I was measuring the current with a Hantek CC65 current probe connected to a scope. The device does of course have local decoupling, 2x 100µf electrolytic on the power rail and local 100nf caps on the ICs. The secondary side of the DC/DC has a 680µf on the output for the inductive load.
However, I have just solved the problem, or to be exact, it solved itself. I tested the DP832 with an electronic load and brought it slowly up to 3A to check that all was working fine. At 2.8A the 5V suddenly dropped to 0.48V and I immediately switched off. I left it powered down for 30 minutes and since powering back up, the problem has gone and it also now supplies a steady 3A at 5V. No idea what the problem was, but it's gone. It no longer dips to 2.5V when the inductive load switches in.
One side note: This particular DP832 is very new and had never been used above 500mA since it arrived. Could it have needed some sort of "burn-in" on any of the components? I have an older DP832 which regularly saw 3A, but I don't remember having had similar issues.
McBryce.