The forum is open to new posts.

Noise at roughly 60 KHz intervals

2»

Comments

  • Clint,
    Thank you for the informative reply. I mis-stated the frequency of the spurs. They are around 18250KHz or 18.25MHz. I am sorry about that and I've corrected my post in this regard.

    I have used large Toroid chokes, but I'm finding good results without them with this amplifier. What I've encountered is interesting working with it, I've discovered things I didn't know. i reterminated the CAT6 in a shielded RJ45 and took care to place the small wire under the foil against the shield of the plug. This termination matched the performance of the unshielded connections at the receiver end. I've tried 3 winding methods on my toroid baluns. One is to wrap the two wires side by side in the same direction around the core. Two is to wind one wire left hand from the start and then to continue the second wire with right hand rotation from the end of the first wire. Three is to do the same as two but the second wire continues with left hand rotation. I have not been able to determine a performance difference with the sense or the capacitance changes of these winding methods for the balun, but maybe I don't know what I should be looking for, other than noise and signal strength. Methods 2 and 3 should reduce winding capacitance between the two windings but probably not do much about individual winding capacitance. Comparing my dipole antenna to the loop, I notice the spurs around 18.25MHz are comparable amplitude between the two. The dipole has 340 feet of RG213 placed in underground conduit and this a very low noise antenna. So as it is my comparison, the noise on the loop is now very good. The dipole has 210 feet of horizontal wire and generally has more gain across the band, so much so that I need notch filters for 2 MW frequencies where the loops do not.

    I appreciate your willingness to share experience and insights into this process. Most KiwiSDR's have noise and 60KHz spaced spurs. I think I'm on the right track to make loop antennas that will improve this performance. At least it is improving here.
    Ron
    KA7U
  • I note that the high-quality shielded Cat cables have the drain wire tack soldered to the metal clip at the plug. Sadly, I have also found that "professional" installers often do not even deal with the drain wire on installations in businesses.
    KA7U
  • ethtool -s eth0 speed 10 duplex full

    I found that forcing the NIC to 10-Base-T makes a very significant improvement. Is there any downside to this i.e., will there be a bandwidth utilization concern? It looks like only 50kB/s or so is needed per connection, but that seems too low to me.




    Additionally, has anyone tried wifi? I'm expecting this to perhaps cause more problems than is solves, but it's a worthy experiment. In my installation the kiwi/BBG is out in my barn with 150+ft of Cat5 run to it. I'm expecting some coupled RF is making its way into the receiver despite the Ethernet magnetics.

    Nick
    W1NJC
  • jksjks
    edited December 2018
    Wow! I never thought of this. If your Debian disto doesn't have the ethtool command just type pki ethtool (pki is an alias for apt -y install) I should probably add the ability to switch the Ethernet interface speed directly to the admin interface.

    There shouldn't be any problem running at 10 mbps assuming your switch/router can handle it. With a very insensitive (aka broken) antenna where the Ethernet RFI is prominent I got this result. A lot of those spurs dropped by almost 10 dB!

    Ethernet at 10 mbps is 1250 kB/s. Assume pessimistically you get half that with the software/protocol overhead, 625 kB/s. Each Kiwi connection is:
    non-IQ  6snd + 12wf = 18kB/s
    IQ     48snd + 12wf = 60kB/s (IQ uncompressed is 4x, and the two IQ channels is 2x = 8x more snd b/w than non-IQ mode)
    
    So 4 IQ/wf connections (4 channel mode) shouldn't be an issue (240 kB/s total). 8snd-IQ/2wf mode (408 kB/s) will probably be okay too. As usual with this kind of stuff it's best to just try it and see what happens.

    image

    Attachments:
    http://forum.kiwisdr.com/uploads/Uploader/bc/c5110e55c99fa99d782105d786b2bf.png
  • Nick,
    I think 10mbits will run the KiwiSDR. I applied the ethtool utility with your parameters to my KiwiSDR's. The first one was imperceptible change, then the 2nd one showed improvement on the KiwiSDR that has the most interference, then the 3rd one and at that point the SDR with the most interference showed a 6dBm improvement and the interference lines seem to have less density on the spur, as though interfering signals are missing. The KiwiSDR antenna with the worst spurs is about 30 feet from the equipment, the 2nd KiwiSDR antenna is about 250 feet from the equipment, and the 3rd KiwiSDR antenna is about 350 feet from the equipment. So antenna 1, 2, and 3 for future reference. Antenna 3 is a 210' top dipole with 350' of RG213/U underground in conduit, and the KiwiSDR shows the least interference, ka7u.no-ip.org:8073. Antenna 2 is a 120cm copper active loop antenna with 250' of shielded CAT6 4 pair, ka7u.no-ip.org:8074. Antenna 1 is a 100cm coax active loop with 50' of shielded CAT6 4 pair, ka7u.no-ip.org:8075. The dipole is not amplified, the two loops both use the same design of differential amplifier and matching baluns and power supplies.

    When there is no amplifier on the loop antennas the interference is there. With amplification the spurs increase. The differential amplifiers are well balanced for hfe and bias, so common mode signals should be seriously attenuated. Which point has been verified by accident with shielded and unshielded RJ5 connector mismatches. So my theory is most of my spur interference is actually received at the antenna and not common mode. Hence antenna 1 has the greatest interference, and the interference subsides greatly in antenna 2, and is practically nonexistent on antenna 3, the dipole. Now your ethtool parameters have improved the situation even more, but not eliminated it. There remain 3 network switches, one shielded and 2 not shielded, and 2 computers, and 2 monitors, running in the radio shack with the KiwiSDR's.

    I think my next best move is to move antenna 1 out to where antenna 2 is located or nearby, and expect that the interference will be reasonably abated. Replace the unshielded network switches with shielded ones should help too.

    You might find that moving your antenna to the barn and keeping the KiwiSDR where it is, will give the best results.
    Ron
    KA7U
    W1NEJ
  • I'm glad you guys are seeing an improvement too. Thanks jks for the bandwidth calculation and also mentioning how to install ethtool as I neglected to include that. I added that command to my /etc/rc.local so that it's persistent across reboots. Perhaps you can comment on this?

    My antenna is a ~625ft Beverage running westerly. The near end terminates at my barn (more of a shed but barn sounds nicer!) which is about 100ft from the house. Inside there is a home brew 9:1 (6:2 turns) transformer with about 25ft of RG-58 to the other side of the barn where the kiwi/BBG is inside a metal box. The SMAs are electrically connected to the box, and the box cover has a bonding wire to the main box. Currently the box is not tied to power Gnd or anything else. I had previously found that tying box Gnd to different power grounds through 1uF made some differences. I am using a good 5V/3A linear PS. My ethernet cable is unshielded. My servers/switches are in the basement, about 200ft from the barn, partially blocked by the earth.

    With EMI/RFI it's always a question of whether it's radiated or conducted, common mode or differential, and on a receiver whether it's from the antenna or directly into the board.

    Ron, your reports seem to indicate that much of the noise is radiated from the networking equipment and coupled into the antennas. Out of curiosity, are you running any PoE equipment? If so, you should try shutting them down and record the result. My original plan was to power the kiwi with PoE, and it worked but the noise was awful. Grounding the SMAs helped a little but not enough. Some serious 68uH chokes and caps on Vin also helped some (if you try this, you might need diodes across the Ls because the transient voltage drop can prevent the system from starting) . The PoE PD(powered device) board was IN the box with the kiwi so that was likely why it was so bad. I want to try again at some point with the PD in its own metal box.

    Nick
    W1NJC
  • Hello...as per my knowledge quality shielded Cat cables have the drain wire tack soldered to the metal clip at the plug. I have also found that professional installers often do not even deal with the drain wire on installations
    KA7U
  • @njc
    Nick,
    I added the ethtool -s eth0 speed 10 duplex full , to /etc/rc.local and it works just fine. A review of /var/messages after a reboot, shows no issues with eth0 starting, stopping and restarting with 10mbits. So this is one of the best spur reductions to date. I have one PoE in the house and it is used to power the Motorola Canopy antenna transceiver that provides my Internet Service. It is about 80' east of the nearest antenna and the other antennas are farther west. The PoE cable is unshielded and runs in the crawl space under the house another 70' to the east end and then up about 20' to the antenna. I might look into a shielded cable and perhaps a power supply mounted at the antenna. But, with this new 10 base T network on the KiwiSDR's, I'm happy with the receiver performance. What spurs remain are not a serious problem. I'll disconnect that PoE tomorrow just to see what improvement might be had before I go to any more rework. I'd do it tonight, but my bride is watching Netflix Hi Hi.
    73,
    Ron
    KA7U
  • Nick,
    I shutdown the PoE on the ISP antenna. No difference in the spur amplitudes.
    Ron
    KA7U
  • @KA7U
    Ron, I'm happy to hear that you are getting good results from this. I hope that others who have this problem (everyone?) can try it and see an improvement. I know @jks thought he might be able to add a setting to the admin UI -- that would be great but I know he's a busy guy!

    Thanks for trying and reporting on the PoE. Interesting that it had no effect. Regardless, it was a smart move to wait until she was finished with Netflix! I still want to try and get mine PoE powered without major interference, kind of a personal challenge despite being unnecessary with the linear supply I'm using. Maybe after I finish the antenna switch project (using the @OH1KK extension) to switch in/out a 1.8MHz high-pass filter and a low-noise amp - but that's a topic for another thread.

    73,
    Nick W1NJC
  • I used ethtool to drop to 10 Mbps, no change noted here, but I am not sure if the problem existed anyway? I use shielded ethernet cables in the shack, and my coax cable runs to the antennas go down the side of the house, then underground 60 to 100 ft through conduit, grounded at each end, then back up and to the antennas.
  • Thanks for the feedback, @ChrisSmolinski . I suspect that the significant distance you have between the Ethernet/kiwi and your antennas has much to do with your lack of interference. Buried cables helps too. I think you would know if it was there. You could look for a signal at ~60kHz and also around 18200kHz. For some reason they seem to be pretty strong up there.
    As discussed previously, the shielded cables are dubious as to the relief (or enhancement!) of interference observed.
  • Ah to be in a location where I could benefit from such fine tuning ;-)

    My most recent test was to bury 10m of 10mm diameter microbore copper tube with CAT5 FTP external, nice and slipery pushed up the middle. I fed the end through the external wall having attached it to two existing copper (clad) earth rods in the garden. This was party to remove the need to for the earth braid I had going under the back door, now the earth rod comes through the wall and the CAT5 for the active loop goes up the middle. Of course true to form my neighbour then powered up some new device that sprays hash across HF and is visible up to about 125MHz, I think it is probably flashing Xmas lights.
    I turned the Kiwi off for a while due to the profanity I kept hearing when trying to use it (with only me in the room), only tuned it back on today to do some tests.

    For info I tried dropping the NIC speed on mine (using the managed switch) no difference seen but 98% of my noise is external anyway and there is only 2m of CAT7 between Kiwi and the fibre switch.
    Here the CCTV cameras are a 100% reliable source of the network spikes but Type 31 ferrite rings and some shielded CAT6/7 where I can ground them out seems to tame that. In fact I had two cheap Chinese (read no filtering at all) cameras sharing one shielded CAT5 through the loft that was giving out lots of noise until I put a ferrite on just one of the two cameras before the cable economiser and the difference was massive, that seemed fairly illogical but I'll live with it.

    When trying to reduce the noise I tend to think of making the unwanted path hard, preferred path easy so I ferrite the bits that radiate and try to give any shields somewhere to dump any pickup.
    Obviously my full runs are like your patch lead lengths but then I do have some really dedicated noise producers within a few feet.
  • I found that constraining the input to a switch (at the router) to 10 m can be an issue if there a lot of 10 m activity at the switch ports.
  • I got a lot of white noise and vertical lines, using a 30 + 45 meter long network cable with a switch in the middle.
    currently using 30 meters of network cable and dipole antenna .. signal was very clean ..
    lowering the settings to 10mbs made a lot of difference too ..
    i am from Brazil

    http://py4aaz.homeip.net:8073/
  • edited May 2020
    I am editing this because whilst changing the Ethernet speed from 100 Mb to 10 Mb significantly reduces the spurs, it creates substantial wide band noise below about 15 mHz that makes the band below that frequency unusable.

    I found the spurs are significantly reduced as shown in these photos:

    Around 18 mHz
    image

    Around 28 mHz
    image

    Entire 30 mHz
    image

    There is a way to change the speed on the Network tab of the admin page, however, this does not always work. In my case it would sometimes work, and other times not work.

    Since I have a 28 port switch where I can set the speed, I just changed port 10 (which is where the Kiwi connects) from 'auto' to 10 Mb to force it to go from 100 Mb to 10 Mb. I can make the change on the fly with no changes to the Kiwi, and the Kiwi automatically sets its speed to the switches setting.

    image

    I find it interesting that with the antenna disconnected from my Kiwi I have absolutely NO spurs (this is with Ethernet set to both 10 Mb and 100 Mb about half way down the screen):



    So I was under the impression that these spurs were coming in from my antenna, and since I live in a fairly dense suburban situation in SE FL where the houses are on from .3 to .6 acre lots I expect that. Everyone has lots of electronics, and in my case I have about 60 devices that have IP addresses from my router around the house.

    While it appears the spurs above 15 mHz, and the wideband interference below 15 mHz is related to the Ethernet speed, I do not understand why they disappear when the antenna is disconnected.

    I am using an Alpha Delta DX SWL antenna which is pretty good from the broadcast band on up. I had no problems picking up weak signals, and it is not *too* noisy.

    Many if not most of these spurs are modulated with what sounds like an RTTY signal. It sounds like a strong RTTY is mixing somehow into the spurs that are from the 100 Mb Ethernet signal. This is 95 % cleaned up when switching the Ethernet connection to 10 Mb.

    Photo of the spur modulated with RTTY signal with Ethernet running 100 Mb and 10 Mb. At 10 Mb the RTTY modulation is gone.

    image

    I know there are quiet Kiwi's, and my Flex 6600 is not seriously affected with spurs. Unfortunately the Ethernet speed fix is not going to fix my problems.

    So for now my Kiwi Ethernet is set to 100 Mb.

    Nothing worth while is easy!


    73,

    Eric
    W1NEJ

    http://boca.homeip.net:8073

    Attachments:
    http://forum.kiwisdr.com/uploads/Uploader/bc/57df8ce0cae674e10fdc33f5184b8f.png
    http://forum.kiwisdr.com/uploads/Uploader/04/7da577c0a5468a4ea9bd0401037643.png
    http://forum.kiwisdr.com/uploads/Uploader/92/dbb9b5c6aa9a27453f9306c2178200.png
    http://forum.kiwisdr.com/uploads/Uploader/a9/fe60c7231782fa429d10603b543626.png
    http://forum.kiwisdr.com/uploads/Uploader/13/f39c5b59a2ba0c3373c2370f53ff1d.png
  • You said Ethernet Switch, I assume you meant Ethernet Router (which has an integral switch).
  • edited May 2020
    No, it is a 28 port D-Link Ethernet switch (DGS-1210-28P). This is connected to my router with an Ethernet cable.

    There is a way to change the Ethernet speed on the Network tab on the Admin Interface. Most Ethernet switches should adjust to either speed.
  • The DHCP in the switch config fooled me
  • Eric,
    i'm on a farm and i have the same noise as you, like rtty. if I leave the network at 100mb.se step to 10mb the noise disappears
  • I did manage to get a wifi USB adapter (MT7601 chipset) working on my BBG/kiwi which solved all of my ethernet-related noise problems. It's also quite convenient to only have to run power out to my barn where the kiwi is. It was quite the challenge for me to get the wifi working though. There is another thread on this forum with some information if you're inclined to try...
  • Hello, see the result of a balun of air nucleus in my apartment.very simple. about 10 turns.cleaned a lot of dirt


    f1jek
  • The effect of using 10Mb/s instead of 100 is incredible on high frequencies.

    Thanks a lot to who had this idea !!!

  • I would add that my Kiwi, which is in attic, is connected to a 5V-supplied Ethernet-Wifi bridge (TP-Link Nano Router) with a very short wire (50cm).
  • It would be interesting to test both the ethernet cable and the supply to the Nano router with a few turns through ferrite.

    This is a remarkable difference so I assume the Kiwi is also physicaly very close to the antenna?
    If the setup is that sensitive it is probably also seeing other, less obvious noise.

    I tend to purchase clip on ferrite for 11-13mm cable then add those to any cables around the Kiwi and close to the antenna.
    Pass as many turns through the ferrite as will physically fit. Run the waterfall a little slower and make sure there is a decent bit of "before" prior to any changes.
    One advantage of the Kiwi is that the user can have a phone right beside the cable or item under test.
    I also leave another device on at the same time with the full span waterfall at half speed, I check that later to confirm any improvement.

    I spent far too long doing that but it does mean now that my setup is reasonably resilient in a noise prone location (terraced houses).
    f1jek
  • Hi,

    The kiwi is in the attic while the antenna is some meters above, on the roof, attached to the chimney.
    I have such ferrite clips, I will try them when I will have some time to go to my parent's house where the Kiwi lives ;-)
  • if the power cable runs close to, near, or close to the network cable, you will experience interference problems.
  • I got rid of my second switch mode power supply for the KIWI last week and transformed my noise levels. I bought the Kiwi with a power supply provided by Martin Lynch and co in the UK as a package. The power supply was very noisy. I swapped it out for an old much quiter supply salvaged from a 1990s desktop computer that long ago went to the dump. That was an improvement, but last week, I bought a nice, inexpensive (£35) linear supply and the LF part of the radio spectrum is now clean where it used to have spurs 60 khz apart which wandered about on Top Band which I like to listen to. They would annoyingly slowly crawl over signals that I was listening to. Now the average noise level on top band has improved by around 9db and the spurs are gone.
Sign In or Register to comment.