High current on cable connecting negative outputs of two power supplies

[PelzigesOhr]

Hi,

I am working on an appliance in which one of the cables decided to desolder itself. Inside the appliance there are three power supplies:

• 24V 600W LCM600Q-T-4-A connected to peltier elements
• 12V 160W RPS-160-12 connected to logic board and pumps
• 5V PSU connected to logic board (probably not involved in my issue, just listed for completeness)

The cable that decided to desolder itself was connecting the negative rails of the 24V PSU to the negative rails of the 12V PSU. See also the blue-highlighted wire in attached minimalistic schematic. Without both PSUs' negative outputs tied together the appliance fails to read the temperature of the peltier elements, so the logic board seems to require this.

I assumed that the wire connecting the PSUs' negative outputs was intended to ensure that the logic can trust that the negative on each PSU is the same. From what I understand this should be the "star" ground point with the "star" being very simple here since it is only one line.

When temporarily re-attaching the cable all is good with the appliance turned on and idling. In idle state the logic and the pumps (the motors in the schematic) on the 12V 160W PSU are running. However, once the peltier elements (powered by the 24V 600W PSU) are switched on, there is a current passing through the wire connecting the two PSUs. Current should be significant since the wire itself is becoming warm to the touch.

Is it expected that a significant amount of current is passing through that wire?

Any idea what could be playing up and how I could fix it?

Regards
PelzigesOhr

[Swake]

Are these 'industrial' type power supplies? If so these might have the GND connected to chassis (and therefor earth cable...).
This might induce an unexpected path for the current to flow. Especially if a component such as a peltier element becomes defective or is mounted to the case without isolation.

[PelzigesOhr]

Both PSU chassis are connected to earth via appliance chassis, but the PSU DC outputs look to be floating.

Here are datasheets (or does the forum prefer attaching just in case these links go dead at one point?):

• RPS-160-12 https://docs.rs-online.com/7dee/0900766b81621f54.pdf
• LCM600Q-T-4-A https://www.micro-semiconductor.com/datasheet/93-LCM600W-4.pdf

I already swapped the LCM600 to see if something was up with that part, but that did not change anything, so I am assuming for now that the LCM600 PSU is not the issue.

The peltiers can be switched separately and the same behaviour can be observed with either peltier. So far I deemed it unlikely for both of them to fail at the same time.

[py-bb]

In your picture the blue wire isn't involved in a circuit. So if any current flows it should be transient and is just "electrons equalising" - like a static shock.

If both appliances' grounds go through say some cabling in the wall you have a "ground loop" - this is an ordinary loop that happens to go through a wire we call earth - earth isn't involved. The circuit is:

Thing A --> earth wire of its power lead --> building wiring earth wire --> earth wire in thing B's power cable -> thing B --> your blue wire.

That should never conduct much current.

Because that's the only circuit that we can infer could be in play this means your diagram is incomplete there must be a path you've missed.


By the way in our "earth loop" example if a fault causes current to go through that loop note that it doesn't go through any RCDs or have anything to do with the earth spike in the ground somewhere, it's an ordinary loop of wire that happens to have yellow/green insulation.

I suspect something you don't intend is completing the circuit but isn't a near-perfect-short (which would cause a fuse say in the input to pop as the thing's power use ramps up to feed the short)

[jwet]

You have indeed created a ground loop.  The blue wire should have no current nominally.  Eliminate the ground connection from the 12v/160W and allow it to track with the 24v supply through the blue wire.  This is at least a good test.  Those two "grounds" going off the triangles are apparently not at the same potential under some condition.