VK1KW WSPR beacon trial (2m)

[VK2OMD]

Following our recent successful trial of a beacon with composite 12WPM and QRSS3 (slow Morse code with 3s dits) Morse identification, Rob and I have run a short trial over the same 150km path using WSPR.

WSPR promises decoding to similar levels as demonstrated in the QRSS3 trial, though with the advantage of hands off automation.

An 'animation' of two hours of capture is at http://youtu.be/kYJh4bruuZQ.



Rob transmitted 100% of the time at the same power level, same antenna, and I used the same receiver, same antenna as last week's QRSS3 experiment.

Although there were 60 two minute indent opportunities during the two hours and there was clearly visible signal in almost all of them, there were no successful decodes.

Visually, there doesn't appear to be significant frequency drift of tx relative to rx, clocks were synchronised and rechecked during the experiment, receive was decoded on two parallel computers with the same results.

The waterfall displays shows significant frequency dispersal of energy which is almost certainly due to propagation mechanism / conditions.

The experiment questions the effectiveness of WSPR for weak signal beacons on 2m, they might work on some paths, but WSPR was an abject failure on this trial.

See viewtopic.php?f=11&t=11191 for info on the recent the QRSS trial.

Owen

[VK2FAK]

Hi all....

Owen..

Interesting experiment, but I wonder if your using the best choice. As I understand it people do use it and have some success on 2m and above but it was not designed for this purpose....I think HF and MF was about it.....a few of the other modes may be more appropriate for 2m....

Maybe the JT65 is better for this.....as you have a good link going there I would like to see a test of that mode....

John

[VK2OMD]

...
Maybe the JT65 is better for this.....as you have a good link going there I would like to see a test of that mode....

John,

Rob and I have demonstrated by experiment a very simple variation to a conventional VHF beacon that permits successfully decoding an identification at a S/N (in 2500Hz to borrow K1JT's 'standard') below 20dB and close to 30dB (depending on a range of factors), some 10-20dB lower than the existing ident scheme. The tested scheme or a variant might be employed on reactivated VK1 beacons in the near future.

As I understand it, there has been very little interest by way of reception reports, though the trial beacon continues to operate daily (except for earlier today whilst WSPR was trialled over the path). The trial beacon should be heard well in Melbourne. Sydney will be a bit more challenging due to path obstructions near Rob's location... but if I can hear it in Bowral with a very poor antenna (4el Yagi at 2m height), I expect it should be heard in Sydney (using Spectran or the like) by a medium performance station.

One pundit suggested that WSPR would be better. The latest experiment casts doubt on that. The two hours of observation did not result in a single decode, in the next four hours there was one decode, albeit unsuccessful as the decoded callsign was FH7/PL7ZDW.

G4JNT has done work on composite beacons that include a JT65 segment, but I have not seen any write-ups of reception compared to other modes. If you Google for it, you may enjoy the read. JT65B has a noise bandwidth of 5.4Hz as I understand it, and that is 20 times that which was used for the QRSS experiment... a disadvantage of 13db however the FEC claws back some of that. The real issue though is that the human brain is more powerful in interpretation of the spectrograms we have seen with QRSS than the software.

Owen

PS: my grabber will be monitoring the 12WPM/QRSS3 signal until a bit later this evening: http://vk1od.net/grabber/grabber.php .

[VK3DXE]

Hi Owen,

Great work you and Rob are doing!

I've listened/looked at times and seen nothing from here, but Canberra/Sydney are in probably my most challenging direction due to proximity of the Melbourne CBD and local noise issues.

I have listened later at night (after midnight) when the noise here subsides, but I'm really unsure of the operating schedule. i.e. Is the beacon still QRV at that time?

As you're no doubt aware, JT65 is more tolerant of drift and frequency inaccuracies, so I'd suggest perhaps a period of JT65, followed by a CW ident and maybe some carrier.

Myself and others have run successful JT65 tests under ordinary (non-enhanced) conditions out to 600 - 700km which still works despite meteor pings, and multipath and doppler from aircraft enhancement. WSPR tends to be more intolerant of drift, doppler and meteors, so perhaps it's not the ideal choice for a beacon?

[VK2KRR]

Ive had a look at the screen grabs. Most of the periods have doppler signals in them. Any time a doppler shifted signal comes across the main signal, it usually will not decode, if it drifts more than around 4 Hz it will also probably not decode. The rest of the signals without interference are weak and patchy. From looking at the 65 transmissions in the youtube I would have thought that maybe only 3 should have been close to decoding.

WSPR is not specified to decode weaker than about -25 to -26dB SNR signals from memory there are other software programs that will work better for weaker signals and more unstable signals or signals with doppler from AE or meteors. Programs as mentioned such as JT65.

The advantages of WSPR over the normal JT65 set up is that it can run fully automatically (hence being called Weak Signal Propagation Reporter).
If paths open the data is automatically sent to the WSPR server for anyone to view what is happening. Where not only can you view the paths but you can access the database and see the reported signals strengths and other information.

WSPR will behave better on longer DX paths that are to long for AE reflections and meteors, such as VK to ZL and some Bight paths.

I have decoded VK1KW on 2m WSPR from here whilst Ive been beaming 180 deg the wrong direction, so it does work. Ive also decoded VK2, VK3, VK4 and VK5 all using 2m WSPR in the past.

If you want to decode ulimately low signals, you would be better off using JT65b on 2m. But you dont get the automatic uploading and stats, history to review etc.

Here is a link to the 2m WSPR msg board https://groups.google.com/forum/?fromgr ... meter-wspr

And also video of Derek VK6DZ WSPR decoding here at 2664 km. This occurred on 4 consecutive mornings here around New Year. A couple of the mornings I could not even hear the Adelaide or Albany beacons, quite amazing.

[VK2OMD]

Hi Owen,

Great work you and Rob are doing!

I've listened/looked at times and seen nothing from here, but Canberra/Sydney are in probably my most challenging direction due to proximity of the Melbourne CBD and local noise issues.

I have listened later at night (after midnight) when the noise here subsides, but I'm really unsure of the operating schedule. i.e. Is the beacon still QRV at that time?

As you're no doubt aware, JT65 is more tolerant of drift and frequency inaccuracies, so I'd suggest perhaps a period of JT65, followed by a CW ident and maybe some carrier.

Myself and others have run successful JT65 tests under ordinary (non-enhanced) conditions out to 600 - 700km which still works despite meteor pings, and multipath and doppler from aircraft enhancement. WSPR tends to be more intolerant of drift, doppler and meteors, so perhaps it's not the ideal choice for a beacon?

I expect WSPR works on some 2m paths and I would not suggest otherwise, but a 2m weak signal beacon needs a scheme that works on most if not all paths. Today the WSPR score was 0 correct decodes and 1 false decode in 6 hours. Also noteworthy was that most of 6 hours today was 100% transmission, something you would not do if you located all WSPR beacons in a 200Hz shared band segment (as intended by WSPR) as people relatively local to a beacon need some silence to hear other beacons. So if you extrapolate our observations to a 33% tx cycle, that is zero correct decodes in over 15 hours.

A path affected by aircraft reflections, multipathing and Doppler even on the atmospheric and tropospheric paths is part of the environment, and a beacon scheme that does not deliver under those conditions is less use that the old CW beacon.

Rob discussed the schedule in the thread at viewtopic.php?f=11&t=11191 .

If you were unable to hear it after midnight, it is probably because it was not on air. If you listen between 9am and 9pm, I would be surprised if you do not at least observe aircraft reflections as there are so many aircraft plying the Syd-Can-Mel path. (I am seeing three, four, five aircraft reflections in each 10min grab at the moment - 20:30 local.)

Having already demonstrated a scheme that is simple and effective, that can easily be deployed on existing beacons with minimal modification (no modification to the modulator, no additional stability requirement), we have been told "WSPR is better". We have shown it isn't necessarily, now we hear that JT65 would address those problems. Well, I look forward to writeup of anyone's experiments with a JT65 beacon here in VK/ZL. We surely ought to reconsider beacon options given the demise of the analogue TV transmitters that supplemented our own beacons over the past 50+ years.

The recent responses to discussion of a certain new beacon along the lines of "pity it isn't GPS locked etc" are trite and unhelpful... beacons will not be converted to GPS locking overnight and it is naive to think that GPS locking is bullet proof, and in any event, it locks only one end of the link (albeit the one with often the most challenging environmentals).

I do have a prototype beacon in development which will be GPS locked and trial a composite modulation scheme of 12WPM 200Hz FSK and QRSS3 20Hz FSK (http://www.vk1od.net/module/smbk/dual.htm). The transmitter delivers 20W from a crystal, the keyer / DDS exciter has been demonstrated using a crystal clock, what is needed now is to reference the DDS from the GPSDO (because the crystal is too unstable for 20Hz FSK) and hook the RF output into the transmitter. The keyer could be replaced fairly easily with a JT65B keyer, but that requires WSJT to decode the signals and the automation to some extent hides the raw signal.

Owen

[VK2KRR]

I should also mention that the other attractive thing about WSPR is that technically, all of the Australian & NZ 2m beacons (and for that matter, world wide) could operate on the one base frequency of 144.489. Each beacon in Australia, if freq locked, could then use a particular TX freq in the WSPR 200 Hz bandwidth, 144.489400 to 144.489600 MHz.

For example, we could have -
Geelong Beacon on 144.489530.
Adelaide on 144.489450
Mt Gambier on 144.489550
Mildura on 144.489435
Esperance on 144.489568

To further limit interference potential a certain transmit percentage could be specified for each. For example 20%, 25%, 30%. But then even if they were all set at 20% TX rate, they TX randomly.

BUT ! Even if all five of the above beacons were to transmit at the same time, and someone could hear all five of the signals, they are of sufficient spacing to where they should all be decoded from the one TX period all at the same time. But with randomisation and propagation conditions, to have all signals being heard at the one QTH on the one beam heading would rarely happen.

You can then set your radio to monitor the 2m WSPR freq 144.489, overnight for example, and the next morning you can review whats been decoded on the screen.

[VK2KRR]

we have been told "WSPR is better". We have shown it isn't necessarily, now we hear that JT65 would address those problems. Well, I look forward to writeup of anyone's experiments with a JT65 beacon here in VK/ZL.

I've also conducted numerous JT65 & previously JT44 tests on 2m & above with some VK5 stations in particular. And also VK6, VK3 etc over the years.
JT65 in particular still does not handle meteor doppler interference well (or aircraft), but it does a much better job than what WSPR would do in the same situation, whilst also being able to decode deeper into the noise.

Also both the timing and radio freq drift are not so critical.

One pundit suggested that WSPR would be better

If you are referring to me with this comment, by suggesting WSPR would be better, I was implying, in general WSPR would be better as a beacon for the reasons mentioned previously. I was not necessarily implying that it will out perform other software versions, as it certainly will fall short in some situations. But the benefits are those mentioned in my previous posts, regarding the database, mapping, single freq etc, and I think the outweigh the negatives.

If you could not decode Rob VK1KW, there are other factors at play that are causing this. Primarily, I feel you have to much interference & dont have a good enough SNR from Rob for WSPR to decode. In a DX situation, once you got used to it, you would be able to tell where the signal is coming from even if it didn't decode. If that signal was coming from Adelaide for example, you can hover your pointer over the signal line and WSPR will display the freq or if your watching on Spect Lab you can also read the freq. Now if Adelaide was running on 144.489450 and you are beaming to the west, there will be a fair chance this will be coming from Adelaide. To confirm if this was the case, if it did not decode for you, you would do a refresh on the WSPR database page & you would probably see that another station (as there would be a few if there was WSPR beacons running) would have a decode logged in the WSPR database at the same time period, corresponding to the same freq and has been decoded with the VK5VF callsign. This could be from someone like VK5BC who may be monitoring or myself or anyone else in VK3 etc. Thus if you really know what to look for you can work out where signals are coming from even if they dont decode on your screen in particular. If all else fails with the decode, there is room at the end of the WSPR TX for a single CW ID to be keyed up, but will need to be strong enough to hear.

I also feel that beacons running modes like WSPR and anything other than CW, will help to encourage the younger generation into radio and VHF UHF DX.

[VK5PJ]

Hello Owen and Rob,
I have parked my yagi towards VK1, the IC-275H is running about 10W of WSPR on 144.489 at 20% for the next three days. While the rig has a TXCO for freq control there are other xtals in the rig for BFO's etc that do cause some drift during tx but I hope it will recover back to the base freq during rx.

I checked the dial freq to ensure I am on 144.489 USB from the GPS locked Marconi Sig gen and HP Counter, then measured the 1KHz offset in the base band of the rx so I am confident that I am as close to the right spot as I can get.

Not sure who will hear it or if it is just a curiousity but as I wil be away from the shack it seemed like a good idea at the time.

regards,
Peter, vk5pj

[VK2KRR]

I've been able to see Peters 2m WSPR signal all night and day here, not every transmission decodes, but this is just due to drift or occassional aircraft doppler interference. Although some of the aircraft reflected signals will decode at quite high strength as long as the doppler shifting is quite steady, within 4 Hz or so.

Also Phil VK5AKK has given me an occassional signal throughout the day, decoding very easily. In the attached image you can see how good Phils signal is in terms of drift, there is none at all on his signal (at 0908 utc, the lone signal lower in the WSPR graph).

Pete's sig does get a slight drift up at times, and as you can see in the image, Pete's sig looks fairly drifty, but this 0928 utc sig did decode OK from this indicating 3 Hz.

VK5PJ & VK5AKK 2m WSPR

[VK2OMD]

One pundit suggested that WSPR would be better

If you are referring to me with this comment,

No.

If you could not decode Rob VK1KW, there are other factors at play that are causing this. Primarily, I feel you have to much interference & dont have a good enough SNR from Rob for WSPR to decode.

The interference you refer to could only be the aircraft reflections, they are just part of the environment for most of us hearing weak 2m signals and a system that is not tolerant of aircraft reflections might have limited usefulness.

The SNR measured on the QRSS3 trial was very often well above the claimed -28dB decode limit for WSPR, yet no decodes.

My own view is that the interleaving incorporated into WSPR encoding to provide protection against burst noise should also provide protection against an aircraft reflection that sweeps quickly through the 'direct' signal, and so might be less a factor that thought. The observable spectral dispersion of the direct signal hints at a real problem given the very narrow FFT bins (1.5Hz) used for WSPR.

The simple fact is that the QRSS3 message was decodable and reasonably quickly led to positive identification of the signal, whereas the 6 hour WSPR trial resulted in one false decode, and zero valid decodes.

So, on this trial, the QRSS3 message was effective whereas WSPR failed totally.

WSPR might have some advantages over QRSS3, but identification of very weak signals over long paths on 2m is not one of them.

Owen

[VK2KRR]

Owen/Rob

I'd like to see Rob turn the TX power up on 2m WSPR to give u a better signal strength to see if you can then decode the signals. If they still dont decode when there is no AE doppler present and a stable signal, then I would think there must be another issue at play which needs to be sorted out before drawing your conclusion. As if you have sufficient signal strength and stable signal there should be no reason you cant decode him.

[VK2OMD]

... As if you have sufficient signal strength and stable signal there should be no reason you cant decode him.

This is not to prove that WSPR cannot work, just that in the particular experiment, it didn't work as well as QRSS3, in fact it didn't work at all.

I have already accepted that it does work sometimes. I have even noted record in the WSPR archives that PD0DNL spotted you on 2m in Jul 2011... quite an achievement. I doubt it, the WSPR database is full of dud entries... a by product of the automated logging... both a strength and weakness of the WSPR system.

I note your recent success with VK5PJ, and I note the average SNR when I looked a few minutes ago at 32 spots was -8dB, they are quite strong signals relative to the QRSS signals we were decoding. That stands as proof that WSPR works sometimes.

I have looked at the distribution of 2m spots in the Dec 2012 archive, and of the roughly 1500 spots over 100km distance, there were plenty of spots below -20dB SNR, again evidence that it does work on some paths, and to lower SNR than your trial with Peter.

But on the day, on the path tested here, it was not as effective as QRSS3.

Owen

[VK2KRR]

it didn't work as well as QRSS3, in fact it didn't work at all.

Thus in my view, it cant be a valid conclusion to say that WSPR didn't work at all if you cant show that with a better signal strength, that Rob's signals received at your location are able to be decoded.

But on the day, on the path tested here, it was not as effective as QRSS3.

To be fair on the WSPR program, if you guys cant show that a stronger signal can decode between you, then to draw a conclusion that its not as effective as QRSS3 is a bit unrealistic I would have thought

Why I mention this, is that WSPR wont work unless everything is right. It can be fairly touchy. Thus its a good idea to get it going and make sure it works using higher TX power/signal strength and then drop it down and see how low you can run it to into the noise.

[VK2KRR]

Owen

I'd just like to point this decode out of Brian VK5BC's signal. Ive circled it on the image in grey. This signal is compareable to some of Robs signals, but decoded OK here. It does look extremely weak on the display, and does look like it would hardly be enough to decode but gave a reasonable -20dB report.

VK5BC decode

[VK2OMD]

Leigh,



Above is a freq distribution of ~1500 2m spots for Dec last where dist>100km.

Clearly spots well below -30dB SNR occur, though they taper off below about -24dB (the -28dB figure isn't a brick wall like often believed).

It is apparent that SNR is not the sole requirement for decoding.

Owen

PS: Some of the SNR figures plotted below -30 are a bit concerning, they may be the result of software problems (I think some earlier versions of WSPR miscalculated SNR and they may still be in use). They may just be a product of the algorithms, but you have to wade through the source code to understand that.

[VK2OMD]

A look a little deeper at the QRSS3 beacon sig from VK1KW:



Above is the waterfall for 10min.



Above is the plotted SNR (blue) on the K1JT scale (2500Hz ref).

It can be seen that the KW is identifiable in the waterfall although the SNR is rarely above -25dB in the even minutes (ie when the QRSS3 mesage is sent).

Owen

[VK2KRR]

I can see the signal in your waterfall attached. But would not say that its readable. I can see 2 characters I would guess could be a K -.- in CW but thats about all. If that showed on my screen there would be no way I could say who it is. Perhaps over a longer time period if it was repeated over and over and with rises and falls in QSB this could be decoded. Not trying to be bias or anything, but thats just my honest opinion on what I see there.

[VK2OMD]

I can see the signal in your waterfall attached. But would not say that its readable.

Well you should stick with WSPR, it seems to best suit your needs.

Owen

[N6GN]

I've not had time for careful scrutiny of the WSPR trial by Owen, but I'd like to say that my first impression of your "Doppler" interference appears to be Aircraft scatter, ACS (random as compared to AES) and is something we routinely see on short/mid DX 2m paths. Even on the VHF bands, where it is possibly more of an issue than on either HF, where most spots are DX, or UHF where the ACS component is generally so far from the frequency-stable component being spotted as not to cause self-QRM, these do *not* generally cause much of a problem with decodes.
We have a great deal of data looking at these spots, even identifying the particular aircraft responsible and correlating with observed frequency shift of the ACS compoonent. As it appears, the longer DX - out past say 400 km- seems to be what is supporting the spots and the frequency shift component is negligible or small enough that WSPR (standard) can follow it fine in a 2 minute interval. These findings correlate with ongoing SF to LA tests on 1296 which produce regular QSOs over non-flat terrain from well equipped stations. It appears that this random ACS mode is available on all VHF through microwave bands though it may take a narrow information bw such as WSPR to utilize it if only modest stations are involved. There are both screenshots and time lapse videos of waterfalls available on links from the 2-meter-WSPR Group threads.

We have also run WSPR-15, like WSPR-2 but scaled by a factor of 8 and having the ability to decode to around -39 dB relative to a SSB BW. Contrary to K1JT's original suggestion, this mode also appears to work very well for VHF and UHF experiments. Spots have been made using it up through 10 GHz.

If you would like to study some of the characteristics we have been seeing in California over the last year on 2m and 70 cm WSPR, have a look at the 2-meter-WSPR Google Group. One need not join in order to read it but you are also welcome to join.

In addition to the ACS mode, we have also seen propagation which appears to be due to wing tip vortices produced by aircraft. Again, these can be correlated to specific aircraft.

In summary, I concur with what I think I hear Leigh saying, that one shouldn't be too quick to write off the usefulness of WSPR for discovering new modes and general weak signal propagation studies at VHF and UHF. We are having a really good time doing this and have regularly seen spots out to ~900 km.

best,
Glenn n6gn