spurious emissions from kiwisdr on 144,010

Comments

• KM4YRI

  December 2016

  @rjschneider, dumb question, but how are you receiving anything at 144 MHz when the KiwiSDR is a 0-30 MHz receiver?  Even with harmonics, I don't see how anything from the 2m band could 'get in' to the 0-30 MHz of usable bandwidth.  

  
  

  Could you post screenshots?
• rjschneider

  December 2016

  i receive this emmissions not with the kiwisdr. I never said this.  I receive them with other equipment, for example a airspy, or with my icom R7000 receiver. I can post a screenshot later
  
• jks

  December 2016

  We've had a couple reports of this issue. But it is difficult because it's not necessarily a simple problem with the Kiwi but rather a more complex systems problem with cabling conduction and re-radiation issues.

  
  

  Your Kiwi might use a coax antenna cable that goes up to the roof near your other V/UHF antennas. The Kiwi cable might even travel in the same bundle of cables or at least parallel to the cables of your other antennas. Digital noise from the Kiwi or your Ethernet-based network can easily be conducted along the cables unless they are decoupled or proper grounding is in place. A perfect example of this is the Kiwi self-interference observed when an E-field probe (Mini-whip) fed with coax is used (there was just a thread about this topic). The digital noise rides up the other braid of the coax and is received by the Mini-whip.

  
  

  It would be very interesting to know how the 2m interference differs when changes are made to the Kiwi cabling and operation of the server software. First, try logging into the Beagle and typing "cdp" and "mst" to stop the Kiwi server from running. Does the interference decrease? Then it is related to the network traffic on the Ethernet or level of digital activity on the Kiwi FPGA or Beagle. Does it decrease then the Ethernet cable is disconnected? Then it is likely noise being coupled into the Kiwi system from your network in general: add toroid decoupling to the Ethernet cable attaching to the Kiwi.

  
  

  How does the interference change when the Kiwi antenna (and/or GPS antenna) is removed? This would tell you if Kiwi noise is being coupled into the power supply or Ethernet cables and re-radiated.

  
  

  How is the Kiwi grounded? Does your 5V power supply to the Kiwi provide a ground path to the power mains ground pin? If you also then ground to another source like an external ground rod or ground at another location then you are creating a huge ground loop that can cause re-radiation problems.

  
  
• rjschneider

  December 2016

  I tried several things.
  1. I added a clip-on-ferrite and some ferrite cores to the 5 V power cord. 
  2. I added a common mode choke  to the ethernet cable
  3. I disconnected the gps
  4. I disconnected the Antenna
  I opened a client session with Firefox on the kiwisdr. On my Airspy with sdr-console (first screenshot)
  Then i close Firefox. after about 30 to 60 sec the noisespectrum changes. (second screenshot)
  Then i log in the thw kiwisdr via ssh and enter cdp and then mst (you see the effect of cdp and mst on the third screenshot
  
  In a last try i started the kiwisdr, with only ethernet and power connected. opened one client via Firefox and then removed the ethernet cable. And even then with no possible ground loop ( becaus kiwisdrs only connection is power), i get the noise peak. (last screenshot, you see when i unplugged ethernet)
  
  
  
  
  Attachments:
  https://forum.kiwisdr.com/uploads/Uploader/60/6e5ce173f5d0ad8d1cdfc9a3f259b7.jpg
  https://forum.kiwisdr.com/uploads/Uploader/a8/514cfdc276299db57a3954b0a5404f.jpg
  https://forum.kiwisdr.com/uploads/Uploader/43/f02321d1a80711a5f4c2d13129c214.png
  https://forum.kiwisdr.com/uploads/Uploader/49/68ff5db3fa45f0cfe8b51b5b49e4e7.png
  https://forum.kiwisdr.com/uploads/Uploader/24/b8d98ebea75691ead55912ae2659f6.png
  https://forum.kiwisdr.com/uploads/Uploader/b5/9ee47b61d9c053baf7ccd62d93ed6b.png
• jks

  December 2016 edited December 2016

  Are you saying with only a decoupled 5V power cable to the Kiwi you get the 2m noise? (no other BB/Kiwi cables attached). Is this with the Kiwi server stopped?

  
  

  If so, there is a test I'd like you to do. It's a pain but if you could remove the Kiwi board and run the Beagle alone with only power connected I'd like to know the result. Be very careful when removing the Kiwi board gently rock the Kiwi board along its long length a little at a time unit the spring tension of the Beagle headers releases. If you bend the Kiwi board too much the ball joints of the FPGA will break and then all is lost.

  
  

  When re-attaching the Kiwi board be absolutely certain to put pin #1 of the Kiwi pins at the far end of the Beagle header. If you're off by one pin the wrong voltages will be applied and the FPGA will be destroyed. Look here if there are any questions: http://kiwisdr.com/quickstart/quickstart.pdf

  
  
• jks

  December 2016

  This is a list of all integer sub-multiples of 144.010 MHz and the known direct and derived clocks of the Beagle and Kiwi.

  So it looks like 3x the USB or SPI clocks could be near 144.010 or 6x the Beagle CPU main clock of 24 MHz. These are only some possibilities of course.
  

  
  

  
  
  
  
  
  
  

  sub-multiple	MHz		source, continuous unless otherwise noted
  	303		BB derived: HDMI?
  1	144.010		Interference
  	100.000		BB derived: Ethernet
  2	72.005000
  	66.6666		Kiwi oscillator: ADC/FPGA
  3	48.003333
  	48		BB derived: USB
  	48		Kiwi/BB derived: SPI clock (only when SPI active)
  4	36.002500
  5	28.802000
  	25		BB crystal: Ethernet PHY
  	24.576		BB crystal: HDMI, should be disabled
  6	24.001667
  	24		BB crystal: CPU clock
  7	20.572857
  8	18.001250
  	16.368		Kiwi oscillator: GPS clock
  9	16.001111
  10	14.401000
  11	13.091818
  12	12.000833
  13	11.077692
  14	10.286429
  15	9.600667
  16	9.000625
  17	8.471176
  18	8.000556
  19	7.579474
  20	7.200500
  21	6.857619
  22	6.545909
  23	6.261304
  24	6.000417
  25	5.760400
  26	5.538846
  27	5.333704
  28	5.143214
  29	4.965862
  30	4.800333
  31	4.645484
  32	4.500313
  33	4.363939
  34	4.235588
  35	4.114571
  36	4.000278
  37	3.892162
  38	3.789737
  39	3.692564
  40	3.600250
• jks

  December 2016 edited December 2016

  Sorry about the formatting of that last message. Not sure what happened.

  
  

  Anyway, I've had a third report now of strong 2m interference saying that it only happens when an Internet connection is made to the Kiwi.

  I'm beginning to wonder if the couple of centimeters of the 48 MHz SPI clock line that goes on the header between the Beagle and FPGA is a problem. That frequency can be changed with software, with some loss of performance. Also I think the strength of the drivers of the SPI signals can be lowered to decrease the slew rate and maybe attenuate any harmonics. I'll try some of these things out.

  
  

  I just bought a near-field antenna probe kit I can attach to my spectrum analyzer so I can go do some EMI hunting. It will take a little while to be delivered.

  
  
• rjschneider

  December 2016

  Yes i noticed that too, now as you say it. I bootet the kiwi up, and the interference got really strong at just the time i connected to kiwi the first time.
  I try this again later. If this is the case, does the BB standalone test make sense?
  I think the driver strenth and measurement with the near field probe is a good starting point.
  
  
  I also have a beaglebone black (the one with hdmi ) lying around. Perhaps it has something to do with the layout of the bb-green.
  
  
  
  
• jks

  January 2017

  Please see this new thread: http://forum.kiwisdr.com/discussion/discussion/497/update-on-the-2-meter-band-interference-issue