Look again. Start with the discrete components in the input stage signal path. Then move onto looking inside the ICs such as the LMH6552 (or similar), and others.
Why should I do this? If there are active discrete components not just dedicated to some auxiliary task, well, then there is a discrete amplifier, but it’s still just an amplifier.
The LMH6552 is just a differential amplifier – albeit a very good one. Neither does the datasheet include any circuit details for its proprietary differential current mode input stage architecture, nor do I get the point why to analyze the internals of an IC, all the more so as it’s not even used in the Rigol DS1000Z.
In a scope like the DS1000Z, we have a total of two amplifiers: the input buffer and the PGA.
Anything else not in the signal path, like all the amplifiers dedicated to the DC offset generation, is purely auxiliary and irrelevant when we’re looking for distortion.
Hint: all oscilloscopes have multiple amplifiers, some in series and some in parallel.
Hint: While you can connect any two-terminal-pair networks in series or parallel in principle, I have never seen any practical application of this with amplifiers in a scope.
So we’re rather talking about cascading amplifiers, when we connect the output of the first one to the input of the next one.
And we use split-path amplifiers if we have to process several frequency bands differently, such as in a scope frontend where we want to have wide bandwidth and high DC accuracy at the same time – a principle introduced by Tektronix in the early 70s of the last century.
So we still have a total of two amplifiers: A split path input buffer and a PGA.
Apart from that, I'm pretty sure even the Rigol DS1000Z will have some clamping diodes at the input - just clamping to the supply rails instead to a specific voltage level that prevents the amplifier from leaving its specified common mode range.
They do have diodes, just like other low-end scopes. But they are nothing whatsoever to do with common mode ranges.
Diodes have nothing to do with low end. Rigol resembles pretty closely what has been published by Tektronix around 1971, who of course have used protection diodes as well - and these scopes were certainly high end back then.
Other than that, not sure why you felt like repeating what is already quoted, i.e. that there are most likely diodes, but only for the sake of input protection. And of course, for a single ended JFET buffer, there is no common mode range. But for any differential amplifier there is.
Anyway as I had a closer look in the meantime, it might be that the DS1000Z only protects the discrete HF path, but leaves the LF path (with the OpAmp) alone. But on the HF path they might have even done the right thing, as the diodes are actually used as limiters here. Only question is, if it always works as intended, given the high manufacturing tolerances of the JFET behind.
This could explain why the distortion effect is worse at lower frequencies, as the OP seems to indicate.
Maybe someone could test this at frequencies >10MHz?Have a look at ... or similar; if you spot what you are thinking of, please point to it.
I had a look at the schematics – and I might post some analysis later.