I realise there have been many great posts on hunting RFI, i've got pretty close to tracking mine down, but would appreciate some insights from the ham.se experts :)

I've done quite a lot of work to hunt this down. The background - using a FlexRadio 6000 series inside a computer rack, radio is grounded to the rack via a short ground wire #6 awg (~40cm), rack is grounded currently to AC ground (work to be done in this area..)

Antenna is a K4KIO Hexbeam relatively low, around 20-feet above a flat metal roof on an insulated tripod, fed with RG8 around 25M, with the supplied Fairrite clamps on the RG8 at the antenna feedpoint. Ferrites (clamps) are on all cables, with the dc power-cable wrapped around a ferrite-ring around 4 or 5 turns at the radio end. An ethernet-switch (Ubiquiti) is outside the rack, and connected via a CAT8 cable (0.5M) to the FlexRadio via a narrow opening in the rack.

Using a switch-mode 40A psu (Manson) that has pretty good filtering within, and disconnecting the antenna, the RFI completely disappears, so i'm fairly sure everything around the radio itself is RF quiet.

Every 61khz precisely i have solid bars on 20M and 15M (40M is clear), that look like this:

15M screenshot

I've used a vertical in the past and the bars were less obvious, so i'd assume this RFI is horizontally polarised. There is an Ethernet cable running under the antenna with PoE (indoors), but i've disconnected the cable and the bars are still there. There are ferrite clamps on each endpoint of that particular cable.

I've gone RFI hunting with a laptop and SDRPlay RSP1 and a random 800mhz yagi (i happened to have) and tried both orientations horizontal/vertical but the SDRPlay waterfall was fairly clear.

My belief is Ethernet interference, not from my equipment, but a neighbour. Wether it's home-internet, or perhaps a solar-inverter's IP interface, but that's just a guess.

Having 61khz bars around 14-21mhz is the signature of this particular RFI.

Apart from this issue, it's pretty RFI clean. Was wondering if anyone has any advice given the findings to date?


  • 1
    $\begingroup$ do you mean we should use cat6 instead, or shielded cat5e? $\endgroup$ – hjf Aug 10 at 11:37
  • 3
    $\begingroup$ Your hypothesis that your RFI is Ethernet-based makes sense. Perhaps the best thing you could do to confirm it would be to do some fox hunting and find the geographic origin of the RFI. If the noise is coming from a neighbor, maybe the neighbor would be kind enough to cooperate with you by flipping circuit breakers and such so you could positively confirm what's going on. $\endgroup$ – rclocher3 Aug 10 at 14:13
  • 2
    $\begingroup$ @t252 If you've solved the problem, you should post your own solution as an answer and accept your own answer, rather than adding a comment and leaving an open question. $\endgroup$ – 比尔盖子 Aug 13 at 6:11
  • $\begingroup$ @hjf I'd say if you do run unshielded (UTP), from an RFI point-of-view, it's best to run low speed, but with properly shielded (and grounded/terminated) cables then it should be okay to run the fastest the equipment can support. Cat5e STP would probably be minimum, up to Cat8 (much more shielding). $\endgroup$ – t252 Aug 17 at 12:43

Further research has revealed the origin of the 61 kHz spacing between the RFI carriers.

100BASE-TX uses a 125 MHz clock, which is divided by 2047 in a scrambling circuit.

This is intended, ironically enough, to reduce EMI by spreading the signal across a wide portion of the spectrum, rather than generating very strong peaks at frequencies such as 31.25 MHz (125 MHz / 4) which would otherwise be generated by the MLT-3 coding scheme used. 125 MHz / 2047 = ~61.064973 kHz, explains the spacing of the offending signals.

Additionally, I have found similar carriers offset by 1/2 of the expected 61.06 kHz spacing in certain portions of the spectrum. This appears to be particularly pronounced in the upper HF and VHF region.

UTP Cat 5 or Cat 5e should absolutely never be used for anything higher than 10 Mbps Ethernet, especially in an environment where any type of HF or VHF receiver will be used.

Note that gigabit Ethernet generates a similar overall spectral profile, though with different modulation characteristics; it appears to generate more of a broadband hash than discrete carriers, but still causes interference.

Take a handheld radio with HF receive coverage and a VHF/UHF rubber duck antenna, and tune it to 14.275 MHz AM. It may be necessary to open the squelch on the radio. Hold the antenna next to an Ethernet cable with an active 100 Mbps connection on it, and there is virtually guaranteed to be a strong signal present.

Move the antenna along (and perpendicular to) the cable slowly, and the signal will increase and decrease in strength several times per inch of cable. This is literally a result of each individual twist in the twisted pair converting some of the differential signal into common mode radiation. The worst offenders can be heard for some distance away from the cable, even with such an intentionally insensitive receive setup.

Note that shielded Cat 5 or 5e (if properly terminated), as well as quality unshielded Cat 6, do not generate nearly as much interference.

I recently discovered that even some unshielded Cat 6 cable is capable of radiating considerable RFI. As already dictated by common sense anyway, shielded cable is simply the only way to go to assure a clean installation.

Credit to KA3RCS's work.

| improve this answer | |

finding rfi source: you must use a battery operated radio, sw & bcast bands. usually switching supplies are at fault. cheaply made, no filtering, made in china. plug in types, led shoplights, inverter controlled motors on treadmills, battery chargers for cordless saws, drills, are very bad.they can be heard sometimes up to a km away.don't forget modems, 2.5 ghz units produce carriers in lower vhf and down to lower hf as well. I use a zenith radio that has rdf that was used for direction finding.

| improve this answer | |
  • 2
    $\begingroup$ Hello and welcome to ham.stackexchange.com! Thanks for stepping up and answering a question with no official answer. However, this site is a question-and-answer site rather than a forum-style site (see the tour for a quick introduction), and we favor well-written answers, so would you please consider editing your answer to make it more complete? Maybe a quick walk-through of how to do RDF? $\endgroup$ – rclocher3 Aug 13 at 15:12

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.