Shortwave – Engineering Radio

Ground Conductivity

An important part of any medium-frequency and high-frequency antenna system is the nature of the ground immediately around the antenna system. The reason for the complex and expensive AM ground system is to create a return path for the portion of the RF wave that is not radiating from the tower. Higher ground conductivity helps this process greatly. The higher the ground conductivity, the more efficient the antenna will be.

Ground permittivity is relative to that of free space which is 1. In general, it is a measurement of how much polarization of the ground material occurs in the presence of an electric field. Relative permittivity is a unit-less number indicating the dielectric constant. It is an important factor in ground wave propagation modes like daytime AM radio. Lower frequencies have greater depth of ground penetration and permittivity goes up. That is why low dial position AM stations have better daytime ground wave coverage per kilowatt then the high dial position stations.

Estimated relative ground permittivity by composition, Courtesy of em.geosci

According to the FCC M3 map, the ground in this general area is supposed to average 4 milliMhos per meter (milliMho is equivalent to a milliSiemen). This is a very rough estimate as conductivity can vary greatly within short distances and can change depending on soil moisture. This is the reason why many AM transmitter sites were built in swamp land. Wet soil has higher conductivity and permittivity. Plus, the land was cheaper.

Estimated ground conductivity by frequency and moisture, ITU recommendation P.527.3

When modeling antennas with NEC, particularly vertical antennas, the ground conductivity plays a significant part in the antenna performance. At lower frequencies, the permittivity (aka dielectric constant) also plays a role. Having actual numbers instead of the FCC estimate gives a better overall model of the antenna.

All of this made me think about how to obtain better data. Fortunately, a radio amateur (Rudy Severns, N6LF, Hardy Lau, DL1GLH and Tobias Wellnitz, DH1TW) have thoroughly documented a good procedure. This is termed “Open Wire Line” method, which uses two metal probes that are the end of an Open Wire transmission line. A VNA is then connected to the probes and the complex impedance is measured. The resistance and reactance are then plotted along with the physical aspects of the probes to calculate conductivity and permittivity. More information can be found at How to measure Ground Conductivity by Tobias, DH1TW.

Once again, I took stock of what I had laying around and found a pretty good way to make a ground conductivity probe. The first thing is to make an insulated holder/handle for the two metal probes. I have a large selection of oak planks, which have been used for many projects in the past. In this case, a small piece of 1 x 2 was cut down to 16 inches. I then used the router to round off the edges and laid out the holes for the probes and N connector. I finished the oak with several coats of polyurethane to keep moisture out of the wood grain. I purchased a three foot piece of 1/4 inch (6.35 mm) mild steel round stock, which I cut in half. Using a 1/4 28 die, I threaded one end of each piece so I could attach it to the oak handle. The probes are spaced 50 mm apart and are 300 mm long. The steel probes were connected to the N connector with two pieces of copper strap.

Name	Quantity	Use	Cost	Comment
Oak 1 x 2″	`1	Handle	$0	Left over
1/4 x 36″ steel round stock	1	300 mm probes	$6.99	Home Depot
1/4 28 nuts	4	Attach probe to handle	$3.99	Home Depot
N Female Connector	1	RF input	$0	Salvaged
Copper Strap	5 inches	Connect probes to RF input	$0	Left Over

The land around my house was farm land for a long time. First field crops then a grazing pasture for a dairy farm. The last farming activity was about 65 years ago, since then the trees have grown in. The soil itself is mostly glacial till; rocks, sand and clay. I do not expect much.

I headed into the woods with my Siglent network analyzer and measure the ground conductance. I then saved that information as a .csv.

I have since saved my ground conductivity sweep profile in the analyzer. I reduced the points down to 180, which gives at least one point in every amateur band from 630 through 10 meters. That makes the marker settings redundant and .csv output much more manageable. I have another sweep profile for just the AM Broadcast band, this one runs from 530 – 1,710 KHz and has 118 points, which provides a measurement on every carrier frequency.

I sent the .csv into ChatGPT and had it analyze the data. When it comes to AI, it is all about the prompt. The more information and instruction provided, the better the output will be, usually.

The ChatGPT prompt:

Calculate the ground RF conductivity and permittivity using the open wire line method. Measurements are made with a Vector Network Analyzer using S11, data is Real/Imaginary, ground measurement probe consists of two steel rods, 6.35 mm in diameter, 300 mm long, spaced 50 mm. The rods are fully inserted into the ground. Include probe end fringing and steel conductor losses. Show all math and proof for the first frequency only. Use this data: cgmfhf1.csv

I have found a need to be careful with any AI. ChatGPT tends to be over optimistic. My first attempt, the output had a ground conductivity of 12 mS/m which looked great until I dug a little deeper. It seems ChatGPT copied the answers from DH1TW’s blog post. When I ran the numbers myself, it was more like 1 mS/m. I scolded that lazy AI and told it to do it again, this time with the proof. That is why I want to see the math for the first frequency. Here is a chart with the ground conductivity and permittivity in my antenna field for the MF and HF amateur frequencies:

Here is the math for the first frequency:

I always ask to see the math for the first frequency so I can check the AI’s work and make sure that it is not being lazy again.

Running some models in MANNA-GAL and I feel like the results are much more similar to what I am seeing WSPR beacon.

Golos Ameriki

The Voice of America; expended goodwill edition.

Radio Liberty and Radio Free Europe were The main sources of Western information behind the Iron Curtain during the Cold War. While the VOA, RFE, and RL used HF, there were also FM relays in the mix. If you wanted to know what the US Government’s position was on any topic, VOA, RFE, and RL were the information sources.

I wrote several articles about this in the past:

So what happened? Why has the current administration shuttered those services?

There are several good reasons why many government broadcasters have reduced or eliminated shortwave:

Fewer people care about the US Government’s position
Changes in geopolitics
Reduced listenership to HF Shortwave Broadcasts due to technical difficulty
Large double-sideband AM transmitters are expensive to operate and maintain and therefore are a target for reducing expenses
HF transmitter sites require a lot of land and physical infrastructure, which is also expensive to maintain
New distribution technology is easier for the end user and less expensive to operate

Many people, particularly young people, do not know the difference between an over-the-air broadcast and an internet stream. Buying a special receiver, putting up some indoor or outdoor antenna, then tuning around several different frequency bands to find something worth listening to, seems like a lot of work. These days, there are few shortwave broadcasts worth listening to, especially in the English language.

The BBC greatly reduced HF distribution of The World Service starting in 2005, favoring more internet-based distribution. Radio Canada International completely went off the air in 2012, scrapping its extensive transmitter site in Sackville NB in 2014. Radio Australia signed off in 2017. Deutsche Welle mostly left the HF band in 2011 while reducing its FM in 2016. The Voice of Russia ended HF broadcasts in 2014.

The only state-owned shortwave broadcaster that has expanded is Radio China International.

VOA Greenville B has some very old transmitters. The newest units date from the late 1980s or early 1990s. The oldest are the two original Continental 420As, dating from 1960, and are original to the building. When I visited there in 2017, two transmitters were on the air, the BBC and the AEG broadcasting at half power to Cuba and Africa respectively. The rest were shut down. The Continentals were difficult to change frequencies on because of the Doherty modulation. The GEs were long in the tooth, but at least serviceable due to the stock of spare parts from site A. I think the overseas sites in the Philippines and Sao Tome are similar.

Many have pointed out, and rightly so, the Internet censorship issue. Terrestrial radio broadcasting is often the best or only way to circumvent the suppression of information. Kim Elliot pointed this out in his Radio World article “Why we need Shortwave 2.0” All of those points are valid.

What can be done? Implementing DRM30 as a worldwide HF broadcasting standard would be a step in the right direction. DRM30 can send ancillary data, including Radiogram type news bulletins. DRM30 is much more energy efficient than DSB AM because there is no carrier, which wastes half or more of the transmitted power on a carrier that contains no information. Instead of a giant transmitter site, with curtain arrays, a more distributed transmission system with several frequencies on the air at the same time uses lower-powered transmitters, simpler antenna systems such as Rotating Log Period Arrays (RLPA), or non-directional vertical towers. This would require some changes to the FCC rules, but now is the time for that.

TV’s ATSC 1 has something called a “Transport Stream ID,” (TSID) which is a unique number assigned to each broadcaster. Wide-band SDRs are capable of scanning across many HF bands. Implementing something similar for DRM30 HF broadcasts would not be that difficult. Shortwave Listeners just program the HF TSID to lock onto the digital broadcast of their choice, if it is available. This would make HF Broadcasting available to most non-technical people looking for information. Most of this can be done with existing technology. However, DRM still (almost 2 decades later) lacks receivers. There is a development on that front as well: RF2Digital support module. The point is that there are many good ways to improve the technology, keep HF broadcasting relevant, and bypass attempts at internet censorship.

What will be done?

I Audited the RF Noise in my House

The largest problem facing analog AM broadcasting (and digital Medium Frequency and High Frequency broadcasting) is RF Noise.

Like most people, I have many modern conveniences that make my life easier than previous generations; electric lights, central heat and air conditioning, appliances like vacuum cleaners, microwave ovens, and whatnot. I enjoy the wireless internet, have an LED TV, use LED light bulbs, and get free electricity from my photovoltaic solar system. These devices can contribute to the high levels of RF noise found in most buildings. RF Noise which is the bain of AM broadcasting. Digital modulation schemes use variations in amplitude to transmit data bits. They are not immune to RF noise, they simply mask it better until they don’t.

I thought it would be interesting to isolate the various noise generators that may be present.

To make measurements, I used the Siglent SVA-1032X spectrum analyzer. This unit has a noise floor of -140 dB. My methodology is to turn everything off except the Device Under Test. Set the spectrum analyzer up for a wide band sweep, then narrow the bandwidth on any detected noise. Turn the DUT off to make sure that the noise goes away. Turn the DUT back on to make sure that the noise comes back.

The first thing I noticed; there is more noise during the daylight hours than at night. This is interesting. I thought it might be coming from my solar system, which uses individual inverters for each panel (so-called microinverters). These are wired to 240 VAC but have an internet gateway device that is in the house and communicates with the inverters using a power line data scheme. It turns out this was a minor contributor below the AM broadcast band.

By process of elimination, here are things that were not contributing to RF noise on Medium Frequency (AM band):

Cable Modem (Motorola MB7420 DOCSIS 3.0)
Router/WiFi gateway* (Netgear R6700v2)
GB Ethernet Switch (Netgear TLSG116E)
Dell Desktop PC’s (three models)
Dell Laptop PC (two models)
Android phones (two models)*
Phillips 4K LED large-screen TV (5PFL5604/F7)
LG LED computer monitor (24MK430H-B)
Refrigerator (Frigidaire FFTR1835VSD)
Stove (GE BP63D W1WH)
LG washing machine (WM3400CW)
LG clothes dryer (DLEX4501)
Bosch dishwasher (SGV68U53UC)
Dehumidifier (GE APEL70LTL1)
LED light bulbs (Sylvania 9W Ultra LED)
Generic incandescent light bulb
Furnace (fancy controller)
Furnace burner motor**

*These are intentional RF emitters

**The furnace burner motor made a small broadband RF signal on startup, likely the igniter which uses an electric arc. Once the unit was running, there was no further RF emissions noted.

The yellow line is the peak hold, the magenta line is the 100 sweep average and the cyan line is the minimum peak hold. I live out in the sticks; there are no streetlights, no stoplights for miles, the nearest cellphone site is four miles away, and houses are spaced far apart.

First, I measured the noise with everything turned off. I then turned things on one by one, noting any changes in the spectrum. For the list noted above, this is the way it looked.

These are a few things contributing to RF noise levels on the MF band.

We have cheap Chinese grow lights to start seedlings for our vegetable garden. We were using these during the daytime hours to augment the low sunlight in early spring. I initially thought this was coming from the solar system. The interference was making a massive noise hump between 750 and 957 KHz. The brand of growlight is BestVA B-1000 LED which was purchased from Amazon.

Next, somewhat surprisingly, the LG computer monitor on my desk was creating a pretty decent rise from 1120 KHz to 1700 KHz. I have three LG computer monitors, this is the newest only this one creates any RF noise.

Then, pretty much every florescent lamp (compact or full-length tube) created a broadband noise increase across the entire MF band and well into HF.

The vacuum cleaner makes a little bit of broadband RF noise when near the receiver. However, you cannot hear the radio when the vacuum is running, so that does not seem to matter.

None of these are surprising. However, I was more surprised that many other electronic devices are not contributing to RF noise in my house.

A little bit about data over power line or power line communications. Searching for power line data can be a bit tricky. First, there is this large voltage 60 Hz (plus harmonics) waveform to deal with. Secondly, there are many different protocols and many different frequencies. I narrowed down my solar system by listening to my Kenwood R-2000 below 300 KHz. Some noise went away when I completely disconnected the inverters. I don’t know the exact frequency, the protocol, the modulation type, etc. But there is something.

Data Over Power line is popular with home automation systems, it can be used to extend Ethernet LAN, and some power companies are using it to control substation equipment, smart power meters, and/or to function as an ISP for their customers. I have heard some HF users complain about BBPL, but I have not experienced it for myself.

Making a notch filter

One small RF project that I am working on; a 770 KHz notch filter. I always figure if I am having this problem, then others may be having it too. This is a relatively simple idea, a resonant LC circuit (AKA a tank circuit) tuned to the carrier frequency. It should have a bandwidth of +/- 15 KHz of the design frequency. Another requirement; use the parts I have available. Finally, the environment in which this is to be used is a high-noise room; with lots of computers, LED lights, etc therefore it needs to have excellent RF shielding.

Something like this would work well for anyone that lives around an AM transmitter site and is having problems with receiver sensitivity or transmitter intermodulation.

The basic design looks like this:

Time for a trip to the local storage facility known as “The Barn.” In my backyard, there is a small agricultural structure that is used for storage of just about everything. In The Barn, I found several parts salvaged from an old Energy Onyx Pulsar AM transmitter. As such, they are more than capable of receiver operation and could likely handle a fair amount of RF power in the transmit mode.

CDM F2B 0.01 uF capacitor with back of N connector inputs

Finding a type F2B 0.01 uF capacitor, rated at 2000 volts and 11 amps, the value of the inductor was calculated. For the inductor, a 20 uH coil with taps will work great. For receive-only applications, much smaller-sized components can be chosen. Also, there are many bandstop filters with multiple poles. Those are great, but I like the simplicity of the parallel resonant LC circuit.

20 uH inductor salvaged from Energy Onyx transmitter

The N connectors were salvaged from I don’t know where and the enclosure used to house a power supply for a Radio Systems console.

For shielding, I sanded the paint off of the enclosure where the lid is attached and tacked some brass screen down with gorilla glue. This will make a good RF contact surface. The outer of the N connectors are bonded to a piece of copper ground strap which also has a grounding lug on it.

Enclosure lid with brass screen to make contact

I used the Libra VNA to tune it up:

The scan shows it is -31 dB on the carrier frequency. It is -17 dB on 760 KHz and -20 dB on 780 KHz. This is good, because I may still want to listen to the station on the remote receiver. According to the smith chart, it is actually resonant on 771.5 KHz, but that is close enough for this application. I think the resonance went up slightly when I put the cover on after the tune-up.

There are several tank circuit calculators online. It is best to have more capacitance and less inductance to keep the Q of the circuit low and suppress the sidebands as well as the carrier.