Okay, so I decided to build a lab power supply from a kit. The kit was based on a 1988 design from Silicon Chip magazine (Australia). Schematics attached.


The build went off (mostly) without any problems, and it functions fine, except for the fact that the 0V (COM, ground, whatever) is at +22.5V with respect to mains earth. The +V and -V outputs are referenced around this voltage.
I discovered this when I went to try to measure the output ripple using an oscilloscope (plugged into the same mains power socket); when I connected the probe ground to the 0V terminal of the PSU, it sparked and then blew a fuse.

Obviously this is undesirable, as I'd like to be able to connect my scope (and other things like my computer) to circuits powered by this supply! How can I fix this? I tried connecting the 0V terminal to the earth terminal, but this just cause the fuse to blow immediately on power-up.
The 0V terminal is connected to the secondary winding's center tap (labelled "15V").

However, it should be noted that the transformer provided wasn't a straightforward center-tap, but a secondary having multiple taps labelled from 0V thru 30V. This project used the 0V, 15V and 30V taps. But that shouldn't make much difference, should it?


I'm a bit stumped. Help?

 

Have you tied the 0 volt tap of the transformer to the 0 volt (common) output of the power supply? If so, you need to change it so that the 15V tap of the transformer is tied to the power supply common. The 15 volt tap would actually be the Center Tap of the transformer secondary winding.

 

Have you tied the 0 volt tap of the transformer to the 0 volt (common) output of the power supply? If so, you need to change it so that the 15V tap of the transformer is tied to the power supply common. The 15 volt tap would actually be the Center Tap of the transformer secondary winding.

Yes. The common output is tied to the "15V" tap of the secondary. "0V" and "30V" taps are fed into the bridge rectifier.

Also, I see how referring to the Common as "0V" could get confused with the "0V" tap on the secondary. I'll update the OP.
Edit: I don't think it's possible to edit the original post (can't find a button). But from now on 0V will refer to the "0V" labelled tap on the transformer, and "common" will refer to the ground/zero volt output of the supply itself.

 

Disconnect the transformer and check the resistance of the windings against the transformer case.
If your transformer case is connected to a winding then this becomes the leg you must bond, OR
you isolate the transformer from chassis and ground and then you can bond the 15 volt leg as ground.

 

Ok thanks, I'll try that. I think it's unlikely though, given the transformer case is earthed but none of the legs showed any continuity with earth.

 

I suspect that one of the LM317 regulators is not isolated from the chassis.

 

I suspect that one of the LM317 regulators is not isolated from the chassis.

@chrisdabrera
As @Lestraveled said, something is not isolated from the chassis.

If using the chassis as heat sink, make sure you have a mica or silicone pad between the heat sink and the transistor or voltage regulator. Also, make sure the screw is isolated with an appropriate washer/isolation ring. The heat-sink tab on these parts is connected to one of the pins.

 

The kit came with TO-220 sized mica pads and nylon bushes to slip over the bolts. I'll check continuity all the same though.

 

What is the resistance between the mains earth and the +/- outputs and 0V, with it unplugged.?

 

I suspect that one of the LM317 regulators is not isolated from the chassis.

If this is the source of the voltage, there must be a second fault in the way the secondary is wired as neutral. I see (at least) two mis-wires in this scenario. Otherwise, the fuse would pop without the center tap being connected to the scope.

Basic principle: The whole power supply is isolated from the scope or it has a mis-wire.
It's time to examine the transformer connections to see how earth ground arrived on the secondary of the isolation transformer.

 

I've just done continuity testing and it looks like the LM337 was shorted to the chassis despite the mica gasket behind it and nylon bushing on the fastening bolt.
Will pop down to the shop tomorrow and buy some polycarbonate bolts (and more fuses!).

What is the resistance between the mains earth and the +/- outputs and 0V, with it unplugged.?

The resistance was too high to measure, but before I removed the metal fastener from the LM337, V+, COM and V- all showed a "diode forward voltage" of between 500 and 1200mV to mains earth (diode tester and continuity tester are combined on my multimeter). Afterwards, this effect disappeared.

I'll post my results tomorrow.

 

It's time to examine the transformer connections to see how earth ground arrived on the secondary of the isolation transformer.

The secondary center tap is currently floating with respect to earth ground, but directly connected to COM. I went to probe the V+ terminal with my scope, and when I plugged the probe ground alligator clip onto the supply's COM output, it shorted, because apparently probe ground is connected directly to mains earth.

The same thing happened when I connected COM directly to mains earth before turning the supply on.

I would like the COM output (and therefore secondary center tap) to be at the same potential as mains earth, otherwise the power supply's usefulness is severely diminished.

However it's now evident there was a short between a regulator and ground. Hopefully fixing that will allow me to tie the COM and mains earth together. But I won't know until I buy more fuses.