LNB re-use:The Hacksaw modification


To be fair, this group tends to include the same models categorised as 'easy', but put to uses which require additional connections, such as gain blocks and converters. Gaining access to the input RF stage, in particular, nearly always requires the waveguide feed to be sawn off. Perfectionists may wish to get their files out too...

When re-using an LNB as a 10 GHz gain block, the image filter that follows the final gain stage may be retained in some applications. For instance, if the gain block is pushed into clipping, it can be used as a LO multiplier, with the integral filter used to select the appropriate harmonic. This is effective for anything up to x 4 multiplication, though the filter will usually require some dielectric loading to bring its frequency down a little. Sticky foam pads are quite good for this - they are pliable enough for the glue bond not to crack, resulting in a reliable modification. For LOs at 9.6 and 10.080 GHz, sticky pads have worked well for me - they are good for minor moves to 10.368 GHz also.

For converter use, one potential problem is the type of mixer the particular LNB uses. Early designs used diode mixers, and this type is again current, but there was a 3 - 4 year period where integrated mixer/LO/IF gain chips were used and these are not suitable for use in transmit applications.


Gain blocks:

It is always sensible to fit a 2p2 blocking cap in series with the 1st stage input - even if you are sure that it is being fed from a dc open circuit source. The biasing arrangement used in most LNBs is very inforgiving to momentary shorts to ground on the gate connection, so be warned. Unfortunately, quite a few models have very short input tracking, and it can be difficult to fit in the blocking capacitor, but it should be fitted non-the-less.

Note that you have a choice of input - the vertical or the horizontal, depending on the supply voltage. The first example below (pcb with discrete probe antennas) shows this clearly, with the right-hand input being the one selected by a 'low' (ie, less than 15v) supply voltage and therefor suitable for a 12v rail.


Cambridge AE12


Grundig variants on a theme:


AEN3S [less connections]

with sticky-foam pad on bandpass filter

foam pad removed


Converter block:

This example was very easy to use. Using an external crystal multiplied 9.6 GHz LO, the IFoutput was 768 MHz - within the range of the LNB IFs high pass filtering. The right-hand picture shows connections which are, from left to right: RF input, LO input, IF output and supply.

Cambridge G88


Transceive converter block:

Dual output LNBs (and more recently, quad output LNBs) are used when one LNB has to supply more than one satellite receiver - so you need both horizontal and vertical antennas active at the same time, and since there is some frequency overlap this requires two complete converter chains in the same LNB. For dual output variants, the same local oscillator is used in both converters. With the advent of high and low satellite bands, the LO is also duplicated in order to give (simultaneous) appropriate LO feeds, which gives four permutations and therefor requires four LNB outputs.

With multi-output LNBs that employ diode mixers, there is no reason why one chain should not be used for receive conversion, and another for transmit conversion, though in this latter case a few more modifications are required. Invariably, the internal ceramic resonator oscillator output is split into two equal amplitude signals using a Wilkinson splitter, so that a pair of receive/transmit chain mixers can be fed. Inputing an external stable LO source to the Splitter input will therefor allow both an rx and tx mixer to be fed easily.

On the transmit chain, you have to be a little agricultural - all of the IF amplifiers need linking out, and the rf stages have to be either 'turned around' (and the gate and drain dc supplies swapped over, or the output taken directly from the diode mixer output. Since the signal level out of the mixer (including LO leakage) is relatively high, it is only reasonable to re-use one of the rf stages, which can usefuly be employed as a buffer. Turning the appropriate rf amplifier device around also implies that the internal band filter image filter be used, so a foam sticky pad is required over this to bring the pass band down into th 10 GHz band. Although the modifications are quite extensive, this provides a low cost and easy approach to 10 GHz transverter construction


High definition picture of an unmodified, late 90's, dual lnb (Grundig AEN 04T)


Cambridge LNBs from the 90's, be they single or dual output, are a joy to work on. The one shown below was modified for use with a tx/rx IF of 288 MHz. On receive, this required some modification to the IF stage too, since these are always designed to include high pass filters with cut-off at about 800 MHz. The right-hand picture also clearly shows sticky pads fitted over the image filters to bring them down onto the 10 GHz amateur band.

Transceiver based on Cambridge AE2

[On Rx, the IF HPFs had to be reworked to cover the relatively low IF of 288 MHz ]


Dual LNBs can also usefully be used as dual 10 GHz interconnected amplifiers - particularly useful where one amplifier is used as a frequency multiplier, feeding a carelessly made pipe cap filter with plenty of in-band loss (!), with the second amplifier used to restore the 10 GHz output to 20mW or so - as below:

This also shows an alternative way to deal with awkward casting shapes. The pcb and sectionised screen casting have been retained, but the main housing has been replaced with a flat piece of aluminium. In this example, which is from a low power beacon transmitter, the drive is at 3456 MHz, and this is multiplied by three in the first amplifier pair, filtered at 10 368 MHz, then amplified up to 20 mW in the second amplifier pair. A 5m length of RG-223 coax (about 15dB loss) then feeds an LNB with the RF devices and their supply connections turned around to produce 20mW at the integral horn antenna. I can hear this 15 miles away 40dB above noise in a 10 Hz bin bandwidth using a 45cm dish at the receiving end.

That's enough LNB mods