This item is concerned with large TV distribution systems and their conversion for DTT. Since c/n inevitably degrades with every successive stage of amplification, it seems a good idea to start off with the best c/n ratio possible. This is especially the case when there is a lot of amplification in the system, for instance 45dB at the head-end and 40dB in repeaters.

I should mention that I'm thinking of a broadband rather than channelised system. To this end, I've experimented with the aerials on three recent conversion jobs. All three were on Crosspool Tx, which has four analogues (21, 24, 27, 31) in Gp A and good old C5 on its ridiculous channel allocation of 67, on half power to boot. The muxes are between 39 and 60. The sites have line of sight to the transmitter, but are not close. Signal strength from 18 ele grouped aerials is +14dBmV on Gp A and +10dBmV on ch67. The muxes are all more or less at -6dBmV. No shortage of signal, then, and no need for an elaborate aerial, you might think. However, there's quite a lot of ghosting on the analogue channels, and there's the ever present threat of co-channel from, I seem to remember, Buxton. In other words, for a LOS location, analogue reception is rather disappointing.

I wondered if something similar might apply to the muxes. After all, now the muxes are on their allocated channels, there's no way of detecting-never mind measuring-what else is on there. Co-channel analogue from miles away? Harmonics from one of the six local taxi base stations? Hash from someone's computer?

I wondered about using a more directional aerial. I tried replacing the TC18E with two stacked TC18Es. They were fixed only 800mm apart, but even so the forward lobe was still very narrow on ch67. I've stacked aerials like this for Crosspool reception before and found an increase in gain of 3dB and sometimes a marked reduction in analogue ghosting.

These latest experiments showed a slight but worthwhile improvement in BER. This is interesting because at first sight with good signal levels the gain increase shouldn't make much, if any difference. So for the price of a TC18, a short bit of mast, a clamp, and a splitter, a worthwhile improvement is possible. By the way, this isn't due simply to the improved gain, because the muxes are equalised before they hit the first amplifier, so the levels at the amp remain the same. At the dreaded 'last outlet', CSI has risen to around 30% and BER after correction is only one order of magnitude better than the 'fail' point. So the improvement is significant. I know it would be better to start from scratch and redesign the whole system around DTT, but this is the real world where cost is king.

Comments anyone?