After much deliberation it was decided to buy the PCB’s ready made from DB6NT in Germany, thereby eliminating any problems arising from our board making techniques, then completing the rest of the module construction ourselves.

The mixer serves as both a receive mixer or transmit mixer depending on the state of the internal TR switching. The 144 MHz IF receive signal passes through a post mixer pre-amplifier while the 144MHz transmit IF has a 50 ohm input adjustable attenuator prior to the mixer diodes, to set the mixer output level.

The Mixer requires 20 - 50 mW of LO drive and 2 watt of 144 MHz IF. It is capable of producing 1.3mW DSB, 0.5mW SSB and It has a DSB noise figure of 8dB and a RX gain of 15dB

DB6NT 3 stage low noise pre-amplifier

The pre-amplifier is a waveguide fed 3 stage low noise amplifier producing around 30dB of gain. Coupled lines are used to couple the NE32584 HEMT stages, for this method of coupling acts as band pass filters as well as coupling, out performing previous chip capacitor coupling designs. Negative bias is generated on board and drain rail voltage regulation is supplied from a regulator built within the waveguide enclosure.

Trevor VK5NC had the task of producing the machined aluminum enclosures for the project. His skills as a fitter-machinist, and a understanding employer who allowed him to use his work facilities after hours, made it possible to produce enclosures of a very high standard.

Ready made boxes are available from Germany for those that do not have access to machinery. A alternative is to make the enclosures from sections of flat plate, drilled and screwed to form the enclosure. Only a limited amount of tools are required to produce this sort of enclosure, which will work just as well as the milled enclosures we used.

The Filter is a necessary component of both the receiver and transmitter filtering out the images and harmonics produced in the mixing process. The required signal on 24.048 GHz will pass through the filter, while the oscillator frequency of 23.904 GHz and lower sideband of 23.760GHz, and images are attenuated.

In the prototype filter designed by Charles Suckling G3DWG, the oscillator was attenuated by 15.5dB and the image was attenuated by 37dB.

The filter waveguide was carefully marked out using the dimensions of the G3DWG design. The slots for the iris plates were cut with a junior hacksaw and a square jig to ensure that the slots were in good alignment.

Both sides of the waveguide are cut just through the surface, so that the iris plate can pass through the waveguide

Assembled G3DWG design filters.

Iris plated, made from 0.5mm thick copper sheet, were first marked out and drilled in one sheet, then cut and trimmed to neatly fit into the waveguide slots.

Once the four iris plates were pressed into place, making sure that the iris holes were centered in the waveguide and in alignment, the waveguide was lightly squeezed on its ends, using a vice, to close up the slots onto the iris plates. This was necessary to reduce the gap around the iris place so that solder would not weep into the cavity.

Soldering the iris plates into the waveguide was accomplished by wiping a minimal amount of solder paste onto the outside of the waveguide around the iris plates, then heating the waveguide on a hot plate until the paste melts. After the waveguide had cooled, the excess was trimmed off by draw filling the outer surfaces. The flanges were then fitted and the filter was then stood on its end, on the hot plate, and soldered.

In the original Toshi project, we had decided to try and use some commercial Qualcom PLL oscillators at 2390.4 MHz, multiplied by 5 to produce the required 11952MHz injection for the mixer. This proved to be a disaster, as all the reasons against the use of PLL’s expressed by John Patterson in previous conference talks , came to fruition. I refer to

When this PLL oscillator and it’s X5 multiplier were tested. The phase noise alone was enough to kill any further thoughts of using a PLL at these frequencies, although a similar system had been used in the Qualcom 10 GHz transverters with success. Still at 24 GHz the multiplication factor was double that of the 10GHx transverter, placing a greater emphasis on having a stable, clean oscillator.

So we quickly changed direction to use the proven G4DDK crystal oscillator-multiplier, as the basis of the oscillator chain. We did use the X5 multiplier section of the original Qualcom PLL system to multiply the G4DDK oscillator output to the desired frequency for the mixer.

By using a 99.6 MHz high stability crystal, X3 to 298.8Mhz, X2 to 597.6 MHz, X2 to 1192.2 MHz, X2 to 2390.4 MHz, then through the Qualcom multiplier, X5 to the final injection frequency of 11.952 GHz.

Stability of the oscillator is of paramount importance and so we chose a very high stability crystal, (40deg), purchased from Germany as suggested by Sam Jewell G4DDK. Remember that to achieve the mixer frequency, the oscillator is multiplied 240 times.This means a 10Hz shift in oscillator frequency will result in a 2.40 KHz drift at 24 GHz.

Operating in all kind of weather conditions as well as from limited portable power sources, presents a unique set of problems not experienced in controlled environment of a home station. To overcome temperature variations, the oscillator enclosure is mounted within a second enclosure with polystyrene insulation between. A crystal heater is used to quickly stabilize the crystal frequency.

To ensure a stable power supply we regulate from 24 Volt battery supply to 13.8 volts. Separate supplies are derived from the 24 volt rail for the exciter and transverter. Internal regulators reduce the supply further for the individual stages of the transverter. This allows us a stable supply from a greater variation of better voltage vastly improving the overall stability of the transverter.

Suitable sized dishes for this project were not going to be easy to find, especially as we needed a quantity of dishes for the 5 transverters that were to be made. In the true spirit of the hobby we decided that we would try and make dishes out of fibre glass, and as no one in the team had any experience with this process, it was going to be a new learning curve.

A pair of commercial dishes that were part of VK3ZQB’s Qualcom 10 GHz transverter, were ideal in both size and depth for 24 GHz. A plaster cast was made of the two dishes, to be used as a mold for making fibre glass forms.

A wooden form was made in which the two commercial dishes were mounted and coated with release a agent. Casting plaster was poured into the form and allowed to set. The form and the dishes were removed resulting in a perfect mold of the original dishes.

To test the accuracy of the two dishes, a target was placed at the calculated focal point of the dishes. The Sun was used as the source of RF. The indication of how successful we were in the dish construction, was shown by a well focused image of the sun right on the focal point of both dishes

Both penny feed and dipole feed systems were used with good results. Trevor VK5NC used the dipole feed made from a length of WR42 waveguide. Colin VK5DK used WR42 and a penny feed while Russell VK3ZQB used a length of hobby tube, available from most hobby shops, that is equivalent to WR34 22-33GHz, and a penny feed. All the feed systems used worked well and it would be hard to say that any one is better than the other. Interconnecting WR34 to WR42 requires a transition and I found that some tuning screws were needed on the WR34 feed.

The bonding of the ground to the upper ground areas of the PCB is critical. Connection of these surfaces has to be accomplished with minimal added inductance. The preferred method of bonding earth plane surfaces is to use through hole plated eyelet’s, a process not easily accomplished by hobbyists.

We used a method of threading copper foil through the board, which achieved a satisfactory bonding and was relatively easy to accomplish with limited resources.

Punching slots in the PCB over the end grain of a wooden block for the foil links.

A small punch was constructed from a slither of Stanley blade, ground to the required thickness and size and mounted into a brass handle, The punch is used to punch a series of slots at exact points where the foil connection was needed. Once the slots were punched, a strip of foil is threaded through the slots in the board and pulled tight.

After the foil was in place, the Teflon board is gently smoothed to flatten the foil and the area around the punched slots. The finished smoothed surface is nearly as flat as the original board before the slots were punched.

Flattening the foil ready for soldering.

The foil is then doped with resin and soldered with silver loaded solder insuring that the foil is well sweated. Excess solder and resin is then removed.

RF lines are treated differently. The component is placed and held in position, observing the rules for static handling, with a coating of resin ready for soldering. Silver loaded solder is then applied and the excess removed leaving just a fine film of solder to hold the component. The less solder used, the better the joint will transfer RF.

Two isolating capacitors of approximately 1pf capacity, are made from a small piece of 0.005" double sided Teflon PCB. The small pieces of Teflon capacitor are carefully soldered to the PCB track, then a piece of copper strip is soldered to the top of the capacitor, across to the second track. The photo of the HPA assembly shows a diagram of the coupling capacitor which I found to be extremely difficult to make. Mostly due to their small size and the difficulty in placing, soldering and seeing them. These capacitors are placed first then tested to insure that they are not shorted.

After all the DC components are fitted and prior to fitting the active RF devices, the assembly is DC tested to insure that both positive and negative regulators are operating correctly, and that the voltages are reaching the correct positions on the RF track. Then the RF devices are fitted and the assembly is once again DC tested to insure that all RF devices are operating. Only then is the PCB fixed into the enclosure.

A silver loaded epoxy is applied to the under side of the PCB, making sure that there is a continuos coating of the epoxy around the RF track. The PCB is mounted into the enclosure and screwed down. A layer of conductive foam is placed over the PCB and a further block of polystyrene foam placed on top of that. Then the assembly is placed under a weight so that the foam will press the epoxy coated PCB firmly into the enclosure. It will remain here until the epoxy dries.

Access to suitable test equipment is essential to successful alignment of these modules. Fortunately Alan VK3XPD, having a fetish for collecting Hewlett Packard microwave test equipment, was able to help. This was Alan’s contribution to the group project and when all the modules had been assembled and DC tested, we lined up at Alan’s shack and performed a production line style RF alignment of all the modules.

Minor fine tuning was done individually as each of us assembled the modules into a working transverter.

The transverter assembly uses a centralized chassis plate with modules mounted on both sides, behind the twin dish assembly, later known as "the Z cups" The 24 GHZ RF chain is mounted directly behind the dish feeds as seen in the accompanying photo

Under the chassis is the X5 multiplier and control unit. The control unit contains the DC polarity sense relay, PTT and voltage control, 24 volt relay supplies and filtering.

Once the transverters were operating the thrill was there to go and see how they performed over some distance.

Our first contact was over a short path of just 20 km, 5-9+ signals. Next we moved further apart taking up locations at Warrnambool and Portland, a distance of 79km. Further contacts were made from Heywood to Tower Hill near Koroit.

After the first field trials where we experienced the sharpness of the dishes on 24 GHz and had gained some confidence in setting up and operating on 24 GHz, Colin VK5DK and Trevor VK5NC set a new VK5 state record with a distance of 86Km.

On the 15/2 VK3ZQB worked from Port Fairy to VK5NC/VK5DK at Mt Gambier, setting a new state record for Victoria and South Australia.

On 17/2 we repeated the exercise established a national record from The Bluff, north west of Mt Gambier to Port Fairy, 158 km.

On the 29/2 that record was extended with a contact from the Bluff to Tower Hill a distance of 171 Km.

On each occasion we used 10 GHz as a sighter for the 24 GHz equipment. Pointing the 24 GHz equipment accurately is extremely difficult with about 2 deg beam-width on the dish and using 3cm to establish contact first, gave a accurate direction to point the 24 GHz gear.

Building equipment for 24 GHz is certainly a test of ones ability and discipline to work on very small SMD assemblies. The saying, "double the frequency = four times the difficulty", is perhaps a understatement, but at the end of the day the rewards are there and I am sure that we will venture to higher bands in time.

This is where we have been, its not the end of the project and the following indicates the direction that we will be taking in the future.

the transverters that we have constructed work well and probably with diligence and perseverance we will extend the record distance still further, but to help our cause we will modify the transverter to improve the system performance dramatically, viz. bigger antenna, 600mm dish and more power.

We have already started construction of the power amplifiers that will increase our waveguide power to approximately 500 mW. Coupled with a single 600mm dish, elevating any fear of squinting that might have occurred with the previous two dish system, our transmit gain will be about 11db bringing our ERP to around 5kW. Our receiver gain will improve by approximately 3db.

Shown here is our recently completed 250mW amplifier stage which will amplify from the original 70mW transmitter stage. This will drive another stage of 8 parallel HEMT stages to produce 500 mW.

Shown is the 500mW amplifier assembly.

Trevor VK5NC has been able to manufacture a waveguide switch that will couple the transverter components into the single dish feed.

It is hoped that two of these "up-powered" units will be completed ready for the coming summer season.