ARRL FILES REGULATION-BY-BANDWIDTH PETITION

AB0WR:"Do you have any documentation available that speaks to capture effect being seen with Pactor and/or mt63? I know from empirical results gleaned from listening to lots of Pactor sessions that it isn't very often that two Pactor sessions can be run on the same frequency without them interfering with each other unless the S1 and S2 signal strengths are vastly different. Especially with Pactor I, hijacking of a session by the ARQ packet from another session is common. I see it happen all the time."

Tim,

I don't have any documentation, but I did take two MT63 baseband signals one with "foxes" text and another with "all good men" text and tested them using MixW as my decoder. When both signals were equal, the text went back and forth between the "foxes" and "all good men" text. Giving the "foxes" signal a 1 dB advantage made it about 90% accurate. Giving the "foxes" signal a 2 dB advantage made it 100% accurate. It appears that MT63 has a 2dB capture in a non-faded environment. This is reasonable because many times I have seen 2 or more MT63 signals on the same frequency, and have been able to correctly decode the strongest signal. It would appear that two MT63 signals can occupy the same frequency, and it is possible for stations to properly decode the strongest signal. I am not sure about PIII as I don't have a way to decode it, but I don't see why there wouldn't be similar results.

In the case where there is a MT63 and a SSB signal, MARS operators have reported being able to decode the both signals simultaneously.

AB0WR:"The bottom line is that if it is "uncomfortable" to share then why recommend it?"

As I said it may be a little uncomfortable, but the probability of experiencing this discomfort is quite small, given the fact that today analog and digital signals are authorized to occupy the same spectrum in the Phone/Image subbands. Allowing data content does not change the modulation. Except for stations that are automatically controlled, I have not seen any evidence that there would be a significant increase in digital traffic in the Phone/Image subbands simply by allowing data content.

AB0WR:" don't want to sound like a smart-a** but if your decoder will only lock on the stronger signal then how can multiple sessions share a channel? If the stronger signal "captures" the receivers, the modes still run a lower spectrum efficiency than SSB."

I am not arguing your efficiency claim, I am simply arguing that it is possible for digital/digital signals to share the same frequency, just like it is possible for analog/analog signals to share the same frequency. MARS operators have shown that it is possible for digital/analog signals to share the same frequency. Frequency reuse is not the exclusive domain of SSB signals.

73,

Mark N5RFX

 

You might get a digital/digital scenario with mt63. I am pretty sure you won't with Pactor. I'll have to do some research on this to see if I can find anything.

Were the MARS tests documented or were these just ad hoc tests? It would be interesting to know the transfer rate that could be maintained by mt63 in the presence of a strong SSB signal. Merely making and sustaining a session wouldn't be sufficient if acceptable data transfer rates couldn't also be maintained.

tim ab0wr

 

Tim,

I think the tests could be considered ad-hoc. There is mention of simultanious MT63 and voice at

http://www.navymars.org/central/reg4/al/MT63.htm

and at

http://www.navymars.org/national/sound%20cards/MT63-ASU.TXT

but no formal documented tests for this specific operaton. What would you like to see in such a test?

73,

Mark N5RFX

 

The primary thing to look at would be throughput rate and error rate.

mt63 is a FEC protocol, right? No ARQ?

While these documents may say that you can get 100% solid copy, I would like to know the signal strength ratios between the voice and the mt63 signal.

As far as two different digital signals on the same frequency, were the test you ran using data streams that were being sent at the same time or were they inter-leaved? While you may see a capture effect while both are transmitting, what happens when the weaker signal is transmitting while the stronger one is not? Will that cause errors in the received data stream? A system using collision avoidance gets away from this by different timings and by packet identification in the headers. The capture effect is not just signal capture by data reception synchronization.

I'm not sure how mt63 and pactor will handle this.

tim ab0wr

 

W6NJ -- Hey Robert !

Despite your being from Hagersville, (!!) it was nice to see you and the wife enjoy the occasional slow speed CW QSO.

We were out on Thomas Point Lighthouse as K3L this summer, using CW and AM as part of a vintage special event station.

I had a dickens of a time trying to find anyone willing to answer my hand key CQ. I wonder if code buffs are so self-conscious about the fate of their mode that they now are too lofty to work QRS.

It was the same story when I was a Novice. "PSE QRS!" one of my favorite lines. That, and I swear I kept copying "so lid copy" ...

Paul/VJB
Annapolis

 

Tim,

Although you are correct that Pactor-II/III drops to an average throughput of zero by the time the S/N drops to -18dB, the throughput climbs to an average of 150 "baud" at the point where the audio can just be heard by the human ear. This is the point where non memory-ARC modes such as PSK-31, MT63, and RTTY can just begin operations. At this point, the average Pactor-II/III throughput of 150 is (CRC) error free and already ~ 5 x faster than PSK-31.

In the past year with the declining sunspot activity, about 60% of my Pactor-II links (using a 500Hz IF passband) fall below the audible level. I find it amazing that text up to 150 "baud" can even appear but that's the end result of SCS utilizing a memory-ARC circuit.

In referral to previous posts, does the fact that data can be transmitted unheard below the S/N ratio mean the channel is in use? Good question? How does it impact spectrum efficiency?

Assume for instance that an intelligent alien world with a transceiver tuned to a MT63 digital channel transmits a signal one million years ago from a distance of one million light years. Although the signal is just reaching earth, the alien civilization is no longer in existence because their signal was transmitted so long ago. Their signal level is far below the audible level of a human ear with a typical amateur transceiver but despite this they are occupying the MT63 HF channel. In your scenario, would this channel be occupied? What if you could hear the signal? If you could, would the FCC have jurisdiction?

Ed

 

Tim,

Yes MT63 is FEC no ARQ.

My original tests were not interleaved both signals were transmitting at the same time.

I created a new test. The weaker signal was on constant, and the stronger signal was on for a period of time, then went off, then after 10 seconds or so came back on. What happened was that during the first part of the test when the stronger and weaker signals were transmitting, I copied the stronger signal. When the stronger signal went off, I started copying the weaker signal. When the stronger signal came back on, I copied it with no errors.

I found an analog phone QSO going on and recorded it for 2 minutes. I then made an MT63 transmission and sent foxes for 2 minutes. I first set the average power of both signals the same and had no problem copying both signals. I then set the MT63 signal 10 dB lower than the analog phone QSO, and could not copy the MT63 very well. I then set the MT63 transmission to 6 dB below the average power of the phone transmission and could copy the MT63 signal with no errors. I have uploaded that recording to

http://home.comcast.net/~mdmiller7/mt63_2m_wphone_6dB.wav

You can copy the analog phone QSO with no problem and can hear the MT63 in the background. Play this into your favorite MT63 program and you should see the "foxes" text. So it appears that MT63 and analog phone can exist in the worst case scenario when both are on the same frequency.

73,

Mark N5RFX

 

Mark,

Impressive.

Your test with interleaved transmissions would lead me to believe that mt63 is going to have a hard time co-existing with mt63 on the same frequency. If the subject matter was different enough you could perhaps distinguish between two different qso's . Otherwise things would get very confusing.

Your demonstration on the SSB and the mt63 is interesting. Did you do this test on-the-air or by combining the baseband signals in an audio mixer?

It's probably a good indicator even if done at baseband but some RF testing, especially under noisy conditions would be interesting to document.

I am concerned that you could set the average power of both the same and you could still copy the SSB signal. Since much of the intelligibilty of speech is in the f1 and f2 formants and these can be anywhere from 5db to 30db below the f0 peaks depending on the vowel or consonant, I would have thought a mt63 signal with the same average power as the SSB signal would have significantly reduced the intelligibility of the SSB signal.

If this testing holds up it will be nice to see that the research indicating that a 6db difference between band limited noise and a SSB signal will result in a fairly high articulation index.

Exactly how did you determine the average power of each signal?

You realize you have the beginnings of a very good artlcle for one of the amateur magazines, right?

It is this kind of research that will advance amateur radio.

As I said in another message, your results show that the spectrum efficiency for mt63, at least, may not be a 1 when compared to SSB. It may be a two. That isn't quite as good as SSB by itself which would be a three but, at least for one digital mode, we are on the way to some concrete results we can base judgements on.

If we are to take this to the next level we'll have to set up some tests using designated "voices" (i.e. specified test people) and word lists in order to actually document articulation indexes for the SSB.

Now, the *big* question. Is mt63 kind of by itself on this? What other modes are this compatible? Are there any? I would expect SSB to cause significant interference to some modes if there is a 6db difference in average levels.

tim ab0wr

tim ab0wr

 

I'm not sure where you are coming from with the alien angle. I think I will leave that one for someone else to answer.

Part of the problem with mixing digital and analog is that SSB enjoys a level 3 (my designation) spectrum efficiency with typical amateur transmitters running at the 100watt level. If you put three SSB qso's on the same frequency with enough path loss that each qso sees a 6db advantage over the other two, all three qso's can progress adequately - perhaps not perfectly, but adequately.

Now, if Pactor, or any other digital mode, can work through a SSB signal with the average power of the digital signal 6-12db lower than the average power of the SSB, then they could probably co-exist. Can Pactor do this? Preliminary indications are that mt63 can although more rigorous testing needs to be done. Pactor I have my doubts about from listening to it on the air but I'm willing to keep an open mind.

The kicker here, and I haven't yet addressed this with n5rfx, is how would this difference in average power maintained today? The answer is that it wouldn't. Two transmitters sitting side by side, one running SSB and the other Pactor, would need to have the SSB transmitter running 100PEP output while the Pactor transmitter would only be running 3watts to 4 watts PEP output in order to maintain the 6db difference in average output power. I sincerely doubt if many Pactor users crank their transmitters back to 3-4 watts PEP on a consistent basis.

This puts digital users in a quandry. If I only run 3-4 watts PEP then vagaries of the RF spectrum such as selective fading, noise, path fades, etc will have a big impact on the throughput rates. Lower throughput rates means longer denial times for the frequency for other digital users. This is not good. But if the PEP level is raised then it doesn't play good with other users on the frequency.

It is for this reason that I believe Region 1 has recommended that digimodes and analog modes not be mixed in the same spectrum. Until new digimodes are developed that can perhaps analyze a frequency and set power output levels based on current occupancy and can also share a frequency with other digital users (ala Packet), segregated operating areas should be maintained for both.

tim ab0wr

 

Tim,

I have time for one final pass before heading off for a remote VHF satellite DX Expedition. The last sentence reiterated above states you are listening to Pactor-I. The fact you don't listen to Pactor-II or Pactor-III suggests you do not have a SCS modem. You specifically state you monitor Pactor-I which is a FSK mode and not a PSK mode like Pactor-II or Pactor-III.

I will add Pactor-II/III can coexist with SSB and even FSB based Pactor-I developed by SCS.

Before I QRT for the season, be aware that there are two versions of Pactor-I. The one originally developed by SCS and the one reversed engineered by others without memory-ARQ prior to general release to amateurs. The one with memory-ARQ can run 6dB lower in S/N ratio than the one without! Which one are you currently using?

Ed

 

Tim,

Thanks for the comments. When the sunspot cycle was at its peak, we would have multiple MT63 signals in the same spectrum on 14.1095 and 28.130. That is why I was pretty sure that mixing two MT63 signals at different power levels would not be a problem. My tests were done at baseband level in a non-faded environment. You are correct that a faded environment will cause some differences.

The SSB analog phone and MT63 tests were not done on the air, but by digitally mixing the basebnad signals. I was not sure if I would have room on my web server to put the wav file which has MT63 at the same average power as SSB, but it appears that I do. The file is at:

http://home.comcast.net/~mdmiller7/mt63_2m_wphone.wav

I used a program called Cool Edit Pro 2.1. This program is now Adobe Audition. In any case you can make a recording then analyze the file. My original use of this program was to determine peak to average ratios of digital baseband signals. When I started, I used some simple waveforms to see what the statistics were telling me. I used a sine wave at first and found that the peak to average ratio was 0dB, then I used a two tone signal and found the peak to average ratio was 3dB, which is correct. The statistics in the program tell you what the average RMS power is of the waveform over any part of the waveform. This is what I used to determine the average power of the MT63 and analog SSB phone signals. I was not interested in the absolute average RMS value, but in the difference between the values of MT63 and SSB analog phone. The program can then mix any number of signals together. During the mixing process you can amplify or attenuate any of the signals. In the MT63 SSB tests I simply looked at the average power of each signal, then calculated how much to amplify or attenuate each signal to raise or lower the average power. To generate the MT63 signal originally I used the program MixW. With MixW I can generate a digital signal, add text to be transmitted, then save the signal as a .wav file. I can then bring this .wav file into Cool Edit Pro and calculate the statistics, mix it with other signals and then replay it so that MixW can decode MT63.

An article would be interesting, but to do this over the air I would have to move to another country that allows me to mix digital signals with data content and analog signals. I do visit Canada from time to time, but I am usually on a business trip, and would not be able to drag all of the equipment with me that I need to run the tests.

Your question "Is mt63 kind of by itself on this" is interesting. As far as the capture ratios, I did run some other tests using RTTY and Olivia. In my tests RTTY had a 6dB capture ratio, and Olivia had a 2 dB capture ratio. I did not mix these two types of signals with SSB analog phone yet. I do have a sample of PIII and could mix this with a SSB analog phone signal, but I would not be able to determine if I can decode PIII. I have put a short sample of phone mixed with PIII at

http://home.comcast.net/~mdmiller7/phone_wPIII.wav

In this file the analog phone has a 1.6 dB advantage in average power over the PIII.

I agree that SSB could cause significant interference to some modes if there is a 6db difference in average levels. I think the probability of this risk is slim. Here is my reasoning.

The purpose of mixing modes is not to expand the spectrum occupancy of any particular mode, but to enhance the operating capability of that mode. My examples in the past have been to allow the current digital voice and digital image users the ability to send data if needed. Since the modulation is the same for digital voice, image and data this should not be a problem. In fact it is going on today even though it is not authorized, and is not causing any problems. Why not bring the regulations up to date with current operating practice? The fear is that automatically controlled stations would migrate into the Phone/Image bands. I agree that this could be a problem, given the fact that this type of operation has the potential to cause interference because there is not a control operator at each end of the link to determine if the frequency is clear. This is why I support the continued segregation of automatically controlled stations to the automatic control subbands. Segregating in this way would keep PIII out of the Phone/Image subbands because it is not likely that PIII will ever become a popular keyboard to keyboard protocol for various marketing reasons. We could still find PI and PII emissions in the Phone/Image subbands, but they have much narrower bandwidths, and I don't think they would find a good home here. That may need more discussion.

As as far as phone emissions moving into the RTTY/Data subbands, this could be easily remedied by limiting the authorized necessary bandwidth in this area. We do have experience in mixing all modes in the RTTY/Data subband of 40 meters because of the different regulations of our neighbors. We constantly mix analog and digital in this area, and with exception of the broadcast stations, I don't find that there is an enormous problem here. Both analog and digital signals seem to survive on 40 meters.

This has been an interesting discussion Tim, and thanks for the test suggestions. I will have to evaluate further testing and perhaps an article as you have suggested. I think I would have a year or so worth of work to do on this.

73,

Mark N5RFX

 

No, I listen to them all. Especially the Pactor II and III signals that start up regularly right on top of my Pactor I sessions and stomp all over my sessions that were already in progress.

As I have pointed out repeatedly here, I have no problem with Pactor. If it can live with SSB so be it. (But I will have to be shown.)

It is the Pactor robots that I have a problem with. At least with mt63 it is a keyboard-to-keyboard mode with operators on both ends. They can listen to the frequency and either choose to forego operation or to work "under" the communication already in progress using low power if the need is high enough.

I take the Region 1 recommendations seriously. They have experience in exactly the arrangment the ARRL has proposed. Their conclusion is that mixing digital and analog does NOT work well.

No one on here has given me any reason to expect our experience will be any different. Does anyone think the Europeans don't use Pactor and mt63 and the other various digimodes?

Good luck on your DX expedition. But you have convinced me, and I suspect very few of the readers, that mixing analog and digital is workable.

tim ab0wr

 

mark,

I dont' have time this morning to fully reply to your post.

One observation. I find it curious that you could have multiple mt63 sessions on the same frequency when all kinds of interleaved copy would be appearing on the screen.

Are you sure these sessions are not offset by some amount that allows individual sessions to pick out the wanted one? An offset of even 50hz may be enough to allow the mt63 detector and transceiver tuning to discriminate between signals and treat the others as noise, especially if there is a power level difference.

A minor frequency difference may not be noticieable by ear but might be enough to allow the DSP to separate them out.

This is one of the reasons that RF testing will be important to do at some point in time.

It is testing the ARRL should have done PRIOR to their proposal so they could publish the results along with a spectrum efficiency analysis. They certainly have the capability. Why don't they have the ambition? Call me suspicious but it is this kind of flim-flammery by the ARRL that biases me against the proposal just on its face.

tim ab0wr

 

Albert,

You are welcome. Reasonable and thoughtful discussion can go a long way. I understand Tim's arguments and am glad that we had the discussion.

73,

Mark N5RFX

 

Tim,

I understand the time problem. As far as the multiple MT63 signals, you are correct this is not a desirable operating mode, but it did occur and I was just passing along my observations. I know during field day I have experienced the same phenomenon with PSK31. While I don't know the capture ratio of PSK31, I have observed that the stronger signal is decoded. I do agree that the weaker signal is decoded when the stronger signal goes away and you have to know that this is happening. You are correct that many times there was a frequency offset with the multiple MT63 signals because you will find that most ham transmitters are rarely on exactly the same frequency, but if you watch your waterfall, and listen to the audio, you can tell that there are multiple signals to decode, and if you happen to start decoding another signal, you have to ignore that text.

I also agree that the ARRL has not done a good job in selling their petition. I feel that this petition has done damage to the concept of mixing modes. The main problem is with the inclusion of semi automatic control. This does lead one to wonder what the real motives were in submitting this petition.

73,

Mark N5RFX