was even faster than P2, although a much wider signal. The problem with SCAMP was that it had no fallback position. Pactor 3 is runs an occupied bandwidth of about 2.4 kHz, but raw speed is over 2700 bps. Instead of 2 tones, P3 uses up to 18, separated by 120 Hz and modulated at 100 baud DBPSK or DQPSK. SCS has some fairly detailed data on Pactor 3 at: http://www.scs- ptc.com/download /PACTOR-III- Protocol. pdf http://www.scs-ptc.com/download/PACTOR-III-Protocol.pdf I wish someone could explain why we can not have a sound card mode that is roughly the same as Pactor 2 at least. Even if there was no ARQ at first. And how different is Pactor 3, than what the SSTV hams are using everyday? Aren't they using OFDM with QAM? If you recall what Tom Rink said back in 1995 on the TAPR HF SIG: As mentioned in the introduction, PACTOR-II uses a two-tone DPSK modulation system. Due to the raised cosine pulse shaping, the maximum required bandwidth is only around 450 Hz at minus 50 dB. ASK, which was also tested in the early stage, provided poorer results in weak conditions compared with a higher DPSK modulation, as different amplitude levels are more difficult to distinguish in noisy channels than more phase levels. Additionally, ASK increases the Crest Factor of the signal. For these reasons, it is not used in the final PACTOR-II protocol. Basic information on these items can also be found in the first part of this series. Although not ASK, doesn't QAM employ amplitude changes as part of the modulation scheme? What happens if you use a multitone DPSK? It seems to a non-engineering person like myself, that a lot of what P2 and P3 are made up of are really a series of PSK100 or PSK200 tones (carriers). Isn't Q15X25 a similar modulation scheme? It even runs at 83.33
Re: [digitalradio] PSK Modes
KV9U wrote: If I understand it correctly, the raised cosine pulses tend to be more efficient with power, reduce the crest factor (Pactor 2 is under 1.5), and perhaps make it easier to have a cleaner signal. Raised cosine is, above all, less bandwidth greedy. Just for clarification I have a question: Is QAM modulation a form of ASK? It would seem so to me but I am not sure. Otherwise, what other modulation forms fall into the ASK category? It may be seen as that. Just depend on what abstraction you make to reach that conclusion. FSK is a form of complementary ASK of two carrierswhich is bad is pure ASK because one state is pure signal and another pure garbage (noise, etc). QAM can be seen as an ASK of four phases at a fixed amplitude. Using two quadrature modulators, you create four states keying them with (1,1) (1,-1), (-1, -1) and (-1,1). 1 is the same phase, -1, reversed phase. Combine them and you get a constellation with points every 45 degrees. off the XY axis. Although the SSTV modes are not automatically adaptive, there is a limited choice of number of tones, but for the most part I believe that they have found 16QAM to be about all you can get to work well on many HF circuits, particularly on the lower frequencies. It depends on the signal to noise ratios. There is a video presentation of Doug Smith on Georgia Tech about Digital Voice which is pretty illustrative. Some Googling should find it. It shows the constellations and the effects of noise on it. The more complex the constellation, the less distance there is between constellation points, and so, less leeway for noise before confusing the decoder. I wonder how a 4-QAM mode compares to say a 4-PSK mode when up against the ionosphere? Should be about the same... There must have been a reason that DRM uses QAM instead of PSK? Any thoughts on why? When you need a modulator for 64QAM for the MSC, it is rather easy to create 4QAM with the four extreme points of the 64QAM constellations. In terms of coding, it would be very interesting is to compare two multitone modems, perhaps a 2 tone and an 8 tone (similar to pactor 2 and 3) and have one with R-S and one with Viterbi and see if there is any difference on various circuits. Pactor II and III use both Viterbi decoding and block encoding with interleaving... That is not the test that needs to be done. The difference between P2 and P3, is that P3 stays with the most robust and capable constallation, 4DPSK, and starts deploying carriers using it. The coding tricks are about the same. What I don't know so far is how does it distribute the traffic among the carriers. Jose, CO2JA
Re: [digitalradio] PSK Modes
Walt/K5YFW wrote: if you may be receiving 1, 2 and 3 hop signals. How does this affect BPSK and QPSK signals from for example PSK31/63/125? the 3 different signals will sum at the receiver, but, having each one a different phase, the sum is destructive with the result that they tend to cancel. If the paths are stable you notice a drop in the signal strength but if paths are unstable (as it is often the case) one signal may win over the others and the phase of the PSK decoder will wander back and forth. The clock recovery is also problematic because of the unstability of the reference. Among the possible solutions to multipath there is the spread spectrum modulation (as in Chip64) where the symbols at the receiver aren't expected at a precise timing, but are decoded in a clockless manner. In Chip64 signal scope you can actually see the signal trace wandering left and rigth due to path hopping or see the ghosted trace of the secondary path. Nino/IZ8BLY
Re: [digitalradio] PSK Modes
Nino: I have not had luck with Chip...not a single QSO so far. On 40 meters local NVIS test it did not work. Maybe the 300 baud chip rate was too fast for it to work. Would it be prefarable to use it on a close to the MUF, single ray link?? I would like to try it on the air. How has been the actual experience with Chip modes? 73 de Jose, CO2JA -- Original Message -- From: Nino Porcino \(IZ8BLY\) [EMAIL PROTECTED] Reply-To: digitalradio@yahoogroups.com Date: Mon, 22 Jan 2007 18:18:07 +0100 Walt/K5YFW wrote: if you may be receiving 1, 2 and 3 hop signals. How does this affect BPSK and QPSK signals from for example PSK31/63/125? the 3 different signals will sum at the receiver, but, having each one a different phase, the sum is destructive with the result that they tend to cancel. If the paths are stable you notice a drop in the signal strength but if paths are unstable (as it is often the case) one signal may win over the others and the phase of the PSK decoder will wander back and forth. The clock recovery is also problematic because of the unstability of the reference. Among the possible solutions to multipath there is the spread spectrum modulation (as in Chip64) where the symbols at the receiver aren't expected at a precise timing, but are decoded in a clockless manner. In Chip64 signal scope you can actually see the signal trace wandering left and rigth due to path hopping or see the ghosted trace of the secondary path. Nino/IZ8BLY __ __ __ __ Correo enviado por ElectroMAIL. Facultad Eléctrica. CUJAE Dominio: electrica.cujae.edu.cu
Re: [digitalradio] PSK Modes
Some of us did try Chip modes when Nino first came out with them, but they did not seem to perform as well as existing modes. I really implore to our treasured programmers to see if they can come up with some modes that can compete with Pactor modes. Especially some ARQ modes that can work on MS OS. We know from Pactor 2, that a raised cosine shaped pulse is likely a very good basic waveform. Then for the most robust mode, a two tone DBPSK modulation is used and as the conditions improve, the modulation changes to DQPSK and then with further improvements to 8-DPSK and even 16-DPSK for maximum throughput when conditions are very good. This is what enables Pactor 2 to send about 700 bits per second at the peak speed and do it in only a 500 Hz wide span. We know this can be done at the higher speeds under good conditions with sound card modes since SCAMP was even faster than P2, although a much wider signal. The problem with SCAMP was that it had no fallback position. Pactor 3 is runs an occupied bandwidth of about 2.4 kHz, but raw speed is over 2700 bps. Instead of 2 tones, P3 uses up to 18, separated by 120 Hz and modulated at 100 baud DBPSK or DQPSK. SCS has some fairly detailed data on Pactor 3 at: http://www.scs-ptc.com/download/PACTOR-III-Protocol.pdf I wish someone could explain why we can not have a sound card mode that is roughly the same as Pactor 2 at least. Even if there was no ARQ at first. And how different is Pactor 3, than what the SSTV hams are using everyday? Aren't they using OFDM with QAM? If you recall what Tom Rink said back in 1995 on the TAPR HF SIG: As mentioned in the introduction, PACTOR-II uses a two-tone DPSK modulation system. Due to the raised cosine pulse shaping, the maximum required bandwidth is only around 450 Hz at minus 50 dB. ASK, which was also tested in the early stage, provided poorer results in weak conditions compared with a higher DPSK modulation, as different amplitude levels are more difficult to distinguish in noisy channels than more phase levels. Additionally, ASK increases the Crest Factor of the signal. For these reasons, it is not used in the final PACTOR-II protocol. Basic information on these items can also be found in the first part of this series. Although not ASK, doesn't QAM employ amplitude changes as part of the modulation scheme? What happens if you use a multitone DPSK? It seems to a non-engineering person like myself, that a lot of what P2 and P3 are made up of are really a series of PSK100 or PSK200 tones (carriers). Isn't Q15X25 a similar modulation scheme? It even runs at 83.33 baud rather than a minimum of 100 baud such as P2. Why did it not work as well as P modes? Or is it because it has no coding such as Reed-Solomon block coding or Viterbi convolutional coding? 73, Rick, KV9U Jose_Angel Amador Fundora wrote: Nino: I have not had luck with Chip...not a single QSO so far. On 40 meters local NVIS test it did not work. Maybe the 300 baud chip rate was too fast for it to work. Would it be prefarable to use it on a close to the MUF, single ray link?? I would like to try it on the air. How has been the actual experience with Chip modes? 73 de Jose, CO2JA -- Original Message -- From: Nino Porcino \(IZ8BLY\) [EMAIL PROTECTED] Reply-To: digitalradio@yahoogroups.com Date: Mon, 22 Jan 2007 18:18:07 +0100 Walt/K5YFW wrote: if you may be receiving 1, 2 and 3 hop signals. How does this affect BPSK and QPSK signals from for example PSK31/63/125? the 3 different signals will sum at the receiver, but, having each one a different phase, the sum is destructive with the result that they tend to cancel. If the paths are stable you notice a drop in the signal strength but if paths are unstable (as it is often the case) one signal may win over the others and the phase of the PSK decoder will wander back and forth. The clock recovery is also problematic because of the unstability of the reference. Among the possible solutions to multipath there is the spread spectrum modulation (as in Chip64) where the symbols at the receiver aren't expected at a precise timing, but are decoded in a clockless manner. In Chip64 signal scope you can actually see the signal trace wandering left and rigth due to path hopping or see the ghosted trace of the secondary path. Nino/IZ8BLY __ __ __ __ Correo enviado por ElectroMAIL. Facultad Eléctrica. CUJAE Dominio: electrica.cujae.edu.cu
Re: [digitalradio] PSK Modes
-- Original Message -- From: KV9U [EMAIL PROTECTED] Reply-To: digitalradio@yahoogroups.com Date: Mon, 22 Jan 2007 14:47:12 -0600 Some of us did try Chip modes when Nino first came out with them, but they did not seem to perform as well as existing modes. So we coincide...it is a pity...but using the ionosphere, you have to play by its rules. I really implore to our treasured programmers to see if they can come up with some modes that can compete with Pactor modes. Especially some ARQ modes that can work on MS OS. We know from Pactor 2, that a raised cosine shaped pulse is likely a very good basic waveform. That is for saving bandwidth, mostly. It might allow better decoding, as well. Then for the most robust mode, a two tone DBPSK modulation is used and as the conditions improve, the modulation changes to DQPSK and then with further improvements to 8-DPSK and even 16-DPSK for maximum throughput when conditions are very good. This is what enables Pactor 2 to send about 700 bits per second at the peak speed and do it in only a 500 Hz wide span. The steepest loss of performance in PSK constellations occurs from QPSK onwards as the distances between the constellation points diminish. It is well treated in Communications Systems by Carlson et al. We know this can be done at the higher speeds under good conditions with sound card modes since SCAMP was even faster than P2, although a much wider signal. The problem with SCAMP was that it had no fallback position. So, fallback is important on HF. Pactor 3 is runs an occupied bandwidth of about 2.4 kHz, but raw speed is over 2700 bps. Instead of 2 tones, P3 uses up to 18, separated by 120 Hz and modulated at 100 baud DBPSK or DQPSK. SCS has some fairly detailed data on Pactor 3 at: http://www.scs-ptc.com/download/PACTOR-III-Protocol.pdf I wish someone could explain why we can not have a sound card mode that is roughly the same as Pactor 2 at least. Even if there was no ARQ at first. I don't know if the least complex of it all is ARQ...most likely, the rest is harder to implement. And how different is Pactor 3, than what the SSTV hams are using everyday? Aren't they using OFDM with QAM? If you recall what Tom Rink said back in 1995 on the TAPR HF SIG: That it is not adaptive as pactor is. As mentioned in the introduction, PACTOR-II uses a two-tone DPSK modulation system. Due to the raised cosine pulse shaping, the maximum required bandwidth is only around 450 Hz at minus 50 dB. ASK, which was also tested in the early stage, provided poorer results in weak conditions compared with a higher DPSK modulation, as different amplitude levels are more difficult to distinguish in noisy channels than more phase levels. Additionally, ASK increases the Crest Factor of the signal. For these reasons, it is not used in the final PACTOR-II protocol. Basic information on these items can also be found in the first part of this series. Although not ASK, doesn't QAM employ amplitude changes as part of the modulation scheme? Yes. A key requirement is having the highest distance between constellation points to have an edge against the noise (or QRM). That's why, in DRM, the FAC uses 4QAM, as it allows to send the reduced but very important info it conveys. But the MSC must use 64QAM, because the amount of data to be sent does not allow otherwise in the least bandwidth. What happens if you use a multitone DPSK? It seems to a non-engineering person like myself, that a lot of what P2 and P3 are made up of are really a series of PSK100 or PSK200 tones (carriers). Isn't Q15X25 a similar modulation scheme? It even runs at 83.33 baud rather than a minimum of 100 baud such as P2. Why did it not work as well as P modes? Or is it because it has no coding such as Reed-Solomon block coding or Viterbi convolutional coding? Certainly...all those tricks add up, and most likely, in a non proportional way...I cannot assure it by heart, but is very likely. One of the gains of the code used in pactor modes is using convolutional encoding with Viterbi decoding. The Viterbi decoder, knowing the history of what has been sent, as the convolutionally coded stream depends on what has been sent previously, makes a soft decode of what is the most likely symbol transmitted. RS coding, after deinterleaving, on the other side, may allow to recover erors WITHOUT retransmission, which may save more bandwidth than what is wasted on the FEC overhead. Also, P2 and P3 avoid the edges of the channel to have the least amplitude and delay differences between carriers. That's why a reduced version of Q15X25 is being more succesful in holding the link. 73, Rick, KV9U 73, Jose, CO2JA __ __ __ __ Correo enviado por ElectroMAIL. Facultad Eléctrica. CUJAE Dominio: electrica.cujae.edu.cu