In this particular build the metal shell of the SMPS serves as a ground. The metal shell is connected to the safety ground wire coming from the wall.
I regularly use wooden boxes. Just because the chassis is wood doesn’t mean that there is no ground.
That's interesting. Makes one wonder if that would work for output tubes as well; or any grid circuit. On guitar amp input stages with the typical bypassed cathode resistor.
I did it once for test automation. Tek scope didnt have quite enough programmable DC offset to make a 12V DC coupled current transient response measurement at the channel v/div we wanted. Put a 12V battery in a brass tube with BNCs soldered on each end. Programmed a short circuit to measure the battery voltage prior to running that test, each time the test was run. Worked for nearly a decade...
Lots of circuits run output tubes using fixed bias. That’s all this is - negative voltage on the grid - the only difference is that it’s not adjustable. But if an output tube is driven hard it might start to draw grid current and that would affect battery life. Works well on input tubes, though, and battery should last as long as its shelf life.