Power supplies in series, can you ground the middle leg?

Question

Let's assume I have two 15V DC power supplies. These power supplies are allowed to be connected in series (per the user's manual).

I have a device that requires a signal from +15V to -15V (motor driver circuit).

Typically, I would put both my power supplies in series, I would then provide 15V (relative to ground) to the motor driver circuit to use as ground. So from potential earth perspective, the motor driver circuit has 0V as its -15V, 15V as its ground, and 30V as its 15V.

However, what would happen if I were to put both of my power supplies in series and then ground (potential earth) the middle leg from image below (i.e. ground negative terminal on power supply 1/positive terminal on power supply 2). Would I need to worry about power supply 2 shorting to potential earth?

I would guess that since each power supply can be used in series, that means the + and - terminals are isolated from ground. Thus, it would be possible to connect the middle leg to ground. This would require that power supply 2 be able to maintain a truly negative voltage relative to potential earth.

Am I correct in this assumption?

Answer 1

As long as the outputs of both supplies are floating (not connected to Ground, or to each other in some other way), connecting the two supplies in series, and calling the mid-point "Ground" is perfectly normal.

Answer 2

This will cause a problem. The reason you can get away with them in series it is because of galvanic isolation, which explicitly removes the continuity between the input and output via a transformer.

You can, however, treat the tap between the two as ground as far as your motor driver is concerned -- assuming it's not itself referenced to earth elsewhere.

Answer 3

If the power supplies make it explicit that their outputs can be connected in series to stack up the voltage, it's been my experience that this works fine. (The outputs are always isolated from Earth ground in my experience for cases like this.)

Let's see how this might actually work:

^{simulate this circuit – Schematic created using CircuitLab}

On the left side, the load is grounded on one end, with the other end attached to the \$-15\:\text{V}\$ rail. Here \$V_1\$ isn't doing anything. But \$V_2\$ is just working normally. It sees the load and does the usual things you expect from a power supply. So \$V_1\$ is inactive and \$V_2\$ is active (sourcing current into the load.)

On the right side, the load is grounded on one end, with the other end attached to the \$+15\:\text{V}\$ rail. Now \$V_2\$ isn't doing anything. But \$V_1\$ is now just working normally. It also sees the load and does the usual things you expect from a power supply. So \$V_1\$ is active and \$V_2\$ is inactive (sourcing current into the load.)

Nothing particularly troublesome, so far.

So here's the last case I want to cover:

^{simulate this circuit}

In this case, the load isn't attached to ground. So the current through \$R_{LOAD}\$, sourced from \$V_1\$'s (+) terminal, must be somehow sunk by \$V_2\$'s (-) terminal. This then means that there must be a path by which \$V_2\$'s (+) terminal can allow such currents to arrive at \$V_1\$'s (-) terminal.

When they say that the power supplies can be stacked, they are telling you that this path exists without impairing the function of the power supply regulation features.

So everything should work fine, I think.