Second build sold out. Message will appear here when store is ready for third build ordering.

Noise advice for MiniWhip?

Hi all.
I moved my Kiwi last year to a Suburban location (avaliable here: http://fsdr.duckdns.org/ ) And it was noise-free (almost) until the middle of December. Above 3 MHz periodic noises appeared, probably from heating systems, boilers. But the biggest problem, a strong, static noise from the AC line around 1600-1700 kHz. It is S8 sometimes. I grounded twice my MiniWhip. One ground at the pole of the antenna, connected to the shield at the connector, another grounding is behind the receiver. Two different grounding rods, NOT connected to the AC Line ground. Maybe I need to ground the coax at halfway? Using 1,6m long grounding rods.

Antenna is ~50-60m away from houses at a gerden, leading 60m RG59 coax to a little house. In the little house the 100-2000 kHz band became full of noise from the AC line, but it was quiet before december. Most of this noise is reduced by the 2 groundings, but the strongest at the end of MW band and 160mb was remained. There is NO noise at the place of the MiniWhip. So the noise is carried to it's AMP via the Coax. I added ferrite cores to the Whip's amp PSU and the Kiwi's PSU. When I disconnect the coax (but the PSU's staying plugged) there is No noise. When I disconnect the Whip's PSU only, the majoritiy of the noise is still there.

So what can I do? Here is the noise: https://vocaroo.com/i/s0A9mT7rcCyk

Comments

  • A spectrogram from the Kiwi in a wide zoom might be more useful than audio but, if I understand your description correctly, it would seem that in any case this is likely to be common mode (TM) noise on the coax, coupling on the feedling TL and converting to TEM/differential noise into the Kiwi. There is at least one other possible mechanism, common mode noise current through the Kiwi enclosure/system but perhaps what you are experiencing is not that.

    From my experience, the Kiwi actually has good rejection of differential PS noise. Most of the issues I've seen have related to common mode current flowing through the "ground" of the Kiwi (the metal enclosure). This should not be thought of as a single ground but as an imperfectly conductive mesh or 'blob' such that CM noise current in from one line, flowing through the structure and out another generates IR drop that appears across the ADC input which also has distributed grounds internally. With an approximately -155 to -160 dBm (1 Hz BW) noise floor, we need to really reduce these currents if they are not to show up along with desired signal (and actually, propagated band noise) currents.

    While it is useful to ID the noise source and kill it where it is generated, eventually this becomes impossible. I would suggest adding a single toroid core with all lines going in/out of the Kiwi. Perhaps a 5 cm HF mix with 4 turns (wound the same direction) of each line: antenna, GPS, LAN and power. This can have the effect of reducing common mode current through the Kiwi more than simply adding inductance/loss, a choke, on each line separately. Just trying this as an experiment to see if anything changes could be helpful in solving the problem.

    Glenn n6gn
  • How about a spectrum instead, there's information in that display that is hard to see in the waterfall.
    WA2ZKD
  • That looks to me a bit like a failing switching power supply - but I have also seen this sort of thing with a very lightly-loaded switcher running in some sort of "discontinuous" mode (e.g. it turns on intermittently to maintain a voltage rather than having enough load to regulate.) Of course, if there is any switcher running the Kiwi or antenna, all bets are off.

    I have also seen noise from the power mains couple into otherwise-clean devices: It usually takes a millihenry-range common-mode choke to effectively quash this - sometimes on both the mains side and also on the DC side of the supply - to adequately decouple at RF. A nice coax choke (a couple dozen turns of RG-174 or similar on a FT-240-75 or 77 toroid) is always a very good thing to prevent RF from finding its way along a feedline - ideally placing one at the antenna (between the radio and the antenna with its own "local" ground) and another one near the receiver, with the shield on the "outside" portion of the coax to the antenna grounded to something other than a noisy station ground.

    * * *

    A bit wider view might verify the type of source, to see if it recurs in a way that is periodic with frequency (e.g. spaced by the fundamental frequency of whatever it is.) Typical plug-in supplies are in the 20-80 kHz range (usually 30-60) while "POL" (Point-Of-Load) supplies inside devices (such as IP cameras) tend to be in the 200-500 kHz range.

    If a noise suddenly appears and there's no obvious explanation for it (e.g. Christmas lights) it is quite likely that it will be a new device, or a switching supply with failing capacitors: Inexpensive wall warts seem to have a 18-36 month lifetime before they go bad - often less - and when they do, they tend to become inefficient (e.g. quite warm) power conversion devices, but surprisingly effective RFI generators.

    Finally, remember this saying: Most QRM begins at home.

    73,
    Clint
    KA7OEI
  • As you mention on the "stat" tab, do you really have HGA22 135.6 kHz notched? Because it's still S9+55 (~ -20 dBm). Wow!
    I actually think your noise floor is pretty darn good compared to most Kiwis.
  • It doesn't seem to be 135.6 but rather closer to 135.4, I doubt the calibration is off 1%, MSF? at 60 kHz looks OK. It looks more like a SMPS coupling into the Kiwi. Rather in the territory for an LM2596. Sure is big.
    At ~1650 there seems to be ~60 kHz periodicity to the family of QRN. Not sure what that means but it's fairly complex.


    .
  • 135.6 is the middle of the passband for FSK decoding, i.e. http://fsdr.duckdns.org/?ext=fsk,135.6
  • sure enough. ~135.43 carrier (low space with 340 Hz shift gives a 135.6 center) with FSK. It must not have been sending when I first looked at it. Sure is big!
    As you say, that it's like that in the presence of any kind of notch filter it must be extremely large field strength.

    HGA22 is shown as 47°22'23.7"N 19°00'17.1"E in JN97ma (Budapest) while the Kiwi seems to be in JN97ED more than 50 km distance away. Surprising, even considering the 100 kW transmitter.
    I have similar power (~100 kW) at 60 kHz (WWVB) only 22 km away from me and signals are smaller.

    But I guess that isn't the question, rather it's the source of QRN in MW. I wonder if the noise is actually coming in on the antenna or whether it's common mode ingress. A directional loop might help discover that...

    Glenn n6gn
  • edited January 2019
    jks: without the notch HGA22 is overloading the full stuff.
    n6gn: the noise is from the power line, probably caused by a failed pole somewhere? Not sure, but I need to protect the antenna against any noise :)
  • Very possibly the source is power mains related noise. The question I have is how does it couple to the kiwi. I suspect a near-field effect rather than far-field (inverse square law) radiation. The latter is hard to do anything about but much rarer. The former fall off much more quickly and can almost always be mitigated.
  • When I was using an active whip I found network cable with CCTV baluns either end was much better at rejecting local noise pickup.
    Another wierd thing I have just done is put network cable through 10mm copper tube and grounded that by burying it and attaching it to an earth spike (or two).
    If the pickup is on the coax going over to cat5(or better) should make a noticeable improvement (in my purely hands-on experience), you could even try different balanced pairs in the bundle see which twist rate is best, also try grounding the unused pairs at one or other end.

    You seem to have a lot of signal there, the antenna must be quite high gain, enough headroom to put some fairly lossy fixes in.
    Here my current active loop(s) are a bit too small but the other set I made up too large (about 8-10db increase in noise floor for for 2db more S/N) the extra S/N is nice but the waterfall looked bad, I need to find just enough gain for optimum signal to noise.

    When I just looked at the SDR it had some noise that looks similar to PC power supply (I suppose any switchmode with a bit of current handling) would be interesting to see what difference CAT5/baluns made.
    Just in case you are tempted the Baluns I tried can be found by searching for "Video Balun 202P" and my prefered CAT5 is "Ligowave CAT5E FTP Outdoor Cable".

    Stu
  • I added grounding to the coax feed on halfway plus a common mode choke from 10m coax near the antenna. The S9 noise reduced to S6 and the LW band became almost clear :)
  • edited January 2019
    In the near future I'll add a second choke near the little house.
Sign In or Register to comment.