Thanks Captain Bullshot - once I realised that the supply is grounded (silly to assume it was floating) the problem is obvious. I had the wrong side of the sense resistor connected to ground. So the loop resistance via scope/probe and mains wiring is in parallel with the sense resistor. I guess switching the probe from x1 to x10 changes that impedance somehow, so inconsistent readings.
Anyway, with the connection reversed, I get much more consistent (and higher) readings, regardless of probe setting, which is what bothered me. Typically 10-20V. With x10, I can also see an occasional nasty HF pulse that gets to about 25V - across 0.22R! Glad that I found this. (It wasn't visible at all, before my silly error.)
I disagree however, Capt B, that you cannot measure the current like this because of the ground. Whatever current goes through into the input cct flows in the sense resistor - period. Kirchoff's Law says that. You just have to connect it correctly. (Of course the potential across that resistor gets subtracted from i/p voltage, so things look a little on the rosy side. But that is just the nature of using a sense resistor - it does a bit of current limiting itself. The .05R 1% that I ordered yesterday just arrived, I'll repeat with that - after coffee).
Tautech - thanks for your response. But this has nothing to do with DSO's versus CRO's. A CRO also has the sleeve of the BNC connected to mains earth (unless of course you've been naughty and disconnected it), and you can blow it or the UUT just the same way - as I found out about 30 years back. Also - there are plenty of places where a x1 probe is needed - for instance, looking at output noise on a linear PSU designed for low noise stuff (typically a very small number of mV). These things disappear into noise on x10.
The DSO is cool for non-repetitive signals, or things where you want to see a 1uS pulse that only happens every second, that kind of thing. But for most bench work, I'll probably stick with my trusty Tek 2225, for its vertical sensitivity (5mV with a x10 amp gives 500uV, although it is a bit noisy down there), its immediacy - and the fact that there is no fan.
As for a differential probe - well, I wouldn't buy one, but it would be a cool thing to build. There is a nice LTC app note, also some circuits online. But probably not needed for this.