Large masked band segments
Paul, VK3KHZ, from PK's Loop Antennas, brought to my attention the http://plonsk.proxy.kiwisdr.com:8073 Kiwi that has these very wide masked band segments (AKA notched, blacklisted).
How was this done? It turns out it's a feature/bug of the existing masked frequency option available with dx labels. Normally when you setup a dx label and set the "type" menu item to "masked" the width of the masked frequency area is the passband of the label's mode (AM, CW, ...) But the dx label panel also contains a "passband" field where you can specify a custom passband which overrides the default.
Well, it turns out you can specify a ridiculously wide passband too. The audio code will clamp an overly wide passband to 12 or 20.25 kHz depending on the Kiwi FPGA configuration, but the masked frequency code will happily take your crazy passband width and block a huge segment of spectrum. So creating a dx label at 1120 kHz with the passband field set to "1180k" would mask the entire NA 530 - 1710 kHz AM BCB (530 = 1120 - 1180/2, 1710 = 1120 + 1180/2). The passband field also accepts a "lo, high" specification (e.g. "300, 2700" for a USB passband). So you could specify a large asymmetrical masked segment as well (label at 6 MHz, passband "-10M, 20M" to get a masked segment from 5 - 8 Mhz). Note the usual 'k' and 'M' suffixes can be used for kHz and MHz respectively.
Rather than specify the middle of the masked range and the width you can also specify the starting frequency and width which might be more convenient. Say you want to mask 26 - 27 MHz. Create a masked label at 26 MHz and set the passband field to "0,1M".
Pretty cool. Thanks Paul!
How was this done? It turns out it's a feature/bug of the existing masked frequency option available with dx labels. Normally when you setup a dx label and set the "type" menu item to "masked" the width of the masked frequency area is the passband of the label's mode (AM, CW, ...) But the dx label panel also contains a "passband" field where you can specify a custom passband which overrides the default.
Well, it turns out you can specify a ridiculously wide passband too. The audio code will clamp an overly wide passband to 12 or 20.25 kHz depending on the Kiwi FPGA configuration, but the masked frequency code will happily take your crazy passband width and block a huge segment of spectrum. So creating a dx label at 1120 kHz with the passband field set to "1180k" would mask the entire NA 530 - 1710 kHz AM BCB (530 = 1120 - 1180/2, 1710 = 1120 + 1180/2). The passband field also accepts a "lo, high" specification (e.g. "300, 2700" for a USB passband). So you could specify a large asymmetrical masked segment as well (label at 6 MHz, passband "-10M, 20M" to get a masked segment from 5 - 8 Mhz). Note the usual 'k' and 'M' suffixes can be used for kHz and MHz respectively.
Rather than specify the middle of the masked range and the width you can also specify the starting frequency and width which might be more convenient. Say you want to mask 26 - 27 MHz. Create a masked label at 26 MHz and set the passband field to "0,1M".
Pretty cool. Thanks Paul!
Comments
Looks like rx.linkfanel has redone his algorithm so as not to reward those stations with a ridiculously high SNR due to the large and/or numerous masked sections.
I don't know if anything has changed in the meantime, but the co-sited Polish KiWi's with largely masked frequency ranges, and moderately low signal levels are back at the top of the list again.
http://plonsk5.proxy.kiwisdr.com:8073/
Regards,
Martin
My impression is the rx.linkfanel guy doesn't read this forum..
Yeh I know, I have posted a new issue on his Github repository, but I doubt it will result in any action being taken.
It's not a big issue, I just like legitimately being at number one, for most of the time :-)
Regards,
Martin.
Well, especially since you worked so hard for it (antenna and frontend optimizations, and all the knowledge you've shared on the forum).
Hello, :)))
Attention. If I add gain... I will move to the front, to the top of the list. Ahaha ;)
Best regards, Philippe
What do you mean exactly? Adding gain as in a preamp? That won't change the measured SNR because the noise floor will rise to offset the increased signal strengths (i.e. no free lunch).
My receiving antenna is too powerful. With the Kiwi 2, I have to reduce the gain by -7 dB. if I don't want to saturate the converter. Even with 250 meters of coaxial cable between the vertical antenna and the Kiwi receiver.
The antenna noise is lower than that of the Kiwi (measured in the absence of propagation). If I reduce by -4dB, it's better for the snr, but it saturates at certain times of the day. So I prefer to have 3dB security, so I set it to -7dB.
Currently, I am testing at -4 dB. (and I left at the top of the list)
At the bottom of the vertical antenna I installed a HiZ to LoZ transformer. I added a buffer to obtain the same output impedance for all frequencies (75 Ohms). To send the signal through 250 meters of coaxial cable (75 Ohms), the impedance must be very stable.
The buffer, installed at the base of the antenna, also compensates for the losses of the long coaxial cable (of a few dB)
Attention. If I add gain... I will move to the front, to the top of the list. Ahaha ;)
Best regards, Philippe
Excellent, anything to improve the deployed KiWi's performance is good news.
So many have poor antennas and so much noise to be almost unusable, one of mine (Kernow) included.
What's your KiWi URL ?
Good luck,
Martin
Well, especially since you worked so hard for it (antenna and frontend optimizations, and all the knowledge you've shared on the forum).
Thanks John,
But you forgot to mention all the hassle I have caused too...
The KiWi is an excellent product, and it's definitely well up in the top five radio related purchases I've made over the past 50+ years.
I very much appreciate your effort, ongoing support and enthusiasm, which is a rare thing these days.
Regards,
Martin
The KiWi is an excellent product
Indeed. And it's even much more than at the time of purchase, thanks to Johns restless work. I often say with a wink that - alongside my e-bike and my camping bus - the kiwi is one of my most important purchases in the last few decades.
What's your KiWi URL ? Good luck, Martin
Dear Martin,
Thank you very much. It is : http://sdr.autreradioautreculture.com:8074
QTH : Trémolat, Sud-Ouest, Périgord, France
Best regards, Philippe
This rx.linkfanel list has been showing some miracles for a long time and, as I see it, creates an artificial, unnecessary rat race in which I do not intend to take part, if it is to ensure a good night's sleep for you, I have turned off SNR measurements in my receivers, and if necessary, they will disappear from the list public. I was forced to mask what is outside the bands due to the geopolitical situation and persistent bots/sniffers looking for sensations due to the proximity of everything. There was a situation where normal listeners, and even myself, being away from home, had no chance to get here because all 16 or 20 channels were sucked by leeches from buzzers and the rest of that crap, as can be seen in the screenshot from the counter from the day it started.
Regards
Hi Artur,
Please don't turn off your SNR measurements for my sake. There are plenty of others that are skewed because of masked frequency ranges. I was not specifically singling yours out, it was just that they were the best example of the underlying problem.
I absolutely understand why you have masked the bands, and in your situation I would have done exactly the same. It is not your fault or problem that it messes up the SNR measurements.
I think it is useful to have some sort of guidance as to how well individual KiWi's perform, especially when performing TDoA plots, as it quickly allows you to ignore the awful ones that have far too much interference to be usable.
However, the current scoring method can only be used as a guide at best, and although it is prone to errors it is better than nothing.
We have six or seven KiWi's at the Weston site, and despite them all being connected to the same RF splitter, they all produce different scores, depending on when the sample was taken.
Some time ago, I suggested a different SNR averaging method to John, in order to try and improve the accuracy. He implemented it, but it just made the reporting even less accurate, so I asked him to revert to the previous method.
It seems that what can quickly be determined by a human looking at the overall waterfall display, is very difficult to define for analysis by machine.
Regards,
Martin
Dear Martin,
Thank you very much. It is : http://sdr.autreradioautreculture.com:8074
QTH : Trémolat, Sud-Ouest, Périgord, France
Best regards, Philippe
Hi Philippe,
Now it has gone dark, your KiWi sdr score is very impressive.
I think this is because your HF noise floor is much lower than mine (less sensitive).
Your impressive antenna is clearly working well.
Regards,
Martin
I'm having to repeat myself yet again: The Kiwi's SNR measurement code ignores masked band segments. They do not skew the result.
rx.linkfanel.net connects to each Kiwi and does his own SNR measurement using an algorithm I don't know anything about. Maybe someone else does.
Thank you very much Martin. I try to do the best.
For the antenna, I found this location which seemed appropriate to me: a small plateau of limestone earth, slightly rounded, well clear towards the 4 cardinal points, far from the first houses. I had to deforest and clean one hectare (the wood had been devastated by the storm of 1999), then create a 250 meter trench for the coaxial cable to Kiwi.
Best regards, Philippe
Hi John,
Ah OK, I think I've seriously misunderstood, how the SNR values are derived.
I thought that rx.linkfanel.net simply took the internal values from the /snr URL extension.
I think I've become confused by the internal and external headings on the SNR columns.
I assumed the internal values were generated by the KiWi itself, as they always seem to match the values displayed by the individual KiWi's.
Regards,
Martin
I hadn't looked at the rx.linkfanel.net/snr site in a long time until now. I didn't realize he shows "internal and "external" values. I agree that the "internal" ones look like the ones from the Kiwi /snr URL.
Hi John,
Ah OK, I thought we had discussed this problem a while ago, but I think we must both of misunderstood what each other was trying to convey at the time.
Never mind, no great concern.
Regards,
Martin
Although I think public Kiwis with ham-band-only frequency masking is kind of annoying, I've had several requests to make this easier to setup. So in the documentation describing the feature there is a .json or .csv file you can use with the admin/DX tab "Import JSON/CSV" button to do so: http://kiwisdr.com/quickstart/index.html#id-user-marker-masked