I don’t really understand your question. Those are not even the same channels or channel types.
The scalings dictionary, as mentioned in the docstring, lets you provide one scaling for each channel type.
If you lower the number, you zoom-in.
If you increase the number, you zoom-out.
In the data browser, you can use the +/- key to zoom in/out interactively.
I’m still not entirely sure what you mean by that, it’s probably an issue of nomenclature. When you’re saying “epoch”, you’re not referring to epoched data, yes? And when you say “event”, you’re not referring to a single (instantaneous) trigger pulse?
So I’ll just assume you’re looking at continuous (“raw”) data and want to ensure that whichever time segment of the data you’re looking at, will be scaled by a meaningful factor?
I don’t think MNE currently supports that, unfortunately. All scaling is always applied to the entirety of time. Your best bet is probably to pass scalings='auto', which will ensure that 99.5% of data (across all time points, for each channel separately) are visible. However, this can still mean that there are significantly long time periods where the scaling isn’t helpful…