The KiwiSDR 2 online store is open for orders! Please visit kiwisdr.nz

Help diagnosing interference issues

KM4YRIKM4YRI
edited February 2018 in Installation, Antennas and Interference
I'm having quite a bit of trouble diagnosing interference issues on my KiwiSDR setup.

I'm not sure if this is intermod from a close-by AM broadcast station, or something in my house interfering.

As of today, I'm using an active MiniWhip antenna that I purchased as a kit on eBay and fabricated the rest myself, making sure to use lots of ferrite toroids and a clean 12V power supply.  Before that, I was temporarily running on a 40M dipole and an ATX Walkabout (MFJ 1899T) extendable vertical antenna intended for 80M to 4M, depending on the positon of the flying lead.

In all cases, I get these horrible vertical bars across the entire 0-30 MHz range, and noise is pretty lit up across the whole thing, but really hot around MW broadcast and a few other spots.

If I turn the power supply off, it goes pretty much dark.  If I unplug the antenna, it goes completely dark.  Therefore I'm confident that it's something radiated, being picked up by the antenna (and any antenna I've used, for that matter).


0-30 MHz:

image

First blip is power removed.  Second blip is antenna removed.
image

Comments

  • KM4YRIKM4YRI
    Around the MW band:

    image


  • KM4YRIKM4YRI
    Around 40m band:

    image

    Around 10m band:

    image

    Zoom-in of closest AM broadcast station:
    image

    Distance to local AM broadcast stations:
    Note that the first few are 1000W stations, further down are 5000W to 50,000 W stations.
    image
  • jksjks
    Diagnosing interference issues is almost an art form. And fixing issues with E-field probes (Mini-whips), which are very sensitive to local noise, can be a real problem. We've spent many hours here getting our PA0RDT-clone to work well and we still have some issues to resolve.

    These screenshots are not ideal. It would be better to make your Kiwi accessible on the Internet so we could have a closer look. In the first image the noise floor is hard to judge because the "WF min" is set too low (too negative) and the darkest color is green instead of dark blue or black. When zoomed all the way out try around -90 instead of the -106 you have now.

    It would also be useful to see the spectrum display in addition to the waterfall. That way the level of the AM stations can be judged. Although it's good that the OV indicator on the S-meter is not lit indicating ADC overflow/overload.

    We would say the number one problem with E-field probes fed with coax is digital noise from the Beagle, Kiwi, or being conducted from the Ethernet or power cables being coupled into the outer shield of the antenna coax. The interference then travels up the coax to the E-field probe itself where it is received and travels down the center conductor along with the desired signals. The solution is to use toroids on the antenna coax right at the Beagle end. You need a lot of inductance at HF, so try and wind as many turns as possible through a number of stacked cores. We use rg174 coax and managed to get 10 turns through 3 stacked cores. Also make sure the antenna coax doesn't run anywhere near any other cables (like the GPS coax) on its way to the antenna or it will pickup noise all over again. See here for more info: http://kiwisdr.com/quickstart/index.html#id-noise

    A good ground is probably the second most important factor. It will either reduce a lot of your noise or introduce much more noise than you started with. Experiment with the ground terminal of the power mains, CATV system or your own ground rod. Also try no ground at all.

    We've also had a big improvement by adding an RF isolation transformer between the bias-T of the Mini-whip and the Kiwi.

    You should also be comparing an E-field probe against a simple long wire to judge whether the noise is local to your environment or being conducted up the coax.

  • KM4YRIKM4YRI
    edited December 2016
    > It would be better to make your Kiwi accessible on the Internet so we could have a closer look. 

    I am turning it on now.  I am going to email you the URL, just so that I'm not posting my IP address recklessly.  EDIT: If you go to sdr.hu, you'll see it pop up if you search for my callsign, 'KM4YRI'.

    Regarding the difficulty with the MiniWhip, I get the same thing with every other antenna that I've tried.  So far that list includes simple wire, dipole, and vertical w/counterpoise(s).  I am admittedly aware of the importance of having a ground on the MiniWhip, but it will probably be this weekend before I have a chance to actually install a decent copper grounding rod in my back yard.  I'm sure that will help, but as I said, I'm seeing this behavior even with a plain-old dipole.

    Re: Toroids, I am using them everywhere, here's a picture of my power splitter box.  I'm using toroids on the RF antenna input from the MiniWhip, the RF output to the SDR, and the power input. You can also see a toroid on the network cable, and there are a few others out-of-frame on the 5V power to the KiwiSDR, and one on the 12V power to the power splitter.  Since that picture was taken, I've installed the cover and tightened everything.  

    You can also see the KiwiSDR under the power splitter, in its own shielded aluminum enclosure.

    image
  • jksjks
    edited December 2016
    Okay, first of all my apologies for not understanding the sophistication of your setup. It is absolutely excellent. Hopefully my long-winded explanation above will help someone else because you clearly understand the issues.

    Now the problem: You've got the worst case of a bad switching power supply I have ever seen. The first clear carrier is on 167.9 kHz @ -66 dBm and appears about every 30.5 kHz up to 30 MHz. The carrier is narrow and stable but has "static" mixed in. This is a classic trait I've seen many times. The UVic beta site had a horrible signal like this around 14 kHz. Fortunately the harmonics died out in the LF spectrum. After almost a year it was traced to a bad power supply in an Ethernet switch.

    The fact that the harmonics go to 30 MHz and are so loud makes me wonder if it's a problem with one of the switchers on the Beagle. Like a bad filtering cap or something. The overall 0-30 MHz spectrum looks similar to the high noise problems we had with an early Element 14 manufactured Beagle. I assume this is a BeagleBone Green? I'd be happy to send you a replacement.

    There is also another huge signal (-56 dBm) at 86.5 kHz that is about 2 kHz wide and has the classic look of a "spread spectrum" (flattop) clock. It has harmonics in the NDB band every 9.5 kHz but dies out at MF.

    KM4YRI
  • KM4YRIKM4YRI
    Thanks for the kind words :-)  The level of fabrication I've gone to is mainly to tame the voodoo of the RF EMI demons, which I've heard can wreak havoc on HF.  This is my first serious HF project, so I didn't know what to expect.


    > You've got the worst case of a bad switching power supply I have ever seen. The first clear carrier is on 167.9 kHz @ -66 dBm and appears about every 30.5 kHz up to 30 MHz.

    That's (mostly) good news.  It's enough information to hunt down whatever's causing it and fix it.  

    > makes me wonder if it's a problem with one of the switchers on the Beagle.

    My initial guess is 'no', but I will investigate.  I forgot to mention that I saw the exact same pattern before I ever got my KiwiSDR, with my Ham-it-up upconverter and RTL-SDR dongle.  Logic would indicate that it's something else in my lab.  Lots and lots of things that could be causing it.  I've got the whole week off for Winter shutdown at work, so I'm going to start unplugging things and flipping breakers until I've found it. 

    I will reply back when I've found the culprit.
  • jksjks
    Okay, please let us know what you find. Your experience will definitely help someone else.
    I really need to start a separate webpage with nothing but interference screenshots and suggested strategies.

  • KM4YRIKM4YRI
    edited December 2016
    Could you try loading it up again and taking a look?  I shut off pretty much everything in my house while chasing it down.  A few Chinese USB phone chargers removed a bit of the static / noise, but the harmonics were still there.  After it was just my laptop and the KiwiSDR + power supply, I finally swapped the ethernet cable with a shielded CAT6.  I'm also noticing that my cable modem is making quite a bit of racket, like it's got a failing capacitor or something.  But it didn't make a noticable difference on the waterfall.

    Anyway, after replacing the ethernet cable, it does seem to be a bit better, but I'm not sure if anything that's left is just background noise or more interference that I need to mitigate.

    EDIT: Well, so much for that.  I started rearranging a few wires, and now it's all cocked up again.  This is proving to be very frustrating.

    EDIT2: It turns out that most of the static is coming from my APC battery packup going to my PC.  Then, if I turn my PC on, it all goes to complete hell.  Time to re-locate the KiwiSDR as far away from these as possible.  And I guess I've got to figure out if the EMI I'm seeing from my main PC is abnormal.  If so, I need to see what I can to do fix or replace as needed.
  • jksjks
    Sorry, I'm just seeing this message now (I'm drowning in email and messages today) and your Kiwi isn't responding on the net.

  • KM4YRIKM4YRI
    edited December 2016
    Turning on my APC battery backup to the PC, with PC still off:


    image

    Now, turning on the PC: 

    image


    :(
  • KM4YRIKM4YRI
    @jks, no worries, I think I'm just going to have to relocate it at this point, and get a good ground to it while I'm at it.  Then I'll reassess.
  • jksjks
    Ouch. Yeah, pretty much any wire coming out of a digital device turns into an antenna when an E-field device is listening. I think this is one reason why so many people use H-field loops. There are lots of Kiwis hooked up to Wellbrook ALA1530 loops with pretty great results.

  • WA2ZKDWA2ZKD
    I evaluated whips, loops, LW, doublets, both homebrew and commercial. I ended up with a Pixel Loop. I have found that adding a choke on the feedline right near the antenna, one made with Type-43 ferrite reduce pickup of local stuff on the feedline.  I also added a HPF to cut the AMBC to reduce that overload. 
  • WA2ZKDWA2ZKD
    Here's a source for Type 43 http://www.mouser.com/ProductDetail/Fair-Rite/5943003801/?qs=Ca1fAiqt1aqdvI%2b3p7fLXQ==&gclid=Cj0KEQiAtK3DBRCBxt-Yxduq5p4BEiQAbFiaPVjDBD0ZimGD3OST73DDPxotP1lkGR_BFFmXfOD5_vwaAmo28P8HAQ




  • WA2ZKDWA2ZKD
    There are various HPF filters such as http://www.stridsberg.com/prod02.htm but you lose the LW and MW band. I was lucky to have one of those and diagnose things. I am now lookign at ways to notch the strongest signals in the MW band. I have one that is only 1 km from my QTH

  • KA7UKA7U
    I used a series LC circuit to attenuate a nearby MW station on 1380 KHz. The Q of the filter was a bit broad and took down some neighboring stations, but it did solve the problem of spurious signals up and down the band. This calculator makes it easy to find values:
    http://www.1728.org/resfreq.htm

    WA2ZKD
  • G8JNJG8JNJ
    Hi to KM4YRI,

    I've got some 1400VA rack mount APC UPS's and they put out a LOT of noise, even when just connected to the mains but with no load connected. 

    Both were more than 100ft away from my antenna (not an E Probe) and they still caused problems, no amount of filtering seemed to help, so for the time being I've turned them off until I can find a solution.

    Regards,

    Martin
  • WA2ZKDWA2ZKD
    House them in a Faraday cage....   it's practical to build one using wood, sheetmetal, and bronze screen. 
  • fotografmfotografm

    I just got a KiwiSDR kit from OKDO in the UK. It has horrible birdies every 30kHz or so on the waterfall even with no antenna connected. These also radiate from the unit and I can see them on my X6100 with sniffer antenna. I tried several low noise switch mode psu's and the problem is still there but no battery supply yet. Apart from this I love the system !

    I just shut down the system from its webpage and the noise still radiates from the board until I disconnect the power supply physically ! This suggests it might be a clock which is left running on the beagle bone green even in shutdown state. I saw the comment in this thread that early BBGs had a noise problem.

    Do you have any suggestions ? I need to overcome this noise problem.

  • WA2ZKDWA2ZKD

    What kind of PSU are you using

  • fotografmfotografm

    I am using a Meanwell PSU which is recommended by kiwisdr as it is a lower noise switch mode psu. I also try it with a raspberry pi psu with the same result. I can't find a 5V battery here otherwise I would try that. It no longer seems possible to buy a single 7805 regulator otherwise I could hang one of those on a 12V battery. The birdies sound like fsk signals modulated with some raspy sounding data. If I can rig up a battery supply then I could go out into an open field with a laptop and see if the birdies are still there.

  • fotografmfotografm
    kiwi1a.png

    The black band is where I unplugged the antenna for a few seconds.

  • jksjks
    edited October 2022

    He sent me waterfall images in email. I've asked him to post waterfall + spectrum images here. I see all the classic stuff: Ethernet noise, SMPS, VDSL/BPL. Without spectrum images it's hard to tell how bad of a problem this really is. People often see anything other than black / dark blue in the waterfall and think the sky is falling without really understanding what the waterfall is showing.

  • fotografmfotografm

    Problem solved !!! Thanks to the extremely helpful comments from jks via email. I switched the ethernet speed from 100Mb/s to 10Mb/s and the birdies went away instantly. I had tried 3 different routers connected to the ethernet port and the problem was consistent. Perhaps the birdies emanated fom the ethernet adapter on the BBG itself ? Anyway - I am happy today so thanks once again for the excellent support !

Sign In or Register to comment.