Hi
Ok, so I'll try this again. Last time I tried it broke the server
With an analog scope, you can keep driving signals in for a *long* ways past the rated bandwidth and still see *something* as a smear on the screen. It depends a lot more on what you have for amps than anything else. The issue on the old analog beasts was always "how fast will they trigger". Once it turned into a smear it was useless.
With a digital scope, you have both the channel bandwidth and the sample rate. These scopes pretty much all share the sampling between the channels. You get 1GS/s on one and 250 MS/s if you run 4 channels. Nyquist says that anything past 1/2 the sample rate is ambiguous. With 4 channels on a 1000Z, that is 125 MHz. With two channels on a 2000A, that is 500 MHz. With only two samples on a waveform, your ability to know what it is ... not so good. A more practical limit is 1/4 the sample rate (4 points per cycle). That gets you to 250 MHz on a 2 channel 2000A or (gulp) 67.5 MHz on a 4 channel 1000Z. With only one channel on each, you could get 500 MHz and 250 MHz.
Next up is the wonder full world of probes. If everything is simple (single pole) bandwidth goes down as you add a probe. Back in the "good old days" they figured out a way around this. Stick another pole in the response and peak it at some frequency. You could do it with the probe or without. If you did it with the probe, things could get a bit odd when the probe was not there. Works great for sine waves. If you have a square wave, it messes with the phase of the harmonics. At some frequencies it distorts the waveform a bit. Taken to far (and some did that) it distorts the waveform quite a bit.
What does this all mean? If a 1000Z had a bandwidth of > 250 MHz it would be pretty useless. This is not a guess, I've used scopes like that. In fact, with a simple roll off (good for square waves) you really don't want to even get that high. Something needs to take care of the alias products. With a more complex filter, you can indeed "do better".
On the 2000A, 300 MHz is pushing a bit hard in the two channel mode. It is arguable that going further is actually a bad thing.
Lots of tradeoffs in this. With digital, uber bandwidth and slow sampling make for a real mess.
Bob