I’ve been trying to figure out if there’s a way, using salvaged electronics, to store a series of high-speed analog waveforms long enough to digitize them with a low-speed ADC, in order to make a cheap DSO. To be more specific, I want to digitize a channel or two with signals at and above 20MHz, and thus a 40MHz Nyquist frequency, with perhaps 1000 samples, with at least 8 bits of precision, stored following an analog detection of a trigger signal, at varying sample rates, and then digitize them with an ADC of 6Msps or less, promptly enough to update an oscilloscope display many times per second. So those 1000 samples might be 25 microseconds, or 250 microseconds (with elements up to 2MHz), or 2.5 milliseconds (at which point you’re only covering up to 200kHz).
I considered a coaxial cable delay line, but 25 microseconds is almost 4 kilometers of coaxial cable, which would impose a fatal degradation of signal-to-noise ratio. I considered driving the deflector coils of a TV tube or computer CRT with higher frequencies, but for 20MHz, the inductance involved would require such high voltages that the coils would almost certainly fail (see TV oscilloscope). I considered recording the analogue waveform on a hard disk (see Disk oscilloscope) and then digitizing it during succeeding revolutions, which might work, but I don’t know enough about the linearity and SNR.
So here’s another idea along those lines: use a laser-printer mechanism, electroconductive drum and all. A 10ppm A4-size 600dpi laser printer must go through 210 mm × 297 mm of paper every 6 seconds. That paper works out to almost 35 megapixels, which means that the laser has to be modulated (and successfully recorded on the drum) at 5.8 megapixels per second.
Unfortunately, this is not nearly fast enough, and of course the linearity of the system is terrible.
Maybe the 64-microsecond acoustic delay lines that 1980s PAL electronics used? Those could be hard to find.
Better: see CCD oscilloscope.