w2aew:
In the parlance of the youths, you're really killin' it. Nice job!
BTW, can you use the setup shown to modulate say 16QAM (or higher QAMs)? What would be different then, just what you feed through the I and the Q? More levels than just 'high'/'low', I assume? I may be answering my own questions, but various light bulbs are going off here. If this is so, the magic may be in building the I/Q generator capable of doing all these combinations of multiple levels. Fascinating...
I really don't know the answer to your question, but thinking about what specifications are available and what specifications change as you go up to higher and higher QAM configurations, I think this is just a noise issue. The bandwidth (of the mixers and the spliiter/combiner which) doesn't need to change as you go to higher QAM constellations, provided it is enough anyway. The Mini-Circuits I/Q demodulator he uses (ZFMIQ-10) has a bandwidth of only 2 MHz, which you might think is insufficient for, say, QAM64 or higher. But,
for a constant bit rate the bandwidth required for higher QAM encodings
decreases!
The inputs to the I and Q would need to have more levels (as you state). I think the real figure of merit in keeping a higher QAM constellation making it all the way through to the receiver is a combination of the losses at every stage (e.g. conversion loss for the mixers, insertion loss for the splitter/combiner, insertion loss for any cables) and some sort of noise related figure, and/or the sensitivity of the receiver and its capability to demodulate signals which are very small (just above the noise floor of the receiver).
I remember from The Signal Path videos that Shahriar tested various Tektronix RTSAs by attenuating the constellation all the way down to only a few dB above the noise floor of the RTSA, and it still worked. After the attenuation became so severe that the RTSA was having trouble demodulating the constellation diagram, he would reduce the QAM scheme down a step and he could attenuate and successfully demodulate still further.