Wideband noise around 1.6 MHz

I have some wideband noise centered around 1.6 MHz. I am at a loss determining direction or location of this noise. FYI, Kiwi is on the attached connected to a vertical some 5 meters from house and isolated from house by a wideband transformer, therefore the apparent loss of signal below approx 500 kHz. This Kiwi is available on la3rk.dyndns.org:8073

Any info/ideas are welcome.

I have a Samsung freezer which raises my noise level approx 5 db whenever the cooling cycle turns on. What would be the best option to kill such noise? Filter where the freezer is connected to the wall plug or isolating the Kiwi from the mains, both live and ground by a common mode choke. Freezer and radio room is on same circuit breaker.

73 de Olaf - LA3RK


  • How close are you to neighbors? Can you replicate the problem on another receiver if you have one? If not can you run the kiwi on battery and use a laptop or something and kill the mains to your house?

    If you can't do either of these, can you try a different antenna on the kiwi, maybe a directional one like a mag-loop?

    The interference and attenuation below 500kHz should be treated as separate issues until proven otherwise.

  • In spite of the low end roll-off, I do see lines in MW that might be explained by an SMPS at ~20 kHz (bipolar transistors?) with rather wide spectral lines. By 1.6 MHz these lines are all merged and the coherence isn't as obvious but if one listens in AM/10 kHz it is easy to hear the mains component.

    At that frequency this is certainly not far-field radiation. I would suspect either or both common mode noise on the feed line, which might be tested by adding a flux-coupled 1:1 transformer right at the kiwi input to see if it affects the level or near-field coupling at the antenna. Such coupling could be due to antenna proximity (near-field) to the source and/or currents in the earth/image associated with the antenna.

    A different antenna, preferably one with high symmetry and good CM rejection, one with a minimum of feed line could be a helpful tool and better yet an entirely portable system.

  • Thanks for advice. Neighbors are fairly close, closest approx 20 meters away. My house closest to antenna and separated approx 5 meters from nearest wall.

    Can tune my KX3 to same frequncy and check, however my K3 is normally connected to this antenna and noise on 80 m is a problem. I have disconnected power to house and listening on batteries, but that is many years back. Maybe time to redo the test.

    Having a Kiwi really shows the problems and results / improvements can be seen.

    Yes, there are a number of smps connected and they may be the source. I have an magnetic broadband balanced loop and the very low frequncies around 30 to 50 kHz are fairly clean even used indoors, many vlf signals like the time signals eg dcf77 in germany are clearly heard.

    73 de Olaf - LA3RK

  • I'm not sure what antenna it is on at the moment but that noise is very similar to what I see here from a high end gaming PC on with glass sided case.

    Good info in previous comments so lots to work with.

    I agree on the "switchmode" and I'd say something that is a few hundred Watts (as there is a fair amount of hash spread across a lot of HF). There was also some powerline type data in there too when I looked.

    Here when it is in that "Power PC" mode I found it was a great way to find points of ingress, grab a bunch of ferrite and start adding it to leads, any leads, DC especially, while watching the waterfall (slow), you may be surprised where it is getting in. Here I have to use the mains supply on one side of the house, filter mains, DC and even any other long leads on the same socket as the SDR.

    Last point ground it out, really well.

  • Antenna at the moment is an Cushcraft R5, 20m to 10 vertical. Base approx 5 m above ground. Coax feed via common mode choke good for 10 MHz and up. I should probably put in a lower frequency choke and ground my coax below outside house or a broadband transformer to break any groundloops. I have a feeling the noise is following coax outside and fed back through antenna.

    No gaming PC in the vicinity as I know of.

    Ground is a difficult matter, mostly shallow earth with rock underneath.

    Seems I need to do some serious hunting.

    73 de Olaf -LA3RK

  • edited December 2020

    Gaming PC is my local shorthand for anything that burns a good few Watts from a switchmode. From a quick listen it seemed to have a similar signature across the band so I tend to think you have a good shot at finding and taming a single source.

    I have used the signal meter extension on 1/64 or 1/128 speed set to a receive frequency that does not get signals, watch when the base level changes, that can give clues (E.G. does it reduce when some local house lights go off ) - if you have local houses.

    If ground is not a good place to sink energy try a mat or number of wires instead, here I have noise inside the house and out, the trick is to have the Kiwi at a point where it is not at the greatest potential between those fields. As I can connect to earth that helps but even then the SDR has to be less than a metre from a fairly good earth. I have a number of feeds going down the garden and each one seems to affect the noise level, more copper out there is generally better. Verticals are harder to work with here, If you can I'd consider trying a receive only antenna, mag loop type --later-- (sorry I didn't read that you already do have a loop).

    73 Stu M0AQY

  • This is a timeshot of signal levels from 0 to 30 MHz. Note the variation of noise level, green line, peaks coincide with Samsung freezer. There is also a variation between night/daytime.

    As to ground, I have a number of radials installed which can be used to create a fair ground outside the house. But key issue is to find and isolate the offending noise source, the Samsung freezer is not the only one.

    73 de Olaf - LA3RK

  • Might be nice to compare the red line with indoor/outdoor temperature in case it is something like underfloor heating distributing existing mains borne noise. Maybe even thermal noise I suppose (though it didn't sound smooth enough)

    It seems to be damped at a time where the house temperature would be most stable and bodies are in place. The "on" pulses do seem to be the longest duration at a time when people would awake but the outside temperature low.

    Just speculation.

    73 Stu

  • "This is a timeshot of signal levels from 0 to 30 MHz."

    Olaf, this is a very smart graphical presentation that could presumably be adapted to a huge variety of monitoring scenarios. Have you documented it somewhere please?

  • Basically a rewrite from microkiwi_waterfall.py in github, https://github.com/jks-prv/kiwiclient Microkiwi_waterfall.py gives you some statistics on a given frequency band. Data from this program is combined with a socalled round robin database/graphing tool, some crontabs and a minimal web page. Needs some python knowledge and local adjustments.

    Free for anyone to use, but I need to find a location to share the code. Current version attached, but you need to collect utilities from kiwiclient link above. Had to include as txt file, py extension not allowed.

    Python 3 recent version needed. Program is runnung on a local raspberry. I do not think such program should be used on a remote kiwi without owner accept.

    As to the variation in noise level approx every 30 min,, it seems to be connected to a Samsung freezer, noise coincides with run times on freezer. Change of levels betwwen day/night is propagation (peak levels) and less human activity night time, less overall noise.

    73 de Olaf - LA3RK

  • edited December 2020

    OK I misunderstood the graph, thought the red was one signal (frequency) and the green another, i must stop posting quickly while at work.

    Thanks for sharing that.

    Oh and just for info what is the power consumption of the Samsung freezer?

  • Sorry, Bosch freezer, 200 litres, marked with a yearly consumption of 214 kWh. Averages to approx 30 watt and around 60 watts when running assuming 50% running time.

  • OK thanks, I'm a bit surprised for a big name it is that visible at ~ 60W but perhaps something like the radiator-shell acts as a decent antenna. At least it would be a good candidate for a mains filter.

  • Thanks again, Olaf.

  • one brute force technqiue on finding noise is to get your kiwi and computer running off a battery, that may require kiwi<->PC ethernet directly. Then, shut your whole house off. Noise still there? blame your neighbor. If noise is gone, bring stuff online circuit breaker by circuit breaker.

  • At least one source identified and reduced. My station pc is powered via the 12 volt supply. Even when off it supplies 5v via usb. The internal voltage controller sent significant noise back on the 12 v bus and was picked up by the outside antenna. Probably following the coax outside, my choke was not sufficiently effective below 2 mhz.

    Found by sniffing with a ferrite rod antenna while watching the spectrogram.

    There is also some noise on my mains ground, my mains filter had a straight through earth wire.

    My SNR ratio is improving, now seeing numbers around 25 dB (P90 level minus mean level, 0 to 30 MHz)

    73 de Olaf - LA3RK

Sign In or Register to comment.