And now we see where the results of all this recent flurry of enquiry about the killer POKE has gone! An excellent video:
https://www.youtube.com/watch?v=7bMJ0NIuWU0
Yes, just as I said, "The Killer Poke is Busted"
Also, don't get too upset about forcing the output of a 74LS TTL IC to ground, remember that a TTL IC is excellent at sinking current, but pretty useless at sourcing it. This is one of the primary reasons why negative logic designs were encouraged (true active low) in that the operating states, where all the action was happening, were generally low on gate outputs with a fan out of 10 or so IC inputs. Also why you will note a lot of enable lines on TTL chips are active low, example chip select lines on ROMs etc.
In addition, passing by all the remarks in the video, the fundamental problem with the "killer Poke" was in fact nothing at all to do with the design of the computer. It was the design of the VDU that was the problem in my view.
When it comes to video snow obviously, if the computer is doing things that alters the video ram data, during scan time you are going to see it as the ram is read out. So most designs, or operating modes, the video ram data updates in the blanking time. The H blanking time is very short, so mostly its done in the V blanking time to avoid image disturbances on the VDU (video monitor).
Back to the monitor:
Because the video monitor in the PET was built into the housing, the designers in my view took liberties. If the pet was designed to drive an existing external monitor, like many computers such as the SOL-20, they would have been forced to create a near industry standard composite video signal.
However, since they custom designed the monitor(VDU), they thought, heck, why bother having an independent vertical and horizontal scan oscillator, we can save on parts and not have that, we will create those signals from the computer's logic circuitry. It was actually a foolish error and I suspect that the designers were unfamiliar with the lesson learnt in the Television industry, where it was realized, even in the early 1930's, that H & V scan oscillators for video monitors (or the part of a TV you can call that) need to be "free running" with their own oscillators, and synchronized to the incoming timing pulses, but don't depend on those incoming pulses (after all it might be noise) to even be present or roughly correct, and they can be grossly abnormal in amplitude and frequency without any detrimental effect on the monitor or VDU part of the television.
So in a computer system, if you don't have those independent scan oscillators (like you have in a standard TV monitor) and rely on H & V "Drive" signals from the computer's logic circuits to coordinate the scan activities (and the fact that the EHT and other auxiliary voltages are derived from the H scan output stage), and for any reason these drive pulses from the computer's logic circuits are corrupted in any way, it will affect the VDU. As noted though a reduction in V drive signal amplitude or frequency cannot harm the monitor in my view, but this is definitely not the case for the Horizontal scan circuits.