Allowed leakage current and current harmonics

[prasimix]

I am in the process of certification to obtain CE (EMC and LVD). Two new requests regarding AC input arose:

1) Leakage current has to be below 3.5 mA (for sinusoidal currents)
2) Current harmonics will be measured (because total power is above 75 W)

Any idea where and how to measure the required params is welcome!
Thank you in advance.

[Alti]

Interesting question.
I do not know the answer but I am guessing the leakage only applies to I appliance class. Can you quote specific regulation/norm that limits the leakage to 3.5mA? Considering that typical residential RCDs are set to trip over 15mA, this 3.5mA sounds quite high..

[Vovk_Z]

Leakage current mostly depends on Y2 capacitors value (for AC/DC).

[prasimix]

I have just 2 x 2.2 nF Y capacitors between AC and PE, and 2 x 10 nF between DC and PE. Therefore the leakage current should be well below 3.5 mA limit.

[TimNJ]

Two types of leakage current:

1. Earth conductor current (mainly driven by Y-caps to earth)
2. Touch current (mainly driven by capacitance across transformer)

The latter is more directly related to user safety since it's the current which flows through an equipment-human-earth loop. The first is also limited per safety standards but usually only becomes a safety issue in particular cases/environments. The "measurement device" is specified in the standard. All it is is a current measuring shunt resistor with an RC filter. Usually 1K shunt resistor (1mV = 1uA) with RC = 10K + 0.015uF (-3dB @ 1KHz). But, the measurement device can vary.

If you want to measure touch current, put this network between an output conductor, say V+, and the input earth connection. Use a true RMS multimeter with decent bandwidth to measure the voltage across the resistor.

If you want to measure earth conductor current, put the network in line with the input earth conductor to your power supply.

Typically a normal condition (NC) and single fault condition (SFC) limit is specified for each. Single fault condition for touch current is either open neutral or open earth. I'm not too familiar with LVD, so I'm not sure. This should be tested at worst case line voltage (240 * 1.1 = 264V)

----

Regarding harmonics measurement...usually this is done with a dedicated piece of equipment (AC power analyzer). May be a way to do it with a current probe and FFT, but I am not sure the accuracy. If the main power supply is >75W rated, it has PFC and most likely will be okay. If you're running a few non-PFC converters side-by-side, may be an issue.

[Ice-Tea]

Interesting question.
I do not know the answer but I am guessing the leakage only applies to I appliance class. Can you quote specific regulation/norm that limits the leakage to 3.5mA? Considering that typical residential RCDs are set to trip over 15mA, this 3.5mA sounds quite high..

I'm not sure to what country/region this applies but it should be noted at least this is not universally true. I'll refrain from generalising but at least here a main 300mA unit is regular, supplemented by a 30mA unit in humid areas.

[prasimix]

Thanks everyone for more inputs and ideas. I measured "Earth conductor current" as @TimNJ noted and it seems that I'm within limits.

The real challenge will be to pass the test for the current harmonics since used Mean Well LRS-150F-48 do not come with PFC, and I don’t have room in the current enclosure for a model that has PFC.
I made a preliminary measurement for half the power of one module and the current has characteristic shape for such a converter without PFC. Rigol's FFT analysis is not very helpful I will have to analyze the data elsewhere, preferably in an application that recognizes/detects harmonics and their values in percentages.

[Jeroen3]

Harmonics are just multiples of 50, which regulations take up to the 50th (2.5khz).
You are looking at 500 Hz/div, so only the 4 left divs are useful.  You will get 3,5,7,9...
Try changing the Hz/div and increase the vertical scale. You also have a linear scale Vrms.

[jbb]

Yeah, that doesn’t look promising re harmonic currents.

If your current probe has OK bandwidth, a scope FFT is a good start. You’ll need to adjust the FFT settings to get reasonable results, then read off using the cursors. Then calculate

It’s probably good to dig into the standard a little for details on harmonic limits. Is it Total Demand Distortion (TDD) or THD?  TDD is a win because it (basically) means you use the full load fundamental current. It’s a huge help at partial load.

[David Hess]

As other have posted, Rigol's FFT is not the best but I can see everything you need there.  Adjusting the sampling parameters will help somewhat; it looks like a lower time/div will help.

[prasimix]

Hm, this doesn't look good at all. Another measurement on Rigol with 50 Hz/div:



FFT performed in EEZ Studio with the same load:



It looks to me that 3rd harmonic alone (150 Hz) is enough to not pass this test.
This is what one can find in the Mean Well report. Interestingly, for current harmonics there is no mention that the test was done in a certified lab.

[Jay_Diddy_B]

Hi,

The harmonics should be measured over 200ms. 10 cycles of 50 Hz or 12 cycles of 60Hz.

I found this reference:

https://www.emcstandards.co.uk/files/61000-3-2_mains_harmonics.pdf

I believe that you need to pass class A.

The class A limits are not a percentage but absolute limits:




Mean Well are self-certifying based on the measurements from their Chroma Power Meter Harmonic Measurement.

Regards,
Jay_Diddy_B

[prasimix]

Thanks a lot. Don't have currently idea how to proceed before visiting the lab once again and that they make measurements. If we failed it seems that I don't have many options then to try to deploy another model that includes PFC, or to build my own PFC that has to be installed before existing Mean well modules (up to three of them).

[prasimix]

Another presentation of the similar results (output power is the same, about half of the total load for a single module or 77 W, Vin=115 Vac via isolation transformer):

[prasimix]

Another measurement, the same setup but with Vin=230 Vac:

[jbb]

Well, I’m not too well versed on the standards side, but there are a couple of things that could be attempted on the power side.
If you’re close to a pass, it might be feasible to add some filtering with a larger differential mode capacitance. Industrial settings may use LC filters on specific harmonics, but that would be bulky and heavy.

Cooking up a PFC rectifier and feeding the modules with DC might work, but could land you in trouble with EMC emissions again.

The most academically interesting solution would be an active power filter: essentially a small mains inverter with special control so that it spits out currents to cancel the lumpy stuff from the Mean Wells. (Typically one would measure either the total line current or the current branching off to the Mean Wells and use frequency selected control loops to force the odd-numbered harmonic currents to zero.). Would be a custom design with microprocessor control.

[TimNJ]

I would not bother trying to roll your own PFC module. While you don’t have to worry about primary-secondary isolation safety, since the PFC deals directly with mains, it will need some level of 3rd party safety testing and validation. In my opinion, much more straightforward to use an off the shelf converter with integrated PFC.

But does your configuration even fail?

Regardless, I’d take a guess that the $25USD pricetag is somewhat related to the lack of PFC. How about Cosel PJA150F-48? You have to be careful with some of these old school type power supplies. Their days may be numbered.

[Jay_Diddy_B]

Another presentation of the similar results (output power is the same, about half of the total load for a single module or 77 W, Vin=115 Vac via isolation transformer):

Hi,

I think that this looks like a pass to me.

at 150Hz you have 524mA this is the third harmonic.

If you are class A the limit for the third harmonic is 2.3A you are 1/4 of the limit.

Because of the symmetry in the waveform all the even harmonics should be zero.

at 250 Hz (N=5) you are 200mA the limit is 1.14A

etc..

I believe that you can justify class A.


This why you have to transition to PFC at around 200 - 300W for class A

Other classes need a PFC at lower power levels

Regards,
Jay_Diddy_B

[prasimix]

Well, I’m not too well versed on the standards side, but there are a couple of things that could be attempted on the power side.
If you’re close to a pass, it might be feasible to add some filtering with a larger differential mode capacitance. Industrial settings may use LC filters on specific harmonics, but that would be bulky and heavy.

Cooking up a PFC rectifier and feeding the modules with DC might work, but could land you in trouble with EMC emissions again.

You're right, especially if we can expect change in both radiated and conducted emissions. I would definitely like to skip another round of the radiated emission testing.

The most academically interesting solution would be an active power filter: essentially a small mains inverter with special control so that it spits out currents to cancel the lumpy stuff from the Mean Wells. (Typically one would measure either the total line current or the current branching off to the Mean Wells and use frequency selected control loops to force the odd-numbered harmonic currents to zero.). Would be a custom design with microprocessor control.

Huh, I'm completely lost here, but even if I decided to build a "trivial" PFC it would surely take me a few extra months, not to mention an academically interesting solution .

[prasimix]

I would not bother trying to roll your own PFC module. While you don’t have to worry about primary-secondary isolation safety, since the PFC deals directly with mains, it will need some level of 3rd party safety testing and validation. In my opinion, much more straightforward to use an off the shelf converter with integrated PFC.

But does your configuration even fail?

Regardless, I’d take a guess that the $25USD pricetag is somewhat related to the lack of PFC. How about Cosel PJA150F-48? You have to be careful with some of these old school type power supplies. Their days may be numbered.

Price tag is more like 25-35 USD, but you have to multiply that by 2 or 3 for multichannel configuration. Cosel PJA150F-48 is way above my budget (my margin on the BB3 is simply miserable). I can see on the Mouser that is about 47 EUR (25+ pcs). I would rather choose Mean Well EPP-200-48 that I can get for about half of the Cosel price but is twice as expensive as the currently used LRS-150F-48!

[prasimix]

Hi,
I think that this looks like a pass to me.
at 150Hz you have 524mA this is the third harmonic.
If you are class A the limit for the third harmonic is 2.3A you are 1/4 of the limit.
Because of the symmetry in the waveform all the even harmonics should be zero.
at 250 Hz (N=5) you are 200mA the limit is 1.14A
etc..
I believe that you can justify class A.

This why you have to transition to PFC at around 200 - 300W for class A
Other classes need a PFC at lower power levels
Regards,
Jay_Diddy_B

Previous measurements was for half the output power of one converter. I made another measurement for full output power for two converters (300 W, Vin=230 Vac) and it seems that I'm still within the budget. Now we need to wait and see how it goes in the lab.

[TimNJ]

I would not bother trying to roll your own PFC module. While you don’t have to worry about primary-secondary isolation safety, since the PFC deals directly with mains, it will need some level of 3rd party safety testing and validation. In my opinion, much more straightforward to use an off the shelf converter with integrated PFC.

But does your configuration even fail?

Regardless, I’d take a guess that the $25USD pricetag is somewhat related to the lack of PFC. How about Cosel PJA150F-48? You have to be careful with some of these old school type power supplies. Their days may be numbered.

Price tag is more like 25-35 USD, but you have to multiply that by 2 or 3 for multichannel configuration. Cosel PJA150F-48 is way above my budget (my margin on the BB3 is simply miserable). I can see on the Mouser that is about 47 EUR (25+ pcs). I would rather choose Mean Well EPP-200-48 that I can get for about half of the Cosel price but is twice as expensive as the currently used LRS-150F-48!

I figured as much about the cost. Cosel not exactly cheap. I've tested Meanwell EPP-200 at work to check its conducted/radiated emissions profile. It's okay. The competitors (XP, Murata, Bel) were generally better, but Meanwell not too bad. (Remember that EPP-200 140-144W power supply with 200W forced air rating.) It is a much newer design so you won't have to worry about obsolescence, although designs like LRS-150F have the cost advantage, so Meanwell may not be so inclined to obsolete it.

Make sure to look at the other harmonics too. I don't think it's as much of a problem for IEC61000-3-2 Class A, but for Class C, sometimes issues at the higher harmonics, like 9, 11, and 13. But the Class A limits are defined  differently than Class C, so you'll probably be okay.

[prasimix]

Make sure to look at the other harmonics too. I don't think it's as much of a problem for IEC61000-3-2 Class A, but for Class C, sometimes issues at the higher harmonics, like 9, 11, and 13. But the Class A limits are defined  differently than Class C, so you'll probably be okay.

I believe that EEZ BB3 belongs to Class A since it does not belongs to other: B is for portable tools; arc welding equipment which is not professional equipment, C is for lighting equipment and D is personal computers and personal computer monitors and television receivers.

[Jay_Diddy_B]

Hi,

A believe that the classes are based on population and usage factors.

The more popular an item is and the more frequently it used the tougher the regulations.

Power tools are intermittent use, so the harmonics that they generate are not really important. so they have higher limits.

TVs have very strict requirements, everybody has one and they are on multiple hours a day.


I noticed that Mean Well tested at 80% max output.  You might be able to do this if you are close to the limit. I am not sure what the IEC61000-3-2 says. It may say 'typical usage' or it may say 'worst case'.

Regards,
Jay_Diddy_B

[prasimix]

It says:

C.15 Test conditions for other equipment
Test conditions for other equipment will be given as required.