See the comment of Egon Otto below...

I have now created a dataset that anyone can play a recorded 32QAM signal on a suitable signal generator... It can be found here: github.com/RudisElectronicsLab/IQ-for-ARB . Enjoy!

Interesting observation, now reading about it and indeed, Wikipedia mentions that "NTSC and PAL analog color television systems, where the I- and Q-signals carry the components of chroma (colour) information. The QAM carrier phase is recovered from a special colorburst transmitted at the beginning of each scan line." Never realised that. I do know that for later digital TV systems, they proposed an interesting proposal in which SD images would use something like a hybrid QAM4/QAM16 system. The QAM4 would be sufficient to decode HD images, while the QAM16 constellation points (requiring a more advanced received and a better S/N ratio) would contain the HD information. Cool idea, but I don't know whether it was ever commercially implemented.

@@rudiselectronicslab861 Yes, And PAL switched the imaginary axis 180 degrees on alternate raster lines which cancelled out phase errors so that the hue didn't change. Just the saturation. PAL used a glass 63.9 us acoustic delay line to provide the "two raster line" average. The color carrier had a frequency that was just a bit higher than the maximum signal BW of the luminance signal. The PAL colour carrier was precisely 4.43361875 MHz. NTSC was commercial from 1954 and PAL was introduced in 1962. And of course all of the first color TV's used vacuum tubes. Damn clever I think.

Thanks! In principle, any scope that has an XY mode (almost all do, also the most affordable ones) should allow you to see the vector diagram. If you don’t have an IQ signal generator, you might want to take a look at this thread where I posted some IQ files you can upload in a two channel ARB signal generator: github.com/RudisElectronicsLab/IQ-for-ARB To get to see a constellation diagram, you’d need a scope with some for of blanking (sometimes called Z channel). This is less common on (current) scopes. Actually, it wars a bit more common for CRT scopes (my old HAMEG 203-5 has it, I may one time try it for IQ constellations!). But perhaps you can think of other creative solutions (I have one on my mind… haven’t tried it though). Apart from blanking, you need to play around with the timing to make the scope display the dots exactly on the time they should be stable. There may be various ways to to so, in my video I show how I do so using (another) signal generator. Good luck!

Thanks! The breakout board is a standard DB25 breakout board which you can get at many shops, connected to a standard DB25 extension cable (ensure that all 25 lines are wired!).

Yes, that may be an option. In my video REL#1, I used a function generator with something like 5-step sequences on each channel and played with the frequency ratio (e.g.; 2:3) to get usable timing and an interesting set of transitions. But such approaches stay far from real IQ signals.

With my new setup, I might be able to record some real IQ signals and record them into a scope, save them as *CSV, allowing them to be uploaded into a signal generator. Will still take some attention to get the record length exactly right, the sync between the two channels, etc...

I have now created a dataset that anyone can play a recorded 32QAM signal on a suitable signal generator... It can be found here: github.com/RudisElectronicsLab/IQ-for-ARB