How stable is a satellite LNB LO?

Last update: 12.12.07

It had never occured to me to check how stable the puck controlled LO was in any of the many satellite LNBs that had come my way. The assumption had been that they would be so unstable and rough that it wouldn't be possible to obtain an audible beat note from any of them. Then Mike, G8VCN, showed me his portable 3cm receiver, using an LNB fed scanner portable. Good for wideband, I thought, still not fully comprehending the shock about to come my way. Even the request for a 3cm carrier didn't really register. Only the words 'It's a nice scanner - it's got ssb. Several of us at the club have got them. We've been listening to the beacon with them' did I start to understand.

Mike's LNB was an old C-120 waveguide flange type. We had only just connected 12v to it, so the drift was quite bad, but you could still tell that the tone of the beat note was quite good as it drifted through the receiver passband.
Within a few minutes, stability was good enough to catch the beacon call-sign fully. The LNB had not been selected for good stability, and a day or so later I thought it would be interesting to go through the stock of my unmodified LNBs to see how they compared. All the units uncovered had 9.75 GHz LO frequencies, so the IF was going to be about 618 MHz. I have no receiver at this frequency, so to bring this down to HF a connectorised ring mixer was put together and the HP8640 sig gen switched on. Most LNB units tested had a similar note/temperature stability but the odd one was pretty awful. However, two types had noticeably better phase noise. These were samples of the early Cambridge digi-dish unit and, best of all, a couple of the LNBs that Lidle were selling a few ago at 3.99.

   cambridge_lnb                       lidle_lnb

The first Cambridge LNB had noticeable drift, even when left on for half an hour, but the the Lidle unit was much better. This Lidle unit can both be heard in this wav file and seen on the waterfall display below (the main signal is the standby beacon GB3CAM, running in the shed, whilst the weaker one to be left is the main unit operating 12 miles away at the QTH of G4AKD) - ignore the bright line in the centre at 0 Hz:

lnb_lo1    lnb_location

The picture on the right, with the test LNB sat on top a piece of test equipment about eighteen inches off the ground, again shows that 3cm is not to be regarded as a line-of-sight band anymore than say VHF/UHF is (the remote beacon being way off to the right about 60 degrees from the lnb bearing, behind the garage and the rest of the village).

A different Cambridge LNB (a G88) gave better performance. It can be heard here, though you can see below that the LO phase noise is worse than the Lidle unit:


The internal LNB voltage regulator IC works well, since another impressive feature is the LO frequency versus supply voltage variation, with only a few hundred Hz change between 10 and 14v being measured.

These waterfall displays were taken after 30 minutes or so settling time for the LNB. It seemed reasonable to repeat the test with the LNB placed outside, at which point, it also seemed interesting to see what, if anything, the domestic satellite LNB instalation was picking up, so the receiver was connected to the spare IF socket of the Sky receiver. This set-up also uses a Cambridge G88, and as expected, was suffering more drift, as can be seen below, but I could still hear not only the spare beacon running in the shed, but also the one 12 miles away at G4AKD. This was without moving the dish away from its Astra alignment:


Here are a brief set of results from all the LNBs tested.

LNB type
       LO error      
Lidle (A)

+ 0.19 MHz
     Cleanest note of all LNBs measured
         both LNBs similar in performance  
                                   Lidle (B)                                 
+ 0.12 MHz
Cambridge G88 (A)

  cambridge lnb

+ 1.24 MHz
Slightly rough note
Cambridge G88 (B)
+ 3.07 MHz
Slightly rough note
Cambridge G88 (C)
+ 2.46 MHz
Good note, not bad stability
Cambridge G96
+ 2.41 MHz
Slightly rough note
Cambridge G170
+ 3.30 MHz
Distinctly rough
Cambridge AE88
+ 0.09 MHz
Slightly rough note
Cambridge (no label)
+ 0.16 MHz
Slightly rough note
- 0.01 MHz
Roughish note   -  (wav)
Thomson (A)


+ 0.32 MHz
Rough note, - (wav)
Thomson (B)
+ 0.21 MHz
Rough note, but good stability
Thomson (C)
+ 0.59 MHz
Rough note, but good stability
Skyware SX1019/2
- 0.19 MHz
very bitty note

618 MHz Converter

Although the initial measurements were made using a signal generator and ring mixer to bring the nominal frequency down from 618 MHz to 18 MHz, this was all a bit cumbersome, so a converter was built and was used for the majority of measurements. This uses a 20 MHz oscillator retrieved from a old PC motherboard. It had a TTL output, so a coupled pair was used to select the third harmonic, and this arrangement worked well, giving 0 dBm output with all other signals down 55 dB. An RF2365 saturated mmic (available on ebay from RFbasic for about 20p each) follows, with another tuned pair selecting the 120 MHz second harmonic. A second saturated RF2365 acts as a x5 multiplier, producing -3 dBm after its 3 stage semi-rigid line filter.

A second pcb has a diode ring mixer fed via another semi-rigid line filter. The LO port is fed from a uPC1678 mmic to bring the LO drive up to 7 dBm.


(No coupling capacitance is shown between lines on the 600 and 618 MHz filters. Coupling relied on close spacing of the trimmers)

It is very tempting  build in an 18 MHz receiver to the converter and apply afc, but that is probably taking things too far. A measurement of drift v temperature would be useful though.

I think that what does usefully come out of these tests is the possibility of a low cost introduction to narrow-band 3cm reception, particularly in a location that has a beacon running (which is now the case here in the Cambridge area). Used with an SDR receiver and waterfall display combination, it would be quite possible to show doppler effects
such as rain-scatter on the band . A local beacon also provides ready frequency calibration, so there is no reason in this situation why such a set up should not be used to keep an eye on at least the beacon part of the band, particularly if the waterfall display is retained.