Discussion in 'Amateur Radio News' started by AA7BQ, Nov 19, 2005.
Is it only ARQ modes like PACTOR III that you are interested in testing?
No sir, not hooked in. I am speaking of modes that would occupy the 3.5khz spectrum since that is where the robots will do the most harm to analog voice. I believe Pactor is important because that is what we'll see with Winlink. SCAMP too should be tested at some point.
I am suprised we have not had a "byte-off" whereby we tableize the performance of all of these under various conditions in QST or CQ (and Newsline). The goal would be to prove or de-bunk the hype. What is the advertized performance? What is the performance under interference condx. How much interference can they take? How do they relate to SSB Analog Voice? What happens to through-put? Comparative occupance rate? Duty cycle?
I am not sure what you mean by "not hooked in", but you are looking for testing of any mode greater than 500 Hz necessary bandwidth? I think I understand. So let me go back to your 3 points so I can understand.
"1. Digital signals with error correction will sync and be usable in less bandwidth than voice."
When you use the term bandwidth, you are referring to throughput? If so, then you want to know when a particular mode's throughput decreases. What did you think the the MT63 tests that I posted? It showed that their is a 2dB capture effect. To continue testing with that mode you would like to have no copy from each transmissions (signals within 2 dB of each other) then move one of the signals away and see when both start to give 100% copy? There really is no throughput test with MT63 as thoughput is fixed based on the bandwidth and the interleaving that is used, so we would have to use a character error rate. Am I on track here?
"2. The Digital results under #1 will be less than acceptale due to the time and number of re-trys (my example of a Pactor modem taking 240 seconds to transmit a short message). What is "acceptable" is subjective, but if you re-wrote the modem's specs to the level of performance under #1 above, almost nobody would buy it."
I think this only applies to an ARQ mode? Correct me if I am wrong.
"3. While digital modems may sync and work at levels below the noise floor (CW can do this as well with SSCW), they cannot exibit this low noise operation in the presense of a strong interfereing signal. They can do #1 or #3 but not both."
I am not sure what I am supposed to test here. Can you explain in a different way? BTW I did not understand the noise floor arguments earlier in the thread, so maybe I am hopeless with this one.
I don't want to waste my time testing things you and others are not interested in, so that is why I am asking a lot of questions. I would like to know what it would take to convince folks that mixing analog and digital signals is not as horrible as some are describing.
Your audio files are too big for me to download with a dial up line so I haven't listened to them.
However, I really do doubt that SSB and digital can share a frequency without some adverse effects. I am sure net operations could be handled on top of a digital signal or other short term conversations. But routine SSB QSO's WILL SUFFER.
I know how tiring and frustrating a hetrodyne can be when you try to ragchew over one. This happens even if the hetrodyne is several S-units lower. Lots of people don't experience this anymore because of DSP and auto-notch, but for those without this we hear it all the time.
A digital signal in the background wouldn't be any different, except auto-notches won't work so a lot of people will hear the effects. The result is that the digital operators will not be affected and will continue, however, SSB QSO's will have to be cut short.
In the long term this could result in fewer and fewer hams using SSB and quite possibly a lot of them dropping out of the ARS altogether. Now this may sound good to the digital elitist's who think we should be going all digital and removing the analog modes from the air, but be careful of unintended results. I can think of several, none of them good.
The Region 1 representatives have decided that analog and digital should not be mixed on HF. They already have experience with this. While your tests may show it is POSSIBLE to mix them, ask yourself why another region with experience would recommend against it.
I suspect they have already analyzed the adverse effects it could have on ham radio's viability in the long term. Remember, SSB drives the market today and most likely for at least another 10 - 20 years. SSB also controls the viability of several amateur radio manufacturers offerings. Doing anything that upsets the status quo and results in an early reduction of SSB operators could reduce the number of hams and equipment offerings.
Thanks for the comments. I agree that it is not desirable to have a digital signal in the background, but why would this undesirable condition increase because we allow digital data content where we now allow digital image and voice content? I agree that if automatic control is not regulated, then there is a possibility that digital data emissions would increase in the phone bands; no argument there.
I keep hearing about the region 1 representatives having decided that analog and digital should not be mixed on HF, but I have not found any papers that discuss the reasoning and the empirical data that shows mixing of analog and digital signals is problematic. Maybe someone can point me to the study that caused this action? The 2006 Region 1 bandplan certainly does not reflect a prohibition of mixing analog and digital modes.
The fact remains that digital and analog signals share spectrum on HF today. Where is the documentation that shows that this is problematic?
A peak to average ratio of 0db indicates the peak and the average are the same value.
ie. 20log (1) = 0db
This isn't possible for a sine wave. That's what you would get for a square wave.
Are you sure the program is analyzing the waveform correctly?
I don't think you can just convert from peak/average to crest factor by adding 3db.
For a sine wave with a peak voltage of 1v, the rms value is .707v. That makes the crest factor 1/.707 = 1.414.
To change this to db you would use 20log(1.414) = 3db.
The difference is 1.586 not 3.
A good description of crest factor can be found at http://www.tinaja.com/glib/muse125.pdf. This is how I learned it anyway. Don Lancaster has always been a favorite author of mine.
One thing everyone needs to be careful about is that although the term "average power" is bandied about a lot *I* always consider it to be "rms" average power unless otherwise specified. Your Mileage May Vary (YMMV).
I'm surprised that your program doesn't analyze a straight sine wave right but shows the right figure for a two-tone test.
The big problem with doing this at baseband is that the receiver impacts are not measured at all. My primary concern would be impacts of AGC on actual received signals.
I'm going to have to think about this one some. I'm not sure I see any way besides doing this at RF. It would certainly be possible to do without violating FCC rules. You just need to work the transmitters at low power into a combiner or into a couple of dummy loads. The individual average power levels could be adjusted just using the S-meter on the receiver (if you trust the agc to be linear). But you will need two transmitters and a receiver to do the test. I'll have to see if I can come up with what is needed.
Charlie, I'm listening from near your Florida address. Let your CW neighbors know you are not a 20wpm extra.
Here are the recommendations approved by the Region 1 plenary that addresses this.
DV05_C4_Rec_13 It is recommended that the following proposed principles for new IARU Region 1 HF bandplans be accepted, and that the principles be included in the IARU Region 1 HF Manager's Handbook:
CW operation is accepted across all bands, except within beacon segments.
Telephony (including AM) is limited to certain telephony segments.
Digital data modes are limited to certain digital segments.
Digitised speech is considered a digital data mode regarding bandplan matters.
The current IARU Region 1 band plan is well known and receives a high degree of respect and adherence within the IARU Region 1; hence major changes to the bandplan are not necessary for the time being.
Proposed by SARL, seconded by RAAG, approved with 4 abstentions
DV05_C4_Rec_14 That the bandplan created by the Bandplan Working Group be approved. (See Annex to recommendation DV05_C4_Rec_14)
Proposed by DARC, seconded by OeVSV, approved with 5 abstentions
A discussion took place on the changes that had been made, and whether they reflected the agreed bandplan principles in DV05_C4_Rec_13. Iceland pointed out that paper DV05_C4_13 was not totally consistent with paper DV05_C4_11. However it was agreed that the bandplan should be considered as proposed.
It is obvious that they know that the bandplan doesn't match with the recommendations to keep digimodes separated from analog.
It looks like that will be addressed in the future.
If I can find any discussions of the issues that I havn't already posted, I'll put them on here.
Look at your math.
"For a sine wave with a peak voltage of 1v, the rms value is .707v. That makes the crest factor 1/.707 = 1.414."
"To change this to db you would use 20log(1.414) = 3db."
100% correct; but, why do you subtract 1.414 from 3? You need to be subtracting dB's from dB's.
I think our problem is terminology. When I use the term peak to average power ratio, I am actually saying peak average power to average power ratio. Just like when you say Peak Envelope Power you are really saying peak envelope average power. Now I think the math will work. Using your Sine wave with a peak voltage of 1V and an RMS value of .707, the average power is .707^2 or .5 . Since the envelope is flat, the peak average power is also .5. So 10log(.5/.5) is 0dB. 3-0=3 so to convert from peak average power to average power ratio, add 3dB. Cool Edit pro gives me a peak average power and an average power in dB. I simply subtract the peak average power from the average power. Does that make sense?
Thanks. I have been looking to for information as to how they have come to the conclusion that digital and analog signals should not be mixed. Actually I am looking for any technical paper that discusses this subject.