Conducted emissions conundrum

[colsramble]

We have an existing commercial product (mains powered with internal SMPS) which we're augmenting with an add-on module that by necessity requires it's own power supply. The existing product has been EMC certified, and the power supply we've chosen to power the add-on module is a Meanwell brick that is itself EMC certified. Unfortunately when we plug them together we're seeing excessive conducted emissions back through the existing product.

We're struggling to find a suitable workaround or address the root cause - the emissions themselves are quite low in frequency so ferrite beads don't help much, and the only real success has been placing an EMI filter inline with both devices.

Changing power supplies and isolating the Earth on both supplies gives some improvement but neither are practical or desirable long term.
 
I'm hoping this is a good place to find some ideas or expertise to help isolate the root cause and/or workaround the problem.

Thanks in advance!

[T3sl4co1l]

Y cap across the new supply.

Tim

[srb1954]

Add an additional common-mode choke on the mains input. These are more effective at reducing the lower frequency conducted emissions than ferrite beads.

[colsramble]

Y cap across the new supply.

Thanks Tim, it's my understanding that both SMPS have Y caps already, but given it's an off-the-shelf device I'm not sure how we'd add one externally? Do you mean external Y-caps between active/neutral and ground on the input of the new supply?

Add an additional common-mode choke on the mains input. These are more effective at reducing the lower frequency conducted emissions than ferrite beads.

Thanks srb1954, yes I think that's pretty much what the inline EMI filter is, common mode choke with CX and CY. A combination of what you're both suggesting perhaps.

We've been hunting around for such a filter that we can just buy and add retrospectively - both devices have IEC connectors for power connection, so something that acts as a dual IEC adapter with inline filter would be perfect but the closest thing we've found so far was over $300 a piece ?!?     

[Psi]

We have an existing commercial product (mains powered with internal SMPS) which we're augmenting with an add-on module that by necessity requires it's own power supply. The existing product has been EMC certified, and the power supply we've chosen to power the add-on module is a Meanwell brick that is itself EMC certified. Unfortunately when we plug them together we're seeing excessive conducted emissions back through the existing product.

Not sure i can offer any help, just some words of understanding with regard to meanwell and EMC

My bedroom is illuminated entirely by hidden LED strip lighting in two large rings around the ceiling.
It is powered by a meanwell brick PSU (one of the 5V 40A open frame ones, mounted in a box).
It totally f's up the alarm clock/FM radio on my bedside table, pretty much jamming the FM station out of existence whenever the light is on.
And before anyone asks, this meanwell psu was bought from mouser, it's definitely genuine.

Note: I'm not 100% sure it's really the meanwell PSUs fault, it might be my MCU switching frequency driving the RGBW addressable LEDs in a nice inductive ring around the room

[T3sl4co1l]

Yup, likely both is a good solution.  Add CMC (say 1-10mH), add Y cap from mains-N (either side of the CMC -- try both ways, see which works better) to secondary side, GND or whatever it is.

Sometimes helps to use a small CMC on the secondary side, alone or in combination with the primary side choke.  How useful this is, really depends on where the noise is, if it's truly common mode between primary and secondary, or if there's some to ground or free space or something like that.  Whether the module is bare PCB or enclosed metal, etc.

Don't forget to re-check the original one in the same ways; it could always be that the new one has exacerbated emissions from the first.

Tim

[TimFox]

In what frequency range are your excessive conducted EMI voltages?  Cheap line filters are effective above maybe 1 MHz, but there are (more expensive) units that are effective at lower frequencies.

[fourfathom]

the power supply we've chosen to power the add-on module is a Meanwell brick that is itself EMC certified.

Have you measured the Meanwell brick by itself (perhaps with a resistive load)?  Just because it claims certification doesn't mean that the one you have actually complies.

[jkostb]

I recently finished a product which had to comply with the EU medical device directive. This means it had to comply with IEC60601. The power supply was a Meanwell power supply (as yoiu probably guessed), because this was the cheapest medical power supply with very good specs. During EMC testing we discovered that the conducted emission was exceeded. During the design we had paid special attention to EMC, board layout and filtering.  All power supplies were heavily filtered with common mode chokes, ferrites etc. We suspected that the root cause must be in Meanwell power supply, so we asked Meanwell to share EMC test report. Meanwell was not willing to co-operate. For this reason we replaced the power supply with a TDK-Lambda power supply and retested the product. This time our product passed the EMC test for IEC60601 with a large margin for conducted emission!  I suspect that Meanwell test their product with a resistive load during EMC test. This is of course not realistic if your product contains switching regulators.  Later I opened a meanwell power supply and compared it to a TDK-Lambda power supply. The TDK lambda power supply was 2x times more expensive but the build quality, filtering was a lot better.

[colsramble]

It certainly seems that the selection of Meanwell is big part of it, so will try some tests by itself - assuming the supply was compliant may be the first mistake. We have got some better results with a similar OEM supply that came with a Mikrotik router, just trying to work out if we can source that ourselves or find something else similar. This was the last thing we thought we'd have problems with - choosing an off-the-shelf supply was supposed to avoid this kind of shenanigans!

[colsramble]

In what frequency range are your excessive conducted EMI voltages?  Cheap line filters are effective above maybe 1 MHz, but there are (more expensive) units that are effective at lower frequencies.

Noise is sub 1Mhz - mostly 300-800Khz which, as you say, is hard to get rid of easily/cheaply.

[prasimix]

Welcome to club . I can confirm that Mean Well or in my case combination of two models has some real issue with conducted emission (more details you can find here). Separately tested with a resistive load both pass tests properly.
If you have a spectrum analyzer on hand you just need LISN (ask @Jay_Diddy_B for details) and you can start with in-house testing.
Buying $300 ready-made filter is still no guarantee that it will helps in you case (I have tried many and none have helped). I'd suggest that you invest that money in different CM chokes, and assortment of X and Y caps and find your combination. If you succeed such a filter will require some extra PCB space but will be many times cheaper than any ready-made filter.

[colsramble]

Welcome to club . I can confirm that Mean Well or in my case combination of two models has some real issue with conducted emission (more details you can find here). Separately tested with a resistive load both pass tests properly.
If you have a spectrum analyzer on hand you just need LISN (ask @Jay_Diddy_B for details) and you can start with in-house testing.
Buying $300 ready-made filter is still no guarantee that it will helps in you case (I have tried many and none have helped). I'd suggest that you invest that money in different CM chokes, and assortment of X and Y caps and find your combination. If you succeed such a filter will require some extra PCB space but will be many times cheaper than any ready-made filter.

That thread makes for an interesting read as our configuration is somewhat similar. Unfortunately I don't have the equipment to test this locally so relying on remote 3rd party who have been trying various choke arrangements. The Meanwell supplies are external in-line adapters and we don't currently have the luxury of modifying or integrating them without a significant change to the overall system design. We've got a few alternate supplies on the way from our supplier in China so crossing our fingers atm!   

[TimNJ]

I recently finished a product which had to comply with the EU medical device directive. This means it had to comply with IEC60601. The power supply was a Meanwell power supply (as yoiu probably guessed), because this was the cheapest medical power supply with very good specs. During EMC testing we discovered that the conducted emission was exceeded. During the design we had paid special attention to EMC, board layout and filtering.  All power supplies were heavily filtered with common mode chokes, ferrites etc. We suspected that the root cause must be in Meanwell power supply, so we asked Meanwell to share EMC test report. Meanwell was not willing to co-operate. For this reason we replaced the power supply with a TDK-Lambda power supply and retested the product. This time our product passed the EMC test for IEC60601 with a large margin for conducted emission!  I suspect that Meanwell test their product with a resistive load during EMC test. This is of course not realistic if your product contains switching regulators.  Later I opened a meanwell power supply and compared it to a TDK-Lambda power supply. The TDK lambda power supply was 2x times more expensive but the build quality, filtering was a lot better.

Resistive load is the standard way power supply manufacturers test emissions. It would be quite unlikely that TDK was using some “other” kind of load. I believe Meanwell does a lot of “self-certification”, so that may have some influence...if you catch my drift. Also check the footnotes in the EMC section. May say something like “Needs 500x500mm metal plane and King Core PNxxxxx ferrites on input to pass Class B emissions.”

[Mr. Scram]

I recently finished a product which had to comply with the EU medical device directive. This means it had to comply with IEC60601. The power supply was a Meanwell power supply (as yoiu probably guessed), because this was the cheapest medical power supply with very good specs. During EMC testing we discovered that the conducted emission was exceeded. During the design we had paid special attention to EMC, board layout and filtering.  All power supplies were heavily filtered with common mode chokes, ferrites etc. We suspected that the root cause must be in Meanwell power supply, so we asked Meanwell to share EMC test report. Meanwell was not willing to co-operate. For this reason we replaced the power supply with a TDK-Lambda power supply and retested the product. This time our product passed the EMC test for IEC60601 with a large margin for conducted emission!  I suspect that Meanwell test their product with a resistive load during EMC test. This is of course not realistic if your product contains switching regulators.  Later I opened a meanwell power supply and compared it to a TDK-Lambda power supply. The TDK lambda power supply was 2x times more expensive but the build quality, filtering was a lot better.

It surprises me Meanwell isn't sharing the tests reports freely, let alone upon request. If they can't pass EMC testing or prove it they're not much better than any other generic unbranded outfit. I really thought better of them as they're reasonably well regarded.

[NiHaoMike]

In my experience, Meanwell does have a reputation for being electrically noisy. In one place I worked at in the past, they kept a specific model of Meanwell PSU as the reference for a noisy supply.

[Psi]

In one place I worked at in the past, they kept a specific model of Meanwell PSU as the reference for a noisy supply.

hehe, doesn't surprise me.

[TimNJ]

I recently finished a product which had to comply with the EU medical device directive. This means it had to comply with IEC60601. The power supply was a Meanwell power supply (as yoiu probably guessed), because this was the cheapest medical power supply with very good specs. During EMC testing we discovered that the conducted emission was exceeded. During the design we had paid special attention to EMC, board layout and filtering.  All power supplies were heavily filtered with common mode chokes, ferrites etc. We suspected that the root cause must be in Meanwell power supply, so we asked Meanwell to share EMC test report. Meanwell was not willing to co-operate. For this reason we replaced the power supply with a TDK-Lambda power supply and retested the product. This time our product passed the EMC test for IEC60601 with a large margin for conducted emission!  I suspect that Meanwell test their product with a resistive load during EMC test. This is of course not realistic if your product contains switching regulators.  Later I opened a meanwell power supply and compared it to a TDK-Lambda power supply. The TDK lambda power supply was 2x times more expensive but the build quality, filtering was a lot better.

It surprises me Meanwell isn't sharing the tests reports freely, let alone upon request. If they can't pass EMC testing or prove it they're not much better than any other generic unbranded outfit. I really thought better of them as they're reasonably well regarded.

I think you'll notice most of the popular (low-ish cost) companies do not publicly disclose their 3rd party reports. They typically just show the certificates, i.e. the cover page of the report. I agree it does come off as rather suspicious.

[jkostb]

The Meanwell power supply which failed conducted emission  in my project was an ACDC adapter GSM60A12. This is a medical power supply. My suspicion with Meanwell power supply started, when I discovered that they were not willing to share their EMC test report. Why on earth does a company not want to share their EMC test report unless they have somehthing to hide? When I contacted TDKLambda and requested for EMC test report I received immediately complete documentation set including EMC test report! From the TDKLambda EMC report I learned that all measurement were performed by an accredited lab. I was really very surprised that our project passed conducted emission test with a large margin without any modification to our electronics, while it failed with the Meanwell power supply. If we did not tried another power supply, we would have spent weeks debugging the electronics, adding filters etc. Be carefull with cheap power supplies!

[Mr. Scram]

I think you'll notice most of the popular (low-ish cost) companies do not publicly disclose their 3rd party reports. They typically just show the certificates, i.e. the cover page of the report. I agree it does come off as rather suspicious.

It should be outright illegal. If you claim a standard you should be able to prove it. No proof, no claim.

[Vovk_Z]

Mean Well power supplies are very popular. Is there some typical improvement for them? For example for LRS series and similar?  Fit some larger X2 caps or CM choke?

[RABeng224]

I’ve been most successful with Ycaps after the common mode choke when dealing with low frequency conducted issues. Can’t go overboard with the value due to regulatory limits on leakage.