SSB

No that isn't the best way to do it.

------------------------------------------ ! ------- ------- ! ---!D Q!---------------!D Q! ! ! ! ! ! ! CLK--!> ! CLK--!> ! ! ! ! ! ! ! ! Q/!----X0 ! Q/!------+--X90 ------- -------

This only takes a single 74HC74

I think that's what I meant, but couldn't quite remember the implementation ;-)

...Jim Thompson

When you multiply two signals together, their phases add. To get one sideband to cancel and one to add, you need two 90-degree phase shifts. (If you use one 180-degree phase shift instead, both sidebands cancel.)

So, to make an SSB modulator, you can use 90 degree phase shifters in any two arms of the mixers. There's no magic about doing it at baseband, so one can just use a divide-by-4 Johnson counter on the LO inputs (as others have proposed), drive the baseband inputs in phase, and use a narrowband 90-degree phase shifter on the mixer outputs. The

Cheers,

Phil Hobbs

write

The scheme you described produces the I and Q stream, but this is not what is needed to generate SSB. However, there are DSP techniques to make Hilbert transformers.

Methinks so, too. In the olden days before DSP became somewhat affordable we had to do that for demodulation in Doppler receivers. We used plain old RC lattices. But what a pain that was for the purchasing department, all those caps with 1% or better tolerance.

Adam: Audio phase shifters can be found in publications such as the ARRL Handbook, at least in older editions. Usually only for audio from 300Hz to 3kHz but one can then calculate the additions for more. If it's a one-off build you could measure out capacitors and combine to get to less than 1%. Otherwise finding the caps on the market can take much longer than calculating the values.

That's interesting. IN this case, I assumed he had some specific need for the very wide bandwidth, but now that you mention it, it might be simply that he doesn't realize SSB for communication would never be beyond that.

And of course, you can get by with relatively simple phasing networks for such a narrow bandwidth, so long as you can live with limited opposite sideband suppresion. Once you take out the carrier, whatever you get rid of in the opposite sideband may just be the icing on the cake.

ANd unlike using the phasing method at the receiver, you can control what the phasing network has to deal with when it's in the transmitter. Either your voice won't be much beyond that 300 to 3000Hz range, or you can even filter it before it gets to the phasing network, so the network only has to deal with that limited range. PUt the phasing network at the receiver, and it has to deal with a much wider range of signal, the incoming radio spectrum.

Michael

But there's no such simple sorting scheme. A DSP approximation to a Hilbert transform will give the proper wideband allpass phase shift, but it's rather compute intensive, too much for a PIC. But even a smallish FPGA could do it, and then you may as well do the carrier 90 degree shift and the mixing/summing stuff in there as well.

Somebody must have done the entire SSB chain in VHDL somewhere.

John

Then there is the crystal filter, or mechanical filter methods without so many critical adjustments.

MUCH simpler networks have been used for 300-3000 Hz. The Barker & Williamson model 2Q4 phase shift network, which was much used in early SSB transmitters contained only 4 resistors and 4 capacitors, I believe. No op-amps !

Steve

Crystal filters is what I used. But 30Hz-16kHz can be financially painful. Even 300Hz-3kHz easily set you back $150 in 1980's Dollars. AFAIK only Collins used mechanical filters. I remember these being even more expensive and guys literally took them apart and reassembled them to dodge the expense for a new one. One of them said "If I can't fix it I'd have to mortgage the house".

Don't overlook the option of doing both. The crystal could do the higher frequencies but lets say 30Hz to 300Hz could be done with sin() and cos() functions. This would ease the requirements on the filter a whole bunch.

Or doing as I just did and buying a broken SSB CB rig for 10 GBP. Contains a nice 10.6975MHz SSB, 2.5KHz crystal filter. Not sure on the spec yet as I haven't measured it yet.

But I guess most people would use a DSP these days. Nowadays you can get lots of MIPS for just a few Dollars. When I started out we used AD2105 and things like that but they cost a lot and the compiler was really pricey.

Where is the fun in that? A crystal or mechanical filter doesn't need to have software loaded every time the radio is powered up, and reduces the required current to operate the unit.

There is an error in the schematic - the RC chain to ground on the first opamp in the top pair should be connected to the +ve input of that opamp, not the -ve input. Look at all the other opamp connections.

With that change, it does provide a reasonable 90deg difference and constant amplitude response.

Regards Ian

And it doesn't quit working when its little backup battery dies ;-)

I still have a new 455 KHz CF 2.1 KHz BW Collins mechanical filter in my collection.

USB, LSB, or center? ;-)

I remember reading about SSB in the 1963 Radio Amateur's Handbook (or was that "Amateurs'?") and it seems that just an ordinary balanced modulator and sideband filter was either cheaper or easier than the phase-shift method.

Cheers! Rich

Actually from performance characteristics i think it is intended for AM stereo broadcast applications.