Peter G8BBZ, thanks for mentioning the Racal RA98C ISB adaptor. Whilst I was vaguely aware of its existence, I had not correlated it with this thread. It is significant, though, in that it shows that back in the valve era, general-purpose, general coverage HF receivers were adapted for ISB, including broadcast relay purposes. I imagine that the Racal ISB assembly would have been more compact and lower cost than say a Marconi HR21/23/24, although above the HR22 in cost and complexity.
Anyway, a (web) search under “RA98” has brought up some useful information.
The RA98 (D-version) is described at:
http://www.pdp-11.nl/racal/ra98/startpage.html.
And this,
http://www.qth.net/pipermail/premium...q2/002874.html, provides some useful additional commentary, including:
"7. For some years the BBC World Service (and I think CBC) used the Racal RA98C ISB adaptor with the RA17L for long range HF feeder circuits at the distant end for rebroadcast of BBC programmes there on local MW or FM. The normal -3dB filter bandwith of the RA98 is 300Hz to 6,000Hz. Using a 6.5 kHz bandwidth in the receiver reduces the overall bandwidth to about 300 -
3250Hz. In the RA98C version the sideband filter bandwidth is widened to about 90Hz - 6,000Hz to give a much enhanced bass response, necessary for rebroadcasting quality."
The 90 Hz number is covered in the previously mentioned BBC Engineering #84. (I think that the Marconi ISB receivers ran 100 to 6000 Hz.) Anyway, we now have at least a partial answer as to what ISB receivers the BBC used for program relay purposes during the 1960s.
The RA98D version, with 300 to 6000 Hz audio bandwidth, looks as if it were designed for multiplexed voice channels, say 300 to <3300 Hz and >3300 to 600 Hz. Nominal voice channel bandwidth is 300 to 3400 Hz (as I understand it established in the 1930s on the basis that intelligibility in the presence of noise did not further improve by going below 3400 Hz), but the 3400 Hz number seems to have been more honoured in the breach than the observance. (perhaps with good reason, in which case the original (and more obscure) meaning of that phrase would apply).
The RA98C version and its use by the BBC also gets a mention here:
https://uk.groups.yahoo.com/neo/grou...ions/topics/17.
As to how the MA299A DSB demodulator & Audio Combiner worked, I can only speculate.
At minimum I think that it would have provided for simple summation of the two sideband audio signals. This would then make the combined output more-or-less equivalent to DSB synchronous demodulation. In situations where both sidebands are relatively “clean”, this would reduce the frequency response “bending” effects of selective fading. I noticed this with the Liniplex F2. Where usable, DSB was the best option. Where interference to one or other sideband made necessary the use of USB or LSB, the frequency response effects were more noticeable, although several orders of magnitude less distracting than the harmonic distortion caused by rectifying demodulators dealing with selectively fading signals.
That the MA299A was also described as a demodulator is intriguing. Perhaps it took an 18 kHz DSB signal from the RA98C (i.e extracted ahead of the sideband filters), along with an 18 kHz “carrier” signal, and used the latter (after any needed phase adjustment) to synchronously demodulate the former. But then the summation function would appear to be superfluous. So it remains an open question.
Possibly the MA299A also included some AF conditioning circuits. Such might have included a 5 kHz adjacent channel notch filter, and possibly low pass filter(s) restricting the AF response to say 3.5 or 5 kHz, to mitigate interference effects.
It does look as if Racal was capitalizing upon the general success, frequency setting ease and stability of its RA17 series receiver by extending its capability not only into the ISB realm, but also into the broadcast relay subset of that realm. One wonders whether the same was done for any other valve-era HF general-purpose receivers, or whether the RA17 was uniquely suited, stability-wise, to that role. Possibly it could have been done with the crystal-controlled Eddystone 880, but in that case there surely would have been some information about it amongst the wealth of Eddystone data available. The apparently obscure Marconi HR120, also crystal controlled, might have been another candidate.
Presumably the Racal Wadley Loop system was clean enough, in terms of oscillator phase noise to handle broadcast-quality SSB demodulation. Phase noise was an issue with synthesizers at one time, though. The Liniplex F1 was crystal-controlled only, for that reason. The F2 made provision for using an external synthesizer, which was made available (as the OSC-1) once such could be made with low enough phase noise at acceptable cost (bearing in mind that the Liniplex was basically an upper consumer level receiver.) The original OSC-1 only went to 22 MHz (receive frequency); its designer told me at the time (1988, as I remember), extending it to 26 MHz would have been too costly. But within a couple of years or so, costs had come down enough to allow that extension.
Looking into the solid-state era, where ISB versions of general-purpose receivers became perhaps commonplace, I imagine that broadcast-quality ISB sections were a subset, possibly not available in all cases. Certainly the data that I have seen for the Eddystone EC958/12 suggests that its ISB section was communications-oriented, although one may suppose that Eddystone would have built a broadcast relay version had it been wanted by a customer.
Cheers,