I have an Armaco KR-38 analog panel-mount voltmeter, which I salvaged out of an antique home-made (or possibly kit) power supply. Made in Japan.
I decided to re-purpose it for a project I'm working on, when I noticed a problem. It was reading more than 2V low, which I verified using a DMM with probes put directly to the KR-38 terminals. The zero calibration screw was set properly. I have three more KR-38s, so I decided to just swap it out with another one. I tested the other three, and they were reading fine, so I mounted another one to my project's metal front panel. Then it was reading low too. I tested the (now liberated) original meter again, and it was reading fine. Sure enough, when I take the nuts off the meter, apply a voltage, and move the meter in and out of the panel, the needle rises and falls.
The metal panel is somehow interfering with the meter reading, but why? Here's the kicker... my project enclosure is the very one which the KR-38 came out of. I reused that too! Surely, these meters were designed with metal panels in mind. I have another analog volt meter (Chinese cheapie from eBay) mounted to the same panel, and it reads fine. As far as I can tell, the sheet metal panel isn't magnetised. I doesn't pick up tiny bits of metal.
Can anyone think of a reason why this is happening?