Continue to Site

Welcome to EDAboard.com

Welcome to our site! EDAboard.com is an international Electronics Discussion Forum focused on EDA software, circuits, schematics, books, theory, papers, asic, pld, 8051, DSP, Network, RF, Analog Design, PCB, Service Manuals... and a whole lot more! To participate you need to register. Registration is free. Click here to register now.

Y2 capacitors downstream of common mode choke in offline SMPS

Status
Not open for further replies.
T

treez

Guest
Hello,

We have seen a report detailing that Y capacitors in an offline SMPS should be placed upstream of a common mode choke. (as in the attached schematic)

However, I always thought that it was better if they were downstream of it. If one only has two Y capacitors, then surely they should best be placed downstream of the Common mode choke, as in page 4 of the following….
Do you agree?

Page 4 shows Y capacitors downstream of common mode choke…
https://www.ti.com/lit/ug/sluu263c/sluu263c.pdf
 

Attachments

  • Y2 capacitors in offline SMPS.pdf
    14.2 KB · Views: 112

It depends on a number of factors - there is no one correct way, if you put then on the LHS then they form an LC filter with the CM choke (noise going from right to left) - most ppl do this as you get a greater filtering effect, but;

If the fets are on an earthed heatsink with minimal separation, then sometimes having the Y caps near to the incoming mains terminals can simply couple some of the noise on the h-sink into the Phase and Neutral, if the mosfets/output diodes are not well coupled to an earthed h/sink then this effect is less.

similarly if there is a lot of magnetic coupling to a metal (steel) earthed case...
 
  • Like
Reactions: treez

    T

    Points: 2
    Helpful Answer Positive Rating
Thanks, also, may i please ask

1)...in some SMPS’s there is no Earth connection, and the Y capacitor is simply placed across the transformer…(but not with a connection to the switching node). Would you say that this results in better or worse filtration of common mode noise from the SMPS?.....ie, when compared to SMPS’s which do have an earth connection and have Y caps from Earth to Live/Neutral.

Also,

2)...in those cases where there is an Earth connection, would it provide even better filtration of common mode noise, to also implement a Y capacitor across the transformer, aswell as Y capacitors to Live/Neutral from Earth?
 

The Y cap across the Tx is an attempt to limit impulses across the Tx from getting onto the HVDC bus ( for mains conducted ) or onto the output lines ( if you are concerned about noise on the output ) choice of cap size and damping is important to avoid ringing at full power, some times it makes things worse. Try and test is the only way with that cap.

For either case mentioned above - testing is required, always best to design in the parts and then leave them out of production if they prove to be not needed.

For power supplies with an earth, a 2 stage CM filter ( i.e. with 2 sets of Y caps on the LHS) always works best, the SRF of the CM chokes should be different and match up to some worst case peaks in the noise, usually one at 200kHz and one at 700-800kHz ...
 
  • Like
Reactions: treez

    T

    Points: 2
    Helpful Answer Positive Rating
Thanks, it looks like it’s definitely the case that passing the common mode aspect of the Conducted EMC test will be easier if the offline SMPS has an earth connection to the PCB, -because then we can connect Y capacitors to Earth and get the advantageous filtering of common mode noise in this way. As discussed, the case of having no earth connection to the SMPS PCB means that you have to have the Y capacitor across the transformer, and as discussed, this can be problematic.
So would you agree that having an earth connection to the SMPS PCB means facilitating an easier time of common mode filtering?
Also, we are noticing in many isolated offline SMPS’S , that there is an insulated earth connection wire connected between primary and secondary…it connects to output ground. Would you agree that this is for the purpose of preventing the SMPS output from floating up to high voltage levels with respect to Earth ground?....and not anything to do with trying to pass EMC?
 

It is good practice to either earth the 0v output - as long as it is labelled so..!! or put a 1W, 1Mohm res from earth to 0v o/p. By the way the descriptor " an insulated earth connection wire connected between primary and secondary " is not the best way to say this, there should never be any wires from pri to sec, better to say the output is connected to earth of the mains lead. That earth wire must be suitably insulated or laid out and restrained so that it cannot contact pri side and output side simultaneously - obviously chafing here might cause a safety issue.

If the o/p is to be earthed in use - then you will have to use an earth on the pri side to filter EMC noise, else you would need very large CM chokes. For low powers where the RFI noise is smaller and the o/p is not to be earthed you can consider not having an earth wire on the mains (must meet double insulation stds) and setting the SRF of the CM ckoke or chokes to the worst offenders of the CM noise - assuming DM noise is low...
 
  • Like
Reactions: treez

    T

    Points: 2
    Helpful Answer Positive Rating
By the way the descriptor " an insulated earth connection wire connected between primary and secondary " is not the best way to say this, there should never be any wires from pri to sec, better to say the output is connected to earth of the mains lead.
Thanks, I see what you mean, at one company we were ripping open Far Eastern offline SMPS’s and looking inside. We often saw an insulated Earth connected wire literally connected from primary to secondary….ie from the SMPS PCB on the primary side to the secondary side of the same PCB….however, the wire was very thickly insulated….looked like 1.5mm plastic insulation sleeving….and the wire was kind of “stood up” proud of the components……and this “standing up” was due to a kind of rigid metal “stork” to which the wire was connected to on both primary and secondary sides… which kind of helped to lift the wire up and above the components. –But in all truth, this may well have been an illegal practice and they just did it any way.
 

It is good practice to either earth the 0v output - as long as it is labelled so..!! or put a 1W, 1Mohm res from earth to 0v o/p
Thanks, I forgot to do this on our 60W SMPS prototype due to rushing. We do have an Earth connection to the PCB…therefore, for the EMC test, I will just have to have a wired connection from the Earth connection on the primary side…..over to the secondary 0V. I will make sure it won’t chafe on the components.
Incidentally, which do you think would give the better EMC scan…….using a direct connection of the secondary 0V to Earth…or connecting secondary 0V to Earth via a 1Meg resistor?
I presume the latter would be better since it reduces the propensity for common mode noise to avoid the Mains input common mode filter(?)

- - - Updated - - -

Hello,
We have just hurriedly designed and built a 60W offline Flyback LED driver for a demo at a show. It’s now working well.
We have booked in for Conducted and Radiated EMC scan as we don’t have the EMC test kit.
(However, in my haste, I forgot to add a 1Meg resistor from Earth to secondary 0V on the PCB).
As such we will have to EMC test it with the secondary output just floating.
In the real application, the secondary output will have a DALI radio module connected to it, and the 0V of this radio module is Earthed. As such, in the real application, the secondary of our 60W offline Flyback will be earthed. This obviously means that the conducted EMC test test results regarding common mode noise would be worse in the real application, as common mode noise can flow to earth via the secondary without having to go through the common mode filter of our 60w offline flyback PCB.

Anyway, supposing we do Conducted EMC testing on our 60W Flyback without the DALI radio module connected, and with its output just just floating and powering the LEDs….then does this give us a valid EMC pass?.......ie, is it valid to test it without the secondary being Earthed? (it will be Earthed in the real application).
 

For power supplies with an earth, a 2 stage CM filter ( i.e. with 2 sets of Y caps on the LHS) always works best, the SRF of the CM chokes should be different and match up to some worst case peaks in the noise, usually one at 200kHz and one at 700-800kHz ...
Thanks, its interesting because its often said that Common Mode effects only manifest themselves above 1MHz…..but as you kindly describe, in fact, they manifest themselves at much lower frequencies than that?
 

Status
Not open for further replies.

Part and Inventory Search

Welcome to EDABoard.com

Sponsor

Back
Top