Antenna projects
2.45 GHz 30 Watt rectenna array (July 2009)
High power rectenna
The rectenna / beamer below was developed for the BBC to demonstrate transmission of power using a radio beam and the concept of an "infra red" rectenna. It appeared on "Bang goes the theory" on BBC1 in September 2009, where my son and I appeared in the program at Thales quarry in Wells.
Originally, the unit was intended to hang below a model helicopter and hover in a 2.45 GHz microwave power beam, producing 3 Amps at 7.2Vdc. However, it quickly became apparent that this was never going to work - the final rectenna was too heavy and a tethered flight was accepted. There was also the problem of the helicopter RC electronics and servos being illuminated by the high power RF.
Rectenna V1, carbon fibre frame to fly underneath helicopter (35 MHz RC), wrong rectifier, not enough elements
and wrong diodes (BAT70). Carbon fibre was not the best choice - it is highly conductive (but very light).
Rectenna V3 with ground plane - best rectifier, best HSMS 2800 diodes, 100 elements
Jem Stansfield demonstrating the final rectenna during filming at Thales quarry
Getting ready for flight, Sam Worskett and Jem, Director Giles and cameraman
The power beam was produced by a magnetron from a domestic microwave oven driving a waveguide horn, producing 100 kW EIRP. The horn was designed as described in Jasik. The safe distance in the power beam was calculated as 10m at 137V/m. During experimentation, the field strength was measured continuously to ensure safety.
Even though the horn was quickly welded up from steel plate, field strength measurements in the beam showed it was producing the expected gain. Sidelobes and leakage were good.
The magnetron has a fat ceramic probe antenna which excites the WR340 waveguide in TE10 mode and is spaced about 20mm from the back short. The output is very broadband, about 40 MHz wide and is heavily modulated with 50 Hz:
Horn /magnetron assembly before seam welding
A research license was obtained from OFCOM to use the equipment, limited to the BBC Research Laboratories at Tadworth and at Underwood Quarry near Wells (Thales radar test site). This made rectenna development rather difficult.
Four attempts were required before a viable rectenna unit was produced. It cost many burnt out diodes and weeks of work. The choice of suitable Schottky diodes is limited - the diode needs to have low capacitance, high breakdown voltage and high dc current capability of at least 100mA. Agilent HSMS 2800 (Vbr = 70V, Cj = 1.6pF) were used in the end but they are not ideal. HSMS 2810 give more output (lower Cs and Rd) but are too delicate with Vbr = 15V. BAT70 were quite good and cheap but produced a low output. Some researchers in the US advised a 10A 100V MBR10100 Schottky, but with 300pF capacitance it didn't produce a glimmer, as expected.
DC output vs load resistance, +26dBm EIRP at 10cm - HSMS2800 used.
There is much contradiction in research papers. One Japanese paper advised a full wave bridge. However I found that a single shunt diode gave the best dc output power. Using a voltage doubler circuit with 2 diodes was not effective at 100mA loads. Another paper described how the single diode behaved as a full wave rectifier due to class E operation. Many people are interested in getting low powers, a few 100 uW to run a processor, but I needed 25-30 Watts.
In 1964, Bill Brown at Raytheon obtained 75 Watts from a 2 ft square rectenna directly driving a drill motor. He had 5 kW of transmit power into a 9ft dish and 4480 diodes. From the papers available, it is not clear how exactly this was obtained, as it would be impossible to fit 4000 diodes/dipoles in such a small size. Some rectennas appear as a mesh or net of diodes, without any dipoles, but this wasn't tried. HP 5082-2800 leaded diodes are very expensive.
I connected each SOT23 diode across the centre of a dipole. There was no LPF to reflect harmonics back as reported in some papers. I used a balanced choke line and 22pF 0603 cap to recover the dc. I tried a matching network between the dipole and diode but it didnt make much difference.
What improved output power the most was reducing the groundplane spacing to much less than a quarter wave. Possibly, it reduces antenna source impedance and doubles the aperture of each element (about lambda squared/8). Virtually all designs give a good voltage until a load is placed on them. Each element gives maximum power transfer when loaded with 600 Ohms.
Rectenna element with rectifier, choke line and dc capacitor
There are 100 elements consisting of half wave dipoles etched on 1.6mm FR4. I found the best length experimentally as FR4 has quite a high dielectric constant. Fat dipoles were used to try and reduce the source impedance and much experimentation and optimisation was carried out.
It proved impossible to simulate the rectenna as it was not practical to measure the input impedance of the diodes in a balanced system at high power levels or under high and varying dc current. The RF input match of the rectifier changes depending on dc load.
I tried varying the spacing of the elements and found lambda/2 was the closest I could go before there was any reduction in power due to mutual coupling to its nearest neighbours.
At 5m range, the power density from the horn/oven is 200 W/sq m. The effective aperture of each element is 0.00186 sq m, (ignoring the image antenna behind the groundplane) so with 100 elements Aeff = 0.186 sq m, the theoretical max power that might be collected is 37 Watts.
Note that it is the dc power from each element that is combined, the RF outputs are not summed coherently, then rectified, as this would produce an antenna with a narrow beamwidth. It is interesting we have an antenna whose radiation pattern is no more directional than a dipole over a groundplane but has a large collecting aperture .
The voltage from the rectenna varies considerably with load, it is effectively a current source. The Esky Lama V3 helicopter requires 7.2V at 3 Amps (2 cell LiPo), so a switching power supply was designed to take 9 to 35Vdc in and generate 7.2V fixed dc out. There are 50 pairs of 2 series elements in parallel giving about 35Vdc (no load). It is not known how well the elements current share and how they can be forced to current share. A bank of electrolytic caps helps reduce the surges taken by the motors.
Slotted waveguide antenna
A horizontally polarised omni, (slotted waveguide array) at 9.4 GHz in 0.4mm brass sheet for
a marine radar target enhancer (RTE)
Designed using MATHCAD, measured gain = 5 dBi, S11 = 15dB, B = 300 MHz. Using reduced height WG16 waveguide
produces a better ripple in the azimuth response.
The antenna can be made for a few pounds in production quantities by photo-chemical machining
(carried out by TECAN in Dorset).
Unfortunately, I could never make the system work to spec
and the project was cancelled by the Customer - the only project that I have
ever worked on that was a complete failure and cost me dearly.
Unfolded x-band antennas on a sheet
Measuring the input match of the pin probe and guide
Prototype spiral c 1985, backing cavity and Marchand balun, 0.5 to 4 GHz.
Very inefficient, but wideband and gives circular polarisation - good for radar warning receivers
where gain is not important. The cavity loading (exponentially tapered 377ohm/square Aquadag paint)
is ineffective here due to skin depth. Better to use a dipped honeycomb structure
Helix at 900 MHz.
These work well to 3 GHz but higher frequencies require
very thin supports although the accuracy of the turns has only a small effect on return loss and pattern.
Air is by far the best dielectric.
This example was matched with a quarter-wave transformer and designed from
the "Antenna Engineering Handbook", Jasik, 1961.
Dual wideband "U-slot patch" antenna
These patches give a 25% return loss bandwidth with a modest ground plane spacing. It was found that 0.8mm FR4 backing required significant rescaling over a pure copper foil implementation, even at 750 MHz.
Even greater bandwidths can be obtained using a complementary ground plane design. Wideband, half-space radiation - and no balun required.
A range of RFID loop antennas for high power investigations
Antenna measurement facility
Using a small rotor and FSP spectrum analyser, the amplitude responses
of small antennas can be plotted up to 13.6 GHz. The plot below shows the azimuth responses
of a 2 slot waveguide array at 0, 10 and 20 degree elevation angles:
It can be seen that this type of antenna does not work too well as an omni as the elevation angle is increased.
Using Sonnet , the antenna can be modelled very effectively, to see if this should be expected
or is it just that the antenna is not working properly?
The simulator shows that ripple should be expected at large heel angles:
Modelling kindly carried out by Hoopy-hoppy
Contact details:
JWD Ltd
Telephone/Fax: +44 (0)1403 261251, Mob: 07973 173808
email:
julian @ jwdltd.demon.co.uk
