Sorry about that, the Image site I used has compressed the neutral out.
Good. That looks proper.
As you say this is only semantics, but I would still like to understand it. So If I understood correctly the mistake I'm making, is measuring the voltage without taking into consideration that the body is the ground point and I should connect the black lead to the body work?
The body of the train car is "Circuit Ground"/Zero BY DEFINITION for measurement purposes.
And the body of the train car is Earth/Ground/Common because it MUST be connected ("bonded") to PE. Perhaps using a ground rod driven in to the crust of the planet between the ties.
Someone on this forum has a signature line that says something like "pick a point and call it ground".
Electrical (and electronic) convention is that SOME point in a circuit (or a system) is "ground".
And also by convention, all voltage measurements are made with reference to ground (except in some special cases which should be explicitly identified).
When you are using a DMM, the black/"negative" lead should be connected to circuit ground/common.
Unless there is some special case where you are measuring a "differential" voltage between two non-grounded points.
So in your case, the DC power rail should measure NEGATIVE 48V relative to circuit ground (which is the train car chassis).
Back in the days of mechanical meters, if you wanted to measure a negative voltage, you would connect the red lead to ground as you have been doing.
Because old mechanical meters could not handle positive/negative. But we assume you are using a modern DMM.
I believe the PSU is a floating output if the symbol between the switching device and the rectifier is a 1:1 transformer.
Yes it looks like the output is completely isolated from the input. With that transformer, and also with those optical isolators.
So what it boils down to is can I have mains earth (that comes from the utility) connected to the body work (which is effectively connecting the mains earth to the the PSU's positive output) without tripping any RCD's?
Yes. It appears that there is no path for any fault current, so the RCD should be happy.
I could try it out but the PSU's are quite expensive and I dont want to damage them by trying
You could simply measure between the output terminals of the PSU and input PE to see if there is any resistance.
It should be very high (100s of kilohms to megohms).
But the PSU diagram shows pretty clearly that the output is isolated. So I would have no doubts that it was safe.