tech stuff – 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.

The Ultimate Receive Antenna

This post is about a Large Loop On Ground (LLOG) receive antenna.

I have, for some time, been lamenting the demise of my K9AY antenna last year. While I don’t spend a lot of time Shortwave Listening these days, it is sometimes a nice diversion on a rainy day.

I have been reading about LOG (Loop on Ground) broadband receive antennas for a few years and the concept sounds interesting. Loop antennas in general, tend to reject RF noise better than dipole or vertical ground plane type receivers.

The idea is the loop antennas are more H field (magnetic) than E field (electrical) receivers. Most unintentional RF noise generators create noise in the E field.

Near Field vs. Far Field Electro-Magnetic energy

Additionally, close proximity to the ground further lowers E field noise due to capacitive coupling with the ground. The ground acting as a sink for the E field noise. Loops also tend to have better rejection of chaotic Near Field energy. Thus the combination of those three things will create slightly lower signal with much better signal to noise ratios over traditional up in the air antennas.

The downside is that it cannot be used for transmitting. The capacitive coupling noted above will quickly dissipate all of the transmitted RF into ground as heat.

The LOG antenna that most amateur radio operators use is either or close to the antenna outlined by Matt, KK4JY: The Loop on Ground Antenna

That is all well and good, but I want something larger to go down through the entire Medium Frequency band in into Low Frequency range. The main reason is I want to start fooling around with the 2200 and 630 meter amateur radio bands. Also, wanting a good low noise AM broadcast receive antenna. With that in mind, I began taking inventory of available real estate and parts.

As for real estate; I own a patch of woods behind my house that is approximately 200 x 200 feet. Good enough for a fairly large square loop. The good thing about particular plot of land is it is far away from the neighbors, the utility lines and other sources of electrical noise.

As for the available parts; it turns out I had almost everything needed. The only thing I needed to purchase is a water tight outdoor enclosure for the matching transformer, from Amazon for $9.00.

Here is a list of items used:

Item	Amount	Use	Comment	Cost if new
Wire, 14 AWG	470 feet (143.26 M)	Wire Loop	Pieced together from left overs	$100.00 (500 ft spool) at Home Depot
Toroid, BN-73-202	1	Transformer core	Left over	$0.95 at Amidon
Wire, 26 AWG magnet wire	18 inches (46 cm)	Transformer winding	Left over	$9.00 (2 oz) at Amazon
Coax Cable, Cablewave FCC38-50J	175 feet (53.34 M)	Transmission line to house	Salvaged from decommissioned AM directional	$53.00 (500 ft spool RG-6) at Home Depot
RF Connector, type N Female	1	Antenna output	Found in junk drawer	$6.00 (6ea) F chassis mount at Amazon
Enclosure, ABS water tight IP67	3.9 x 3.9 x 3 inch (9.9 x 9.9 x 7.62 cm)	Transformer housing	Purchased from Amazon	$9.00 at Amazon

The wire is copper THHN 14 gauge. If buying new, this is the single largest expense. I soldered and heat shrunk several lengths from left overs on spools. This antenna is going to be in the shade, so I don’t have to worry too much about UV degradation of the insulation. If that happens, I can find some more insulated wire.

The big bonus is the Cablewave FCC38-50J phase stabilized cable, most of which is buried out to my old K9AY antenna, I simply extended it about 50 feet. Having the line already buried to much of the manual labor out of the project. The buried section of line acts as an RF choke, fully eliminating any electrical noise coming from my house at the antenna side of the line.

Large Loop On Ground antenna, installed in woods behind my house

I did have to go around and clear away some brush and dead trees along where the antenna wire was laid on the ground. I wanted to make sure that the wire was directly on or within an inch or two of the ground.

LLOG antenna transformer and transmission line

The transformer took about 20 minutes to construct. It consists of six windings on the antenna side and two windings on the transmission line side. I used a type 73 core because I am more interested in the medium frequency range. It seems to work well up to about 15 MHz. The transformer is for galvanic isolation, keeping the loop current balanced and separated from the transmission line. The shield is not grounded at the antenna side, but is grounded where it comes into the house. On the transformer enclosure, I used a salvaged N connector to match what I had for the transmission line. Any low loss cable, including RG-6 or RG-11 will work very well.

WKNY signal, 18 miles distant from transmitter

After a few hours of low intensity work, I had everything together. I connected the antenna to my trusty Kenwood R-2000 receiver and tuned to the nearest Class C AM on 1,490 KHz. Wow! What a difference! This little 1 KW high band station 18.6 miles (30 km) away barely came in before, especially when the ground is dried out. The relative signal strength meter on the receiver shows an S9+5. I am located on the edge of the predicted 0.5 mV/M contour while the spectrum analyzer shows a -65 dB signal, which is netting a 44 dB SNR, which is very good. All of this is after a Mini Circuits ZSC-4-2 power divider, which has an insertion loss of -6 dB per port.

Wide view of the entire AM band, about 1pm

One of the reasons for the “decline of AM;” overcrowding. This was taken during the daytime, when there was no skywave propagation. Barely an open frequency.

With the spectrum analyzer and made a few interesting discoveries. First, the noise floor in the Medium Frequency range goes from about -100 to -110 depending on the frequency. The noise floor on High Frequency is -112 or lower across the entire band. I clearly get all of the NYC AM stations as well as most of the stations in and around Albany. Before sunrise, I tuned around the 630 (472 to 479 KHz) meter band and heard a few stations sending CW (W4TS @ 164 miles (264 KM), K2ORS @ 157.2 miles (253 KM)). A few overnights of WSPR monitoring on 475.6 KHz netted 194 spots from 11 different stations. The furthest was 822 miles (1,322 KM) away in central Illinois. All of these stations are transmitting 5 watts EIRP or less.

LLOG 24 Hour WSPR/FST4W reception map; 33,592 spots, 1,421 unique

This antenna works well to about 15 MHz, where the signals start to drop off because of the transformer material that I used.

Using the Network Analyzer, I found the resonance is 2.865 MHz with a -36 dB return loss. This makes sense, as the wire laying on the ground will have a lower velocity factor due to the capacitive coupling with the earth. In this case the VF of the antenna is 73% when the soil is dry. The velocity factor should go down when the ground gets wet, or covered with snow.

It might be interesting to do some ground conductivity measurements…

Designing filters for 630 meters

This has nothing to do with broadcasting. It does, however, have a good deal of geeky goodness.

I have started a new project, getting on the air on the 630 meter Amateur band. For those who do not know, 630 meters is from 472–479 kHz which is below the AM (or Standard) broadcast band. It was formerly part of the Maritime Mobile allocation. For US Amateurs, these frequencies were added in 2017 so it is a relatively new experience.

There is no commercially available equipment for this band, so it depends on the potential operator to make his or her own equipment which is where the fun begins.

To start, I thought I’d repurpose a WSPR beacon to 630 meters to do some antenna experimentation. Like all transmitters, the output of this unit needs to be filtered to reduce or eliminate out of band emissions. The Amateur radio service falls under Part 97, which has somewhat different requirements than Part 73 or 74.

47CFR 97.307(d) states:

For transmitters installed after January 1, 2003, the mean power of any spurious emission from a station transmitter or external RF power amplifier transmitting on a frequency below 30 MHz must be at least 43 dB below the mean power of the fundamental emission.

That is a fairly low bar. I am going to shoot for something better. WSPR beacons center around 475.6 kHz. The harmonics are at 951, 1428, 1902, 2378, 2853, 3329, 3804, 4280 and 4756 KHz. The first two are in the AM broadcast band. A quick look at the Zachtek WSPR beacon show these harmonics:

unfiltered ZackTek WSPR Desktop transmitter, 630 meter band

Definitely does not meet the out of band emissions standard set forth in FCC 97.307. Typical of solid state amplifiers, the odd harmonics are greater than the even.

I used a filter design program called Elsie to design suitable filters for 630 Meters. There are two types of filters that can attenuate the harmonics; low pass and band pass. A low pass filter passes all emissions below the cutoff.

That is fine, however, it does not eliminate the possibility of interference and inter- modulation from frequencies below the band. Both types of filters are also good for receivers in the presence of AM broadcast band towers, which can desensitize receiver front ends when located nearby.

A band pass filter cuts off frequencies above and below the pass band.

This is a nodal inductor-coupled band pass filter. I like this design because it has deep shoulders and has better performance with the harmonics in the AM broadcast band.

Prototype low pass filter:

Quick prototype board. Capacitors are Cornell Dubilier silver mica dipped, the inductors are wound on T130-3 material. The SWR and return loss:

Smith chart:

The Smith chart shows that it is slightly inductive on the desired frequency. The way to mitigate is to either add some capacitance (not easy) or reduce the inductance (somewhat easier). I tried tuning it by changing the spacing on the windings of L1 and L2. There was no change.

Low pass filter response:

The second harmonic on 951 KHz is -58.37 dBc. Harmonics 3 – 7 are 40 dB below the fundamental. This is adequate but the band pass filter below is better.

Prototype band pass filter:

The above filter is a little rough, but it was a good test of the filter program’s design parameters. The capacitors are Cornell Dubilier 0.02 uF 500V. The inductors are wound on T130-3 iron powder toroid cores. The results are good:

By adjusting the spacing of the windings on L3 (center of the board), I can tune the VSWR and Return loss for best values.

The same ZachTek WSPR transmitter noted above, running through the prototype filter:

ZachTek 630 Meter band pass filter response

The second harmonic on 951 KHz is -62.86 dBc. The rest of the harmonics are less than that.

Of the two, the band pass filter has better performance characteristics. The return loss/SWR is lower and can be tuned by adjusting the spacing of the toroid windings on L3.

I decided to take the next step and make a PCB. I have KiCad on my Linux machine, which works well. Sometimes some of the foot prints need to be edited so the dimensions are correct, but that is easy.

What I like about this board design is that it will work on any amateur band below 7 MHz with different component values. I had five boards fabricated and built one of them out. The T130-3 toroids are wound with 14 AWG magnet wire. The capacitors are same used in the prototype, 0.02 uF, 500V mica dipped.

Board mounted in aluminum enclosure:

Band pass filter mounted in diecast aluminum enclosure

This build has similar measurements to the prototype board above. Based on what I found while making this, I made a few tweaks to the circuit board in KiCad. I would consider selling these, if there is enough interest.

Locking AM station carriers to GPS

This is not a new idea, many people have discussed it in the past. The National Radio Systems Committee (NRSC) has a guidance paper, NRSC-G102 which gives a detailed explanation of why synchronized AM carriers are beneficial. There was even a move by some to have it included in the AM revitalization plan of a few years back. The NAB opposed this idea, saying it would be too expensive. That is unfortunate because out of all of the revitalization initiatives, GPS locked carriers had the best potential for an actual technical improvement. While it may be expensive for some very old tube type transmitters, for more modern solid state transmitters, GPS referenced carriers can be implemented as little as $200.00 US.

The FCC rule (73.1545(a)) for AM Carrier frequency specifies:

AM stations. The departure of the carrier frequency for monophonic transmissions or center frequency for stereophonic transmissions may not exceed ±20 Hz from the assigned frequency.

40 Hz is quite a bit of movement on a 20 KHz AM (18 KHz in ITU region II and III) channel. The reason for trying this is simple; there are many co-channel and first adjacent channel AM stations which at night, interfere with each other.

Above is a typical spectrum analysis of an AM station on 940 KHz. This was a 10 minute peak hold for an NRSC-2 spectrum mask measurement. The carrier is approximately 20 dB greater than the audio, which means that most of the interference between co-channel AM stations is created by the carriers beating against each other. By locking carriers to the same reference, that carrier interference will be greatly reduced. NRSC-G102 goes into great detail on the listenability of interfering stations with synchronous AM carriers (Page A-3).

Stations drifting off frequency also cause greater adjacent channel interference.

Almost all transmitters made in the last 30 years have an option to use an external frequency generator or 10 MHz reference. The required drive levels vary. The easiest way to implement this is by using a GPS locked programmable frequency source such as the Leo Bodnar LBE-1420. It can be programmed to any frequency from 1 Hz to 1.1 GHz, has a frequency stability of 0.000001 PPM (10^-12), and an output level of 3.3 V peak-to-peak. This drive level is not enough for some transmitters. For those situations an additional amplifier such as a Mini Circuits ZHL-3A+ is needed.

Here are a few AM transmitters that except an external RF source including a GPS disciplined oscillator.

Nautel J-1000

An external RF source can be plugged into the EXT RF IN connector (J-6) on the Remote Interface board. The source must be on the carrier frequency ± 5 Hz and have a peak-to-peak voltage of between 5 – 15 V (sine wave or square wave), 50 ohm impedance.

An external 10 MHz signal can be connected to the RF synthesizer board 10 MHz REF INPUT (J2). The external 10 MHz frequency reference must be precisely 10.00 MHz and have a peak-to-peak voltage of between 2.2 – 8.0 V (sine wave or square wave).

Jumpers on the Remote Interface Board and RF synthesizer board need to be configured appropriately for each source. Consult the manual pages 3-9 and 3-11.

Nautel ND series

An external RF source can be connected to ABA1J1 on the external interface board. The RF drive must be on the carrier frequency ±5Hz with level of between 5 – 12 V peak-to-peak (sine wave or square wave) and have a 50-ohm impedance.

Do not remove the crystal from the RF Drive board as the PDM frequency for the modulator is derived from it.

To select the RF drive source for the transmitter the links on the RF Drive board need to be changed. Consult the manual pages 3-3 and 3-14.

Nautel XL series

An external RF frequency source can be connected to the Exciter Interface board, J7. The external drive signal must be between 5 – 12 volts peak-to-peak (sine wave or square wave). Consult manual pages A1 and B4.

Nautel XR series

An external RF source can be connected to the remote interface board’s digital EXT RF IN (J6). This replaces the internal carrier frequency oscillator for one or both exciters (A/B). The external RF source must be the carrier frequency, within ± 5 Hz, have peak-to-peak voltage between 5 – 12 V (sine wave or square wave). Consult the manual pages 7-1.

Broadcast Electronics AM2.5 – AM10A, AM5E

The transmitter has an external RF input on the top of the unit (EXTERNAL RF INPUT). The input is designed for an external stereo generator or reference oscillator with a signal level from 5 to 15 volts peak-to-peak. To use this input, program jumper P7 on the exciter circuit board in position 1-2. Consult the manual page 2-19.

Broadcast Electronics AM500 – AM1A

The transmitter has an external RF input on the ECU rear-panel (EXTERNAL STEREO RF INPUT (J1). The input is designed for an external stereo generator or reference oscillator with a signal level from 5 to 15 volts peak-to-peak. To use this input, program jumper P7 on the exciter circuit board in position 1-2. Consult the manual page 2-20.

Harris Gates AM series

An external RF source can be plugged into J-1 on the Oscillator board. The source must be on the carrier frequency ± 20 Hz and have a peak-to-peak voltage of 5 volts. Frequency source selector P-6 must be set to external. Consult the manual, Oscillator Board Schematic.

Harris DX series

An external frequency generator can be connected to J2 on board A3. Jumper P5 should be set to either 20K ohms or 50 ohms depending on the source impedance. Jumper P6 can be set to either external source or automatic source selection. The drive level needs to be 4 to 4.5 volts peak-to-peak square wave for high impedance inputs or 0 to +25 dBm for 50 ohm impedance sources. Consult the manual, page A-2.

DX-50 oscillator board, A-17 external source connected

Newer DX series oscillator boards which have automatic source selection will fail over to the internal oscillator if anything happens to the externally generated RF signal.

Harris DAX

An externally generated carrier frequency or 10 MHz reference signal can be connected to connector J11 for the external carrier or J10 for the 10 MHz reference on the External I/O board. External carrier or 10 MHz reference must then be enabled via the VT100 screen. The external carrier frequency or 10 MHz reference must be above 2.0 volts peak-to-peak. Consult the manual page 3-15.

Harris 3DX

An externally generated carrier frequency can be connected to J12 (RF CARRIER) jack on the external IO board. The drive levels need to be 4 to 5 volts peak-to-peak, square or sine wave. On the carrier frequency +/- 5 Hz. The input is impedance is selectable for either 50 ohms or 10 K ohms.

An externally generated 10 MHz reference frequency can be connected to J10 (10 MHz REFERENCE). 10 MHz reference level needs to be 1 to 5 volts RMS, square or sine wave. The input impedance is selectable for either 50 ohms or 10 K ohms.

Programming for these options is done on the exciter setup page. Consult manual page 2-43.

Harris SX series

SX series transmitter have either an oscillator board or a frequency synthesizer board. Both will accept an external frequency source. The oscillator board is A16J1 and it needs a 5 volt peak-to-peak carrier frequency signal. Frequency source selector P-6 must be set to external.

The frequency synthesizer board external frequency input is also J1, however, it requires a 10 volt peak-to-peak signal. Frequency source selector P-6 must be set to external.

Conversion table for various RF power levels into a 50 ohm impedance

Volts, Peak-Peak	Volts, RMS	dBm	mW
2.2	0.77	10.8	12
3.3	1.23	14.35	27
5	1.76	17.9	62
10	3.5	23.9	250
12	4.24	25.5	360
15	5.3	27.5	562
20	7.07	30	1000

A 10 MHz reference input is preferred over direct carrier frequency generation simply for the ease of implementation. With direct carrier frequency generation, the frequency output of the GPSDO needs to be double checked. One misplaced digit and severe damage to the transmitter can result.

AM Tube type transmitters, plus early solid state transmitters such as the Harris MW1A may have instructions for implementing AM stereo. Since the AM stereo exciters generated the carrier frequency, those instructions would be a good guide on how to connect an external frequency generation source. However extensive modifications may be needed to the oscillator section depending on the transmitter.

Honestly, this is cheap enough that I think all new AM transmitters should come with this from the factory.