New Source of RFI

After some recent RFI sleuthing, which required at least a dozen trips up and down the stairs between the radio room and the electrical service panel, I finally identified the source of some wideband noise centred around 1,500 kHz.

This is what five seconds of the buzz looks like on an SDR waterfall...

It turned out to be the 25-year old Beam central vaccuum system. The interference is present even when the vaccum is NOT in use.

This is what one of the ports looks like...

Most RFI sources have relatively small radiating elements, such as the cable between a mobile phone and its wall-wart charger but in this case many metres of wire have been embedded in the walls to reach each of the ports.

Fortunately the vaccum is on its own circuit so keeping that breaker off when the system is not in use does not affect any other fixtures in the house.

Meyerton and Pyongyang on 3,320 kHz

3,320 kHz is occupied by transmitters in Meyerton, South Africa and Pyongyang, North Korea but at my location in grid square CN88 they are not heard at the same time because there is never a time when all three locations are even close to being in darkness. By the time each UT day begins, Meyerton has been in darkness for several hours. Not too long after that, the sun sets here, a darkness path exists and a tropical band opening is created. A couple of hours after that, the sun rises in South Africa and the signal is lost. A few hours later, while it's still night time here, the sun sets in North Korea and the signal from Pyongyang propagates well across the northern Pacific. Eventually the darkness path is lost when the sun rises here. This conveniently lets me compare two very different paths on the same frequency every UT day.

Starting in late March, I recorded 3,320 kHz for 10 consecutive days and then again for nine more days, but not quite consecutively.

The first 10 days are shown here...


Averaged together, they look like this...


The remaining nine days are shown here...


Averaged together, they look like this...

The 2019-03-19 curve terminates early because of a mistake I made with a circuit breaker in preparation for the arrival of an electrician.

During the period of these recordings I paid a lot of attention to the popular space weather indices but could not find any that correlated with the observed signal strengths. The second period had more varied signal strengths than the first but I could not even find any reason for that. The only trend I can even claim to see is that signals tend to be strongest during twilight at the receiving end of the path.

CHU on 7,850 kHz

Following on from the 3,330 kHz observations described in my previous post, over the past 10 days I have been recording the signal strength of the 7,850 kHz carrier from CHU. All of the individual 24-hour UT periods are shown here...

The curve for 2019-03-05 obviously stands out from the rest but none of the usual solar-terrestrial indices show anything unusual for that day.

The average curve for the entire period, shown below, differs substantially from the VOACAP prediction for the same path but at the most general level they do compare favourably. Both show peaks when one end of the path is in darkness and the other is in daylight and both show the signal to be much weaker when the entire path is in darkness than when the entire path is illuminated by the sun.

On February 28th and March 1st the A index was above 20 but the curves for those two days are not conspicuously different from the others. Whatever differences they may have from the average are minor compared to the unexplained event on March 5th.

CHU on 3,330 kHz

I have been aware of the VOACAP propagation prediction software for many years and have wondered how accurate its results are. My casual observations have not been adequate for drawing any meaningful conclusions so for 10 nights this month I recorded the strength of the CHU carrier on 3,330 kHz. The results are shown here...

This is the result of averaging over all 10 nights...

As expected, the signal was strong during period between when the sun sets here on the west coast and when the sun rises in Ottawa. Most interestingly, the gradual rise in strength throughout the night agrees well with what VOACAP predicts for the path on the 80 m band. I don't know if this is what most people would call a sunrise enhancement.

10 Nights of Medium Wave Propagation

For 10 nights in late January and early February I recorded the carrier strengths of eight medium wave broadcasts from transmitters in western Canada and the western US.

Here is an example showing the individual signals from a single night...

Since top end of the medium wave broadcast band is more than three times higher in frequency than the low end of the band and since the band spans the electron gyrofrequency range, I thought it would be worth dividing the frequencies into "low band" (below 700 kHz), "mid band" (near 1,000 kHz), "high band" (above 1,400 kHz) ranges...

Here are what the same looked like two days later...

It's probably worth noting that the mid-band curves are averaged over only two signals, commpared to the low-band and high-band curves which are each derived from three signals, so it should be more variable. The mid-band signals also happen to be from transmitters to the west of here while the others are from the south or near southwest.

To get a better idea of what to expect on a typical night, this next graph shows averages over the entire 10-day period...

The high-band curve is higher because one of it's three component signals is quite a bit stronger than the others. Not surprisingly, the shape of the low-band curve is the one that most closely resembles the NDB curves of my previous longwave post.

Since KSFG and KFBK both serve the Sacramento, CA area (almost identical distance and direction from here) and are both near the top end of the MW band, I singled them out for closer examination...

I am not quite willing to interpret the dusk and dawn peaks at either end of the KSFG curves as greyline enhancements because they are only based on 10 nights and because similar features do not show up on any of the other curves from the same nights.

UPDATE REGARDING THAT LAST POINT: Two local friends (Thanks Walt, Hew.) have pointed out that those KSFG peaks could easily be explained by the switch between day and night power at those times.

Comparing Northern Alberta to Coastal BC on Longwave

For five consecutive nights last week I made observations of six NDBs. The methodology was a refinement of that used for my December post on the same subject. This time the raw data came from two NDBs in Alberta and four along the BC coast. Each carrier was sampled every 5 seconds from 0000z to 1600z and then for each frequency 120 consecutive linearized values were averaged to smooth the data with advice from Nick Hall-Patch on the best way to do this based on his medium wave work. Here is the result from the first night for all six stations...

As expected, all of the signal levels rise rapidly at dusk when the entire path is in darkness and they fade back into the noise at dawn. The apparent 20-minute periodicity is presumed to be an artifact of the 10-minute averaging. For completeness, here are the other four nights...

Since the four stations on the BC coast are clustered fairly close to the longitude of my receiver location in grid CN88, I expected their curves to have some commonality not shared by the two Alberta stations to the east of here so I averaged the two groups into two separate curves for each night...

Seeing no reason to believe that spatially grouping the signal sources was of any value, I looked at how individual beacon signals behaved from night to night...

I see no source of profound geophysical insight from the above but at least they confirm some basic DXing wisdom. "Listen, listen, listen." For example, if I had tuned for 368 Sandspit BC on the 23rd at 1200 I might have concluded that it could not be heard here but checking back at 1345 would have shown it to be putting in a decent signal. Even returning at 1200 on any of the following nights would have found it to be 7 to 15 dB stronger.

This next graph simply averages each of the above signals over the five nights...

In my logbook I noted that the A index was at 19 on 2019-01-25 at 1505 due to a recent brief period of active conditions in the middle of this study so I produced these to graphs to see if any one day was significantly different from the others...

If the event did have any effect over these paths it's buried in the noise but this might be interesting to do again during a major geomagnetic storm, if there are enough NDBs left on the air by then.

The Best MF and LF DX Forecasting tool?

One week ago today my wife and I came home to a house that had been without electricity for several hours due to a major wind storm and was still in the dark.

Did I immediately I walk around the property checking for damage, such as that which might result from fallen branches? No.

Did I rush to the freezer to check for thawed food or melted ice cream? No.

Did I head straight for my best battery-powered radio, an Eton E1 with a choice of wire aerials? Yes.

As mentioned in a previous post, I have only been “seriously” listening to longwave from this location for a few weeks, after spending the majority of this past solar cycle above 530 kHz. At my previous home, in a much more urbanized neighbourhood less than 30 km away, my Perseus SDR found 224 NDBs during my most active period from 2007 to 2009.

Since getting back below 530 earlier this month I have most commonly been making 2-minute long top-of-hour recordings with the same Perseus. Careful examination of any one of the resulting files typically takes at least 20 minutes and yields 13-19 NDBs, including 10 “daytime regulars”. In all of the 24 such recordings for 2018-12-06 UT I found 35 NDBs (best DX: MB).

Last Thursday I logged 75 of them (best DX: HI, QC) during the approximately 2 hours of live DXing that I managed to get in before the power came back to my neighbourhood. By my standards, that was a resounding success; it was by far the most productive longwave session I have ever had. I attribute it to the absence of any discernible RFI, especially considering the limitations of the receiver I was using. The narrowest bandwidth of my Eton is 2.3 kHz. It’s passband tuning works well and helps make the E1 one of the best portables ever but it’s selectivity is still no match to what a Perseus can do. It’s too bad that I was not prepared to run the SDR without AC. Even better would have been to have used Mestor to start it recording at dusk.

What does the above have to do with a DX forecasting tool?

Earlier this year we had a short and very localized scheduled power outage. I used the opportunity to check noise levels on a wide range of frequencies but found nothing dramatic other than the absence of a few persistent buzzes. The outage last week was a completely different event. At one point during the listening session I used my mobile phone to quickly check the electrical utility’s service status and found that the vast majority of customers within 20 km of here were without power. Therein lies the tool.

Over the past 100 years, much research has been done on how, when and where a transmitted radio signal can be detected at large distances from its source. Most of the work to quantify this has been concerned with frequencies above 2 MHz. Others may have come to different conclusions but, from my limited experience, the resulting propagation indicies have been of little or no value for longwave. Background noise from both natural and artificial sources is what matters most. Regarding the former, the concept of a DX season stems from the reduced QRN in winter as opposed to the static crashes from numerous distant lightning strokes in the summer months. Regarding the latter, the time effort and money that dedicated hobbyists, especially MW listeners, put into DXpeditions reflects the value of escaping RFI at any time of year.

Last week’s wind storm gave me almost all of the benefit of a modest DXpedition but at zero cost from the comfort of my own home. Please note that I said almost all. A proper DXpedition is something that you plan for so that you can make the most of it. I was not prepared but I could have been. Power outages are not uncommon in SW BC, mostly because of how high winds can knock down trees which then fall on power lines. Environment Canada and other sources did provide severe wind warnings but I failed to see their silver linings.

From now on, my primary tool for choosing the very best times to search for weak signals below 3 MHz is going to be my local weather forecast. When ferry sailings get cancelled due to weather I will make sure my SDR is poised to run for hours off of SLA batteries.

Quiet Baseline for Longwave - Part 1

Ionospheric propagation is probably my number one radio interest, particularly the idea of prediction so that I can choose the best frequency or time or, ideally, the best date for a particular path.

We have these general rules about moving up and down the bands according to how high the sun is above the horizon and where we are in the sunspot cycle and propagation software that incorporates these factors has been around for years but I would like to do better than that, especially since I am interested in low power communications.

One problem with programs like VOACAP is that almost all of the work has gone into the HF part of the spectrum from 3 MHz to 30 MHz but these days I am mostly interested in the lower frequencies, partly because we are at the bottom of the sunspot cycle and partly because of the new 630 m and 2200 m allocations.

After almost a year and a half of renovating the home I moved into in 2017, I finally have a couple of descent wire aerials installed. My first project is to evaluate my local noise level under quiet solar and geomagnetic conditions and to get some quantified baseline of what to expect as "normal" propagation so that I can recognize unusual conditions when they occur.

Aircraft NDBs are wonderful for studying frequencies below 3 MHz because, unlike medium wave broadcasters, they transmit 24/7 without changing power or their antenna patterns and they constantly identify in slow Morse.

I spent much of this month collecting longwave data by recording NDBs with a Perseus SDR. My original idea was to record from 135 kHz to 535 kHz for two minutes every hour for at least three consecutive days with quiet solar and geomagnetic indices and then chart the carrier strength for each signal I found. Since this is winter, QRN from northern hemisphere storms should be low enough to give me a good chance at capturing some of the weaker signals.

Unfortunately, the human element of this did not go well. The best I could do was get through three hours of data in one hour. Manually estimating and recording signal strengths just took too long, especially for weak signals where I sometimes had to spend 10 minutes replaying a file to pull a single ID out of the noise. For 72 files it would have taken me about 50 hours just to do the data entry.

Then I remembered that the Perseus has the ability to place markers on up to eight frequencies and record the corresponding signal strengths to a text file at regular intervals along with a timestamp.

Eight was not ideal because a typical evening file had 15 identifiable signals. Still, about 10 of those were nearby “daytime regulars” of little interest for studying ionospheric propagation so eight markers should be enough for the more variable distant signals.

One weird thing I discovered is that, even when automatically logging markers from a previously recorded file, the timestamps were for the current UTC of when I was playing back the file, rather than when the file was recorded. That was a nuisance but I was able to work around it. The marker signal strength file is always named markers.log and old versions get overwritten by new ones so I included the timestamp of the original recording when renaming the file. That made it available for use when analyzing the data.

The above graph shows my result from a single 24-hour period (2018-12-06 UT). The vertical scale is in dBm but I only care about the shapes of the curves. The following graph is the result of averaging the respective ToH values over 4 consecutive UT days (2018-12-06 to 2018-12-09). Correction to both charts: The ID for 284 should be QD.

Q: What can I conclude from this?
A: Not much, other than "Distant signals are stronger at night." but this is nothing new.

That's why this post is Part 1.

Before this line of enquiry can be of any use to me, more work will be required:

1. The 284-MB signal from The Pas, Manitoba indicates a problem. It's one of the weakest regular nightly signals here but both graphs show it as being surprisingly strong during the day and not changing much at night. That's because that part of the spectrum was consistently noisy and the marker's bandwidth was between 200 and 300 Hz with the settings I had to use to span all eight frequencies. A future attempt at this should record only four stations and use the other four markers to record only the adjacent noise so that it can be subtracted.
2. Recording only the markers, as opposed to recording raw spectrum, generates very little data so next time I will sample the carriers much more often. I should not have chosen smoothed curves when producing these graphs because they are misleading considering the hourly sampling windows.
3. The eight stations were chosen for geographic diversity. I was hoping to see trends like a peak moving from east to west but that was too ambitious for only eight signal sources. Next time I should do something like comparing a group of four Manitoba stations to a group of four Alberta stations averaged over several consecutive nights of similar quiet solar and geomagnetic conditions.

The work to this point has not been a waste of time because it has served as a "shake out" to see what is realistic. I hope to come away from subsequent experiments with a general idea of how a typical night of longwave behaves. A bonus would be to come up with a credible strategy for targeting the states and provinces that are still missing from my log, especially since I decided to reset it to zero at this new QTH.

Once I get that sorted out I would like to compare propagation on nights of quiet solar and geomagnetic conditions to nights of more active or disturbed conditions as we claw our way out from the bottom of the sunspot cycle. Even if nothing meaningful comes of it I should at least have good data from which to draw that conclusion.

Wildlife and the DXer

The deer around here have no idea how much effort I have put into keeping them from trampling my feedlines or getting tangled in my aerials.

ABSC Nostalgia

A recent comment on an earlier post got me thinking about other experiments I did with kits of parts from the American Basic Science Club so I opened up a box and extracted these items.

Loop Comparison

LW, MW and HF listening did not become practical at my current location until I got my first Wellbrook active loop antenna (an ALA1530). Even when I just had it sitting out on the wood-frame balcony outside my home office it was a huge improvement over the indoor random wires I had been using up to that point. The noise level was low enough that I could rely on it for reception of much more than WWV and CRI. When I moved it away from the house to be mounted NE-SW on a nearby fence the noise level dropped a further 5 dB to 10 dB on most of the frequencies I tested. I call this Loop A.

A few years later I added a second Wellbrook loop (an ALA100) with four turns of wire wound on a wood frame and mounted on a different fence panel so that it favours NW-SE. I call this Loop B. It's centre is about 5 metres away from Loop A.

The plan was for Loop A to favour Europe and the South Pacific on HF while nulling the MW broadcasters in northwest WA. To compliment that, Loop B was to favour Asia, Alaska, the Lower 48, and the Caribbean on HF while nulling the MW broadcasters in the lower mainland of BC. That has all turned out as expected and the spacing between them has also been adequate for creating a deep electronically steerable null on MW using a DX Tools RF+ Quantum Phaser (not nearly as lethal as it sounds). I also hoped that on HF either loop could serve as a reference for the other when connected to an MFJ-1025 noise canceller. Unfortunately that turned out to be a complete waste of time and money because I did not find a single signal that benefited from use of the device.

After a recent post to the Shortwave Radio Station Listening group on Facebook I was asked about how well Loop B works. Such things are difficult to quantify so the best I can add to what I have said above is to refer to the following two frequency response curves from yesterday afternoon.

Loop A

Loop B

At almost all frequencies the output of Loop B is lower than that of Loop A but this is to be expected because, even with four turns, Loop B is built with much less "capture area" than recommended by the manufacturer. What else do the two images reveal? As expected, my only local MW broadcaster is clearly weakened by being in the null of Loop A and multiple Vancouver stations are brought down by being near the axis of Loop B. Aside from that, the two curves show differences in the sizes and locations of their peaks and valleys but there is no way to know, from this experiment alone, the extent to which that is due to differences in their frequency response or to differences in their location and orientation. For example, the electromagnetically noisiest place in my house is the service panel where AC power, cable TV and ADSL internet all converge and it is almost directly in the null of Loop A while the orientation of Loop B favours reception of noise from my nearest neighbours.

Switching from Loop A to Loop B

This last image shows what happened in the vicinity of a broadband noise source when I switched from Loop A to Loop B. I am sure that the frequency shift means something but I currently have no clue about what that might be or how to use the information.

Signal Tracer and Injector

On Christmas Eve I finished building this simple piece of test equipment from QRPkits. It's a combined signal tracer and injector. It doesn't do anything that I cannot accomplish with an oscilloscope and a function generator but it's much more convenient to use.

As with the Accuprobe described in my previous post, this device uses heat-shrink tubing as a case. At Joanna's suggestion, instead of using a hair dryer as a heat source, I simply place the unit on top of a space heater (as shown in photo) with the air flow directed vertically. This worked much more quickly and provided more uniform heating.

Accuprobe for Measuring RF Voltages

Today I finished building my latest piece of test equipment -- the N2CX Accuprobe Plus from Pacific Antenna. It was an easy, high-quality kit. The only odd bit was using heat-shrinkable tubing in lieu of a tubular case. Without a proper heat gun I had to improvise with a hair drier but it worked well enough.

For testing, I compared the results obtained with it against the RMS values I calculated from the peak-to-peak readings off of an old Kenwood oscilloscope when using a function generator feeding my Elecraft dummy load for weak signals and a Yaesu FT-817 as a source for QRP power levels. The accuracy was what would be expected from the specs and much better than the readings made with my old Heathkit probe. However the latter would still be the device to use for voltages over the Accuprobe's 35 V limit.

Elecraft DL1 Dummy Load

Around the same time that I got the ZM-2 shown in my previous post, I also bought an Elecraft dummy load kit that you can find on this page.

One nice thing about it is that it includes a diode, capacitor and two convenient test points for measuring power with any high input impedance voltmeter. The manual comes with a graph for converting the voltage readings to watts. Using an inline meter to measure forward power as a reference, I got considerably more accurate results by using an old Heathkit RF probe across the load. However he test points are convenient and for my needs would be perfectly adequate to get relative readings for peaking the output of a QRP transmitter.

Emtech ZM-2 Antenna Coupler

Several years ago I bought this kit but only got around to building it last week. The manufacturer calls it an antenna tuner but I insist upon referring to it and other such devices as an antenna coupler because it doesn't actually tune an antenna. Aerials can be tuned by changing the sizes of their elements or the values of their lumped reactances but this device is not part of any antenna. It's function is to couple a feedline to an aerial so that energy can be transferred from one to the other. I am also willing to call it a transmatch because it does its job by transforming or matching impedances.

The ZM-2 is a nice little unit that if very easy to adjust without the need for an external SWR meter. The empty holes in the corners are for screws that I still have to buy to hold the faceplate to the plastic box.

ABSC Regen Meets Ocean Hopper

After receiving generous constructive feedback from members of the Regenerative Receiver Initiative Facebook group and others, I have decided to abandon the idea of reviving my ABSC regen in its original form and, instead, to merge it with the even older Knight Ocean Hopper (OH) design. The above circuit is my current idea of how to create something similar to the OH while still reusing as many components as possible from the ABSC radio.