Re: [digitalradio] Re: PSK-ROBUST - Path Simulation Results vs field tests
Gary grwes...@... wrote: Maybe there is something that the path simulators are missing. It could be Gary and I think it's fair to say that propagation simulators are just not capable of capturing all the variables of the HF channel. They do, for the most part, capture the fundamental characteristics of the real thing and that has been proven to be useful when comparing the performance of different HF modems. As far as NVIS, the simulator creates a multi-path condition that tries to emulate both ground waves and sky waves arriving at the receiver; a condition that can destroy throughput with certain modes regardless of how strong the signals are. NVIS field tests would have to meet this criteria in order to prove or disprove whether an NVIS path simulation was useful in determining which modes work best for this kind of propagation. The trick is recognizing the presence of both ground waves and sky waves. I doubt if there's enough ground wave signal strength to cause any real problems on the longer NVIS paths; it's the relatively short paths that give the most trouble and of course, that depends on terrain, antennas etc. Thanks for the input Gary. I'd be interested in any NVIS recordings you have. Feel free to send them and don't worry about file size (within reason of course : ). 73 Tony -K2MO . --- In digitalradio@yahoogroups.com mailto:digitalradio%40yahoogroups.com, Gary grwes...@... wrote: Your question is one that I have also. In our recent NVIS testing with fldigi/flarq we found BPSK250 provided better throughput than other modes we tested, and most notably MFSK32 which we thought would be our safe, robust mode. This was with a variety of band conditions including strong signal, weak signal, selective fading, lightning QRN, grungy power line noise, and all the normal stuff we experience here in the Midwest. Path distances varied from 40 miles to 150 miles which were the distances we were interested in. There were a couple ideas we kicked around as possible reasons why BPSK250 worked so much better than we expected. One was that when the signal took a hit from something like a lightning burst, BPSK250 recovered and resynchronized very fast. The second was even more speculative in that maybe the higher phase modulation rate (250 times per second) was faster than Doppler path modulation allowing the BPSK decoder to ride through. Anyway, we expected BPSK250 to be useless on NVIS but every time we have tried it, it has worked. (and better than most modes.) Maybe there is something that the path simulators are missing. Dunno... Just throwing some ideas out. Gary - N0GW
Re: [digitalradio] Re: PSK-ROBUST - Path Simulation Results vs field tests
For ARQ there is an additional criterion, viz. the value to look for is the thin border line between 95% and 100% copy. 95% copy is generally not enough for efficient ARQ operation. That is why pskmail does not use S/N levels as the only parameter for switching modes, additionally we use block size adaptation and ARQ success. Using this the MFSK and THOR modes get their fair share of the cake under NVIS conditions. Rein PA0R Gary, Thank you for the feedback. Using ARQ systems like you did we have also noticed that faster PSK modes show better performance as in less repeat request than slower ones. Your first theory is the one we believe is the most likely contributor for the ARQ links in general benefiting from faster psk speed (up to a certain point of course). But like you we are a bit puzzled by the apparent robustness of these modes in NVIS conditions. If you are so inclined, could you please perform a comparison between psk250 and psk250R as available in the beta version 3.20 of Fldigi? The FEC of course reduces the speed by almost half but the error rare (ARQ repeats) appears to be lower than even the half psk speed (i.e. psk250r seems more robust than psk125) as Rein reported in his post on this subject. Another perspective would be appreciated. 73s, John --- In digitalradio@yahoogroups.com, Gary wrote: Your question is one that I have also. In our recent NVIS testing with fldigi/flarq we found BPSK250 provided better throughput than other modes we tested, and most notably MFSK32 which we thought would be our safe, robust mode. This was with a variety of band conditions including strong signal, weak signal, selective fading, lightning QRN, grungy power line noise, and all the normal stuff we experience here in the Midwest. Path distances varied from 40 miles to 150 miles which were the distances we were interested in. There were a couple ideas we kicked around as possible reasons why BPSK250 worked so much better than we expected. One was that when the signal took a hit from something like a lightning burst, BPSK250 recovered and resynchronized very fast. The second was even more speculative in that maybe the higher phase modulation rate (250 times per second) was faster than Doppler path modulation allowing the BPSK decoder to ride through. Anyway, we expected BPSK250 to be useless on NVIS but every time we have tried it, it has worked. (and better than most modes.) Maybe there is something that the path simulators are missing. Dunno... Just throwing some ideas out. Gary - N0GW http://www.obriensweb.com/digispotter.html Chat, Skeds, and spots all in one (resize to suit)Yahoo! Groups Links
[digitalradio] Re: PSK-ROBUST - Path Simulation Results vs field tests
Gary, Thank you for the feedback. Using ARQ systems like you did we have also noticed that faster PSK modes show better performance as in less repeat request than slower ones. Your first theory is the one we believe is the most likely contributor for the ARQ links in general benefiting from faster psk speed (up to a certain point of course). But like you we are a bit puzzled by the apparent robustness of these modes in NVIS conditions. If you are so inclined, could you please perform a comparison between psk250 and psk250R as available in the beta version 3.20 of Fldigi? The FEC of course reduces the speed by almost half but the error rare (ARQ repeats) appears to be lower than even the half psk speed (i.e. psk250r seems more robust than psk125) as Rein reported in his post on this subject. Another perspective would be appreciated. 73s, John --- In digitalradio@yahoogroups.com, Gary grwes...@... wrote: Your question is one that I have also. In our recent NVIS testing with fldigi/flarq we found BPSK250 provided better throughput than other modes we tested, and most notably MFSK32 which we thought would be our safe, robust mode. This was with a variety of band conditions including strong signal, weak signal, selective fading, lightning QRN, grungy power line noise, and all the normal stuff we experience here in the Midwest. Path distances varied from 40 miles to 150 miles which were the distances we were interested in. There were a couple ideas we kicked around as possible reasons why BPSK250 worked so much better than we expected. One was that when the signal took a hit from something like a lightning burst, BPSK250 recovered and resynchronized very fast. The second was even more speculative in that maybe the higher phase modulation rate (250 times per second) was faster than Doppler path modulation allowing the BPSK decoder to ride through. Anyway, we expected BPSK250 to be useless on NVIS but every time we have tried it, it has worked. (and better than most modes.) Maybe there is something that the path simulators are missing. Dunno... Just throwing some ideas out. Gary - N0GW
[digitalradio] Re: PSK-ROBUST - Path Simulation Results vs field tests
Hi Tony, Thank you for the information. This is the issue with field tests, there are always several variables. I will proceed with some other field tests, trying to eliminate some of the variables. In my case I have an inverted V on 40 and 80M at only 9 meters peak over the ground for the server. So if I use a low dipole for the client next time (2 or 4 meters high) instead of the vertical I assume I should be able to safely eliminate ground waves over that distance (95 miles). Am I correct in my understanding that there is still multipath and therefore selective fading in pure NVIS (no ground wave) conditions? Thanks again, Regards, John --- In digitalradio@yahoogroups.com, Tony d...@... wrote: John, The first thing that comes to mind is whether there were any ground wave signals mixing with sky waves during your field tests? It's been shown that NVIS throughput can fail when the sky wave echoes interact with ground waves. The sky waves take more time to arrive at the receiver so you can imagine what the difference in timing does to copy when the two signals interact. This is what the NVIS simulations were based on; two channels, one with no delay (simulated ground wave) and the other with a 7 ms delay (simulated NVIS sky wave). January's path tests showed that PSK-R appeared to be less robust than BPSK under NVIS simulation while the white noise tests clearly showed PSK-R the winner in terms of sensitivity. Your field tests seem to reveal the same results in terms of which modes have the edge in sensitivity, but not necessarily the edge in terms of dealing with multi-path timing delays. I could be wrong though and there may have been strong evidence of ground wave interaction? It can be difficult to tell; some paths are more obvious than others. Hellschreiber is the only mode I know of that can visually indicate this sort of thing, but that's not an option with PSKMail. Hope to hear from you soon John. Tony -K2MO n 4/1/2010 9:45 AM, vk2eta wrote: To Tony (K2MO) in particular, but not exclusively: Following your simulation results on these modes in January I have done a few tests in the field and I have to say that I don't understand the results. Please note that I am not trying to make a point, but to understand why the theory does not seem to match the practical side. My tests simply revolve around examining the bahaviour of the Pskmail server adapting speed to the conditions. We have in the latest version a table of modes that the server can use by shifting up and down, one mode at a time. It does so by relying on the s/n report gathered from Fldigi and the number of repeats due to damaged ARQ frames. The list is arranged in an empirical order of speed vs robustness and is the following for regions 2 and 3: THOR8 MFSK16 THOR22 MFSK32 PSK250R PSK500R PSK500 The MFSK/IFSK family of modes are normally the modes of choice for NVIS. This week I did some tests at 95 miles in a strait line from my server on 40 and 80M between about 1PM to 2PM local time so obviously in NVIS conditions. What I noticed every time I would connect in MFSK16, the server would progressively shift the TX mode up into the PSKR modes, up to PSK500R, but never to PSK500. I also noticed that there would be no fallback from PSK250R to MFSK32 after a shift up from MFSK32. So my interpretion is the following: If the PSKR modes had a weakness in NVIS conditions, I would see the server moving continuously between MFSK32 and PSK250R: good reception in MFSK32, speed up to PSK250R, poor reception, return to MFSK32, etc... Also since it did not go up pass PSK500R to PSK500 it indicates that in these particular cases the PSK500R modes was starting to show signs of limitations and the server calculated that there was not enough s/n margin to shift the speed up. Selective fading is very visible especially on the PSK500R mode of course. So my question is: in the simulation you performed, are there parameters that maybe would need to be looked at to explain why these modes seem to behave well in these conditions or are there other variables to consider? Also trying to get a more formal comparison, how would you design some practical tests that minimize the effects of variation in propagation in the field? On this point I was thinking of sending a set text in different modes and repeating the test several times, interleaving the modes so that in average it would be unlikely to be just propagation. Mode1, Mode2, Mode3, Mode4 then again Mode1, Mode2, Mode3 etc... repeated say 5 times. Then taking the average result for comparison. Best regards, John (VK2ETA) --- In digitalradio@yahoogroups.com mailto:digitalradio%40yahoogroups.com, vk2eta vk2eta@ wrote: Hi Tony, Thank you for the simulation results. I
Re: [digitalradio] Re: PSK-ROBUST - Path Simulation Results vs field tests
I have done several tests with DA5UWG on 80m, and the 200 km path still shows lots of multipath. Both antennas are low (mine is a full wave 80m loop at 10m, DA5UWG has a dipole at 12m). Sometimes the mode goes up to PSK500 for a few frames, but it always switches back to PSK500R, PSK250R or MFSK32. There is also a clear variation pattern involving the time of day. Multipath is heaviest around 17:00 local time. I have found PSK500 or PSK250 to be useable only on single hop (Eindhoven-Stockholm) or groundwave ( 50 km) paths. During such occasions they save a lot of time :) I live near an airport, and when a plane is overhead the download from PI4TUE (20 km) switches from PSK500 to PSK500R. Upload remains PSK500R beacause the high noise level at PI4TUE prevents PSK500. The robust modes are generally better than the raw modes, that is why PSK500 is the only PSK raw mode in the mode table of pskmail. This mode table was established using the trial and error method over several months and paths... 73, Rein PA0R Hi Tony, Thank you for the information. This is the issue with field tests, there are always several variables. I will proceed with some other field tests, trying to eliminate some of the variables. In my case I have an inverted V on 40 and 80M at only 9 meters peak over the ground for the server. So if I use a low dipole for the client next time (2 or 4 meters high) instead of the vertical I assume I should be able to safely eliminate ground waves over that distance (95 miles). Am I correct in my understanding that there is still multipath and therefore selective fading in pure NVIS (no ground wave) conditions? Thanks again, Regards, John --- In digitalradio@yahoogroups.com, Tony wrote: John, The first thing that comes to mind is whether there were any ground wave signals mixing with sky waves during your field tests? It's been shown that NVIS throughput can fail when the sky wave echoes interact with ground waves. The sky waves take more time to arrive at the receiver so you can imagine what the difference in timing does to copy when the two signals interact. This is what the NVIS simulations were based on; two channels, one with no delay (simulated ground wave) and the other with a 7 ms delay (simulated NVIS sky wave). January's path tests showed that PSK-R appeared to be less robust than BPSK under NVIS simulation while the white noise tests clearly showed PSK-R the winner in terms of sensitivity. Your field tests seem to reveal the same results in terms of which modes have the edge in sensitivity, but not necessarily the edge in terms of dealing with multi-path timing delays. I could be wrong though and there may have been strong evidence of ground wave interaction? It can be difficult to tell; some paths are more obvious than others. Hellschreiber is the only mode I know of that can visually indicate this sort of thing, but that's not an option with PSKMail. Hope to hear from you soon John. Tony -K2MO n 4/1/2010 9:45 AM, vk2eta wrote: To Tony (K2MO) in particular, but not exclusively: Following your simulation results on these modes in January I have done a few tests in the field and I have to say that I don't understand the results. Please note that I am not trying to make a point, but to understand why the theory does not seem to match the practical side. My tests simply revolve around examining the bahaviour of the Pskmail server adapting speed to the conditions. We have in the latest version a table of modes that the server can use by shifting up and down, one mode at a time. It does so by relying on the s/n report gathered from Fldigi and the number of repeats due to damaged ARQ frames. The list is arranged in an empirical order of speed vs robustness and is the following for regions 2 and 3: THOR8 MFSK16 THOR22 MFSK32 PSK250R PSK500R PSK500 The MFSK/IFSK family of modes are normally the modes of choice for NVIS. This week I did some tests at 95 miles in a strait line from my server on 40 and 80M between about 1PM to 2PM local time so obviously in NVIS conditions. What I noticed every time I would connect in MFSK16, the server would progressively shift the TX mode up into the PSKR modes, up to PSK500R, but never to PSK500. I also noticed that there would be no fallback from PSK250R to MFSK32 after a shift up from MFSK32. So my interpretion is the following: If the PSKR modes had a weakness in NVIS conditions, I would see the server moving continuously between MFSK32 and PSK250R: good reception in MFSK32, speed up to PSK250R, poor reception, return to MFSK32, etc... Also since it did not go up pass PSK500R to PSK500 it indicates that in these particular cases the PSK500R modes was starting to show signs of limitations and the server calculated that there was not enough s/n margin to shift the
Re: [digitalradio] Re: PSK-ROBUST - Path Simulation Results vs field tests
Rein, what is the cause of the 1700hr heaviest multipath? Is that a ionospheric condition of some peak airport traffic issue ? Andy K3Uk On Fri, Apr 2, 2010 at 7:35 AM, Rein Couperus r...@couperus.com wrote: I have done several tests with DA5UWG on 80m, and the 200 km path still shows lots of multipath. Both antennas are low (mine is a full wave 80m loop at 10m, DA5UWG has a dipole at 12m). Sometimes the mode goes up to PSK500 for a few frames, but it always switches back to PSK500R, PSK250R or MFSK32. There is also a clear variation pattern involving the time of day. Multipath is heaviest around 17:00 local time. I have found PSK500 or PSK250 to be useable only on single hop (Eindhoven-Stockholm) or groundwave ( 50 km) paths. During such occasions they save a lot of time :) I live near an airport, and when a plane is overhead the download from PI4TUE (20 km) switches from PSK500 to PSK500R. Upload remains PSK500R beacause the high noise level at PI4TUE prevents PSK500. The robust modes are generally better than the raw modes, that is why PSK500 is the only PSK raw mode in the mode table of pskmail. This mode table was established using the trial and error method over several months and paths... 73, Rein PA0R Hi Tony, Thank you for the information. This is the issue with field tests, there are always several variables. I will proceed with some other field tests, trying to eliminate some of the variables. In my case I have an inverted V on 40 and 80M at only 9 meters peak over the ground for the server. So if I use a low dipole for the client next time (2 or 4 meters high) instead of the vertical I assume I should be able to safely eliminate ground waves over that distance (95 miles). Am I correct in my understanding that there is still multipath and therefore selective fading in pure NVIS (no ground wave) conditions? Thanks again, Regards, John --- In digitalradio@yahoogroups.com digitalradio%40yahoogroups.com, Tony wrote: John, The first thing that comes to mind is whether there were any ground wave signals mixing with sky waves during your field tests? It's been shown that NVIS throughput can fail when the sky wave echoes interact with ground waves. The sky waves take more time to arrive at the receiver so you can imagine what the difference in timing does to copy when the two signals interact. This is what the NVIS simulations were based on; two channels, one with no delay (simulated ground wave) and the other with a 7 ms delay (simulated NVIS sky wave). January's path tests showed that PSK-R appeared to be less robust than BPSK under NVIS simulation while the white noise tests clearly showed PSK-R the winner in terms of sensitivity. Your field tests seem to reveal the same results in terms of which modes have the edge in sensitivity, but not necessarily the edge in terms of dealing with multi-path timing delays. I could be wrong though and there may have been strong evidence of ground wave interaction? It can be difficult to tell; some paths are more obvious than others. Hellschreiber is the only mode I know of that can visually indicate this sort of thing, but that's not an option with PSKMail. Hope to hear from you soon John. Tony -K2MO n 4/1/2010 9:45 AM, vk2eta wrote: To Tony (K2MO) in particular, but not exclusively: Following your simulation results on these modes in January I have done a few tests in the field and I have to say that I don't understand the results. Please note that I am not trying to make a point, but to understand why the theory does not seem to match the practical side. My tests simply revolve around examining the bahaviour of the Pskmail server adapting speed to the conditions. We have in the latest version a table of modes that the server can use by shifting up and down, one mode at a time. It does so by relying on the s/n report gathered from Fldigi and the number of repeats due to damaged ARQ frames. The list is arranged in an empirical order of speed vs robustness and is the following for regions 2 and 3: THOR8 MFSK16 THOR22 MFSK32 PSK250R PSK500R PSK500 The MFSK/IFSK family of modes are normally the modes of choice for NVIS. This week I did some tests at 95 miles in a strait line from my server on 40 and 80M between about 1PM to 2PM local time so obviously in NVIS conditions. What I noticed every time I would connect in MFSK16, the server would progressively shift the TX mode up into the PSKR modes, up to PSK500R, but never to PSK500. I also noticed that there would be no fallback from PSK250R to MFSK32 after a shift up from MFSK32. So my interpretion is the following: If the PSKR modes had a weakness in NVIS conditions, I would see the server moving continuously between MFSK32 and PSK250R: good reception in
[digitalradio] Re: PSK-ROBUST - Path Simulation Results vs field tests
Your question is one that I have also. In our recent NVIS testing with fldigi/flarq we found BPSK250 provided better throughput than other modes we tested, and most notably MFSK32 which we thought would be our safe, robust mode. This was with a variety of band conditions including strong signal, weak signal, selective fading, lightning QRN, grungy power line noise, and all the normal stuff we experience here in the Midwest. Path distances varied from 40 miles to 150 miles which were the distances we were interested in. There were a couple ideas we kicked around as possible reasons why BPSK250 worked so much better than we expected. One was that when the signal took a hit from something like a lightning burst, BPSK250 recovered and resynchronized very fast. The second was even more speculative in that maybe the higher phase modulation rate (250 times per second) was faster than Doppler path modulation allowing the BPSK decoder to ride through. Anyway, we expected BPSK250 to be useless on NVIS but every time we have tried it, it has worked. (and better than most modes.) Maybe there is something that the path simulators are missing. Dunno... Just throwing some ideas out. Gary - N0GW --- In digitalradio@yahoogroups.com, vk2eta vk2...@... wrote: I also noticed that there would be no fallback from PSK250R to MFSK32 after a shift up from MFSK32. .. So my question is: in the simulation you performed, are there parameters that maybe would need to be looked at to explain why these modes seem to behave well in these conditions or are there other variables to consider? Best regards, John (VK2ETA)
Re: [digitalradio] Re: PSK-ROBUST - Path Simulation Results vs field tests [1 Attachment]
On 4/2/2010 5:24 AM, vk2eta wrote: Hi Tony, Thank you for the information. My pleasure John, wish I could have helped more. Thank you for your field tests. I should be able to safely eliminate ground waves over that distance (95 miles). It would also be interesting if you could work stations closer to home where ground waves become even more intrusive; that kind of test would correlate well with the NVIS path simulations that include both ground wave and sky wave paths. Am I correct in my understanding that there is still multipath and therefore selective fading in pure NVIS (no ground wave) conditions? Yes, and that selective fading is detectable in the waterfall at times. It's probably more apparent near sunrise / sunset as the structure of the ionosphere changes. Tony -K2MO John --- In digitalradio@yahoogroups.com mailto:digitalradio%40yahoogroups.com, Tony d...@... wrote: John, The first thing that comes to mind is whether there were any ground wave signals mixing with sky waves during your field tests? It's been shown that NVIS throughput can fail when the sky wave echoes interact with ground waves. The sky waves take more time to arrive at the receiver so you can imagine what the difference in timing does to copy when the two signals interact. This is what the NVIS simulations were based on; two channels, one with no delay (simulated ground wave) and the other with a 7 ms delay (simulated NVIS sky wave). January's path tests showed that PSK-R appeared to be less robust than BPSK under NVIS simulation while the white noise tests clearly showed PSK-R the winner in terms of sensitivity. Your field tests seem to reveal the same results in terms of which modes have the edge in sensitivity, but not necessarily the edge in terms of dealing with multi-path timing delays. I could be wrong though and there may have been strong evidence of ground wave interaction? It can be difficult to tell; some paths are more obvious than others. Hellschreiber is the only mode I know of that can visually indicate this sort of thing, but that's not an option with PSKMail. Hope to hear from you soon John. Tony -K2MO n 4/1/2010 9:45 AM, vk2eta wrote: To Tony (K2MO) in particular, but not exclusively: Following your simulation results on these modes in January I have done a few tests in the field and I have to say that I don't understand the results. Please note that I am not trying to make a point, but to understand why the theory does not seem to match the practical side. My tests simply revolve around examining the bahaviour of the Pskmail server adapting speed to the conditions. We have in the latest version a table of modes that the server can use by shifting up and down, one mode at a time. It does so by relying on the s/n report gathered from Fldigi and the number of repeats due to damaged ARQ frames. The list is arranged in an empirical order of speed vs robustness and is the following for regions 2 and 3: THOR8 MFSK16 THOR22 MFSK32 PSK250R PSK500R PSK500 The MFSK/IFSK family of modes are normally the modes of choice for NVIS. This week I did some tests at 95 miles in a strait line from my server on 40 and 80M between about 1PM to 2PM local time so obviously in NVIS conditions. What I noticed every time I would connect in MFSK16, the server would progressively shift the TX mode up into the PSKR modes, up to PSK500R, but never to PSK500. I also noticed that there would be no fallback from PSK250R to MFSK32 after a shift up from MFSK32. So my interpretion is the following: If the PSKR modes had a weakness in NVIS conditions, I would see the server moving continuously between MFSK32 and PSK250R: good reception in MFSK32, speed up to PSK250R, poor reception, return to MFSK32, etc... Also since it did not go up pass PSK500R to PSK500 it indicates that in these particular cases the PSK500R modes was starting to show signs of limitations and the server calculated that there was not enough s/n margin to shift the speed up. Selective fading is very visible especially on the PSK500R mode of course. So my question is: in the simulation you performed, are there parameters that maybe would need to be looked at to explain why these modes seem to behave well in these conditions or are there other variables to consider? Also trying to get a more formal comparison, how would you design some practical tests that minimize the effects of variation in propagation in the field? On this point I was thinking of sending a set text in different modes and repeating the test several times, interleaving the modes so that in average it would be unlikely to be just propagation. Mode1, Mode2, Mode3, Mode4 then again Mode1, Mode2, Mode3 etc... repeated say 5 times. Then taking the average result for comparison. Best regards,
[digitalradio] Re: PSK-ROBUST - Path Simulation Results vs field tests
Interesting observations John. I can't offer any useful insight, but look forward to hearing from those that can. Andy K3UK --- In digitalradio@yahoogroups.com, vk2eta vk2...@... wrote: To Tony (K2MO) in particular, but not exclusively: Following your simulation results on these modes in January I have done a few tests in the field and I have to say that I don't understand the results. Please note that I am not trying to make a point, but to understand why the theory does not seem to match the practical side. My tests simply revolve around examining the bahaviour of the Pskmail server adapting speed to the conditions. We have in the latest version a table of modes that the server can use by shifting up and down, one mode at a time. It does so by relying on the s/n report gathered from Fldigi and the number of repeats due to damaged ARQ frames. The list is arranged in an empirical order of speed vs robustness and is the following for regions 2 and 3: THOR8 MFSK16 THOR22 MFSK32 PSK250R PSK500R PSK500 The MFSK/IFSK family of modes are normally the modes of choice for NVIS. This week I did some tests at 95 miles in a strait line from my server on 40 and 80M between about 1PM to 2PM local time so obviously in NVIS conditions. What I noticed every time I would connect in MFSK16, the server would progressively shift the TX mode up into the PSKR modes, up to PSK500R, but never to PSK500. I also noticed that there would be no fallback from PSK250R to MFSK32 after a shift up from MFSK32. So my interpretion is the following: If the PSKR modes had a weakness in NVIS conditions, I would see the server moving continuously between MFSK32 and PSK250R: good reception in MFSK32, speed up to PSK250R, poor reception, return to MFSK32, etc... Also since it did not go up pass PSK500R to PSK500 it indicates that in these particular cases the PSK500R modes was starting to show signs of limitations and the server calculated that there was not enough s/n margin to shift the speed up. Selective fading is very visible especially on the PSK500R mode of course. So my question is: in the simulation you performed, are there parameters that maybe would need to be looked at to explain why these modes seem to behave well in these conditions or are there other variables to consider? Also trying to get a more formal comparison, how would you design some practical tests that minimize the effects of variation in propagation in the field? On this point I was thinking of sending a set text in different modes and repeating the test several times, interleaving the modes so that in average it would be unlikely to be just propagation. Mode1, Mode2, Mode3, Mode4 then again Mode1, Mode2, Mode3 etc... repeated say 5 times. Then taking the average result for comparison. Best regards, John (VK2ETA) --- In digitalradio@yahoogroups.com, vk2eta vk2eta@ wrote: Hi Tony, Thank you for the simulation results. I will report any field results for PSKR modes in NVIS conditions. Regards, John