Perhaps not the best spot to ask, but I'd like to pick the collective brain resource available on this forum anyway.
I'm fully aware of the fact that asking for this sort of advice is a bit like asking "which book should I read?". Anyway, here goes...
I've been using a DSO-2090 USB-connected oscilloscope from Conrad for some time now. My unit only goes up to 40 MHz (probably less), but I've been fairly happy with it because it does have enough features to perform all kinds of tasks. The main problem is that it's a USB-connected scope which makes me hesitant to put it anywhere near AC mains - even if they have full isolation (which I doubt), I'm just not willing to let my main desktop machine fry and burn because of some connection mistake (and boy, do I make mistakes!).
So reason #1 to look for an alternative (without necessarily getting rid of this one), is mains isolation. An accidental 220V input on the scope, even when in the millivolts range, should not destroy anything - in my book.
Reason #2 is that the USB scope software is Windows only, which means I need to fire up a VM just to use the scope. Many times I just don't bother.
Reason #3 is that my desktop computer is not on the same desk as the electronics lab corner. Running this on the laptop is an option, but not a terribly convenient one due to the VM, find a place for the laptop each time, yada, yada, yada.
Reason #4 is that I'd really like to switch to a mixed-signal oscilloscope, which means you can view both logic signals and analog signals on one screen and on the same time scale. Trouble with that is that my (again, USB-connected, same sort of issues) ZeroPlus logic analyzer is quite good, because it does sample compression and has decoders for I2C, SPI, and RS232. On a serious high-end scope, all that is no big deal, but it can come at a hefty price.
Reason #5 is an awful one, which merely pushes me to something I can't afford by a long stretch: it'd be fantastic to actually see 868 MHz signals. There, I said it - a scope with 1 GHz bandwidth. Now ignore it, because I know that this will always remain a pipe dream.
And reason #6 is that Dave Jones made me fall off my chair when he reviewed the AgilentInfiniivision 3000 X. What an incredible scope, but also what a horrible price tag - especially in the higher ranges and with a couple of options added in.
Agonizing trade-offs are unavoidable, it seems. And the worst bit: it's not just the choice, but the knowledge that any choice will prevent me from ever buying something else again and could haunt me for the rest of my life (for a very interesting story about the paradox of choice, see this TED Talk, BTW).
My budget? Let's just say "fairly high", since JeeLabs is a mainstream work context for me by now. I'll keep it vague, because I'd prefer to talk in terms of value-for-money, and incremental benefits. Maybe a low-end version of a high-end instrument is wise, so it can be upgraded and extended over time? Then again, maybe it's not because technological advances are simply moving too fast? New or second-hand? 2-channel or 4-channel? How much memory? I'm lost. I do think that features such as variable hold-off, mixed-signal, protocol analysis, and adjustable persistence exist for a reason.
What are your experiences?
How high did/would you aim?
How did/would you decide on the trade-offs?
Is "now" a good time?