SOLVED: Identifying interference with oscilloscopes and software-defined radio

[djhuft]

Hello,

I run station #1500 and am trying to reduce local RF interference so that my receiver performs better. My station statistics suggest that I am receiving lots of interference, so I have been trying to identify and hopefully eliminate some of them. I live in a row of apartment buildings, so I have many neighbors with lots of  electrical equipment. Finding and fixing everything will be nearly impossible.

There has been much written on interference, particularly this thread: http://www.wxforum.net/index.php?topic=20439.0.

I previously used my System Blue's web interface for trying to diagnose interference, it was poorly suited for the task. The time-domain signals that it displayed could not give me any frequency information, and also were not useful for tracking the interference over long time periods. Therefore, I am documenting two methods I found to be much more powerful and insightful.

Oscilloscope Method
Earlier in the summer, I was using an oscilloscope to examine the signals that my System Blue was receiving. I soldered the four optional SMA connectors to the PCB (1). These connectors provide access each of the four amplified antenna signals. I connected these signals to my oscilloscope (2) and used the scope's FFT mode to view their frequency spectrum. From this, I was able to identity several spurious frequencies.

After a lengthy process of elimination, I found that one very significant source of interference was the network switch that my receiver was connected to. It was a large 24 port TP-Link Gigabit Ethernet switch, and it was actually conducting the interference through the Ethernet cable into my receiver. I solved this problem by instead connecting my receiver to a port on one of my WiFi routers. Apparently, that router was less noisy than the network switch and did not tend to conduct noise through its Ethernet jacks.

After fixing this, I still had a few sources of interference. I still received a constant signal at 25 kHz and a flickering one at 60 kHz. I tried turning every piece of electrical equipment in my apartment off, but the interference was still there..


Software Defined Radio Method
After much experimentation, I stumbled upon an even better way to troubleshoot the interference. Basically, it involves connecting an inexpensive software-defined radio (SDR) dongle to my receiver, and viewing the received signals on a PC. Similar to the oscilloscope method above, you must install the optional SMA jacks on your System Blue board and connect the dongle to them (3).

I am using an R820T2 RTL-SDR dongle (approx $20). This particular model is capable of receiving between 24 – 1766 MHz, which is obviously not the correct frequency range.

To receive the low-frequency signals that I am interested in, I use the Ham It Up Upconverter. This is connected between the System Blue and the RTL-SDR dongle. It shifts its input signals up in frequency by 125 MHz, where they can then be received by the RTL-SDR dongle. (4)

The final thing required is software to view the signals. I am using GQRX (5). It has an instantaneous frequency display, similar to the FFT plot on my oscilloscope. More importantly, it has a nice waterfall display that allows me to view the received frequencies as they change over time. The time scale on the waterfall can be adjusted from a few seconds to many hours, which is extremely useful.

LED Lighting Causing Interference
Here is a short capture that I took while turning on and off a fixture with six LED light bulbs. The timescale of the waterfall is about 30 seconds. The area circled in red, just above 70 kHz, shows emissions that appeared each time I switched the bulbs on. Perhaps I should find different bulbs, or switch to incandescent.



Long Time Spans
Here is a very long capture that I made starting late at night. In this case, the time scale on the waterfall is 10 hours! Notice the increased emissions near the top of the waterfall. These appear at about 8:00 and 9:20 AM. One might suspect that they might be caused by my neighbors switching on electrical devices after getting up in the morning. I was still sleeping at the time, so I could not have caused them.

Also note that there are significant emissions at 25 kHz and 60 kHz. These are the signals that I originally identified using my oscilloscope! They seem to be present all the time.

Conclusion
I still have not found the source of my two main interfering signals, but in the process of searching for them I think I developed some useful and powerful methods for investigating interference. Hopefully others will find this information useful.

If anybody has questions or wants clarifications on anything I discussed, please ask. I'll try to post more screenshots if I find any other interesting interfering signals.

Footnotes:

(1): Mentioned in the "Digital filter option" section of the assembly instructions
(2): Be sure to switch your station to fixed gains, not "automatic mode" if you want to make measurements, otherwise you will see sudden changes in signal amplitude whenever the gain is automatically adjusted by the receiver's microcontroller.
(3): I drilled an extra hole into my receiver's case, then installed a small SMA cable from the connector on the PCB out to the outside of the case. This gives me a convenient connection for my experimentation without the need to open the case.
(4): Software-defined radio receivers have become extremely popular in the past several years, and many different models are available.  Other models should certainly be suitable, and some are able to receive signals below 1 MHz without the need for a separate up-converter.
(5): Similar programs exist for Windows, for example PothosSDR

[djhuft]

Here is a photo of my test setup:


And here is a close-up of the connection that I added which connects to the internal amplifier:

[djhuft]

I recorded a bit more since last post. Keep in mind that you are seeing the spectrum between DC and 120 kHz.

10 hours of recording through the night:


Another 10 hours from the daytime:


There are lots of interesting signals here, in my opinion:

• Both images: There is broadband interference across the spectrum periodically.
  
• Both images: There is a signal that sometimes appears and drifts slowly from low 60 kHz range to steady-state above 70 kHz. Maybe this is some sort of thermal effect inside the interfering device?
  
• Both images: The signals at 25 and 60 kHz are very strong and still present all the time.
  
• Both images: There is a strong signal at 45 kHz that comes and goes.
  
• Second image: There is a broadband interferer between 8 and 15 kHz that occasionally appears.
  

I'm not sure what to conclude...

The next time I am away from the apartment, I plan to shut off every circuit breaker, except for the one powering my laptop and Blitzortung receiver. Maybe some of these strange signals will disappear. If not, I probably need to look outside my apartment.

Also, maybe I should make a 5V battery backup for my Blitzortung receiver, then pray for a neighborhood-wide power outage.    I could get a lot of interesting data then!

[djhuft]

One final experiment for the night...

I still have my software defined radio dongle connected to my Blitzortung receiver's amplifier output.

1. I turned all the (LED) lights on in my apartment, and switched on as many gadgets and devices as possible (my PC, fans, radio).
   
2. I powered my Blitzortung receiver from a USB battery pack.
   
3. I disconnected my laptop charger (the laptop that I used to record this data)  from its outlet. The charger generates significant interference, which tends to confound my measurements.
   
4. I set up my software-defined radio dongle and GQRX to record a waterfall plot that was five minutes long.
   

Then I started recording...

After a couple of minutes, I turned off all the circuit breakers in my house.
(Lower on the waterfall is earlier in time.)

Lots of noise is eliminated, but not everything! Those pesky 25 kHz and 60 kHz signals are still there..

Next, I left everything off for a couple minutes, then I turned the breakers back on.

You can see interesting emissions as my devices powered back up. I suspect that this coincided with routers booting up, etc.

[opadavis2]

Thank you for this information. I recently started up my Blue and am getting very poor signals on both E and H antennas. I already have 4 of 6 inch BNC to SMA cables coming in by mail and look forward to troubleshooting. I think I have room to mount all 4 BNC bulkhead connectors to the back panel. 
Peter

[pasense]

You should consider to use the low pass filters. With them I could cut out the high frequency noise and could concentrate on hunting the noise sources around 10-30 kHz. Soldering the tiny filters in was far easier than I feared, but they are of course a bit expensive.

[Breitling]

You should consider to use the low pass filters. With them I could cut out the high frequency noise and could concentrate on hunting the noise sources around 10-30 kHz. Soldering the tiny filters in was far easier than I feared, but they are of course a bit expensive.

Hi pasense. I don't want to hijack this magnific thread, but I have doubts about soldering the low pass filters (I already have them). Any link or post where to see the PCB place and soldering tips?

Thanks !!

[pasense]

Hi Breitling,
two answers: I took a quick lock at the signal from your station, and the major noise is at 15-16 kHz, so this cannot be removed by the filters. On the other hand, the smaller peak at 66 kHz can be suppressed. 
I used the following procedure to solder the filter ICs: First, I applied a bit of flux with a pen dispenser on the legs of the filters and on the pads on the board. The soldering iron had a flat tip of 4 mm width, like the tip of an old screwdriver. This tip then covers all 4 legs on one side of the package. The soldering iron was on a middle temperature, not on max and I wetted its tip only with a small amount of solder. Then I used a pincer to hold the filter at the two sides were there are no legs and placed them exactly on the pads on the board and pressed the legs on one side down with the flat tip of the soldering iron. The solder on the pads melted immediately and the legs moved down a bit. I then withdrew the soldering iron sideways, so there were to solder bridges between the legs and waited of few seconds until the solder had cooled down. Then I rotated the board and soldered the four legs on the other side. In this way I did'nt need a very fine tip, the broad one did even better, it allowed to solder all four legs in one go. The amount of solder which is already on the pads is sufficient to fix the filters and connect them. 
I should add that I am very short-sighted, so without my glasses I can see very good at short distances and I did'nt need a magnifying glass in order to control the whole operation. .

[djhuft]

You should consider to use the low pass filters. With them I could cut out the high frequency noise and could concentrate on hunting the noise sources around 10-30 kHz. Soldering the tiny filters in was far easier than I feared, but they are of course a bit expensive.

I am seriously considering buying those filters. I read in the instructions manual that they are enabled in software by soldering closed a jumper. I'm very curious what options appear in the web interface when you do this. Is it possible to adjust the frequency manually, and what are the limits??

Thanks!

[cutty]

You should consider to use the low pass filters. With them I could cut out the high frequency noise and could concentrate on hunting the noise sources around 10-30 kHz. Soldering the tiny filters in was far easier than I feared, but they are of course a bit expensive.

I am seriously considering buying those filters. I read in the instructions manual that they are enabled in software by soldering closed a jumper. I'm very curious what options appear in the web interface when you do this. Is it possible to adjust the frequency manually, and what are the limits??

Thanks!

The filters are set for cutoff frequency in settings. these settings are NOT visible unless the devices are installed, you have the latest firmware, and the microprocessor is enabled with that jumper you mentioned. The 'idea' is to work between 3K-300KHz, but the actual high end is 328 in settings. On the "Low end", you can virtually turn a channel off... so say "0 to 328K" effective. 
There's no need for these filters if you don't have the higher frequency noise. And eliminating noise with them between 3K and 30K actually kills a lot of the sferic information we're looking for. And they do work...extremely well.  However, I run mine wide open...why not? Same as not having 'em in there... ... my noise is all in that 'critical area' 3-30kHz, unfortunately... and it's sporadic.

[djhuft]

Longest capture yet, approximately 40 hours.

Here I set the waterfall to its longest time span possible, 48 hours, in hopes that I could see some daily patterns.


I have noted the approximate times of night and day here, which may correlate with some interference. Nothing really stands out though.

One interesting finding is that the strong 25 kHz signal I have been mentioning has been consistently present, until now. Here you can see the following, circled in purple:

• Disappeared at about 06:50
  
• Reappeared at 11:30
  
• Disappeared again at 12:00
  
• Reappeared at 14:30 and has remained since then.
  

(I added the time information from tooltips that I can see if I hover my mouse in GQRX.)

Anyway, it is possible that this signal can go away! I have no other clues about its origin just yet.

I would really like to make a portable sensor that I could use to do direction finding and triangulation to pinpoint the source. I'll keep you posted..

[djhuft]

There's no need for these filters if you don't have the higher frequency noise. And eliminating noise with them between 3K and 30K actually kills a lot of the sferic information we're looking for. And they do work...extremely well.  However, I run mine wide open...why not? Same as not having 'em in there... ... my noise is all in that 'critical area' 3-30kHz, unfortunately... and it's sporadic.

Thanks a lot, Cutty! That's exactly what I was looking for. It's very helpful to know what frequency band is primarily of interest. I may put the filters in to block my interference at 60, 57, and possibly even as low as 45 kHz. It's too bad the the chips are over $10 apiece.. It's an expensive experiment to install them.

One other question: are the filters independently adjustable, or is there only one setting that applies to all channels?

[kevinmcc]

One other question: are the filters independently adjustable, or is there only one setting that applies to all channels?

They are independently adjustable.

[djhuft]

They are independently adjustable.

Thanks, Kevin.

[djhuft]

A discovery.

I found the source of my 60 kHz interference - it is actually the atomic time station WWVB.

I figured this out by listening to the signal in GQRX. I tuned in and listened in single side-band (SSB) mode, and realized that it sounded like an on-off-keying modulated signal.

I then looked up a list of very low frequency (VLF) stations, and found this very useful page: http://www.mwlist.org/vlf.php There are three stations listed that transmit at 60 kHz, particularly WWVB, which is located in Fort Collins, Colorado and transmits at 70 kW. I live in Spearfish, South Dakota, which is about 240 miles away. There is no doubt that I should receive this station loud and clear.

A quick test confirmed my suspicions.. I rotated my ferrite antenna and watched for the signal to null out. Sure enough, when I pointed the axis of the ferrite stick in the direction of Fort Collins (my south-southwest), the signal disappeared. Mystery solved.

The Wikipedia page about WWVB is an interesting read. This station transmits the signal that most radio-controlled clocks in the US use to synchronize their time. You can listen to what the transmission sounds like in this YouTube video.

Anyway, it took me a long time to figure this out... I bet any old Ham operator worth his salt would have known right away.

I suppose I'll install the low-pass filter ICs on my receiver to block this station.

Now I'll try to figure out the origins of the 25 kHz signal. That one definitely sounds like electrical noise..

[djhuft]

25 kHz Source Found!

When I last posted, I had a slight suspicion in the back of my mind that the 25 kHz signal might also be VLF station. I now confirmed that.

The station is NML4, a US Navy transmitter broadcasting from Lamoure, ND. I verified the direction again by rotating my antenna and watching the amplitude. As I expected, it is coming from the northeast (or southwest, but I don't know of any other stations in that direction).

I calibrated the frequency of my SDR dongle and upconverter system against WWVB so that it was as accurate as I could make it. The oscillators in the SDR and upconverter are not perfect, so this was a worthwhile step. The resolution of the SDR is not great, but I measured the frequency of the interferer to be very close to 25.2 kHz, just what I expected.

---

So, I suppose this mostly wraps up my investigation.. I ordered digital filters and will be installing them shortly.

The interference from NML4 is unfortunate. I suppose I will aim most of my antennas to avoid it. Adjusting one or more channel's filters to block 25 kHz and above is probably counterproductive.

It is interesting to note that another user encountered a very similar situation in 2014:

I recently got a RED station set up in the Pacific Northwest area of the US (Seattle, WA), and it turns out that the Jim Creek Naval Radio Station (callsign NLK) is broadcasting on 24.8 kHz at 1.2 MW just 50 miles to the north of me.  Needless to say it's coming across loud and clear on my station.  I carefully oriented my H-field antenna so that one channel was nulling-out the transmitter to the best of its ability, but the signal is still obviously there and the gain isn't even very high (8*2).  Is there any secret tricks for combatting this or should I just give up and sell the station to someone who can make better use of it?

None of the developers had a great solution, and just recommended doing nothing.

[Tobi]

@djhuft: You did some great work! A long time ago I did use a soundcard to check and identify different sources of noise like you did with the SDR. Worked great. It's normal to see submarine or time sync transmitters.

[RichoAnd]

Hello,

I run station #1500 and am trying to reduce local RF interference so that my receiver performs better. My station statistics suggest that I am receiving lots of interference, so I have been trying to identify and hopefully eliminate some of them. I live in a row of apartment buildings, so I have many neighbors with lots of  electrical equipment. Finding and fixing everything will be nearly impossible.

There has been much written on interference, particularly this thread: http://www.wxforum.net/index.php?topic=20439.0.

I previously used my System Blue's web interface for trying to diagnose interference, it was poorly suited for the task. The time-domain signals that it displayed could not give me any frequency information, and also were not useful for tracking the interference over long time periods. Therefore, I am documenting two methods I found to be much more powerful and insightful.

Oscilloscope Method
Earlier in the summer, I was using an oscilloscope to examine the signals that my System Blue was receiving. I soldered the four optional SMA connectors to the PCB (1). These connectors provide access each of the four amplified antenna signals. I connected these signals to my oscilloscope (2) and used the scope's FFT mode to view their frequency spectrum. From this, I was able to identity several spurious frequencies.

After a lengthy process of elimination, I found that one very significant source of interference was the network switch that my receiver was connected to. It was a large 24 port TP-Link Gigabit Ethernet switch, and it was actually conducting the interference through the Ethernet cable into my receiver. I solved this problem by instead connecting my receiver to a port on one of my WiFi routers. Apparently, that router was less noisy than the network switch and did not tend to conduct noise through its Ethernet jacks.

After fixing this, I still had a few sources of interference. I still received a constant signal at 25 kHz and a flickering one at 60 kHz. I tried turning every piece of electrical equipment in my apartment off, but the interference was still there..


Software Defined Radio Method
After much experimentation, I stumbled upon an even better way to troubleshoot the interference. Basically, it involves connecting an inexpensive software-defined radio (SDR) dongle to my receiver, and viewing the received signals on a PC. Similar to the oscilloscope method above, you must install the optional SMA jacks on your System Blue board and connect the dongle to them (3).

I am using an R820T2 RTL-SDR dongle (approx $20). This particular model is capable of receiving between 24 – 1766 MHz, which is obviously not the correct frequency range.

To receive the low-frequency signals that I am interested in, I use the Ham It Up Upconverter. This is connected between the System Blue and the RTL-SDR dongle. It shifts its input signals up in frequency by 125 MHz, where they can then be received by the RTL-SDR dongle. (4)

The final thing required is software to view the signals. I am using GQRX (5). It has an instantaneous frequency display, similar to the FFT plot on my oscilloscope. More importantly, it has a nice waterfall display that allows me to view the received frequencies as they change over time. The time scale on the waterfall can be adjusted from a few seconds to many hours, which is extremely useful.

LED Lighting Causing Interference
Here is a short capture that I took while turning on and off a fixture with six LED light bulbs. The timescale of the waterfall is about 30 seconds. The area circled in red, just above 70 kHz, shows emissions that appeared each time I switched the bulbs on. Perhaps I should find different bulbs, or switch to incandescent.



Long Time Spans
Here is a very long capture that I made starting late at night. In this case, the time scale on the waterfall is 10 hours! Notice the increased emissions near the top of the waterfall. These appear at about 8:00 and 9:20 AM. One might suspect that they might be caused by my neighbors switching on electrical devices after getting up in the morning. I was still sleeping at the time, so I could not have caused them.

Also note that there are significant emissions at 25 kHz and 60 kHz. These are the signals that I originally identified using my oscilloscope! They seem to be present all the time.

Conclusion
I still have not found the source of my two main interfering signals, but in the process of searching for them I think I developed some useful and powerful methods for investigating interference. Hopefully others will find this information useful.

If anybody has questions or wants clarifications on anything I discussed, please ask. I'll try to post more screenshots if I find any other interesting interfering signals.

Footnotes:

(1): Mentioned in the "Digital filter option" section of the assembly instructions
(2): Be sure to switch your station to fixed gains, not "automatic mode" if you want to make measurements, otherwise you will see sudden changes in signal amplitude whenever the gain is automatically adjusted by the receiver's microcontroller.
(3): I drilled an extra hole into my receiver's case, then installed a small SMA cable from the connector on the PCB out to the outside of the case. This gives me a convenient connection for my experimentation without the need to open the case.
(4): Software-defined radio receivers have become extremely popular in the past several years, and many different models are available.  Other models should certainly be suitable, and some are able to receive signals below 1 MHz without the need for a separate up-converter.
(5): Similar programs exist for Windows, for example PothosSDR

Nice explanation - I could not do better :-)
I use myself Oscilloscopes, has 5-6 pieces

Oscilloscopes with FFT analyzer https://www.vellemanusa.com/products/vie...s&lang=enu,

UNI-T https://www.globalmediapro.com/dp/A2H3Y2...pe-100MHz/

HP 3590A Wave Analyzer http://www.hpl.hp.com/hpjournal/pdfs/Iss...968-12.pdf,

EMU0202 Audio Interface http://www.creative.com/emu/products/pro...?pid=15186,

The built-in Scope / FTT analysis,

SDR radios with converters, long wave radios,

EMV Spy http://www.box73.de/product_info.php?products_id=2763,

Krohn-Hit model 3343 https://doc.xdevs.com/doc/Krohn-Hite/KRO...enance.pdf

And lots of other equipment

_________________
First of all, "our" frequencies of interest are filled with noise, 3-30 (100) KHz, and most of it comes from lightning discharges.

There is ALWAYS natural noise, and most in areas with lots of lightning.

If we want high accuracy, we need signals with short rise time (high frequency) - It is achieved only close to the flash discharge (10-20 to 100-200km)
When we get closer, we are disturbed by the "precharges", and are we farther away, the higher frequencies are attenuated, and finally there are only signals around 10-12KHz
Dedicated receivers for long distance have a narrow range around 11KHz 

Most spikes you see in an FFT analysis mean so little that there is no reason to dampen them further

I can not see you need extra filters

Finally I show pictures of what different filters do.
12KHz is very hard, but here it works well on middle and long distance.

/Richo

[JupiterJoe]

This is all so interesting. I'm afraid it's above my level of brain power! (So far, but I'm trying to learn!)
I do have a new guy question though. On so many stations, mine included, I always see a large 1khz spike(shown on Richo's attachments). What is that caused from, if known?
Thank you all for so much information! I do love learning about this.

[djhuft]

First of all, "our" frequencies of interest are filled with noise, 3-30 (100) KHz, and most of it comes from lightning discharges.

There is ALWAYS natural noise, and most in areas with lots of lightning.

If we want high accuracy, we need signals with short rise time (high frequency) - It is achieved only close to the flash discharge (10-20 to 100-200km)
When we get closer, we are disturbed by the "precharges", and are we farther away, the higher frequencies are attenuated, and finally there are only signals around 10-12KHz
Dedicated receivers for long distance have a narrow range around 11KHz 

Most spikes you see in an FFT analysis mean so little that there is no reason to dampen them further

I can not see you need extra filters

Finally I show pictures of what different filters do.
12KHz is very hard, but here it works well on middle and long distance.

/Richo

Thanks a lot, Richo! It is very interesting to know this information about the physics and various frequencies.

I just captured the following screenshot with my oscilloscope:

The large peak in the FFT plot is actually centered at 25.2 kHz, and its amplitude is about 10 mVrms.
In comparison, my thresholds are set at +/-200 mV, so you are correct, the interference is indeed making only a very small difference. I should have considered this sooner.

It is interesting to consider the effect of filtering on timing too. I can understand why it is important to balance signal rise time requirements with filtering in order to get accurate timing information. After all, radio signals travel 1 km in 3 microseconds. .

Very interesting! Thank you.