Archive
 
 
  Topic: SSTV working very poorly (8 replies)
#1     Sun Dec 03, 2006 5:25 am
K7EK
Spanaway, WA
 
Join Date: Nov 2006
Posts: 15
Subject: SSTV working very poorly

Not sure exactly why, however I have observed that SSTV pictures
are not received very well when sent over CQ100. Aside from the normal
operator headspace and timing issues such as failure to correct their
slant and operating SSTV with an open microphone, it appears to me that
there are a number of things at work. Some of it is the internet with
packet loss. Other things appear to be at work too. Not sure if it is the
codec that CQ100 uses or what. In any event, SSTV pictures are mostly
fragmented and smeared. It looks to me like SCOTTY DX,
albeit very slow, is the best mode so far. I have personally had the best luck with
that mode; although it was by no means perfect, SCOTTY DX provided
the cleanest pictures for me.

I am hoping that the CQ100 folks take a good look at the SSTV problem
and address it in the near future. Not sure if it is the codec, internet timing
and latency issues, problems in the QsoNet servers, or just what. Judging
from the overwhelming popularity of SSTV and the various digital modes,
I hope that a solution is soon at hand. Note also that there are occasionally
some problems with PSK31 and Baudot modes, however nowhere as
bad as on SSTV.

I love CQ100 and hope that we can eventually operate all the different
modes, especially SSTV, without any problems. Thanks and a tip
of the hat to the QsoNet folks for their great innovation. Keep up the
great work!

Best regards,

Gary, K7EK
__________________
#2     Sun Dec 03, 2006 6:20 am
K2YAW
 
Join Date: Nov 2006
Posts: 5
Subject:

I'm working with: http://mmhamsoft.amateur-radio.ca/ MMSSTV and I am having good luck so far. It took alot of fine tuning/tweaking. Also for the rest of us Gary, it might help us if we know your OS? and connection type? I've only had CQ100 a few days, & only have been working with SSTV in the past 24 hrs. .... maybe will hear from some others on your post also... I would bet an internet prob/and or packet loss.. maybe? And how long has SSTV been a prob. up north?

regards, just another user... Ed, K2YAW
____________________________________________________
using WinXP home sp2 on a dell notebook & highspeed cable
__________________
#3     Sun Dec 03, 2006 12:44 pm
W7RJR
Spokane, WA
 
Join Date: Oct 2006
Posts: 114
Subject: SSTV problems

I wonder how SSTV works on other VOIP services such as Echolink, Skype, etc?? I wonder how it might work when peer to peer without a server involved? Anyone have a handle on P2P program we can use, such as Peerio?

Anyone wish to try? w7rjr@arrl.net
__________________
#4     Sun Dec 03, 2006 4:42 pm
K7EK
Spanaway, WA
 
Join Date: Nov 2006
Posts: 15
Subject: re: SSTV working very poorly

I am using Windows XP Professional on a 1.3 ghz computer with 512 mb
of memory. The internet connection is Comcast High Speed Broadband.

I am using many of the most popular SSTV programs including MMSSTV,
Chrompix, MixW, MSCAN, etc. Results are pretty poor on all of them. The
best pictures have been exchanged on SCOTTY DX. I have done extensive
tweaking of parameters, especially on MMSSTV without much improvement.
The pictures are usually smeared and/or fragmented, especially on those
with lots of bright colors.

W7RJR had a good point. I would like to do some testing on other internet
mediums such as Echolink, eQSO, and maybe a peer to peer direct
connect. I would be very interested to see how the pictures look over those
systems. If I get something going I will report it to the forum.

Thanks to all for their input. I would like to hear from the architects of
of CQ100 to get their views.

Best regards,

Gary, K7EK
__________________
#5     Sun Dec 03, 2006 6:16 pm
K2YAW
 
Join Date: Nov 2006
Posts: 5
Subject:

I would like to try that SCOTTY DX... & get a post back on the results.
Myself being a tech-no code. I have a question as to if we can send SSTV on any of the 5 bands on this program?.... I've only used 20 meters so far!
And the post above mine W7RJR - Bob, has a wealth of knowledge on SSTV. He got me started on CQ100 with SSTV. (thank you Bob)
and he does post an interesting question!

However most of the qso's I'm hearing today, consist of poor sstv?
so where the problem lies covers a big area, packet loss? server trouble?
I'll stop here & see how sstv works on Echolink, or peer to peer...
73's de... Ed
__________________
#6     Mon Dec 04, 2006 11:17 pm
K9FB
Kokomo, Indiana
 
Join Date: Oct 2006
Posts: 12
Subject: SSTV and the CQ100

As was posted by another commentor on another thread, many of the problems encountered using SSTV on the CQ100 seem to be limitations of the voice codec and timing issues caused by variablities of packet propagation over the Internet. One primary goal of a voice codec is to compress the data as much as possible to conserve bandwidth while maintaining maximum voice quality. The process of data compression literally throws away data not required to reconstruct a reasonable copy of the voice at the other end. Codec designers carefully tailor this process for reconstructing the human voice, not AFSK signals. Therefore, different parts of the audio frequency spectrum are likely to compressed quite differently. Linearity isn't nearly as important in the higher audio frequencies as in the lower ones to reproduce recognizable speech. If the non-voice modes work at all it's a real bonus, but there's no guarantee that they will work well.

I notice that in SSTV, solid colors come across as very pure and noise free, but at high levels of picture detail, noise and smearing are often seen. I theorize this is because solid tones compress well, but wildly varying tones do not. Furthermore, internet congestion can delay the arrival of audio packets and cause sync discontinuities. On HF, one or two missed sync pulses can be tolerated because the next sync pulse will arrive at the next proper place in time. That is not necessarily true over VoIP.

While it's a lot of fun to experiment with the non-voice modes, voice compression codecs were never designed to do this sort of thing well. The CQ100 was specifically designed to communicate via voice and CW, and I suspect a much higher quality audio compression codec and a more sophisticated audio buffering system would be required to achieve the results we would all like to see in the complex synchronous analog and digital data modes.

73,
K9FB
__________________
#7     Tue Dec 05, 2006 4:14 am
K7EK
Spanaway, WA
 
Join Date: Nov 2006
Posts: 15
Subject: re: SSTV working very poorly

Well, since my initial posting I have had time to send some pictures with my friend Bob, N1UAN, on Echolink. The results were very similar to that which was experienced on CQ100, although I must say that the quality was generally better on Echolink. I sent pics in SCOTTY 1, SCOTTY 2, SCOTTY DX, MARTIN 1, MARTIN 2, MP115, ML280, and ML320, P3, P5, and P7. The picture quality was best in SCOTTY DX and MP115. As a matter of fact, those two modes provided pictures without the characteristic smearing of bright colors. Reproduction of callsigns and other bright text was greatly improved on Echolink using SCOTTY DX and MP115. There was still a small amount of the speckling around the brighter text and other objects but the general quality of the pictures received on SCOTTY DX and MP115 was much improved over that which is generally experienced on CQ100. I have not yet had a chance to try any other VOIP systems, however from the tests I have just completed on Echolink, I am beginning to believe that there will be no way to get "closed circuit" pictures over the internet using conventional amateur radio SSTV software. There are just too many variables which wreak havoc on the timing and recovery of conventional SSTV encoding/decoding to be effective. What I believe is that we need is a new SSTV mode, written specifically for the internet, to address the various problems daily encountered there. Alternatively, some rewrites of conventional SSTV software to take internet issues into consideration may also be a possibility.

I would suggest that anyone experimenting with analog SSTV on CQ100 try SCOTTY DX and MP115. SCOTTY DX takes a long time to send pics but the resulting quality is well worth the investment of time. The best pictures have been passed with SCOTTY DX. MP115 transmits pictures that are only slightly poorer quality than SCOTTY DX, but in about 1/3 of the time. Please experiment with those two modes. You will be surprised how much better the results will be than on the others. I was very surprised to find that the so called "high definition" SSTV modes produced pictures worse than SCOTTY 1. At first I thought that the "P" modes would have produced superior pictures. I was wrong.

I have also attempted to use DIGTRX, HAMPAL, SSTVPAL, WINDRM, etc, to pass "digital SSTV" pictures to N1UAN. In every case we failed. There's just too many problems with doing it on the internet over CQ100. It is hard enough to pass good digital SSTV pics on HF under good conditions. Throw in the various internet issues and expecting to pass digital SSTV pictures effectively may be asking too much. We could never get sync'd in either direction, no matter what timing parameters/mode we used, so no data transfer took place. Perhaps if we could actually get sync'd up, some successful transfers could take place. I would like to continue playing with DIGTRX and other "digital SSTV" programs if anyone is game. I encourage you to download DIGTRX or one of the digital SSTV variants and try it on CQ100. With enough of us pursuing it, perhaps we can find answers. If you are willing to give it a try, please let me know. I would like to work with you.

My opinion of "digital SSTV" is that this is nothing more than a file transfer protocol, not in real-time as done in conventional analog SSTV, which paints a picture on the screen, line by line, not unlike FAX. DIGTRX and other so called "digital SSTV" programs send the picture first, then print it on the screen after the fact. One can literally send any file using "digital SSTV", not just pictures. There is the possibility of requesting bad/missing blocks of the picture with a "BSR" routine. This is invoked by a simple button click. This type of approach to sending pictures via CQ100 may be the way to go instead of trying to paint the picture line by line in realtime. I believe that if we could ever get sync'd up, perfect pictures should be possible with DIGTRX. Once again, perhaps some modified timing parameters or modes could be devised, which would be optimized for use on the internet. Any digital SSTV programmers reading this? I believe that DIGTRX or something similar may be the only way to send perfect pictures on CQ100 if we could manage to get two stations synchronized, thus allowing a transfer to commence.

I believe that with changes to the CQ100 codec and maybe some tweaks on the supporting servers we could possibly improve the quality of SSTV pics to some extent. The different codec used on Echolink kind of hints at that as the pictures received there were markedly improved over that of CQ100, albeit not perfect by any means.. Meanwhile if there are any programmers out there, perhaps a new "internet SSTV mode" might be something to consider, whether it be something we could add to MMSSTV or DIGTRX. I am hoping that this new virtual frontier will inspire someone to step forward and begin working on solutions so that we may overcome the problems that have been experienced. It would be great to get perfect or near-perfect picture quality over CQ100. Thanks for reading this. I would like to hear from others on this topic.

Best regards,

Gary, K7EK
__________________
#8     Tue Dec 05, 2006 1:02 pm
W7RJR
Spokane, WA
 
Join Date: Oct 2006
Posts: 114
Subject: Re: re: SSTV working very poorly


K7EK wrote:
I believe that with changes to the CQ100 codec and maybe some tweaks on the supporting servers we could possibly improve the quality of SSTV pics to some extent. The different codec used on Echolink kind of hints at that as the pictures received there were markedly improved over that of CQ100, albeit not perfect by any means.. Meanwhile if there are any programmers out there, perhaps a new "internet SSTV mode" might be something to consider, whether it be something we could add to MMSSTV or DIGTRX. I am hoping that this new virtual frontier will inspire someone to step forward and begin working on solutions so that we may overcome the problems that have been experienced. It would be great to get perfect or near-perfect picture quality over CQ100. Thanks for reading this. I would like to hear from others on this topic.

Best regards,

Gary, K7EK


Thanks Gary for your interesting observations and experimentation with SSTV over VOIP. I know a lot of folks have been tweaking settings in MMSSTV and trying different modes with limited success. I believe no amount of tweaking will derive a good quality image using audio codecs, at least for now.

Observing the waveform in MMSSTV I believe to be a valuable resource in understanding what goes wrong. Several people have sent tones at the key audio frequencies, namely 1200, 1500, 1900 and 2300 hertz.
These tones all come back relatively on target (or at least close enough) which indicates there is not a bandwidth issue. However, observing the overall waveform as transmitted and comparing it to what is received back you can see that there is a substantial degradation in the analog to digital conversion. The square wave looking data segment we send out comes back bell curved and spread out. The very distinctive timing marker at 1200 hertz is almost non-existent.

When compression occurs in a codec, redundant data is discarded. Also, there is loss in the conversion process not heard by the human ear but well discerned when viewed on a scope. I am not suggesting that everyone give up on this, but to understand that different methods will likely have to evolve before we can achieve better results.

73
__________________
#9     Wed Dec 13, 2006 4:42 am
JA1SCW
Atsugi, Japan
 
Join Date: Nov 2006
Posts: 10
Subject: SSTV working very poorly

Hello everyone,
With my little experiment for PSK31 and SSTV, I realized that there are some fundermental issues to be settled if we enjoy SSTV over CQ100.

Poor S/N at higher freq.: This can be easily reconfirmed by using PSK31 signal varying freq. lower to higher. You will get S/N of 40 dB at 500Hz although only 5 dB at 2700Hz. (Note: noise is not floor noise but noise nearby PSK31 idling signal)

What does this mean? what we are doing with MMSSTV, regardless signal format such as S1 M1 and so on, is just sending signal into very noisy environment as a result. If you are able to monitor signals in spectrum you will obviously monitor signals sinking into noise.

Respecting the current codec including signal compander adopted to the CQ100, what can we do under given conditions? I thought the answer should be moving frequency down to somewhere appropriate by doing best compromise. Firstly I did was to set S1 at 900Hz. It exhibited(reduced) very well about artifacts around picture edges which would be genarated by S1 at 1900Hz. In other words, changing freq. of S1 to 900Hz improved picture quality recognizably.

Another big issue is picture holizontal shift by packet loss. I however have no solution nor idea for this but just do crossing fingers. Well, to aboid this problem, we may need to introduce non-realtime protocol which allows resending data upon request from receiving side. I don't think this is acceptable for SSTV and/or ham radio application.

In the past, until about 18 month ago, I made frequent QSO with author of MMSSTV. However, I have no contact with him these days. So, it might not be ralistic solution to modify the current MMSSTV software so that we can define freq freely. Of course I will ask him if I see him.

Thank you for reading my not good English till the end. 73! de JA1SCW
Pictures are at http://sky.ap.teacup.com/ja1scw/ . have a look!
__________________




Copyright ©2013 Cormac Technologies Inc.