Hmmm.... switched on my MSO5000 to check something about it for someone (it only has about 20 hours of use), after about ten minutes on it powered itself down and I was left with a repeating click and short LED panel flash every second or two.
Took it apaaaart, there's a near short (0.4 ohms) on the PSU connector's 5V pins to ground on the main board.
PSU seems to supply voltages fine.
After giving it a visual, there was not much to be seen so I made up a cable to my bench PSU to power all the 3 supplies (5V, +7.5V, -7.5V), sure enough the 5V was shorting. By the way, the connector is the same as a normal PC ATX connector cut in half, so I made up a cable.
After a bit of manual probing of the obvious power supplies, I got nowhere, the Schottkys all seemed fine.
Got out the IR cam, and the perp showed up like a Christmas tree. Looks like the ISL8203M 3.3V SMPS is dead. I didn't immediately look at this because #1 eyeball was looking for inductors: this device interestingly enough has integrated inductors, quite a feat, I've not seen this before.
Success!
Turns out the ISL8203M is configured as 2 x 3.3v in parallel two phase mode.
I lifted the device with flux & hot air alone, it wasn't too hard to achieve. I cleaned up the lands with full fat solder, and wicked off the excess, leaving just enough solder for the device's pads to adhere. Added pen flux to the PCB lands & device pads, and placed the device. A bit more hot air and the device was placed. I then cleaned up the edge castellations. It's a three minute job once the scope's disassembled and you have access to the PCB.
I didn't go to the effort of disconnecting the LCD display, but I did unscrew the PCB from the chassis so I could check what was underneath the package (not much). When applying heat, I lifted up the PCB from the chassis with a wedge to allow airflow underneath.
Take care with the LCD, there's nothing holding it into the chassis except light friction once the front panel's off, so there's a risk of wrecking the flat flex cables if care isn't taken.
One thing I noticed after the repair was that there was a very large offset on channel one immediately after this fix (about a couple of volts). I am pretty sure this is due to flux cleaner solvent creeping into the channel one can: after leaving it switched on for an hour or two, the offset gradually disappeared as the solvent evaporated.
Edit: Postmortem measurements showed that there were shorts on the device between PGND, SW1 and VOUT1 on the ISL8203M device I removed.