Allowed leakage current and current harmonics

[prasimix]

I got a report this morning from the lab regarding current harmonics. Mean Well LRS-150F-48 (i.e. two of them max. load) failed miserably and now I have to find a way how to pass this testing. One possibility is to add PFC on the AUX-PS module, another one is to try Mean Well with built-in PFC. Your suggestions are welcome. This is a summary of report, a whole report can be found in attachment.

[jbb]

Well, are you sure you’re texting against the appropriate requirements? It’s not consumer equipment so there may be exemptions.

It looks like you’re fairly close to a pass. While adding a switch mode PFC would be challenging, a good old fashioned LC line filter might actually do the trick here. Do you have any power factor requirements to meet, or just the harmonic limits?

[prasimix]

Yes, just current harmonics (there was no mention of PF). I already added filtering to pass conducted emission (see here), do you think that passive filtering could improve situation?

[prasimix]

I got additional clarifications: if each power module is measured individually (i.e. DCP or DCM only) with the corresponding Mean well then I pass the tests. In combination, they break the limit.

[prasimix]

I continued to study current harmonics. The thing that is worrying is the big discrepancy in the distribution of harmonics between what I can measure on Rsense and the results from the lab.
However, I managed to make a difference. This time I measured the voltage (cyan) and current (yellow) at the input at the same time. Without any correction for two almost max. loaded channels I have this:



If I add a 10 uF capacitor to the AC input in parallel I get the following correction:



FFT analysis gives the following results and shows a significant attenuation of higher harmonics where several of them break the limit:



The question is whether it makes sense to further complicate this with making a single pole filter, or a high-pass filter, or staying just on the capacitor and believing that I will pass the tests.

[Kleinstein]

Adding quite some capacitance to mains also adds demand on the discharge circuit. The other alternative would be a choke - likely one with a heavy iron core, e.g. like the old PFC chokes found in PC supplies in the 1990s.  AFIAK they are no longer sufficient to easily meet the limit for the PCs. So PCs switched to active PFC as part of the circuit.

[prasimix]

Yes, 10 uF on the AC input is massive. It'll require a massive bleeder. I opened one old PC PSU. Could this be the kind of choke they put for similar purpose? Don't know why it has 4 wires, possibly it is CM choke? Perhaps something smaller could be used in future production instead of adding PFC.

[Kleinstein]

I also don't know why they have 4 wires. It could be an additional use for common mode filtering. I am afraid a PFC choke would need about that size. The old PC supplies where in the 200-300 W range and thus comparable or even lower power than a full equipped BB3.  Chances are they already build the chokes to a price.

No sure if one can still get them, as they are more like technology of the past.  There are a few PFC chokes still available, but not sure they are sufficient. The filter inside the meanwell supplies may also hinder the function somewhat.

[nightfire]

A bit late to the party, but I also throw my 2 Cents (or 0.02 EUR) into the hat: In my experience as a user of equipment (and designer of electric distribution throughout a house etc.) and person that does Appliance testing, Leakage currents are nowadays quite important.
I have in mind that EN60950 (and the successor EN62368) mandates those rules. In the national VDE 0701-0702 (germany) I have to use for checking appliances after they have been bought and are in use, leakage current on the PE is allowed to be at 3.5 mA, touch current 0.5mA.

As a RCD with 30 mA is mandatory in newly built electric installations, the combined leakage of poorly designed devices/appliances can impose some problem, because they are pre-loading the circuit. As a RCD with 30 mA rated differential current is able to act between 15 mA and 30 mA, it can be "blown" by several poorly working  stuff that add up a bit.

Probably a ready-made PSU with PFC would be the easier way to go. Also, if commercial buyers are the target, in due to regulations they are required to ensure that the total power factor towards the energy network provider is within a certain tolerance- and every device that causes unwanted network feedback is no good.

Question: How are the 3.5 mA determined/to measure? "Only" TrueRMS, or with the so-called 50Hz filter according to EN 61010 table A1? Here the body sensitivity is taken into account (DC and high-frequency AC are less harmful to the human body as AC 50Hz) and with the described circuit filter the measured trueRMS value is somewhat normalized.
A PAT tester from the big manufacturers as Gossen Metrawatt, Metrel, Megger, Fluke etc. has this built-in, so the values those testers will show are somewhat lower that if a leakage current is measured by multimeter.

[prasimix]

Yesterday, I got two Mean Well EPP-150-48 that has built-in PFC. With the same declared max. power, they are surprisingly small compared to LRS modules:



It wasn't a problem to install them, and 4 pieces easily fit into the existing space. This means that if they were used for a future 4-quadrant power module (symmetrical power is needed), as many as two 4-quadrant modules could be placed in the chassis, or two more 1-quadrant (DCP or DCM) modules could be placed next to one 4-quadrant module.



As expected the current harmonics are far below the allowable limits:

[jonpaul]

You are trying to fix a compliance  issue with a bandaid

Suggest to nix the chinese Meanwell and get a good quality PSU with builtin PFC and leakage spec.



Jon

[Jay_Diddy_B]

Yesterday, I got two Mean Well EPP-150-48 that has built-in PFC. With the same declared max. power, they are surprisingly small compared to LRS modules:

It wasn't a problem to install them, and 4 pieces easily fit into the existing space. This means that if they were used for a future 4-quadrant power module (symmetrical power is needed), as many as two 4-quadrant modules could be placed in the chassis, or two more 1-quadrant (DCP or DCM) modules could be placed next to one 4-quadrant module.

Have you looked at the conducted emissions of the EPP modules with PFC?
I don't know if they will be better or worse. I know that they will be different than the modules that you were using.

Regards,
Jay_Diddy_B

[prasimix]

Yes I checked, and with the filter I added for the LRS-150F-48 it looks maybe a bit better.

[Kleinstein]

The EPP 150-48 supplies look like they have no PE at the input and thus have the class y caps to both L and N. This can be a problem with the touch current if multiply supplies are combined.

[prasimix]

How I can test or measure that with the current setup?

[Jay_Diddy_B]

Hi,
There is a PE connection, it is doubling as a mounting hole:











(I am not sure I like the idea of relying on a mounting hole for PE)

Jay_Diddy_B

[Kleinstein]

Ground just at the mounting hole must be a thing of cost cutting. Anyway it looks like PW would be mainly used for RF filtering, not as protection for electrical safety.

One can directly measure the leakage current from the isolated output to ground. I am not sure about the limits with respect to the frequencies. There would be some 50Hz part and likely something in the 100 kHz range. The higher frequency part can be additionally reduced from an common mode choke at the regulating modules. So ideally one would measure the leakage current with a complete setup, not just the SMPS module.

[prasimix]

It should be better now:

[TimNJ]

The "touch current" should be measured from user accessible parts (system output terminals) to earth. Use the setup and procedure defined in IEC60990:2016.

You insert the below measurement device between the output terminal and earth.

Depends what standard you are going for, but for IEC62368, you declare the "energy class" of your equipment. I'm not sure entirely how you classify, but you should go for ES1 if possible. So, 500uA RMS for touch current. This is the approximate current that might flow through a human body if touched.

For earth leakage current ("protective conductor current"), the measurement is performed in line with the input earth conductor. This is the current the power supply dumps back into the earth conductor, usually as a result of Y-cap leakage. I believe this should be 3.5mA max.

[TimNJ]

By the way, above may be somewhat confusing. You insert the pair of terminals with 'V1' between them in line with the current path, and make a measurement at 'V2'. The 'V1' terminals are equivalent to points A and B on the second image.

[prasimix]

Yesterday I passed the test for current harmonics but with the new Mean Well EPP-150-48. My attempt to do this with the LRS-150F-48 by adding a large capacitor failed because the harmonics 5, 7, 9 and 11 are still above the limit.

[Kean]

Well that is some progress at least.  Congrats!