MF Operation at DF0WD (and elsewhere..)

by Wolfgang Buescher, DL4YHF

last updated: January 2018.


  1. Initial Reception Tests (at DF0WD)
  2. Transmission tests and antenna matching
  3. Using an Icom IC706 as 'exciter' on medium wave
  4. A small 'Linear' power amplifier for MF
  5. Initial activity on MF (starting in June 2012)
  6. 2014-03: Various reception tests, VO1NA in slow CW (QRSS)
  7. Ball Variometer / Kugelvariometer im Eigenbau
  8. Links (medium wave related)

Initial reception tests

During initial reception tests on the UK's former 500 kHz band, it became obvious that the 'station ground' (connected to mains ground) carried a lot of noise, most likely from switching mode power supplies, compact fluorescent lamps, and similar. The receiver made a buzzing sound, and weak amateur radio signals were completely masked.
The following RX antenna configuration gave a significant improvement:

An important feature of this LC parallel configuration is its ability to act as a transformer. The antenna, and its counterpoise ("Earth") are completely isolated from the noisy "mains" ground.
The first coil was wound on a plastic tube, 15 cm diameter, 20 cm length, using 60 turns of 1.8 mm enamelled copper wire. The spacing between two turns is ensured by winding a thin plastic rope (greenish colour in the photo below).

(antenna loading coil, still without variometer for 'receive-only')

Primary coils (60 turns) and secondary coil (2..3 turns, with red plastic coated wire) are separated by about 1 cm. For simplicity, foam plastic from an iso-mattress (sic!) was used.
The spacing between primary windings is important to achieve the largest possible Q factor, because this coil will be used for a transmit antenna one fine day .. when German radio amateurs may operate on MF, which was not the case in April 2012. For reception, a much smaller coil (thinner wire) will do the job just as well.

The original 'receive-only' setup consisted of a short wire, which was connected to the topmost tap of the primary coil (largest impedance). To connect longer wires without degrading the 'Q' (which helps to keep the strong MF broadcast signals away from the receiver), different taps were made on the primary coil. They will later be used to find the best impedance match for a longer transmit antenna.

(closeup of the antenna loading coil with multiple taps)

With the above receive-setup, crossband contacts with various stations from the UK and Ireland on the 'old' band around 500 kHz were made.

Transmission tests and antenna matching

In June 2012, German radio amateurs were positively surprised by the authorities (Bundesnetzagentur, BNetzA) when the frequency range 472 to 479 kHz was allocated on a secondary, non-interference basis with a maximum ERP of 1 watt. On this occasion, a big thank you to everyone involved !
The loading coil shown above was turned into a variometer (to make it suitable for transmission with a low power level), and a few homebrew accessories were added:

(very first transmit matching test, with too many alligator clips)

In the foreground, from left to right:
SWR meter (measures forward and reflected power) combined with an impedance matching unit (RF transformer with various output taps, from 12.5 to 50 Ohms);
RF current meter ("feed through"-type with another ferrite toroid acting as current transformer).

The variometer consists of the old loading coil wound on a plastic tube with 15 cm OD, and a smaller inner tube with 11 cm diameter. The wire for the inner (rotatable) coil is old 'Tensolite' wire (most likely Teflon covered, silver plated Litz wire).
The outer coil is tapped to achieve resonance with the inner coil rotated for maximum inductance (about 350 uH in this case), which ensures the lowest loss in the coil.

(photo with antenna matching accessories)

The loading coil compensates the capacitive reactive part of the antenna impedance (-j * 1 kOhm here), by connecting +j * 1 kOhm in series. The result is purely resistive, and consists of the ground loss, environmental loss (trees), coil loss, and a tiny bit of radiation resistance (in this case, way below 1 Ohm). The output tap on the 'impedance transformer' (integrated in the SWR meter housing) was set for 26 Ohms to achieve the lowest possible SWR, thus with a 30 watt transmitter it should be possible to push 1 Ampere into the antenna wire. It was.

(schematic of antenna matching transformer and loading coil)

The transmitter used for the initial tests was built in a rush (before discovering that the old IC706 can be used as an exciter for 472 kHz):

(photo of the 472 kHz transmitter prototype, with variable crystal oscillator)

An Icom IC706 as 'exciter' (in german: "Steuersender") on MF

By accident, it was found that the author's old IC706 would transmit around 472 kHz. It's not sure which of the modifications in this radio's former life did the trick, but quite certain it is just a matter of configuration (these "modifications" used to be on or other sites).
When trying this on an IC706 or similar shortwave transceiver, be sure to use the lowest power setting. A radio designed to operate between 1.8 and 29 MHz will not have the necessary 'large inductors' in the final and the driver stage ! Trying to transmit with such a radio outside the specified range will void your device's warranty, and of course you will have to try this at your own risk and expense !
At the lowest power setting ("L" in the IC706's "Q1 RF POWER" menu), the radio produced about 2 watts radio frequency on 475 kHz on a 50 Ohm dummy load. But the slighly clipped waveform indicates that the rig doesn't really like to transmit on this frequency:

(waveform of IC706 loaded with 50 Ohm on MF)

Anyway, the external MF "power" amplifier will remove harmonics, so this less-than-ideal waveform is not a problem. 10 Vrms on 50 Ohms is approximately 2 Watts of RF, but voltage and 50 Ohm impedance are too large to drive the power MOSFET's gates directly.
Thus low-impedance matching is required to drive the MOSFETs:

("linear" MOSFET gate driving circuit, adapted for 2 watts from IC706)

The complementary push/pull driver was left unchanged. Note that the 22 nF capacitors parallel to the MOSFET gates are not a typo: Together with the rather low inductivity of the step-down transformer (trifillar wound ferrite toroid), they form a resonant tank (with low Q) near 475 kHz.
With 2 watts of driving power, and a sufficiently dimensioned step-down transformer (from 50 to a few ohms), it may be possible to drive the MOSFET gates directly.
Due to the "linear" operation, the efficiency of this PA is not spectacular (compared to a switching mode PA with rectangular drive) but who cares if the required RF output power (to reach approximately 1 Watt ERP) is only 50 watts.
The MOSFETs are HUF75343, slightly oversized (75 A max. drain current, 55 V max drain voltage, up to 270 W power dissipation but that's not realistic for a TO-220 housing without liquid nitrogen cooling). But the decision was easy because from an older project (LF PA), a bunch of transistors was available, and some safety margin helps in cases of maltreatment (bad SWR, transmitting with no antenna connected at all, etc..).

("linear" MOSFET output circuit)

The output power could be increased with more secondary turns on the output transformer, if a larger ferrite toroid was used (in the author's prototype, an FT114A-61 was used because nothing else was found in the junk box). Through the secondary taps, the output power can be selected without sacrificing the PA's efficiency. When tested with 14 turns secondary, the PA delivered 30 watts RF, and consumed 3 amperes DC input current. This was at the amplifier's clipping point, i.e. more input drive didn't significantly increase the output power anymore. For CW, this is acceptable, and SSB isn't an option on MF (even though it would be technically possible with this amplifier, when "moderately driven").

An RF vox circuit, a timer, and two relays for RX / TX switching were added to the PA board to complete the 'MF station':

(IC706 driving the experimental PA; with automatic RX / TX switching)

With a 'linear' power amplifier (in addition to a 'linear' exciter, such at the IC-706 in this case) all kinds of soundcard-based modes are possible, including those with a non-constant RF envelope.
Here for example, a Chirped Hell (aka Fourier Hell) transmission by DF0WD, with approx. 20 Watts transmitter output power, received at ON7YD. You can see some unwanted sidebands near the figure '3' and the letter 'G', when the transmitter was driven into compression:

(Chirped Hell transmission "73 GN" by DF0WD, received at ON7YD's MF grabber)

Sidenote: Similar effects can be seen on shortwave, when operators overdrive their transmitters in PSK31, trying to squeeze 100 watts average power out of a 100-watt radio ;o)

Next plan: Install a preselector for reception on the PA board, since there is already a 'receive-only' path on it (between the two relays).

Initial activity on MF (starting in June 2012)

With the IC706 (HF transceiver), using 30 .. 40 W from the amplifier shown above, and 1.2 A antenna current, reception was reported through the WSPR network from a number of receivers in western europe (DL, F, I, PA, ON, G, GM, EI). Even during daylight conditions, the MF signal with an estimated ERP of 500 mW (which was later found to be much less (*) ) was copied at GM4SLV in Shetland, over a 1000-kilometer distance.
Contacts in normal CW in the late summertime evenings were difficult due to QRN but possible - the first successful two-way QSO from DF0WD was DL2HRE.
Nice CW QSOs with 'armchair'-copy on both sides were made with DJ9IE and DK8KW. These stations are well inside groundwave range, and there was no QSB (fading) at all.
After midnight (in mid summer), the band "opens up wide", and with some luck the QRN goes down significantly. Under such conditions, two-way contacts on MF in normal CW with 'armchair copy' were possible over several hundred kilometers distance (like DF6NM and DK7FC). A first 'highlight' was EI0CF - thanks Finbar ! - crossing the 1000-km limit with an estimated ERP of 0.8 Watts (*) on this side.

(*) ERP (Effective Radiated Power) much lower than expected ?
Comparing the ERP with other active stations on MF, and by judgment from 'calibrated' receivers (and their estimation of DF0WD's ERP on MF), it was later found that the trees surrounding the antenna had quite an impact on the 'true' radiated power. Moderately increasing the TX power from 30..40 Watts to something around 100 W seemed easier than improving the antenna efficiency, or felling a couple of trees around the club station ;-)
... using a high-efficiency linear PA with switching mode modulator - details further below)

In winter time, the ionospheric absorption drops faster, and 'DX' signals appear from the east before sunset. With a moderate ERP, and place not plagued by local QRM, contacts were made in relatively slow CW (but not QRSS) with Malta (9H) and Romania (YO).

In January 2013, radio amateurs in other European countries got access to the 630 meter band, like the UK (requires an NoV) and the Netherlands. This will hopefully boost activity beyond 2013-01-01, when the first stations in PA, and one hour later G, GW, GM, and EI got 'on air' just a few minutes after their local midnight.

2014: Improved linear amplifier, with improved efficiency

When it became obvious that the ERP was 'much lower than expected', and some stations had difficulties to copy DF0WD on MF, a new final amplifier was built. To use the existing '13.8 V, 20 A' power supply, a more efficient design was required. The choice was that, instead of operating the MOSFETs in the 'linear' region (which resulted in a poor efficiency), the RF power output stage should run in switching mode, and produce a maximum power of approx. 150 Watts; ideally from a DC supply of not more than 15 A (at 13.8 V) to have some margin for the IC-706 (exciter) from the station's '20 Ampere' power supply. With a DC input power of approx. 200 W, and an RF output of 150 W, the design goal was "at least 75 %".
To keep the CW modulation sidebands as low as possible, and to be prepared for modes which require envelop shaping (like PSK31), the new power amplifier also had to be 'linear'.
As a proof of concept, the switching-mode 'linear' PA was built from various 'modules' on a large copper-clad board. The entire board is at ground level, and there are no traces etched (or cut) into it. All modules (like sync rectifier, auxiliary negative voltage supply, PWM, SWR sense, current sense, protection) are built on smaller, individual PCBs.. in fact, the PCBs are neither 'printed' nor etched, but cut with a sharp tool.

Photograph of the PA in an 'early stage', running in a two-hour stress test (during which much water was boiled to cool the dummy load resistor on the left side) :

Click on the image for a full-screen view
( isn't it strange how fast a workbench can turn into a mess, with cut-off wires floating all over the place ? :o)

Details about the new PA (with switching-mode 'power modulator', put into service in January 2014) are in another document; see 'Switch-Mode Linear'.

2014-03: Various reception tests, VO1NA in slow CW (QRSS)

VO1NA received via DF0WD on 477.7 kHz in QRSS10, 2014-03-22 .
Times are in UTC. Local sunrise (in Germany, RX side) at 05:22 .
Click on the image for a full-screen view

... to be continued ...

Ball Variometer (Kugelvariometer im Eigenbau)

The ball variometer shown below was initially just a 'bad weather project', build partly for fun, and to get a fellow ham QRV on medium wave.
The easier to build 'PVC pipe variometer' shown further above on this page has a limited tuning range, and it's difficult to switch the many taps on the loading coil remotely (when the variometer is located far away from the radio shack, under the roof, in the garden, on a mast, etc).

Variometer prototype, made almost entirely of plywood.
"Remote control" with pushbuttons to increment/decrement inductance at the end of the video.
If your internet browser is as stupid as mine and says
  'Kein Video mit unterstütztem Format und MIME-Typ gefunden' /
  'No video with supported format and MIME type found',
try this plain file link: kugelvariometer_sm.mp4.

The final length and height of the antenna was unknown, thus the capacity (in picofarads) was unknown, and the number of microhenrys required to resonate the antenna. With approximately 5 to 6 picofarads per meter of a relatively small antenna (inverted L with 10 meters vertically, and 40 meters horizontally) may have roughly
    10 * 6 pF + 40 * 5 pF = 290 pF .
Reactive part of the antenna impedance:
    Xc = 1 / ( 2 * pi * f * C ) = 1 / ( 2 * pi * 472 kHz * 290 pF) = 1162 Ohm
This capacitive part must be compensated with an inductor ("loading coil"), connected in series with the antenna wire. At resonance,
Xl = Xc; Xl = 2 * pi * f * L, thus:
    L = Xl / ( 2 * pi * f ) = 392 uH .

Because the antenna may be shorter than 'initially hoped' (tree closer to the house than guesstimated..), the variometer needs some headroom (L), so the design goal was 500 instead of 392 uH.
After playing around with online calculators for single-layered coils (especially this one by Serge Stroobrand, ON4AA), the inner coil should have a diameter of at least 120 millimeters, with 50 turns of 1-mm enabelled copper wire evenly distributed over the coil former's surface (which will be a sphere instead of a cylinder here, but let's ignore that for a moment).
In the interest of a high Q, there must be some spacing between wires. You can try this out for a helical coil with the above coil designer:
Depending on the proximity of wires, the thickest possible wire will not always give the largest possible Q for the same coil size !
  D=120 mm, N=50, L=115 mm, d=1 mm, f=0.472 MHz -> L=209 uH, Q=352
  D=120 mm, N=50, L=115 mm, d=2 mm, f=0.472 MHz -> L=203 uH, Q=284
(D = mean coil diameter, N = number of turns, L = coil length, d = wire diameter)
In this case, 1 mm wire diameter, and a spacing of at least one wire diameter gave an acceptable Q as described later.
For the inner sphere I used two wooden disks, produced with a simple scroll saw (Dekupiersäge). Each disk was slotted from the 'pole' to the center, slot width equal to the plywood thickness, to stick them together easily.
1.5 millimeter holes to hold the wire were drilled near the outer edge of the disks (see drawing further below). Then the wire was 'screwed' into the holes from a spool with the same diameter as on the sphere's equator, beginning near the equator (latitude 0°), and ending near the 'arctic circles' (latitude 66°).
Wire loops closer to the poles don't add much inductance so omit them. Threading the wire through 4 holes for each turn is easy but time consuming. I wasn't very patient and used too much force for the prototype, resulting in a few kinks in the wire (nothing to worry about, as long as the wires don't touch):
Inner part ("rotating ball") of homemade variometer.
First prototype, almost ended up as decoration for a christmas tree.
After winding, the wire was secured by painting (lazy people like me just dip the entire assembly into varnish.. ugly but fills the small gaps in the wire holes).
The outer coil is slightly larger than the inner, just large enough for 'production tolerances' and to withstand a few kilovolts
- see suggestion for a 'slighly larger variometer' further below.
Again, lots of holes with 1.5 mm diameter had to be drilled into the plywood plates to hold the wire (1 mm diameter). The outer cabinet consists of an upper and lower part:
Lower half of the variometer's 'stator' and rotating ball with axle
Again, a sufficiently long run of wire (*) was first formed into a coil (diameter approximately as on the equator by winding it onto a PVC pipe, which simplifies 'screwing' the 1-mm wire into the 1.5 mm holes beginning at the equator. After finishing the inner ball (rotatable), and both hemispheres of the outer ball (fixed), all parts of the coil were connected in series.
Measured result:
    Lmin = 45 uH (*)
    Lmax = 388 uH
    Rloss = 5.8 Ohm at 472 kHz (**), Q = 176 at Lmax
(*) At the minimum position, the magnetic fields from both ball-shaped coils should cancel each other. Here, they don't (at least not completely) due to the unavoidable spacing between both layers.
The 'Q' was measured indirectly by connecting a large Russian 300 pF capacitor (rated at 3.5 kV) in series with the coil, and measuring input voltage and current at resonance. The actual 'coil Q' may be a bit higher because there will be some loss caused by the capacitor itself, which is ignored here).
**A note on coil loss:
In most Europen countries, there's a 1 Watt ERP (or even EIRP) limit on MF. To radiate that 1 Watt, you may have to push 1 to 2 amperes into the wire. If you're lucky, the earth loss will be 20 Ohms, so it really doesn't pay off to spend a fortune into building a loading coil with Rloss below 5 Ohm, or a Q significantly higher than 200 (remember, Q = Xl / Rloss).
Consider this:
An antenna with 1 Ohm radiation resistance needs 1 Ampere into the feedpoint to "radiate" 1 Watt (hopefully into free space).
To push 1 Ampere into a total resistance of 50 Ohms (radiation, coil loss, earth loss,..), you need a 50 Watt transmitter. Only 5.8 Watts will be dissipated in this variometer's Rloss. So don't burn money for a high-Q coil if a smaller coil's loss is just a fraction of the total loss, and you're limited by ERP anyway.
To get closer to the 'design goal' (500 uH at the 'max' position), additional holes were drilled for 10 turns close to the equator of the lower half. You can see them in the video at the begin of this chapter. When connected in series with the other coils, the final inductance range was
    Lmin = 82 uH
    Lmax = 515 uH
If I was to build another ball variometer for MF, I'd use a few more turns on both spheres, and a slighly larger disk diameter.. something like this:
Suggested 'slightly larger' ball variometer with Lmax = 500 uH
Rough guesstimate for the total length of wire required for the above variometer, "if the spheres were cylinders):
Inner sphere: 60 turns * 14 cm diameter * pi = 26.4 meters
Outer sphere: 60 turns * 15.5 cm diameter * pi = 29.2 meters
Total : Less than 55 meters of 1 mm enamelled copper wire, because only turns near the equator have those diameters.

As already stated in the introduction, if you don't need an antenna tuning range of a few octaves, and the antenna impedance (reactive part) is known in advance, a simple 'tube' variometer with a large fixed part, and a small variable part, is easier to build than a ball variometer.

Regardless if the variometer is a simple 'tube-' or a ball variometer, it often needs to be controlled remotely because the capacity between antenna and ground (earth) changes with weather, soil conductivity, etc. There may be some kilovolts(!) on the wire between the loading coil (variometer) and the antenna, so you don't want to run that wire from the shack to the roof or through the garden.
At the end of the video on the begin of this chapter, you see a large wooden gear between the variometer and a very small gear motor (the type used in RC models). A DC gear motor by 'Faulhaber' only consumed 5 mA at 2 Volts (!) to rotate the inner coil at low speed, so a simple passive circuit was used to drive it remotely:

Links (medium wave related)

DX cluster with filter for the 472 kHz band (OH8X DX Summit)
DM4TR MF Grabber: 472-479 kHz live spectrogram by Thomas, DM4TR, in JO61DE
DK7FC MF Grabber: by Stefan, DK7FC, in JN49IK
Grabbers in Birmingham: A long-lasting service for LF and MF by Dave, G3YXM
TF3HZ live spectrograms (covers a part of the 630 meter band)
The Shetland Grabber: 472-479 kHz live spectrogram by John, GM4SLV (temporarily offline?)
WSPR Spot Database for MF and LF ("old" interface but imo easier to use)
The RSGB LF Group (at Yahoo), also used by MF operators
GW3UEP's site devoted to 500 kHz CW with a variety of homebrew CW transmitters
Operating Portable on Medium Wave by Finbar, EI0CF - Amateur Radio at it's best !
Homepage of the Montenegro LF / MF Group(seems to have disappeared; or redirects to a far-east provider)
G4WGT Multi-Grabber page: now includes the new 630 meter band.
Temporary 500 kHz (or 475 kHz) grabber by Rik, ON7YD
WebSDR at the Universtiy of Twente; with gapless coverage from 0 to 29 MHz .

back to DL4YHF's main ham radio homepage