0
\$\begingroup\$

While I understand the problems trying to demodulate a DSB-SC signal using a direct conversion scheme, mostly the phasing issues arising from the fact that the sidebands are symettric and in quadrature, I still don't get why SSB receivers can't just pass band filter one of the sidebands and reinject the carrier, even if it's not frequency and phase synced with the original.

The scheme I see in most of receivers is an heterodyne approach: downconverting both sidebands, then filtering, and finally reinjecting the BFO. But, doesn't this present the same problems as reinjecting a carrier at an offset so as the desired sideband gets downconverted to audio frequency (after bandpass filtered it)?

Why do the other two major methods (phasing and weaver) need to have phasing syncronism while in heterodyne systems this is not required, but instead you have to downconvert twice?

Thanks and regards :)

\$\endgroup\$
  • \$\begingroup\$ Are you in fact asking if it is easier to use a SSB receiver to demodulate DSBSC? \$\endgroup\$ – Andy aka Feb 6 '18 at 10:00
  • \$\begingroup\$ Not quite. I'm asking why heterodyne receivers need to downconvert DSB-SC before filtering the desired sideband instead of filtering the sideband and then multiplying it by the carrier (even if it's not perfectly synced). Also, if a SSB receiver receives a SSB signal, it's still necessary this first downconversion to IF. What are the difficulties in doing it all in direct conversion? \$\endgroup\$ – Angel Feb 6 '18 at 11:02
  • \$\begingroup\$ I understand in VHF bands and above, narrow crystal filters are difficult if not impossible to archieve but, below 10m, I thought it was easier to make a crystal BPF at RF than at IF. Isn't it so? If not, is that the reason for the fist downconversion, or there are phasing issues I'm not considering? \$\endgroup\$ – Angel Feb 6 '18 at 11:06
  • \$\begingroup\$ Well without downconverting, you'd need to make an incredibly selective filter to reject the other sideband AND everything else. It's easier to make a 5kHz BW filter at 500kHz IF than it is at 30MHz or 1 GHz carrier, let alone make it tunable to different carriers. \$\endgroup\$ – Brian Drummond Feb 6 '18 at 11:13
  • \$\begingroup\$ In fact, you can receive SSB voice with a direct conversion (ie, product detector) receiver. Of course you also receive anything in the opposite sideband out to the limit of your audio bandwidth, and anything that sneaks through as distortion before or after. But it works and many a ham or electronics hobbyists has spent many an evening exploring with such a receiver. \$\endgroup\$ – Chris Stratton Feb 7 '18 at 5:20
1
\$\begingroup\$

The whole point of a superhet is that you can do the IF filtering at a FIXED frequency.... A 3KHz wide filter tracking across say the 40M band with sufficient accuracy to pick off the lower sideband while rejecting the upper one 600Hz away is a BIG ask, less so at a fixed 455KHz or so.

SSB really does not care over much about carrier phase (after all the carrier will ultimately end up at DC if the tuning is exact!), but you do at some point need to remove the unwanted sideband and that is usually most easily done (in a classical superhet) by mixing to a frequency that puts the passband of a crystal filter in the right place to pass either the upper or lower sideband.

Modern SDR practise is often to use direct conversion to a quadrature pair at zero IF and then uses either weaver or a hilbert transform to extract the sideband of interest. The thing SDR has going for it here is that working in the digital domain we have both REALLY GOOD mixers (Multiplier blocks in FPGAs are rather linear) and exact gain matching when playing with complex numbers, both of these things are problematic in the analogue world. With analytic signals there is nothing magic about zero frequency, it is trivial to move it around as required.

\$\endgroup\$
  • \$\begingroup\$ So, in an heterodyne receiver, the point is to ease the filtering and the whole post-mix chain. Ok but, at the end, you still have to product detect the audio with another mixer. That's the point that puzzles me the most because, in the product detection, you would still need a BFO with the exact phase and frequency as the TX BFO. If not, the fading and attenuation caused by frequency and phase mismatches are the same as in direct conversion. Am I right? But, in the heterodyne schematics I've seen, I don't see any phasing method at the product detector. \$\endgroup\$ – Angel Feb 7 '18 at 12:39
  • \$\begingroup\$ You don't need phasing method at the detector in a superhet because the IF filter has already stripped off the unwanted sideband. Consider say a 7MHz LSB signal, with an IF filter 3KHz wide centred on 455KHz. My LO is 7,456.5Khz putting the LSB signal extending from 456.5Khz down to 453.5Khz neatly matching the width of the IF filter. Thus at the output of the filter we have 1 sideband not two overlapped as would be the case in direct conversion (Where we would need quadrature and phasing method to remove the unwanted sideband). \$\endgroup\$ – Dan Mills Feb 7 '18 at 12:59

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.