Introduction

The Tri-Band Receiver ("TBR" here after) described in this Web site is the result of a home building project conducted during the second half of 2003. 
The general idea behind this project was to build a receiver capable of copying the digital mode amateur radio traffic.
I downloaded so much from the Internet during the past years, that it is appropriate that I should do something in return now, hence this Web site.  The work on the TBR is not entirely finished yet, so this Web site will be updated has further enhancements will become available.

The TBR is a three band HF receiver for the 80m, 40m and 20m amateur bands, with the following features:
* Coverage: 3500...3600kHz, 7000...7150kHz and 14000...14350kHz
* Good sensitivity: around 1µV MDS (>6 dB S/N)
* Good IP3
* More than 50dB of AGC
* Controlled by a PC
* Single 12...13V power supply (battery)

 

Overview

The TBR is designed as a direct conversion receiver: a set of HF pre-filters followed by a "Tayloe" quadrature product detector, a polyphase filter for side band rejection and a series of audio amplifiers and filters.

The HF and AF part has been largely inspired by the excellent work of the QRP2001 team (see QRP2001: just another radio?), my eternal gratitude to them.

The quadrature local oscillator signals are generated by dividing the signal of a 56MHz VCO. The latter is stabilised using a phase locked loop and a fractional divider

A simple (yet isolated) "com port" interface connects the TBR with the host PC for the graphical user interface (GUI).

Given the description here above, the block diagram should be easy to understand.

 

HF pre-selection & Tayloe detector

See HF circuit for the circuit diagram.

The antenna signal is first led through a set of HF band pass filters, small relays are used to switch the signal according the received band.  
Each filter is designed as a cascade of three coupled resonators, with an input impedance of 50Ω and an output impedance of 200Ω. 
The measured insertion loss (some 6dB) is a bit on the high side, though this loss is partially compensated by the impedance transformation (yielding +2dB voltage gain).

The heart of the system, the D. Tayloe mixer is implemented with the FST3253 IC.  This chip includes a dual 4:1 high speed video switch, featuring a very low on resistance (2Ω per switch as two are used in parallel), a high bandwidth and above all a high switching speed. 
I will not dwell on the principle of this truly remarkable circuit, much has been published already on the Internet.  For a mere 4 euro a quardature detector can be implemented with outstanding dynamic range and intermodulation characteristics. Suffice to say that it works as promised. 
The fairly huge 390nF capacitors, together with the HF source impedance of 200Ω, form an RC circuit, limiting the audio bandwidth to 2000Hz, which is compatible with the "panoramic" screen of the popular digital modes decoding SW such as Hamscope.

The four outputs of the mixer are fed to a set of differential amplifiers build with the low noise op amp NE5534A  (yielding 4nV/√Hz from 300Hz onward).  Their gain is set to 37 or 31dB.  Their output, together with that of a set of amplifiers unity gain inverting amplifiers (each a quarter of a TL084) provides the drive for the polyphase filter.  A small potentiometer is set in one of the loops, to match the gain better than 1%, and to achieve the maximum unwanted side band suppression.

There are certainly lower noise opamp chips than the NE5534A, e.g. the AD797A achieves less than 1nV/√Hz noise figure.  The AD797A costs however some 14 euro, compared to the 0.70 euro for the NE5534A.  As an increase of sensitivity beyond the 1µV is not so meaningful for the low HF bands, the extra cost does not seem worthwhile.  Anyway, both NE5534A devices were socketed, so I can always change my mind about this.

 

AF amplifiers, filters, AGC and output

See AF circuit for the circuit diagram.

I initially used the spreadsheet  by James Verduyn to dimension a 7 stage polyphase filter for frequencies from 200Hz to 2400Hz, taking into account the highest value of precision capacitors I could lay my hand on:  28 pieces of 4.7nF polypropylene capacitors of 2% accuracy (Conrad ref. 458880-44), the resistors in the filter being of the regular 1% accuracy.  The impedance is probably a bit on the high side, but this really baffling circuit actually works: according to my measurements more than 50dB unwanted side band suppression (after trimming the gain) is achieved.  What is quite disappointing though, is the high pass band attenuation of 20dB. 

A few months ago Pim Niessen (PA2PIM) contacted me to point out that I should reverse the order of the resistors of the polyphase network (i.e. going from low to high impedance rather than going from high to low impedance) in order to achieve a much lower attenuation.  Eventually I took the time (and courage) to open the TRB once again, changed 24 resistors and yes the attenuation dropped to 9dB.  I gained 11dB without paying anything in return.  Here the link to Pim's excellent article.

The 2nd amplifier stage is build with a quarter of a TL084 (a fairly low noise quad J-FET input opamp IC) in a non-inverting mode (in order to achieve an high impedance input), it is set to amplify 38 times or 32dB.

The signal is now led to a 7 element passive elliptic low pass filter, with a 3dB pass  frequency of 2200Hz and a 60dB stop frequency of 3500Hz.  It is immediately followed by a 5 element elliptic high pass filter, with a 3dB pass frequency of 240Hz and a 60dB stop frequency of 130Hz.   The characteristic impedance of both filters is 680Ω. 
The filter was synthesised using a very old freeware under DOS called PZ.EXE of unsure origin, as a low pass and high pass section.  The resulting component values where adapted to the E12 series of values, then the entire filter was simulated using CircuitMaker (6.2), an excellent tool provided as freeware for students.

The inductors of the low pass filter are RM6 (3B7, A=160 material) pot cores, while those of the high pass filter are of the EF20 type (A=1200 material).   I spend quite some evenings winding the cores, using a self made gadget based on a small DC motor fitted with a gearbox.
The value of the inductors was adjusted using a simple Colpitts oscillator (fitted with precision capacitors) and a frequency counter.  As I was not able to get precision capacitors, most are of the standard 10% polyester type, and even worse, the larger value capacitors are simply standard Aluminium elcos (though I took care to polarise them).

The choice of passive filters was inspired by the wish to have a high dynamic range and low noise implementation.  The filter is admittedly a bit bulky, measurements show that the pass band is not so flat as it should be and that the insertion loss is a bit higher than expected (about 9bB) .

The 3rd amplifier stage is build with a next quarter of a TL084, also in a non-inverting mode (the input impedance being set to 680Ω), it is set to amplify 16 times or 24dB.

The AGC circuit is based on a enhancement channel N-MOS FET as voltage controlled resistor (the signal across its drain and source is less than 200mVpp), it is controlled by a little circuit with a PNP transistor in a rectifier / amplifier mode.  The fairly large capacitor organises the fast attack / slow decay characteristic of the AGC, useful for SSB reception.   A potentiometer is used to adjust the set point of the circuit.  The control voltage is further used as RSSI signal.  Find here the graph of RSSI value versus the input signal, the values are further converted into the typical S-values by the GUI.

The 4th amplifier is again build with a quarter of a TL084 in a non-inverting mode, this one is set to amplify 84 times or 38dB.  This gain is typically much reduced by the action of the AGC.  
The output signal of the 4th amplifier drives, together with the output signal of yet an other quarter of a TL084 (used as an unity gain inverting amplifier), a 600Ω audio transformer, organised in a 2:1 voltage mode (-6dB).  The voltage is stepped down in order to drive the fairly low input impedance of the sound card (1kΩ) and some length of cable.  The prime reason of the audio transformer is of course to provide a ground isolation between the PC and the TBR (the latter being grounded through the antenna input only).

This spreadsheet gain plan gives an overview about the gain distribution and dynamic range of the TBR. 

Note also, that due to the limited power supply rejection ratio of the TL084 (some 70...80dB), it is of the utmost importance to spread the high gain amplifiers over the two devices.  A lesson that was learned the hard way.

 

Local oscillator and divider

See local oscillator for the circuit diagram.

The local oscillator is a VCO running at 54.8...58.8MHz, it is of a fairly classic Colpitts design using a J-FET, the J310 being reputed for its low noise.  The resonator is lightly loaded and high Q elements were used (air variable capacitor and fairly large silver clad inductor), also the VCO range was kept deliberately small (only 7%, see    VCO calculations), a dual varicap (BB204B) was used for this.
A cascade of two buffers, each based on a J310, amplify the signal to a level compatible with the CPLD logic.
The entire oscillator and the first buffer is shielded (salvaging a tin can !) from the rest of the circuitry. 
The "native" stability of the VCO is good, I noticed a drift of a few kHz per minute after warm up.

As a divide by 1, 2 or 4 (depending of the band) followed by a divide by 4 (quadrature signals) plus the associated selection circuits (frequency band and side band) would make a handful of high speed logic circuits, a CPLD solution was chosen. 
The EPM7032SLC-10 (Altera, see MAX 7000 Family) fits the bill with its maximum clock frequency of 125MHz and is available at a reasonable price (Conrad ref. 170511-44). 
The circuit (see altera.gif) is fully syncrounous, and as a bonus the quadrature signal outputs are registered, yielding a minimal timing difference (save for the wanted 90°) of probably less than a fraction of a ns. 
A divide by 6 circuit was also implemented to feed the fractional divider with a signal around 9.3Mz, well outside the receiving bands.
The CPLD "program" was designed, compiled, simulated and downloaded into the device using a really excellent freeware package from Altera (see MAX+PLUS II Software, one needs to register to get a key which is valid for 3 months on a single PC).
Though the CPLD chip can be programmed in situ using the ByteBlaster II tool (see Dev Kits/Cables) it was nevertheless put in a socket for convenience, since it comes in a 44 pin PLCC package.

Fractional divider & phase locked loop

See frequency loop for the circuit diagram.

The frequency stabilisation of the local oscillator is performed by a rather traditional phase locked loop.  A reference signal of 3125Hz, derived from a TCXO running at 12.8MHz is fed to a phase/frequency comparator, which output is then filtered and fed back to the VCO.
The choice of the PLL was inspired by the non-availability in this part of the world (at least at a reasonable price) of a DDS IC (such as the AD9851) and high speed, low jitter crystal oscillators at a reasonable price  (especially since I needed two DDS ICs for the quadrature LO signals).  

Initially I tried to work with the venerable 4046 IC as phase/frequency comparator, the reference signal being derived from the 12.8MHz clock by the timer function of the controller chip.  I however never succeeded in having the comparator II working according the specifications.
So, I used a MC145170 synthesiser IC instead, as the reference divider and the phase/frequency comparator.  I had good results with this circuit in a few earlier VHF receiver projects, the claim of a comparator with a linear transfer function and no dead zones comes as an interesting bonus.  Note that the reference is fed through the frequency input pin, while the signal derived from the VCO is fed through the oscillator input pin, this is because only the R-counter of the component can divide by 1.

The fractional divider is build in SW, implementing a 32bit DDS within a AT90S2313 microcontroller chip (see Atmel AVR 8-Bit RISC Homepage).  A very short program (see DDS.asm for the source) and a 8-bit sine look up table is all what it takes to "synthesise" a 3125Hz sine wave (with more than 100 steps per sine) out of a 9.3 to 9.6MHz square wave (the VCO frequency divided by 6), the actual clock of the microcontroller. 
This set up enables to program the frequency to an accuracy of better than 1Hz (see frequency calculations) for all bands, disregarding of course the accuracy and stability of the TCXO itself.
The content of the update register  (between 47934902 and 46765758) is calculated by the controller (see further) according the desired LO frequency and band to yield a 3125Hz signal at the input of the phase/frequency comparator.

Code is written for use with AVR-Studio rel. 4.05 (see AVR Studio 4 a excellent programming and simulation freeware), and programmed into the device using PonyProg (see PonyProg - Serial device programmer).  The microcontroller chip is socketed as it needs to be removed to be programmed.
The microcontroller accepts a "xxxx" command through the UART at 2400 baud, where 'xxxx' is 4 bytes (most significant first) representing the 32 bit update value, the first byte is marked by a "1" parity bit (the 9th bit), while the subsequent bytes have a "0" parity bit.
The microcontroller feeds a high speed 8bit DAC (MAX7624 and MAX474 combination) yielding a stepwise approximation of a sine wave.  A passive 3125Hz band pass filter (three coupled resonators) gets rid of all the (sub-)harmonics (yielding a very pure sine indeed), its output signal is squared (the 2nd half of the MAX474) and fed to the phase/frequency comparator.

The loop filter is fairly traditional (see loop filter calculations) fist order one with a loop bandwidth of 31Hz and critical damping, its output is amplified twice in order to yield a 3...8V signal for the VCO, well away from the low Q zones of the varicap. 
One could argue that this amplifier (one half of a TL082) would introduce some noise and that a direct connection between the loop filter and the varicap would be better, even if it meant that the varicap voltage would then be in the range of 2...4.5V.
I could not detect the added noise (limitations of the instruments), although there is a faint ripple of less 200µVpp at some 130Hz (estimated) in the VCO line.  To the best of the ability of my instruments, I could not detect any correlation of this signal with to the 50Hz hum, the burst frequency of the DC/DC converter, or the software loop frequency of the controller.  Some further work has definitively to be done here.
 

Power Supply

See power supply for the circuit diagram.

This is most probably the most challenging part of the design as it includes an inverter to produce the -10V for the AF amplifiers.
Little to comment about the linear regulators for +10V and +5V and the RF filter, unless to note that the 7805 is mounted on a small Aluminium bracket connected to the case for extra cooling.

The DC/DC converter is based on the MAX774 and a P-MOS producing a raw 12.5V, which is further regulated by a 79L10 for a clean -10V.   This type of DC/DC converter produces bursts of fixed duty cycle to charge the reservoir capacitor, then wait until the latter is discharged a bit.   Loaded with the amplifiers in the circuit, this produces a burst rate of some 48Hz.  Although there is no ripple detectable (meaning < 100µV), there are very strong input currents involved, so some precautions are needed:  the use of toroid inductors (rather than the more open mushroom type) and the shielding of the entire DC/DC converter circuit (salvaging another tin can !). 

I initially thought I could feed the TBR with a cheap battery adaptor.  Nothing came of it, like in all direct conversion receivers, hum is the number one problem.  Cheap battery adaptors are indeed cheap and come with little regulation and hum suppression. 
I eventually made a well stabilised power supply for +13V including a 100000µF (yes: 0.1F) reservoir capacitor.  Although no ripple was detectable, still better results were obtained when the TBR is fed by a 12V battery, an old 7Ah sealed lead acid type I still had spare.

 

Controller, PC Interface & reference

See Controller for the circuit diagram.

The controller is build around the AT90S2313 (see Atmel AVR 8-Bit RISC Homepage), and an isolated serial interface.  The device is clocked at 12800kHz (a bit high, but it works). 
A Smith trigger (1/2 TL082) handles the lock detect signal from the MC145170.  A simple ADC (RC circuit plus discharge P-MOS) is used to convert the RSSI signal.

A simple program (see triband.asm for the source) initialises everything (including the MC145170) and then continuously accepts  the commands from the PC translating them into a appropriate divisor for the fractional divider and the status the control lines (BS0, BS1, U/L).  Code is written for use with the same AVR-Studio rel. 4.05.

The controller will accept a 8 byte string through the UART at 4800 baud (8bits no parity) containing the desired frequency information in ASCII, will convert it into a 4 byte string containing the DDS update 32bit word.
The input format is as follows: "X123456<cr>", where X is a single character being "A", "B" or "C" respectively for the 20m, 40m or 80m band.   Upper case (A,B,C) indicate the USB, while lower case (a,b,c) indicate the LSB.  The "123456" digits indicate the portion of the frequency in Hz, this number needs to be added to 14 000 000, 7 000 000 or 3 500 000 respectively.
The result is then multiplied by 2, 4 or 8 respectively and is then used as the divisor in a 64bit division where 2^32 * 25 * 3125 * 3 is used as the dividend. The result of which is the 32bit DDS update word which is send serially (through a SW UART) at 2400 baud most significant byte first (the MSB has the 9th bit set, while the three next bytes have the 9th bit cleared).

Every 20ms the device will scan the lock detect output of the phase detector, and AD convert the RSSI signal in a 8 bit entity, it then will be converted into an ASCII string ("000"..."255") pre-pended with the pll lock status ("u"/"l") and appended with a <cr>, hence a 5 byte message will be send through the UART to the host at 4800 baud (8 bits, no parity).

A bi-colour led was included: green = OK, red = VCO not locked, no light = processing a command.

I was lucky to find at a local Ham fest a pair of KSS TCXOs running at 12.8MHz, so I used one in the TBR as master reference.  Its accuracy and long term stability is supposed to be better than 2ppm.

The prime reason of the isolated serial interface is here again to provide a ground isolation between the PC and the TBR.  It is a kind of traditional set up, stealing some power out of the regular RS232 signals for the PC.   Not entirely within the specifications, but it works OK at 4800Hz, though I would not push it beyond this bit rate.

 

Graphical user interface in Visual Basic

I developed a basic but workable GUI in Visual Basic 6.0 using MS Visual Studio (alas, this one is not freeware), see the SWradio
This small program features:
* a window for the frequency in Hz,  including buttons for increasing/decreasing the frequency per digit,
* a window for the RSSI,
* a USB / LSB selection,
* a band  selection : 20m, 40m and 80m
* a rudimentary scan feature : 2kHz steps every 3s,
* a PLL locked indicator,
* a window (which is somehow editable) of a directory of stations,
* a slider for the volume control,
* windows with the actual messages over the com port.

Download SWradio Install for the full installation package (12Mbytes, sorry this is MS Visual Basic, I could not find a way to reduce it further), unzip it in any temporary folder and double click the setup.exe.  It will install all the necessary files and create a subfolder "Triband" in "Program Files".  The file "SWradio.sta" containing the station's memory should appear in the same folder.  It will also create an entry in the start menu.
SWradio has been working under Win98SE, Win2000 and WinXP, and works fully in parallel with other SW, such as Hamscope.

The SWradio.sta file contains the directory of the stations.  For now, the only function that is available within SWradio is the creation of a new entry.  All other editing things can be done using e.g. Notepad, since it is a plain text file.  It is structured as follows:

    1                        ' number of entries minus one
    20m band                 ' a text identifying the station
    000000                   ' the frequency offset in Hz
    0                        ' the band (0=20m, 1=40m, 2=80m)
    0                        ' the side band (0=USB, 1=LSB)
    20m RTTY                 ' a text identifying the station
    070000                   ' the frequency offset in Hz
    0                        ' the band (0=20m, 1=40m, 2=80m)
    0                        ' the side band (0=USB, 1=LSB)

For those knowledgeable of Visual Basic and who want to experiment further, see SWradio sources for the source files (zipped).  Suggestions for improvement are most welcome.

 

Realisation and pictures

The TBR is build in a nice Aluminium box measuring 168 * 103 * 56 mm (Conrad ref. 523232-02) fitting a sandwich of two Euro sized (160 * 100 mm) boards.

The first board contains the HF and AF parts, the second board all the rest.  Both boards are double sided (but not metalised through).  The top (component side) is not etched and serves as ground plane, pins are simply being soldered on the top side for a ground connection.

Except for the FST3253 chip all components are of the conventional through hole mounting type.  I prefer not to mess with the etching of the PC boards, so I have it done at a local shop,  I believe that pushing to full SMC is stretching the technology (theirs and mine) a bit too much.
The FST3253 is mounted on the solder side of the first board, as it is a SOIC16 component, the soldering did not require too much of a steady hand.

The schematics were drawn using the TinyCad freeware, the artwork was draw using the CirCad (free evaluation).  Find here a zipped file containing the drawings in their original format.

Find here the artwork drawings of the component side and the solder side (large zipped files in bmp format).   Note that these were the initial artwork drawings I used for the prototype.  So, use them only as illustrations, as quite a lot of reworking has been done since.   If there is much interest (and I can spare the time) I promise to do an effort to update the drawings.

Testing was relatively straight forward, as most of the circuitry worked first time, though many hours were spend though getting the performance (especially the hum) right.
I used a small DMM, a RF counter (home build) and an old oscilloscope (dual, 10MHz, 20mV) as measuring instruments.
I also made a small RF generator based on a colour burst crystal oscillator, an amplifier, a diode doubler and a band pass filter.  This little gadget delivers a signal of 0.5mW, which is subsequently attenuated by a 30dB and 20dB pad in cascade, yielding some 500µVeff in 5OΩ.

Find here a few pictures of the current prototype:
* the HF and AF board,
* the LO and controller board,
* see also a view from the side, note here the metal sheet between both boards,
* and finally the cover with the controls.   From left to right and top to bottom: the AF output (3 mm stereo), the HF input (BNC), the Com port interface (RJ45), the led and the 13V supply (5mm adapter).

The total cost of the TBR turned around 250 euro, accounting also for the components I had in already my junk box or I salvaged from discarded equipment people tend to bring to me (mostly the inductors).  This sum includes moreover the etching of the PC boards and the two small gadgets for programming the AT90S2313 and EPM7032S respectively.  I purchased most of the regular components in a local shop in Antwerpen (Rato), the more exotic components were purchased through Conrad a well known mail order service in Western Europe.

 

Tests and results so far

I am currently performing some intensive testing on the prototype (when time permits).  Find here already :
* the spectrogram of the 3125Hz loop signal (after the fractional divisor, band pass filter and squarer). 
* the spectrogram of the output signal when tuning the TBR to 7040kHz and applying a 7039kHz, 30µV signal at the input
Both spectrograms where taken with the Spectraplus shareware.
* the composite spectrogram input signals varying from 3µV to 3mV, using the spectrum window of HamScope.

Find here a screen dump of an Hamscope session (see HamScope), the TBR being tuned at 7038kHz).  A nice BPSK signal received at 480Hz, amongst two MFSK signals
at 1000 and 1400Hz respectively and an AMTOR signal (?) at 2100Hz. 

These results were obtained using the following configuration:
* a 15m outdoor antenna connected to a 5m length of RG58, (the heater of the room serving as counterpoise),
* the TBR itself, fed by a fully charged 12V battery,
* a desktop PC (Athlon XP 1800+) running WinXP, the control software and Hamscope,
* and the cabling to the line input and the com port of the PC.

No "birdies" where found on any band when running the TBR with a shorted antenna input, which is a very appreciated characteristic of the direct conversion receivers.

The accuracy and stability of the frequency setting is remarkable thanks to the TCXO, no difference in frequency setting was needed when testing with the crystal controlled rf generator, even not after several weeks.

As one can see, there is still an issue with the THD, which is a bit less than 30dB. 
I used the RightMark Audio Analyzer 5.4 (freeware) to test the sound controller on my mother board (connecting line-out to line-in), the results were not very positive.   Eventually, I got a second hand Creative Labs Ensoniq PCI-128 board in my system.   Although RMAA was very happy about the change, no improvement whatsoever was noticeable on the THD.  This seems to point in the direction of the AGC circuit as the culprit.

 

Still to be done in (near) future:
* improve the lock time of the pll, as the initial lock takes more than 10 s
 => it appears that this is due to the 100µF decoupling capacitor situated at pin 5 of the MAX474, I rather keep it like that in order the keep the noise and hum down at that point.
* adapt the message format between the TBR and the PC, to that of an industry standard (e.g. Kenwood), so that the Rig Control window of Hamscope can be used.

 

Antenna

Confronted with the issue of the antenna, and being aware of the saying that a good antenna is the best HF amplification one can get, I experimented for a while with a random wire antenna running over the garden out of a first floor window.

Later I made an tuned magnetic loop antenna for 20m and 40m from copper tubing, the sensitivity was rather disappointing, but this was maybe due to the fact I used it indoor behind a window that was actually smaller than the loop itself.

I further experimented with offset fed dipoles and end fed dipoles for 20m and 40m out the same first floor window.  Since the length of the dipole did not match my garden and had to bend it, it became a crooked end fed dipole.  There are no tall trees in this part of the suburb (and an huge antenna mast is a no go), the dipoles were suspended just some 3-4m above the lawn, yielding a rather skyward looking radiation pattern (that is what is written in the literature). 
After some more reading I came to the conclusion that a ground plane vertical in the garden would be a good solution, as it would at least yield a low elevation radiation pattern.

As is would be a major project, I wanted the antenna to be multi-band and capable of transmitting, though I would use it for receiving only on the 20m and 40m bands for now.

So, I selected a spot in the garden compromising between the lawn, the shrubs, the neighbours, etc.  With the use of the excellent antenna simulation software MMANA I designed something that should yield acceptable results for the bands above 40m.  Of course, no use to go on the 80m band (or on the 160m band for that matter), as an antenna height of about 7m was the maximum I felt I could go without causing some concerns in the neighbourhood.  Those who want to experiment further, can possibly find some use in GPgarden.maa .

A galvanised steel pipe of 25mm diameter and of 2m length was hammered into the soil until only 35cm stood out. It serves as the mechanical base and as electrical ground for the antenna.   A heavy-duty plastic (PVC) pipe of 33mm diameter and 80cm length was “sleeved” onto it, and then a lager onen (40mm diameter) was added for extra strength.

Four radials were made of 7m lengths of 2.5mm˛ power wire and buried less than 1cm under the topsoil. They were laid approximately into four quadrants, they had to bend here and there to spare the shrubs and the lawn. The radials electrically are connected to each other, to the galvanised steel pipe, and of course to the mantle of the feed coax.

The radiating element of the antenna itself is made of 4 telescoping Aluminium tubes of 2m length each, their diameters are respectively 25mm, 20mm, 16mm and 13mm.  A set of hose clamps of the appropriate diameter are used to clamp the one tube into the other when erecting, care was taken to slid the tubes for a few centimetre with a hand saw.
As about 40cm of each tube remains into the previous, the total length of the radiating element is 6.8m, this comes now inside the above mentioned plastic tube base construction (its inner diameter is 25mm), so that its lowest part stays about 5cm clear of the galvanised steel pipe, one more hose clamp sees to that.
The whole construction is freestanding and tops at 7.2m height. According to my approximate calculations, it can withstand a gale force wind (force 8 or 20m/s or 75km/h) without breaking under to the combined wind shear and gravity forces. The weak point is the plastic tube base construction (if only I could have found a glass fibre tube of the appropriate dimensions instead of the double layered PVC tube). Since the radiating element can rapidly be dismounted (it weighs less than 2kg) this is not really a problem.
See here a drawing of the entire construction, see also the pictures in the garden erected and collapsed.

Obviously, this kind of antenna need some impedance matching, I decided rapidly that a remote antenna tuner placed on the radiating element would be best, especially since I needed some 40m of coax from the antenna base to the receiver.
MMANA gives a good indication of the impedances at the various frequencies (see spreadsheet), which determines the design of the antenna tuner.   The latter consists of a high pass L-C arrangement made of a heavy-duty coil put in parallel with the antenna and a variable air capacitor that is put in series with the feed.  According to the very nice impedance calculator TLW (a Windows based transmission line calculator that comes with the ARRL "Antenna book"), its efficiency is always better than 90% for every Ham band from 40m to 10m.

The variable air capacitor is a relatively big one consisting of two sections of 500pF, which are put in series (in “butterfly mode”), yielding a 30-250pF variable that can withstand voltages up to 1000V.
The coil is made with 2.5mm˛ section bare copper wire, equally spaced and wound on a 40mm tube (removed afterwards). Tiny wood strips were epoxied to the turns for additional stiffness. The coil is tapped at 2, 3, 4, 5, 6, 8, 10, 12 and 14 turns, its inductance can be adjusted rather coarsely from 0.3µH to 6µH. The selection of the taps is made by a heavy-duty ceramic rotary switch (the indexing mechanism was removed).
Two heavy duty DPDT relays where foreseen in order to be able to change the topology of the tuner. A 100pF in series with the antenna is needed according to TLW for a correct tuning in the 30m band. The relays are not used for now as the TBR does not receive in this band.
I used a pair of radio control servos (that I still had in my junk box) to rotate both the variable capacitor and the rotary switch. These little devices are used for model airplanes etc. and while relatively inexpensive, they are perfectly suited for this purpose. The particular servos I used were the JR-507C type and when fed with 5V and controlled with a width modulated pulse (between 0.7 and 2ms), they can develop a fairly strong rotating force over a full 180°.

To control all this remotely, I build a simple device consisting of a micro controller (again the AT90S2313) and a few power transistors (see the schematics and the firmware source).  It accepts a few simple commands through the UART and position the servos and relays accordingly, see here its picture.  The power (up to 0.75A in the worst case) comes together with the control signals over a single pair (a power line cable, which runs next to the coax) of any length up to 50m. The whole is fed near the receiver with 12V DC, the control signals are nothing more than polarity reversals following at 600-baud the TD line from the COM port of the PC (see the schematics).  A small circuit was build for that purpose, as was the small GUI (see the screenshot and the zipped source) with a few simple controls (written again in VB6), see also its picture.
To avoid of a possible source of interference near the antenna, the microcontroller is put in sleep mode (including its crystal oscillator) when not in use, it has to be woken up with a break signal of 50ms on the COM port prior to each command.

All the remote parts are mounted on a frame made from un-etched copper clad epoxy board, forming the HF earth connection, and then put in a watertight plastic electricity box of roughly 25 * 20 * 10 cm. The whole being clamped with two U-bolds on the plastic pipe antenna base construction.  
For those who are interested, herewith the artwork files of the single layer PCB of the controller (local and remote part).

Though it has not been tested in transmission (not being a licensed amateur), the construction was designed to withstand 150W of HF power.  SWR measurements were not undertaken, since I didn’t have the means to do it, nor are they of too much importance for a receiver.

I will report in due time about the results with this antenna.

 

Contact

My name is Jean Taeymans and I live in Belgium near Antwerpen.  I was trained as an engineer (electronics) and am working in the telecommunications industry.  Though I never got a amateur licence (I might some day), I have always had a keen interest in radio communications.

The information included in this Web site is freely available for non commercial use, though people who use it (wholly or partially) for their own home build project, should have the decency to inform me about any improvements they may have thought about.

Comments, questions, advise, etc. are most welcome, click here on my e-mail address: jean.taeymans@telenet.be.