Ka-Band Frequency List


The picture above shows the HughesNet LNB. This unit was from eBay, and is actually really great value for money. The LNB was intended to work in the 19.7GHz to 20.2GHz band for satellite based Internet, as part of the 'WildBlue' systems which are sold in the USA and Canada. The DRO operates at 9.75GHz is filtered and fed through a Wilkinson divider to the 2 sub harmonic mixers. The LNB is dual polarity and is switched with a 27MHz signal applied to the IF output. The RF devices are p-HEMTs from Mitsubishi and NEC. Front end NF is around 1.2dB at 20GHz.The oscillator can easily be retuned to 19.2GHz, giving a coverage of 20.2GHz to 21.2GHz, with virtually no loss of input sensitivity, despite a band pass filter being in place. The dimensions for the waveguide input face of this LNB can be found here. There is a reasonably high resolution picture of the LNB's PCB here.

Another modified LNB is the Inverto unit originally intended for Saorsat in Ireland. The modification here includes the removal of the bandpass filters for each polarity, and the addition of an LO injection point. This allows the full Ka band to be covered. IF gain is good from 600MHz to 2.5GHz.

Inverto extLO

UFO-9 GBS satellite payload diagram.


UFO-9 suffered an anomaly starting at 12:30z on 31st August 2006. The actual 20.7GHz beacon disappeared a few minutes after its initial anomaly. The satellite is now in a graveyard orbit and has been identified visually to confirm this. This satellite was a handy 20.7GHz test signal! The UFO-2 satellite which takes over from UFO-9 does not have any Ka capability for GBS or beacons. UFO-9 is listed as an uncontrollable space asset - its probably beyond recovery and should be considered inactive.

20700UFO9 Gbsshot1

Gbsshot2 Gbsshot3


Since the launch of the Kepler space-craft, some effort has been expended in putting together a 32GHz down converter from commonly available parts.

32ghz1 32ghz2

The photos above show two views of the down converter assembly that is mounted at the dish feed. A scalar feed and associated 32GHz circular waveguide transfers the signal to the down converter unit which came out of a microwave link. On the back of the down converter are the DC-DC converters for FET bias as well as -24VDC generation for the phase locked oscillator. In this system, the local oscillator runs at 34.2GHz and high-side mixes, giving an IF in the region of 1GHz to 2.5GHz. The LO is derived from a 111.0389610MHz reference, which is multiplied up 308 times to reach the final LO frequency. The 111MHz reference is GPS locked for stability, and results in around 40-60Hz of signal wander at 32GHz. Sun noise is detectable albeit at very low levels of noise increase.


A prototype Ka band "LNB" is shown above. The LNB consists of a ~5GHz phased locked oscillator from M/A Com, the output of which feeds a X4 multiplier.

This gives an LO around 19-21GHz. The LO is fed to a mixer / IF amplifier that came out of an old DMC microwave link operating at 22GHz. The input filter has been suitably modified and the IF section completely replaced with a pair of MMIC broadband gain blocks. The entire unit runs from 24V DC.

The 'front end' is the usual Direct-TV Ka band LNA.

The direct-tv Ka LNB has been chopped in half to produce two LNA units. The LNA has retained the existing Ka band feed, and now provides LHC and RHC outputs on SMA connectors. Useful gain has been measured from 17GHz to 22GHz. The pictures below show the modified LNB.

Kalna1 Kalna2

Ka-band LNB now arrived, pending modification to use the LNA stages. After careful study of the LNB design, it may be useful to know that the LNA stages should be useable at 24GHz.

Kalnb1 Kaconv diagram

The block diagram above shows the basic concept of the converter - the local oscillator is generated from a 10MHz GPS locked oscillator, which is then multiplied up to 20GHz, resulting in a DC to 2 GHz IF. The 20GHz LO is generated with a modified M/A-Com phase locked oscillator, which generates 5GHz from a 100MHz reference. This 5GHz signal at around +10dBm is fed to a Celeritek X4 multiplier, resulting in a local oscillator signal at 20GHz (10dBm). The incoming RF signal, is mixed with the 20GHz local oscillator to give an IF in the 600MHz to 1.8GHz region. For testing purposes, the local oscillator has been set to 19.1GHz.

Test config

The IF is amplified using the usual MAR-6 specials, and fed to a communications receiver, which is also locked to the same 10MHz GPSDO as the local oscillator chain. Note: The AOR 8200 shown in the picture below was used for initial testing - this receiver is not locked to a GPS frequency standard. The receiver was used as an alignment aid for peaking the dish on the satellite.