RFI from 3-8 MHz

On kiwisdr.w1euj.com, I've got this blanket of pulsing noise over these frequencies. I'm looking for the most effective ways to figure out the source.

Location: KiwiSDR setup is located in the barn, which is 100 feet from the house (buried power line to it) and 200 feet from the house and horse barn on the adjacent properties. This is the only thing in the barn that is powered.

Kiwi: The Kiwi is in a metal trashcan, with the power, Ethernet and antenna cables running out the rim. The antenna input has a Mini-Circuits SMA LPF for 30 MHz.

Antenna: A large wire Vertical Delta loop in the trees behind the barn (about 30 feet away) with 50 foot apex, feedpoint at the bottom corner. 4:1 balun at the feedpoint, running to LMR style cable along the ground for 30 feet to the entry point on the barn. Runs through a grounded Polyphaser arrestor, then the wall, then inside for 5 feet to an Array Solutions reciever protector, then 1 foot of cable to the inside of the trashcan.

Power: A linear power supply (2-prong AC connection and inside a plastic box), with the DC lines then wrapped (10 turns) on a large ferrite ring before then running a very short distance into the trashcan.

Ethernet: This is unique. I an running a pair of Ubiquiti Bullet M5 radios to link the barn to the house. They are connected to small yagi antennas mounted on the corner of the barn closest to the house - about 20 feet away from the Kiwi and connected to it via the CAT6 cable. The Bullet in the barn is powered from a linear power supply. Both the power and Ethernet lines are wrapped through several ferrite rings. 


  • Please post a picture with spectrum at a couple zoom levels.
  • edited January 2018
  • In AM, it sounds like a higher pitched version of driving over rumble strips.
  • Hi

    Remember that the Ubiquiti Bullet M5 have a buildin switchmode psu :-)


  • I have to think that the quickest experiment is to shut down the house for a moment, bring my IC-706 on a battery and see what 40 meters (the worst affected band) sounds like. If its the same then sucks to be me. If not then I've got some chance to locate and fix the issue.

  • Hi,

    We used some Ubiquiti kit at another WEB SDR site some time ago and had similar problems.

    From memory the Ubiquiti is powered over the Ethernet cable via a PoE adaptor. It does use internal DC-DC switching convertors, but they were fairly well screened.

    I think both the head unit and PoE adaptor had got RJ45 sockets with a metal screen connection. So you can use foil screened CAT5 cable if you are not already doing so. If not, try swapping over to it so that there is an electrical bond between the Head unit and PoE adaptor.

    The next problem was that the PoE adaptor didn't have a separate ground connection, so if your PSU is 'floating' (no third pin on the mains connector or no connection between the DC connector and a mains ground) there is no electrical bond between the PoE adaptor and ground. 

    It may be worth trying an electrical bond between the Ubiquiti kit and the antenna ground on your KiWi. This may (or may not) improve things so you may need to experiment a bit to get the best results.

    Ferrite chokes can work well, but they do need to be optimised for the frequency range in use. The type of ferrite and number of turns makes a huge difference to the level of performance you may be able to obtain.  Steve G3TXQ has a good guide to this on his website - http://www.karinya.net/g3txq/chokes/


    Martin - G8JNJ

  • I took a screenshot starting around 200 kHz (see below). The interrupted carriers spaced about 7 kHz apart look very similar to ones shown in the last image of the thread where someone had RFI from using Ethernet-over-powerline: http://forum.kiwisdr.com/discussion/902/dealing-with-excessive-noise-solved-ethernet-over-power-caused-noise#latest

    The RFI with the horizontal streaks sort of looks like what this guy had with a bad streetlight: http://forum.kiwisdr.com/discussion/951/wideband-pulsing-noise-noise-blanker-absence-i-m-in-trouble-nb-added-in-v1-152#latest


  • Thanks for the feedback. I've been running the Bullet POE using a lab power supply - I noticed a lot of issues with the noise with the original POE injector.

    I'm going to run a long cable out to the barn to eliminate the Bullets and see what changes.
  • Update: I've replaced the Bullet link with a CAT6 running (pinned to the ground with lawn staples) 200 feet from the house to the barn, all of the old RF link equipment was inactivated. Linear supply replaced with a 3-prong type with EMI filtering on the AC input and multiple choke types on the DC output. Noise is still there, pretty much unaffected.

    I've definitively eliminated our house as the cause, its down to the neighbors and local sources.
  • Hi,

    I've also had a listen and it does seem quite bad, especially around 6MHz.

    I notice you are using a delta loop antenna. Has it got any sort of Balun attached to it ?

    If not I'd suggest trying a decent 1:1 choke balun. Try some thin coax wound 17 times though a FT240-43 or FT240-31 sized ferrite core.

    Is the noise still present if you disconnect the antenna but just touch the screen of the antenna lead onto the outer of the KiWi SMA antenna connector ?

    If it's still present it could be some sort of ground loop between the antenna and DC & Network cables.

    As before try wrapping each cable through a ferrite core using the same type and number of turns as before.

    If you still can't find the source. Try making a small 'sniffer' loop by connecting the inner of a length of coax back to the screen of the coax so that it forms a loop with a few inches diameter. The loop should have a continuous screen around its circumference, but it should not be connected at the 'open' end where the inner protrudes and connects to the screen. Connect the other end up to your KiWi and pass the loop over various items of equipment, cables, cases and anything else that's in the vicinity.

    You should be able to see the noise and gradually (you may have to reduce the size of the sniffer loop) home in on the exact source.

    If this doesn't work build a small tuned loop and DF the source by walking around an taking bearings to triangulate the source of the noise.

    Here's the loop I use.

    More info here



    Martin - G8JNJ
  • I have a 4:1 balun on it now, but that could stand an upgrade. All other cables at the Kiwi end have been wrapped through that type ferrite core, pretty much as you described but a few less turns on the Ethernet. I don't get any noise by the shield-to-shield test you mentioned, its very quiet. The noise is also not present when the coax is disconnect at the antenna end FWIW.

    More chokes on order, but it looks like I'm on a hunt for this summer.

    Thanks Martin!
  • edited June 2018

    OK, it's quite possible that the 4:1 Balun is a or 'Voltage' balun (check to see how many cores it's wound on. If it's only one, and has DC continuity from primary to secondary, it will be), which will have very poor common mode rejection.

    To be honest even if it is a Guanella or 'Current' Balun wound on two (or more) cores, if it has a greater than 1:1 transformation ratio, the chances are that the common mode rejection will still be quite poor. This is because each core is effectively connected in parallel (from a common mode impedance perspective). So the greater the transformation ratio, the more cores end up being paralleled with the subsequent reduction in common mode impedance.

    The best option would possibly be to add an extra 1:1 choking Balun (coax wound on a suitable core) on the input of your existing 4:1 Balun.

    Use the correct type of core material and number of turns to maximise the choking impedance at the frequencies where the interference is at its worst.

    Steve, G3TXQ, has a useful design chart on his website http://www.karinya.net/g3txq/chokes/


    Martin - G8JNJ
  • A balun made to target the 3-8 MHz region (per Steve G3TXQ) was made and installed this evening. Also fabricated one for use at the barn end of the cable. Its seems like it made some difference, 80 meters a bit cleaner, but overall not significantly better. I have some other things to look into, but back to the hunt soon! Oh man, when I do fine this noise source and have it turned off this is going to be a fine setup!
  • Hi,

    OK that sounds promising :-)

    I'dd certainly try a 1:1 Balun on the antenna coax next.

    The great advantage of the KiWi is that it's got it's own built in spectrum analyser, so if you add a small directional antenna such as a loop, it's very easy to DF noise sources.

    Just use a very long length of coax between antenna and KiWi and a WiFi connected tablet, so that you can wander around and take bearings.


    Martin - G8JNJ
  • Yep, installed a 1:1 on the antenna coax as well, per same specs and installed at the same instance as the one on the loop. Agree on the approach with the wandering loop - happen to have a long length of RG-58 to work with.

    A couple other tests come to mind - one is that I've not been able to run a test where the Kiwi was COMPLETELY isolated and on some sort of battery power. Anybody have a known method to make clean 5 VDC from battery for the Kiwi?
  • Perhaps a 12v Gel cell and an analogue 5v 3A regulator ? Not one of the modern switchers or LDO designs but something like an LT323AT or similar ?


    Martin - G8JNJ
  • Ordered a high current linear regulator off Alibaba, should be here soon enough.

    Added #43 toroids to all the things, and even one more at the antenna end. Again, it seems like improvement but not the solution. Getting a friend's loop to help diagnose this. In the meantime, there was a development on how the noise is acting - its gone from constant to pulsing. I can see the promised land through the gaps.
  • Source found - the neighbor with the barn has an Invisible Fence system (Model ICT-801 transmitter) on the same wall of the barn that my loop is parallel to. I've read that installing a Corcom 10VK6 filter on the transmitter output should knock this interference down. The neighbor is willing to help - so I'll be looking for an installer that can put that filter on.
  • lucky to have a QRM owner that will work with you!
  • A good find.

    It may be worth trying a different power supply in place of the existing dog fence 'wall wart' before trying anything more complicated.

    A lot of these supplies have little or no RFI suppression components fitted and may be the source of the noise which is being re-radiated by the 'fence' cables.


    Martin - G8JNJ
  • I'll run that by the installer. I really am lucky to get this far with the neighbor and they do depend on it for their dogs/horses safety so I am going cautiously and backed up with whatever the installer says or is OK doing.

    My last post on the subject will be how it turns out with whatever gets fixed/fitted and the resulting waterfall, wrap it all up with a bow for future sufferers.
  • Bad news on this issue. The 'source' of the interference is not the Invisible Fence; rather its something located along the same wall and incidentally coupling into the loop (via the lightning protector) but is primarily radiating from something in the same vicinity as the fence equipment. Unplugging the loop and equipment from the AC mains made minimal difference. Of course, the AC adapter swap and filter fit on the lines make no difference either.

    My neighbor was VERY helpful in working with me to turn this-and-that off (overhead LED lights, pumps, WiFi, etc) but no source was found. Its a barn, ain't much in there - its a mystery what else is tucked away in the rafters that she wouldn't otherwise know/have control of. Whatever it is, digging around further would take more patience that I'd be able to muster myself, and the fix likely more costly than the option of relocating my antenna and equipment.

    I left it where I would get back to her if it disappeared and ask what changed. Until then, the noise stays till late fall when I get the change to relocate the antenna. Its a shame - those trees are perfect for the job, and the new setup isn't likely to be as good.
  • Could you turn on the spectrum so we can see the shape of it? Even screen dump just the spectrum, when pulses on and off.
    Are there any (battery backup) safety lights there? Freezers?
  • edited June 2019
    This issue was mostly fixed by changing to a Wellbrook loop that is 400 feet away from the original location of the delta loop. The new loop is rotated to null out that interference; its not perfect - the VLF range is covered with some incredible buzzing - but the HF is working out very nicely.

    I had to make the change anyway, ice storms brought down a pine tree, destroying the antenna and damaging the barn where I had the Kiwi.
  • Sorry to hear about the barn/antenna, but on the other hand glad to hear you can put a Wellbrook that far away from the source.
Sign In or Register to comment.