Date   

Re: HB9VQQ testing Kiwi SDR

Rob Robinett
 

Hi Bruce,

In case you and others in this group missed my posts' to Roland's WSPR Beacon Project Telegram group on this subject,  I'll copy those remarks here,  but the subject of how to construct high performance (and what that means needs to be defined), easy to deploy and cost effective WSPR receive site is far more complex than can be covered in one post, but here is my perspective:

Roland wants to deploy a wspr rx system which mirrors his tx system and feels that to do that the cost of each site needs to be $200 or less.  At $400 for a Kiwi + Pi4, on cost alone that system doesn't qualify, so he could have stopped there and settled on the Airspy + Pi and compared it against other single channel SDRs.  However he also wants to know how well his low cost system compares in performance to the Kiwi + Pi used by many of the top spotting sites.  But in making that comparison he put the Kiwi system at a severe and undocumented disadvantage by not adding the AM and LNA requried to properly utilize the Kiwi which is used at those top sites.  

As one of many examples of the disadvantages potentially suffered by the Kiwi in such a comparison, without an AM blocking filter the Kiwi (and and wide band SDR) is likely to be frequently going into overload.  The Kiwi GUI displays those events with the flash of a red OV next to the S-meter, but if you are comparing over a 24 hour period there is no place to find out on the Kiwi the number and severity of those events.  In order to document this problem, wsprdaemon records the frequency and severity of OV events to a log file which the operator should examine to see if there is sufficient filtering to minimize overload events.

As a second example, if there is no blocking filter and no overload events, there is evidence that the Kiwi system is not receiving a strong enough RF signal and it needs a  LNA ahead of it.

While a single channel SDR probably doesn't need blocking filter(s), it too can suffer from impairments introduced by its installation environment, and those are much harder to identify than when using a wideband SDR.

People write whole books on the subject of building a good receive station and I worry that the effort to deploy low-cost plug-and-play rx systems will result in many 'deaf' stations.  Certainly by listening only to the bands and wspr cycles when/where WSPR Beacon stations are transmitting, many potentially interesting spots from non-Project stations will be missed.  So I think it is a shame to invest a lot of time and money in rx stations which could be far more useful to the wider community.

But Roland deserves credit for funding this and I hope his rx stations do make a contribution to our understanding of radio propagation.  But his tests don't prove the Airspy is a 'better' WSPR receiver than a Kiwi.  The Airspy just works better in the way he has been testing and the way he wants to deploy his systems.

73,

Rob



==========

"I understand your application requirements are different, but for many users the Kiwis are cheaper and better solutions when they are properly deployed.  My point here is to clarify to others that your comparison tests are for your specific application:  band hopping with no external filters or amplifiers.

Rob Robinett AI6VN, [Oct 10, 2021 at 1:35:29 PM]:

Given your cost and single band at a time requirements, then the Airspy seems the better choice.  But the Kiwi is not 'bad', it has different features and testing by many knowledgeable hams has shown how to optimize its performance to match that of other top quality receivers like the Airspy.  So I don't think it is appropriate to characterize the $40/rx channel Kiwi as 'worse' than' than the $170/channel Airspy when your test criteria precludes optimizing the Kiwi's RF feed.   I think you have identified the Airspy as the right SDR for your application, but that doesn't mean it is the 'best' SDR for all applications"



On Wed, Oct 13, 2021 at 1:49 PM Bruce KX4AZ <bruce@...> wrote:
Rolf,
I wouldn't be at all surprised if the Airspy HF+ Discovery outperforms the KiwiSDR for WSPR spot counts, give the different architecture and narrower bandwidth on the Airspy.  In my informal comparisons of the two (attached to a simple y-splitter), the Airspy had a measurably higher S/N ratio on the weakest signals.  It appears that modest signal amplification (very carefully applied) can compensate for the difference.  I'll leave it to the experts on this forum to (like KA7OEI) to detail the pluses/minuses of amplification.

I just started setting up a new Pi4 today and hope to get the wsprdaemon running with the KiwiSDR one of these days.

I also wanted thank you for your post, calling my attention to yet another groups.io that I'll need to join!  Might be a good place to discuss my current 10 microwatt WSPR experiments.
73,
Bruce KX4AZ



--
Rob Robinett
AI6VN
mobile: +1 650 218 8896


Re: HB9VQQ testing Kiwi SDR

Jim Lill
 

From my experience on both my own and KD2OM kiwi,  above 10 MHz or so, the Kiwi may need a bit on gain in the path depending on what antenna is used.  You may note that both Steve KD2OM and Tom WA2TP both rank high in wspr reporters and use kiwi.

-Jim

WA2ZKD

On 10/13/21 4:49 PM, Bruce KX4AZ wrote:
Rolf,
I wouldn't be at all surprised if the Airspy HF+ Discovery outperforms the KiwiSDR for WSPR spot counts, give the different architecture and narrower bandwidth on the Airspy.  In my informal comparisons of the two (attached to a simple y-splitter), the Airspy had a measurably higher S/N ratio on the weakest signals.  It appears that modest signal amplification (very carefully applied) can compensate for the difference.  I'll leave it to the experts on this forum to (like KA7OEI) to detail the pluses/minuses of amplification.

I just started setting up a new Pi4 today and hope to get the wsprdaemon running with the KiwiSDR one of these days.

I also wanted thank you for your post, calling my attention to yet another groups.io that I'll need to join!  Might be a good place to discuss my current 10 microwatt WSPR experiments.
73,
Bruce KX4AZ


Re: HB9VQQ testing Kiwi SDR

Bruce KX4AZ
 

Rolf,
I wouldn't be at all surprised if the Airspy HF+ Discovery outperforms the KiwiSDR for WSPR spot counts, give the different architecture and narrower bandwidth on the Airspy.  In my informal comparisons of the two (attached to a simple y-splitter), the Airspy had a measurably higher S/N ratio on the weakest signals.  It appears that modest signal amplification (very carefully applied) can compensate for the difference.  I'll leave it to the experts on this forum to (like KA7OEI) to detail the pluses/minuses of amplification.

I just started setting up a new Pi4 today and hope to get the wsprdaemon running with the KiwiSDR one of these days.

I also wanted thank you for your post, calling my attention to yet another groups.io that I'll need to join!  Might be a good place to discuss my current 10 microwatt WSPR experiments.
73,
Bruce KX4AZ


HB9VQQ testing Kiwi SDR

Rolf Ekstrand
 

Greetings y'all 

HB9VQQ is presently performing a test of  Kiwi with Rpi4 running WD  against  an Airspy HF Discovery running Wsjtx with Rpi4   on the same antenna through a splitter      https://groups.io/g/wsprbeacon/topics

It looks like the Airspy set up  at times outperform the  Kiwi, but as things usually goes, sometimes things can be deceiving. It is interesting though.   Comments anybody? 

73 Rolf K9DZT 



Re: Web Noise Plots

KD2OM
 

Why do the colors of the lines change? The lines from my three receivers change, it seems to vary based on the noise level.

73
Steve KD2OM


Re: Audio input decoding

Martin
 

Hello Rob,

that is great to hear!

A few more details on my setup:
I am using a QCX connected to a very cheap USB soundcard connected to a raspberry pi 3B.
The loopback device is just software created with the following command:
sudo modprobe snd_aloop
The result of this is a virtual device and data is copied from one end (DEV=0) to the other (DEV=1) and also in reverse direction. So I would call this a virtual audio channel but I am not sure if this answered your question.
I am no experienced programmer and found the topic of audio devices under linux very confusing. So I just tinkered around until I got it working finally.
However, one end of this loopback device is the sink for the GNU radio script and the other is the source for wsprdaemon.

I am happy to support by testing something or providing more details.

Best regards,
Martin


Re: Audio input decoding

Rob Robinett
 

Hi Martin,

Your use of the AUDIO input feature has stimulated me to raise its priority in debugging of WD 3.0.
Also, have you tried using a virtual audio channel as input to WD?

Thanks,

Rob

On Fri, Oct 8, 2021 at 12:23 PM Martin <martin.classen@...> wrote:
Great, thank you for providing and maintaining wsprdaemon!
It is a very nice standalone solution.



--
Rob Robinett
AI6VN
mobile: +1 650 218 8896


Re: Audio input decoding

Martin
 

Great, thank you for providing and maintaining wsprdaemon!
It is a very nice standalone solution.


Re: Audio input decoding

Rob Robinett
 

Hi Martin,

Thanks for that welcome report.  In WD 3.0 I will study how to incorporate your fixes.

Rob


Re: Audio input decoding

Martin
 

Hello,
 
baseband audio is working for me, however, two things needed tweaking beforehand:
 
1. the name of my soundcard is "plughw:CARD=Loopback,DEV=1" which does not fit into the naming scheme of wsprdaemon
2. the recorded audio was stereo per default but wsprd requires a mono recording
 
The sox option for mono recording (--channels 1) is specified in the function rtl_daemon but is missing in audio_recording_daemon.
I fixed this together with my souncard name directly in wsprdaemon.sh.
 
By the way - the background for my soundcard loopback device is the following:
My transceiver is a QCX (an excellent kit from QRP Labs) which has an audio center frequency of 700Hz. To get the 1500Hz required by wsprd, I am using a GNU Radio script to do the frequency shift of 800Hz and this script does the handover via the loopback device.
 
Best regards,
Martin


Re: Radio "limbo"

Bruce KX4AZ
 

OK, after 24 hours run time at -20 dBm output, the clear "winner" was W4HOD in Alabama (USA), spotting me twice in 24 hours.  My setup is still running, and at 1742 UTC today another station in the same area "snuck in" with a spot - WA4CQG, at the same time as another spot from W4HOD.  I have exchanged emails with the group of hams in that area to ascertain what is so special that's going on, since several stations in that area have been spotting me during my experiments throughout this week, all on 20m.  In another words, it's not just a one time event like a meteor passing by etc.  What they have in common is using beam antennas, with the one for W4HOD pointed toward Europe, so roughly in my direction.  Also, W4HOD is using an Airspy HF+, a very sensitive (albeit narrowband) SDR.  So those directional antennas may be a factor...although to some extent that takes them out of my notion of a WSPR "superstation", open to all directions.  Nevertheless, 10 microwatts is (for me) a new spotting record, I believe.

I'll keep the -20dBm level going for now to see if any other spotters emerge outside of Alabama.  And it's time for me to make some RF output power measurements to keep things accurate.

Finally, since this group is about wsprdaemon, and not necessarily my low power WSPR experiments, I will say that I await the arrival of a Pi4 so that I can commence running it with the KiwiSDR I recently set up.


Re: Radio "limbo"

Bruce KX4AZ
 

Ouch, I hate to bombard this group yet again with a post, but my UTC timing was off off today too!  I commenced transmitting at 10 microwatts 1506 UTC today (October 6).  Promise this will be the last correction.


Re: Radio "limbo"

Bruce KX4AZ
 

Correction and an oops...I hit the wrong button on the step attenuator when changing power a few minutes before 1600 UTC, and so had the power running 10 dB too high, at 0.1 mW, and K9AN "snuck in" there with a spot at 1058.  So the real -20 dBm output level actually commenced at 1606 UTC today (6 October).


Re: Radio "limbo"

Bruce KX4AZ
 

After nearly 24 hours running at 50 microwatts (-13 dBm), there were still 8 stations who heard me, though the total spot count and geographical diversity is narrowing.  And some stations were hanging by a thread, with a just single spot for the entire day.  Interestingly there is a trio of Alabama spotters, all in the same general area that are hanging in there too.

So it's time to lower the bar a bit further.  As of 1600 UTC today (6 October 2021) I've reduced the power output to 10 microwatts (-20 dBm!).  Stay tuned for further developments.  


Re: Radio "limbo"

@OE3GBB
 

Hi Bruce,

well, we sure have to believe the statements on power sent :-). I myself on TX am using the WSPRlight flexi and I checked the power with a spectrum analyser. The settings seem to be about correct. I did not correct for filter and koax or antenna gain.

On RX I am using a raspsdr in 4 channel mode for public websdr (see OE3XOE at https://www.receiverbook.de/)  and a flydog-sdr with raspPi4 and wsprdaemon as a WSPR reporter on 12 bands. I am constantly about the no. 15 of the best reporters  worldwide. http://www.wa2zkd.net:8088/rank.html

If you only use a raspsdr on several bands for reporting WSPR you will not be able to decode all the signals. I now have my two most busy bands decoded on the raspsdr and 10 bands on the wsprdaemon.

73 de Gerhard OE3GBB

 


Am 05.10.2021 21:00, schrieb Bruce KX4AZ:

On Tue, Oct 5, 2021 at 01:40 PM, @OE3GBB wrote:

For info, just checked WSPR:

I could receive at OE3XOE several DX stations using powers of 5 and 10 mW:


Gerhard,
Thanks for posting that - it's always nice to see others testing the low end of WSPR RF output.  When I operate as a spotter, now and then I'll send an email to stations that are declaring their power to be 10 mW or less....just to satisfy myself that they haven't accidentally mis-stated it.  For myself I verify the output power now and then, with either a spectrum analyzer, or the RF voltage measured across a dummy load using an oscilloscope.  That's something I will definitely need to do for my current round of experiments, once I find the threshold power setting that yields just one spotter station in a 24 hour period. Just like to have that little extra accuracy beyond what the step attenuator tells me.

Separately I will embarrass myself a bit here by admitting I have not yet used wsprdaemon for spotting.  I recently set up a new KiwiSDR (grid locator EN74), and once I get another Raspberry Pi I hope to give it a try.

73,
Bruce


Re: Radio "limbo"

Bruce KX4AZ
 

On Tue, Oct 5, 2021 at 02:50 PM, Jim Lill wrote:

If you did a NEC model of your EFHW (or most any EFHW) you'd be shocked by how lossy they usually are.  Knowing that loss, band per band, you could further reduce your actual ERP.

Jim,
Yes, no argument there, and thanks for posting the link with the article on EFHW characteristics.  I just find it fascinating how such a small power going into a temporary antenna setup like this can still be spotted.
Bruce


Re: Radio "limbo"

Bruce KX4AZ
 

On Tue, Oct 5, 2021 at 01:40 PM, @OE3GBB wrote:

For info, just checked WSPR:

I could receive at OE3XOE several DX stations using powers of 5 and 10 mW:

 

Gerhard,
Thanks for posting that - it's always nice to see others testing the low end of WSPR RF output.  When I operate as a spotter, now and then I'll send an email to stations that are declaring their power to be 10 mW or less....just to satisfy myself that they haven't accidentally mis-stated it.  For myself I verify the output power now and then, with either a spectrum analyzer, or the RF voltage measured across a dummy load using an oscilloscope.  That's something I will definitely need to do for my current round of experiments, once I find the threshold power setting that yields just one spotter station in a 24 hour period. Just like to have that little extra accuracy beyond what the step attenuator tells me.

Separately I will embarrass myself a bit here by admitting I have not yet used wsprdaemon for spotting.  I recently set up a new KiwiSDR (grid locator EN74), and once I get another Raspberry Pi I hope to give it a try.

73,
Bruce


Re: Radio "limbo"

Jim Lill
 

On 10/5/21 2:49 PM, Jim Lill wrote:

If you did a NEC model of your EFHW (or most any EFHW) you'd be shocked by how lossy they usually are.  Knowing that loss, band per band, you could further reduce your actual ERP.

On 10/5/21 2:45 PM, Bruce KX4AZ wrote:
On Tue, Oct 5, 2021 at 02:09 PM, Jim Lill wrote:

what type antenna do you use on 40M?

For transmitting I am using an EFHW that covers 40-10 meters, so about 65 feet long, straight horizontal line at about 12 feet above the ground, the ends pointed roughly NW/SE directions.

Since I reduced the output power to -13 dBm (50 microwatts) earlier today I have been spotted by 4-5 stations.


Re: Radio "limbo"

Jim Lill
 

If you did a NEC model of your EFHW (or most any EFHW) you'd be shocked by how lossy they usually are.  Knowing that loss, band per band, you could further reduce your actual ERP.

On 10/5/21 2:45 PM, Bruce KX4AZ wrote:
On Tue, Oct 5, 2021 at 02:09 PM, Jim Lill wrote:

what type antenna do you use on 40M?

For transmitting I am using an EFHW that covers 40-10 meters, so about 65 feet long, straight horizontal line at about 12 feet above the ground, the ends pointed roughly NW/SE directions.

Since I reduced the output power to -13 dBm (50 microwatts) earlier today I have been spotted by 4-5 stations.


Re: Radio "limbo"

Bruce KX4AZ
 

On Tue, Oct 5, 2021 at 02:09 PM, Jim Lill wrote:

what type antenna do you use on 40M?

For transmitting I am using an EFHW that covers 40-10 meters, so about 65 feet long, straight horizontal line at about 12 feet above the ground, the ends pointed roughly NW/SE directions.

Since I reduced the output power to -13 dBm (50 microwatts) earlier today I have been spotted by 4-5 stations.

381 - 400 of 856