Transmitting into a dummy load .. for a year .. on purpose.

[Onno VK6FLAB]

Just under a year ago I started an experiment. I set-up a beacon for WSPR, or Weak Signal Propagation Reporter, transmitting at 200 mW into a dummy load using eight bands between 80m and 10m.

I also set-up an RTL-SDR dongle, connected to an external 20m HF antenna and made it monitor 18 amateur bands between 630m and 23cm.

I left this running 24/7 for most of the year, though there were times when I detached the antenna due to local thunderstorms and there was a seven week period where there were no reports. It's highly likely that I forgot to reconnect the antenna, but I don't recall.

For this analysis I used the online WSPRnet.org database where I uploaded my spots as they were decoded. I noticed that there are reports that I have locally that are not in the database, though I'm not sure why. They're incomplete and not in the same format and merging these is non-trivial for reasons I'll discuss.

Lesson learnt, the "rtlsdr-wsprd" tool needs to be patched to output the data in the same format as is available from the online database and I need to actively log locally.

The results are puzzling, at least to me right now.

Let's start with the low hanging fruit. There are no reports of my WSPR beacon being received by anyone other than me. That doesn't guarantee that nobody heard me, just that nobody reported that they did.

In the database there's just over six thousand reports of my station receiving a WSPR transmission from my beacon during the past year.

The reports cover all bands, though not equally. The 80m band represents 6 percent of reports, where 40m accounts for 20 percent.

The reported SNR, or Signal To Noise ratio, varies significantly across the data. For example, the 12m band shows a range of 42 dB.

Digging into this does not reveal any patterns related to date, time of day, season, other band reports or any other metric I was able to imagine.

In my exploration, missing records and time-zone differences aside, I discovered that the local data does not appear to match the database.

For example I have records where the software decoded my beacon ten times in the same time-slot, but none of them exist in the database. For others, there's only one matching record, which leads me to believe that the WSPRnet.org database only accepts the first report for any given combination of timestamp, transmitter and receiver, but I have yet to confirm that.

So, let's talk about getting more than one result for a specific time-slot. As you might know, a WSPR signal is transmitted every 120 seconds, starting at the even minute. Each transmission lasts 110.6 seconds.

The decoder will make several attempts to decode multiple, potentially overlapping signals. It is my understanding that the way this happens is by essentially removing a known decoded signal and then attempting to decode what's left, repeating until either there's no more signals to decode, or time runs out, since there's probably only really 9.4 seconds in which to do this. Potentially this means that a faster computer will decode more signals, but I've not actually tested that, but it's probably something worth pursuing.

Back to our decodes. If the first decode is removed from the received data and the next decode gives you similar information, same callsign and maidenhead locator, with SNR and frequency differences, then you might imagine that there's so much of it there that the only way that might happen is because the receiver is overloaded.

I'm still looking into this, because if that's the case, then we'd need to determine if the receiver was always overloaded, or only sometimes.

It's curious, since there's over a thousand other signals being received from other stations, several over 18,000 km away, so it's not like the receiver is completely swamped.

Another hypothesis is that the decode is coming from a different band, like a harmonic. This is potentially caused because from a band and timing perspective, the receiver isn't linked to the transmitter in any way.

The transmitter hammers away 24/7 one band after the next, switching every two minutes, the receiver listens for half an hour on a band, then randomly picks the next, until it runs out of bands and starts again.

The receiver is listening on more than twice as many bands as the transmitter operates on, but that doesn't mean that it cannot hear the transmitter on a harmonic of one of the bands.

Again, I don't know if this is the case, or if something else is happening. One thing I'd expect, is to see reports on other harmonics outside the bands that the transmitter is using, but I'm not seeing that.

Perhaps the overload is limited to just the band we're actively monitoring and the other signals are coming in regardless of the overload. I'm still trying to determine if that's the case.

As I said, merging the data from the two sources is non-trivial, time-zones and formatting are not the same and I'm not in the mood for manually fixing 2,500 or so records, not to mention attempting to determine which SNR is real for the multi-decodes.

So, what did I learn?

For starters, the world didn't come to a sudden and laborious stop when I transmitted into a dummy load.

The experiment was interesting and worth doing.

I should test using shorter runs until I've determined the mechanisms involved. For example, one amateur suggested that I might be decoding information that's coming in via the coax, rather than from an antenna. That said, doing so would also require significantly more effort to incrementally analyse this data, so I'd have to find ways to improve my workflow.

The SNR is all over the place, not something that I expected.

All bands are represented in the data.

There does not seem to be any relationship between date, time, other stations and the signal strength seen for the local transmission.

I need better record keeping.

No doubt there's more.

If you have questions, feel free to comment.

The experiment also leaves plenty of questions.

Why do the SNR values vary so much?

I can't imagine that the variation relates to propagation, since we'd have reports from other receivers, so is it something else, even though we're talking about equipment that's indoors, are we observing variations in electronics temperatures for example?

Alternatively, if the measurements represent overload of the receiver, why don't we see other harmonics and how is it possible that we can receive and decode very weak signals from other stations?

If the signal is arriving via an unexpected path, like the coax, rather than the antenna, what could we do to stop that from occurring and what effects does it have on our current dataset, and could we account for those effects?

I suppose, leaving the ultimate question for last: Is the data that I've collected over the past year useful, beyond potentially "this is not how you do this", or is it essentially meaningless?

Onno VK6FLAB

[Darrel Emerson]

An interesting test.

I wouldn't expect harmonics to be decoded, because the modulation frequency deviations would be doubled, tripled or whatever. Just a thought.

73

   Darrel aa7fv

--
Please follow the HamSCI Community Participation Guidelines at http://hamsci.org/hamsci-community-participation-guidelines.
---
You received this message because you are subscribed to the Google Groups "HamSCI" group.
To unsubscribe from this group and stop receiving emails from it, send an email to hamsci+un...@googlegroups.com.
To view this discussion visit https://groups.google.com/d/msgid/hamsci/CACybYRU33Y_kLk99ODKe-0gKVVkDi%3DN4WgsPsBhdfZww%2BE%2Bhww%40mail.gmail.com.

[Onno VK6FLAB]

If the receiver is tuned to a harmonic band of the transmission in progress, would that not arrive at the decoder as-if it had been transmitted on that band and subsequently decoded?

Alternatively, if the receiver is overloaded, could that not manifest as a signal showing up on an unintended frequency in the A/D samples?

Note that I don't know, I'm asking.

--
finger painting on glass is an inexact art - apologies for any errors in this scra^Hibble

()/)/)() ..ASCII for Onno..

To view this discussion visit https://groups.google.com/d/msgid/hamsci/CAGNQrOrcQfHbn5dZ4JCm3Q%2BYtqPPpi7ByoqhwBnXtZmP9skRQA%40mail.gmail.com.

[Darrel Emerson]

Hi Pete,

   If the original transmission was WSPR with 4 tones 1.6 Hz apart at the fundamental RF frequency, wouldn't we end up with 4 tones 3.2 Hz apart at the second harmonic? That would not decode with a normal WSPR receiver tuned to the second harmonic, which would still be expecting  1.6 Hz tone spacing. Apart from anything else, 2 of the tones would now be outside the normal rx WSPR passband, however you tuned the receiver.

  Does that make sense?

  There may be other spurious mechanisms that might give a successful decode,  but I don't expect that straightforward transmitter harmonics would work. Is that right?

73

     Darrel

To view this discussion visit https://groups.google.com/d/msgid/hamsci/CACybYRVXKO-edw-y14d9f3LTG6XN53mJDUJ%2B9zaYWeTdVPfQPw%40mail.gmail.com.

[Onno Benschop]

Not sure who Pete is, I'm Onno VK6FLAB.

I understand your explanation, but I'm not sure if it applies. When I hear the local AM broadcast station harmonics on HF, they don't appear to be frequency stretched in the way you describe.

Are were talking about different phenomena, or is there something else going on?

--
finger painting on glass is an inexact art - apologies for any errors in this scra^Hibble

()/)/)() ..ASCII for Onno..

To view this discussion visit https://groups.google.com/d/msgid/hamsci/CAGNQrOro8ZTEM01C2yf3DT47kaJ8_h-SFs1_zVy7Wgx21P97fw%40mail.gmail.com.

[Ethan Miller K8GU]

Onno and Darrel,

I don't have any concrete numbers to provide (I could perform a
controlled test, though). However, I can informally confirm that
RTL-SDRs have pretty variable sensitivity at HF and do not have good
"adjacent band rejection" (could be ADC aliasing, saturation, or
spurious mixing products or ?). If you're not using a bandpass
filter, you will certainly receive all kinds of spurious decodes.
Matt, WM9C, and I have been playing with RTL-SDR v3 and v4 for QRSS CW
(which sit adjacent to the WSPR channels) reception. He had a lot of
issues on 20m and eventually inserted a BPF in front of the RTL-SDR
(he has v4, which has better sensitivity at HF) which cleared things
up. My v3 is just deaf (a known deficiency, but I had a couple of
them on-hand) on 30m and sensitive to the nearby 5-kW MW (AM) BC
station when it switches pattern/power from night to day.

73,

--Ethan, K8GU.

> To view this discussion visit https://groups.google.com/d/msgid/hamsci/CACybYRW2T5RyRccb30TsC3YZwPx4THPJsGoHbBUC4V1zBGV83A%40mail.gmail.com.



--
http://www.k8gu.com/
Repair. Re-use. Re-purpose. Recycle.

[Darrel]

Hi Onno,
  First of all, very, very sorry for calling you "Pete".  That name somehow appeared in the "From" header in my email program and I wrongly assumed that was your name.  Very embarrassing, and again, my apologies.

  About your point on AM broadcast stations. The point is that WSPR is a form of FM, not AM.  At a harmonic, AM modulation survives. 

  In the WSPR modulation, for the duration of one symbol, only one frequency is transmitted.  The decoding algorithm has to decide which of 4 possible frequency offsets is the strongest and it chooses that one.  If (to keep things simple) the fundamental WSPR tone frequencies were 1.000000, 1.000001, 1.000002 or 1.000003 MHz, then the WSPR decoder would expect to find one of those possible frequencies and would decide which is the strongest one. 
  At the 2nd harmonic, the transmitted frequencies become 2.000000, 2.000002, 2.000004 or 2.000006 .  If you tune the WSPR decoder to the 2nd harmonic, it's now searching for tones only at 2.000000, 2.000001, 2.000002 or 2.000003 MHz.  Needless to say, it will be totally confused.  

  With AM modulation, for a single modulating audio tone the carrier varies as (1+p.cos(m))  where m is the modulation frequency and p is the modulation index (between 0 and 100 %, or between 0 and 1).
   The 2nd harmonic is generated by squaring the voltage of the fundamental.  The modulation term becomes squared too, giving you modulation of

which can be expanded to


The term in cos^2 is distortion, but the original modulation is there as 2.p.cos(m)

For low modulation index p, the term involve p^2.cos(m)^2 becomes vanishingly small, and we are left mainly with the original modulation involving just cos(m), the undistorted modulation.

  The summary is simply that WSPR is a form of FM, and at the second harmonic the original frequency modulation is completely corrupted.  With AM, the original modulation is distorted but still there.

   Does any of that make sense?  

73
   Darrel

To view this discussion visit https://groups.google.com/d/msgid/hamsci/CACybYRW2T5RyRccb30TsC3YZwPx4THPJsGoHbBUC4V1zBGV83A%40mail.gmail.com.

[Ryan Tolboom]

Very cool experiment! I don't know if it helps or if you've already seen it, but since you mention subtracting signals, I've done a bit of digging into exactly how the Steve Franke K9AN WSPR decoder (the one bundled with WSJT-X) does it: https://using.tech/posts/dissecting-wsprd/#_subtract_signals. It might be of some help.

73,

Ryan N2BP

[Darrel]

Hi Onno,

    One minor correction from me.  I'd said:

  The summary is simply that WSPR is a form of FM, and at the second harmonic the original frequency modulation is completely corrupted.  With AM, the original modulation is distorted but still there.

"Completely corrupted" was over stating the case.  Of course, with analog FM modulation, at the 2nd harmonic the frequency deviation will just be doubled.  That may screw up digital modulation and demodulation, although an analog FM signal will not necessarily be distorted at its harmonics.

73
   Darrel

[David Eckhardt]

I've often pondered the variation in reported SNR.  I live some 25 miles south of the WWV installation in Wellington, Colorado.  At night what there is no scatter off the ionosphere, all I receive is ground wave (GW).  That presents a pretty constant report of SNR.  However, during the daytime, SNR varies as much as 10 dB, certainly not what you report.  Watching the scatter on an o'scope, I can certainly understand this variation.  

But, consider my own signal reports from other more remote reporting receivers.  With relatively stable conditions, I see as much as 15 to 20 dB variation in reported SNRs.  Sure, the ionospheric scattering varies quite a bit!.  I detect that with my direct conversion receiver with the LO locked to GPS via a Leo Bodenar GPSDO.  

Your experiment was pretty interesting.  Most dummy loads, even my Bird kW load, radiates enough through the short coax that local decodes should be present.  The old Heathkit Cantenna and other like it radiated like crazy.  I once owned a Cantenna.   I even worked Denver from Fort Collins on 6-meters into a Cantenna.  However, there may have been enough radiated energy  from the open plate compartment of the amp than the Cantenna.  But there was no way I had line-of-site to Denver.   I never measured the "shielding effectiveness" of the Cantenna, but I know for a fact they radiate like crazy!

Using only 5 mW to the antenna I've had a number of local decodes and a few out to several hundred miles on WSPR on 80 and 160.   

What dummy load, coax type, and coax length did you use over your year-long experiment?   

Dave - WØLEV

--

Please follow the HamSCI Community Participation Guidelines at http://hamsci.org/hamsci-community-participation-guidelines.
---
You received this message because you are subscribed to the Google Groups "HamSCI" group.
To unsubscribe from this group and stop receiving emails from it, send an email to hamsci+un...@googlegroups.com.

To view this discussion visit https://groups.google.com/d/msgid/hamsci/b035d33d-9f0a-407c-be87-308e846ce32f%40gmail.com.

--

Dave - WØLEV