When I first saw the Sym-1's circuit for the one-line oscilloscope text display I laughed about how simple it was, but I thought it was a great, resourceful idea.
I used the X, Y, & Z-axis inputs to do
raster-scan graphics on my scope a couple of years ago. I used the 6522's SR output, using the clock line to feed a pair of cascaded 8-bit counters whose parallel outputs go to a pair of 8-bit D/A converters, the first for the horizontal (X) position and the second for the vertical (Y). The data line of the 6522's SR output went to the Z (brightness) input. The idea was to get a 256x256 display. A third line taken from another 6522 pin reset the counters for the beginning of each frame scan, which is important for initialization and for the case that you don't want to use the full 256 lines.
Feeding the X-axis input from a D/A like this instead of using the oscilloscope's own time base means that if the computer can't keep feeding data at a steady rate to the 6522, you don't mess up your timing, because a hesitation on the computer's part (due to an interrupt to take care of something else, for example) just means the dot also hesitates instead of going ahead of the data. That dot will appear brighter, but all the dots' positions remain unaffected.
The dual op amp I originally used to follow the D/A converters was just a cheap LM358, which proved to be too slow. (I could have done some calculations to find that out beforehand, but just didn't.) So I replaced it with an LT1124, which also has the common dual op-amp pinout so the socket wiring did not need any changes, even though this is a rather high-performance op amp.
I had a few problems which I was able to partly remedy. It turns out the Z-azis input of my A.W. Sperry oscilloscope was capacitively coupled, which is no good for this kind of thing where you definitely won't have or want a 50% duty cycle, or exactly half the pixels being on and the other half being off. I took the user manual which has service data in it, took the cover off the 'scope, and shorted the Z-input coupling capacitor and removed a resistor. That fixed the biggest obstacle.
The next problem was that for displaying a lot of lines (which an analog oscilloscope is not intended for), the brightness had to be turned all the way up, which made the dot on the screen too fat for 256 dots across or down. I decided to forgo a lot of vertical resolution and go down to 64 lines. This means each line gets four times as high a percentage of the illumination time, making it brighter. I also went down to 128 dots across. Now I can't display nearly as much info, but at least it's bright enough, and clearer. I was able to put several lines of readable text on it. I never got far enough to write the software to put graphs on it and mix different sizes of text in with the graphics. Only time and interest are obstacles. It wouldn't be hard.
Next was the flickering. With the computer running at 5MHz, displaying a lot of lines means the number of frames per second is slow enough that the flickering is a little annoying. I need to change a few wires on the breadboard to do interlacing to reduce the perceived flicker, and then I'll have to change the software to feed it appropriately.