That 200 ns is a maximum on the minimum pulse width. If you're confused, you're not alone. Since this function (ALE) is purportedly an edge-triggered one, it shouldn't really matter how wide the pulse is.

I mentioned in an earlier post, that the slow rate at which you're operating may present a problem. However, you don't really need to use your timers if you use the EOC strobe as an interrupt rather than a polled input.

What you do while waiting for a conversion to end is entirely up to you, but if you have nothing else to do, you might just want to poll that input in what would otherwise be an infinite loop. It will, after a time, become active. If you don't enable the interrupt, you can just wait for the signal to become active.

I suspect that what the writeup to which you refer is meaning when they say, "source must remain stable" is the analog voltage you're trying to sample. The longer you wait, i.e. the slower you operate the converter, of course, the more opportunity there is for the voltage to change or for noise to contaminate the signal. That's one advantage of operating it at or near the maximum clock rate.

This converter uses "successive-approximation," which is a sequential method involving a reference voltage, a ratiometric DAC, and a comparator. It compares the input with a portion, starting with half, of the reference voltage and makes subsequent comparisons on the basis of the outcome of the previous comparison, each subsequent comparison is steered using the same principle as binary search. In theory, that would require eight comparisons for an eight0bit ADC, so, clearly, there's more than that going on. Monolithic converters were not so common back in the '70's when this device was developed. Converters in general were quite slow, as this one is, but they've clearly done something to speed up the operation as much as possible.

As I said before, I think you'd benefit by operating your converter with the fastest clock that you can. That might mean speeding up your 8751 in order to speed up the system timing. It will, after all, operate as fast as 12 MHz. The converter has a maximum on its clock of 1.28 MHz, so the ALE or, as you initially attempted, the ALE divided by two, will be a fine clock at 12 MHz, and that will be 3x as fast as you're currently going. It won't guarantee that your inputs won't change, but there's been no discussion of the character of your inputs. I suspect you'll be disappointed by the performance of this conversion system since you're wanting to use it for ranging on a moving device.

RE