As far as I know, the SSB HF radiotelephone links developed in the 1930s generally used a pilot carrier. Fully suppressed carrier HF SSB may have been more of a military development post-WWII, although I am not sure about this. But some of the very early LF links were of the fully suppressed carrier type.
At some stage it appears that carrier levels were standardized; 16 dB down for SSB and 26 dB for ISB. Concomitantly receivers sometimes had adjustable gain in the carrier channel to allow matching to the transmission mode. For example, the Marconi HR24 had a switched attenuator; 0 dB for ISB, 10 dB for SSB and 32 dB for AM (DSB). I should guess the 32 dB number for AM derives from the fact that the carrier is 6 dB above peak maximum power, which would put it 32 dB above the ISB pilot carrier. This kind of three-channel ISB receiver – and I think others of its era – had quasi-synchronous demodulation in that reconditioned carrier for driving the demodulators was obtained by filtering, amplifying and limiting the incoming carrier. I imagine though that there would have been some valve-era SSB/ISB receivers that were fully synchronous in that they used locked oscillators or PLLs to regenerate the carrier. As mentioned upthread, the locked oscillator technique was used by Press Wireless in its exalted carrier receiver, and was proposed by Norgaard for his SSB/ISB receiver.
To what extent DSB pilot carrier and DSB suppressed carrier were used at HF were used I do not know. DSB suppressed carrier falls into the realm of “synchronous communications” as proposed by Costas of GE. In this case PLL techniques were used to recover the carrier from the sidebands alone, all the required information being contained therein. For communications traffic at least the carrier was redundant, although it would still have been needed, if only in pilot form, to establish an agc reference level for broadcast relay links. I am not quite sure – or really I can’t work out - what a Costas loop would do if presented with an SSB signal. Apparently it could deal with DSB corrupted by selective fading, in which case it effectively steered the instantaneous phase of the regenerated carrier. If the loop is endeavouring to keep the Q output at zero, then with an SSB input it would seem that it would need to recreate a symmetrical DSB signal in order to do that. But that is not a conclusion that I feel too comfortable with. Another way of looking at is that provides some crumbs of comfort is that whereas DSB is pure AM, SSB is AM modified by PM. So what the loop might be doing to SSB is applying “reverse” PM to return it to pure AM. Insofar as the loop is swinging the phase of the regenerated carrier to keep the Q output at zero, it does not look impossible, just improbable. By the way, an interesting commentary on the development of the Costas Loop may be found here:
http://williamhaywoodjones.blogspot....gineer-by.html.
The Crosby exalted carrier system evidently had wider use than I had at first guessed. For example, it was used by TMC (Technical Materiel Corporation) in its TDRS triple diversity receiver, see:
http://www.virhistory.com/tmc/tmc_pa...mc.ssb_143.pdf. Background information on TMC is provided here:
http://www.virhistory.com/tmc/tmc_pages/, which page shows that there was a connection to Press Wireless.
RCA included a Crosby exalted carrier receiver in its work on multi-channel radio-telegraph systems, as recorded in RCA Review for 1948 December (
http://www.americanradiohistory.com/...w-1948-Dec.pdf). Its performance, in terms of error rate, was comparable to SSB, both much better than conventional AM, which is confirmation that it worked well enough. RCA appeared to see the exalted carrier technique as having interim value until existing AM circuits were converted to SSB. But it also noted that an SSB receiver could also be used to receive AM with the same result of avoiding selective fading distortion. Eventually that seems to have become the preferred approach for point-to-point working, and I imagine that it happened once SSB telegraph and telephone circuits became commonplace, along with SSB receivers.
Cheers,