Power supply grounding issue

Im am working with the LFMWLT200-1003 switching power supply from eospower and getting some grounding issues. I have the pin J4 connected to earth ground, but when power is connected, the casing around the supply reads 70V. It seems the shields are not correctly connected to ground or I am connecting ground incorrectly.

In addition, there is no continuity or direct path between the housing/shield and the earth ground pin.

Any input to help trouble shooting would be much appreciated.


Sounds like you may have the J1-1 and J1-2 connections mixed up.
Just to be sure, the J1-1 is the AC Neutral line, which should be your electrical white wire.
The J1-2 is your AC line wire, which should be your electrical black wire.
I don’t know of any other reason why the chassis would have a live AC voltage on it.
You may have to consult with an Electrical Engineer if there are further troubles.
Good luck with your project.

The voltage Im reading is 70V dc from the housing to the ground pin.

That does seem puzzling. I have no idea why the outside chassis would have a +70VDC voltage on it. Maybe one of the engineers on the TechForum may have some ideas.

As an open-frame supply it’s not a touch-safe appliance and so not necessarily expected that the heat sinks would be bonded to AC earth potential, though I’d consider it a prudent design choice to do so.

As an isolated supply there shouldn’t be any significant DC potential between either the output GND or V1 terminals, and input earth terminal (J4) though it’s possible one might measure something like that due to some sort of rectification effect in conjunction with a high-impedance voltmeter and the AC leakage currents that do flow across the isolation barrier. It’d also be possible for external equipment to impose such a voltage between the two.

The serious issue is whether or not there is current flow, AC or DC, between the output and input of the supply. (heat sinks/shields don’t count as output terminals) I’d suggest checking for both AC and DC voltage between both output terminals and J4/earth, and then proceeding to measure current flow between the two. If you find any voltage present, you might want to use a resistor to protect one’s meter fuse in case something’s seriously awry.

Under no circumstances should you be seeing more than 1 mA of current flow (AC or DC) between either output terminal and earth GND without using any added series resistance. Depending on the spec to which the supply is built, as little as 10uA of such current flow might be unacceptable. I’d have to look into the spec a bit closer to see which 60601 limit applies. Regardless, if you see more than 1mA of current flow in that fashion, it’d be a very serious defect and I’d be consider anything upwards of 100uA cause for concern.

Again, we’re talking about measuring between input and output terminals here. The heat sinks/shields don’t count, though I wouldn’t blame a person at all for looking at different options if those heatsinks do prove to be electrically hot relative to input earth.

Please make those measurements and report what you find.

200 mV between the output terminals and earth ground, and 0 current.

That’s along the lines of what I’d hope to see.

Again, being an open-frame supply it’s an IP00 product, and one doesn’t expect exposed parts not to be electrically live to some degree or another. As a “medical” supply however, the input-output isolation (including under fault conditions) is subject to much more rigorous scrutiny.

If one looks at a photo of things, it appears that (on the output side, at least)they’re not using isolated-tab semiconductors. If the same is true of the input side, that’s probably why that heat sink isn’t connected to earth ground–a failure of the mounting insulation (i.e. a single fault) would pretty much short the AC line into the earth ground lead, which is a huge no-no for medical supplies. It’s possible that it’s faulted and that 70V DC you’re reading is an average value at one of the supplies switching nodes, or it might simply be capacitive coupling across a thin little insulation pad that’s causing it.

It’s not necessarily a defect either way, but it might be a good indication of an EMI problem preparing to bite you on the sitter. Might want to get out the o-scope and see whether that’s actually DC or AC wearing a mask to figure that out. Preferably using a probe rated for a kV+ if you like your 'scope.