Zur deutschen Seite

K Band Transmitter

Fabrication Responsibility:
Danny Orban, ON4AOD (Belgium)
Stefaan Burger, ON4FG (Belgium)
Peter Pauwels, ON1BPS (Belgium)
Philip Sanders, ON7IZ (Belgium)
Average Power Dissipation:
15 Watt

K Band Transmitter Oscillator

Fabrication Responsibility:
Danny Orban, ON4AOD (Belgium)
Stefaan Burger, ON4FG (Belgium)
Peter Pauwels, ON1BPS (Belgium)
Philip Sanders, ON7IZ (Belgium)
Average Power Dissipation:
1 Watt

Table of Contents

  1. History
  2. How much power do we really need ?
  3. Concept
  4. Oscillator
  5. Lower Half of the Transponder
  6. Power supply
  7. Upper Half of the Transponder
  8. Antenna


Back in April 1993, we started to think seriously about doing 'something' on Ka-Band for the Phase-3D satellite. The original idea was to construct a beacon. We would generate a couple of Watts at 8 GHz and triple this with a varactor. This was introduced to Karl Meinzer (
DJ4ZC) in July 1993. Karl convinced us to construct a transponder rather than a beacon. The biggest problem was to generate the necessary power: 30 dBm on Ka-Band is a lot. The possibility to really make the transponder came with the availability of a 1 Watt transistor. The actual transponder is being built by Stefaan Burger -ON4FG- and myself (ON4AOD). The bonding and testing of the final amplifier is being done by Peter Pauwels -ON1BPS- and Philip Sanders -ON7IZ-. The flight version is powered on and working fine since June.

How much power do we really need ?

Because of Doppler shift, the transponder will be switched on at apogeum only. So the path loss needs to be calculated for 47000 km. From this distance and the diameter of the earth we know that we need an antenna with a -3 dB beam width of 13.7 degrees. Converted into gain, this gives us 23 dB. We assume a receiver system noise figure of 2 dB. By the time this project will be operational, we will be able to get a noise figure well below 2 dB with amateur means. The receiver bandwidth was assumed to be 2400 Hz. Reducing the receiver's bandwidth to 1 kHz reduces noise power by 3.8 dB. The sky temperature is assumed to be 50 Kelvin. Some documents claim a sky temperature of 10 Kelvin at Ka-Band. This would result in 0.9 dB less noise power. The receiver antenna gain can be higher than the 35 dB used. 40 dB seems reasonable. Calculating a link budget with these parameters tells us we need 800 mW of power.

On top of the calculated link budget, we have a number of parameters that will further weaken the signal. Atmospheric losses losses are an estimated 0.15 dB/km for water vapour and 0.01 dB/km for oxygen. Rain is obviously another major cause of loss. A good estimate for absorption on 24 Ghz is 1 dB/km in the case of 5 mm/h rainfall, and approaching 10 dB/km in the case of 30 mm/h.


The transponder is built the same way as our current mode L and S transponders on OSCAR 13. Basically these are HELAPS type transponders without the amplitude restoration. The circuit is divided into four pieces:


The oscillator generates a 65 MHz LO signal. The circuit sits in a separate box to avoid heat transfer from other components.

Lower Half of the Transponder

Here we come in with the 10.7 Mhz from the IF matrix and the 65 MHz LO signal from the oscillator module. The outputs are 470 MHz IF and 1200 MHz LO. This pcb also contains the clipping circuit and the bandpass filter for the IF. This circuit is on standard epoxy pcb. Most filtering is done with helicals.

Power supply

In the same box as the lower half of the transponder, also sits the switched power supply. This converts the 28 V from the spacecraft to the several voltages needed. This power supply will also do the proper power on sequence for the finals. The power supply design is from Werner Haas, DJ5KQ.

This box sits on top of the next one.

Upper Half of the Transponder

This part is fed with 470 MHz IF and 1200 MHz LO. The output delivers 800mW on 24.048 GHz into the antenna. The circuit is partly built on pcb and partly in waveguide. The part from 470/1200 MHz to 7000 MHz LO and 2800 MHz IF is done on 0.79 mm Teflon. All filtering is done in microstrip. The tripler from 7 GHz to 21 GHz is partly microstrip - partly waveguide and the mixer is waveguide. Both the 21 GHz LO and 24 GHz output filters are waveguide. The final stages are again in microstrip: 0.25 mm Teflon with 6 mm aluminum on the back side to facilitate the mounting of the connectors and to allow sufficient cooling for the power transistors. These pcb's are fixed to an aluminum plate that covers the whole back of the module. This plate radiates the heat. All connectors at 24 Ghz are SUHNER 3.5 mm. All others are SMA.

The mixer delivers about -10dBm. From there we amplify with HEMT's to 11 dBm. This signal is amplified by two modules to 26 dBm and by the final to 30dBm. The drivers are HEMT's from Toshiba. They are actually specified up to 18 GHz but work well at 1.5 cm. Depending on the device, they give between 5 and 10 dB amplification. The modules are FMC2223P1-02 and FMC2223P5-01 from FUJITSU. They are internally matched devices that have internal power supply biasing networks. They will deliver 21 dBm at 12 dB gain and 28 dBm at 9 dB gain. Since they are matched for the 22.4 to 23.6 GHz region, we lose some of the performance. Fuji was so nice to test a batch at our band for us, and have come up with an actual decrease in gain of 3 dB. This can be partly compensated by external tuning. The final stage is the prototype of a new type of transistor from RAYTHEON. It's a 1 Watt PsHEMT, with a typical gain of 7.0 dB (at 2 dBc) from 20 to 25 GHz. We mounted the devices on standard 5880 Rogers substrate and obtained very good results.


Although a dish looks promising, there is a problem with the feed. Both horn and waveguide feeds are mechanically weak and might not survive the launch. A Cassegrain feed is good but not suited for this small a dish. The hyberboloid would block too much of the primary reflector and thus reduce efficiency.

A horn with a 13.7 degrees beamwidth at -3 dB, a gain of about 23 dB, would have an aperture of 7 by 10 cm and a length of 22 cm. We have the flight version of the horn. It is a 26.5 dB gain horn. It is compensated for equal E and H planes and reduced sidelobes. Because of this the - 3dB points are close to those of the 23 dB horn. Also, with 40 cm in length, the feed point is close to the bottom of the module where the output is. This saves on coax.

Last updated: Feb 6, 1996
by Ralf Zimmermann,