SMPS noise; Class 1 vs Class 2; grounding the negative SMPS output

Thread Starter

ebeowulf17

Joined Aug 12, 2014
3,307
I'd like to get some general feedback on an EMI issue we're struggling with. We've got an electronics board which uses Cincon CFM1001S SMPS to convert ~200-240VAC to 5VDC for all the low-voltage systems. We've recently discovered that this system is creating unacceptable levels of EMI on the mains supply that feeds it, failing to stay within EN55014 (nearly identical to EN55022) limits. We've tried a variety of EMI filters (including Corcom 5VK1, 3EZ1, 3EQ1, 6VN1, & 6EU1) which still weren't sufficient to reduce noise to acceptable levels.

The power supply supposedly meets the EN55022 noise requirements, but is failing miserably for us. We've done variety of tests to confirm that the noise is coming from the SMPS and I'm quite confident that it's the source. The SMPS supplier tells us that the reason we're having trouble is because we connect the DC negative output to ground (one point only, star grounding.) Here's what they said.
Engineering replied the main reason the customer is seeing high emissions is because the CFM1001S is designed as a Class II (double insulated) product but being used in a Class I application.
Does this make sense to you guys? It seems strange to me.

If it is true, are we just out of luck? Part of our electronics system requires the common frame ground to also be the DC negative as part of a water level sensing scheme. We'd have to do *MAJOR* mechanical and electronics re-designs in order to isolate the DC system. Besides, I'm not sure I'd want to - I've heard that with an isolated DC system, it's possible to have fault modes which result in lethal high voltages piggybacking on ostensibly "safe" low voltage lines (like, instead of a floating pair of wires with 5VDC between them, you've got a pair of wires with 240VAC on them, and a 5VDC offset between them.)

My current thinking is that this particular SMPS is especially noisy and not a good fit for us, and that we just need to try some alternatives. This is not a pleasant thought, since the manufacturer specs don't seem to mean anything and we'd basically be trying supplies at random while paying top dollar to spend time in an EMI test lab.

I'd love to know if tying DC common to frame ground is really a big no-no in terms of noise. If so, how do other manufacturers handle this? I feel like frame ground/DC common bonding is pretty common.
 

Thread Starter

ebeowulf17

Joined Aug 12, 2014
3,307
So, we've done another round of lab tests, and I've made my best attempt at testing stuff in house with an oscilloscope.

We ordered 5 power supplies and every one of them was so much cleaner and quieter than the Cincon model that it was laughable. We did still need a little bit of filtering with most of the alternate power supplies, but it was really minimal (small, cheap filters compared to what we tried when working with the Cincon.)

In the image below, the top supply is the Cincon, the next one is a Delta supply that passes without any additional filters, and the remaining three are variations on a CUI supply that's right on the brink of passing without a filter.

I guess this answers my original question - even if our grounding scheme is less than ideal, it doesn't explain the >20dB difference in EMI performance between these power supplies. Every other power supply works reasonably well in our system.
IMG_3880.PNG
 
Top