Where to best connect Y cap?

[Faringdon]

Hi,

Customer sent us offline flyback which fails EMC.
85-265,24VOUT, 66KHz, 20W.
It has no Y caps, just CM choke.
Its in an earthed metal enclosure.
Served by L/N/E there is an earthing screw restring to which we will connect one of the Y caps.

We have only room for 2 Y caps....we can get one across the transformer.
The other one we can connect to primary HVDC or  to one downstream terminal of the CMC....we cannot connect to the other terminal of the CMC, as no room.
So where would you best connect this Y cap?...from Earth to HVDC, or from Earth to downstream terminal of CM Choke?
If i could connect a y cap to each downstream terminal of the CM Choke, then i would deffo do that, but i cannot as no room......so given that the CM filter will be unbalanced if i connect the 1 Y cap to the CM choke node, then where would you prefer to put the 1 Y cap?

[floobydust]

Why do you expect us to give fixes for your "customer" SMPS?
No Y-cap tells us a monkey did the design, do you agree?

You profit from this, even if it's your design that failed EMC.
No schematic, no pictures - and help with an EMC problem? Sorry, life is not that easy.

You know where to put the cap.

[jonpaul]

We use solder to connect capacitors

(:-:)

j

[abeyer]

You know where to put the cap.

I've only seen some big electrolytics that come in that form factor, never a Y.

[tggzzz]

Hi,

Customer sent us offline flyback which fails EMC.
85-265,24VOUT, 66KHz, 20W.
It has no Y caps, just CM choke.
Its in an earthed metal enclosure.
Served by L/N/E there is an earthing screw restring to which we will connect one of the Y caps.

We have only room for 2 Y caps....we can get one across the transformer.
The other one we can connect to primary HVDC or  to one downstream terminal of the CMC....we cannot connect to the other terminal of the CMC, as no room.
So where would you best connect this Y cap?...from Earth to HVDC, or from Earth to downstream terminal of CM Choke?
If i could connect a y cap to each downstream terminal of the CM Choke, then i would deffo do that, but i cannot as no room......so given that the CM filter will be unbalanced if i connect the 1 Y cap to the CM choke node, then where would you prefer to put the 1 Y cap?

How much are you offering to pay us for this advice? That's only fair since apparently you are being paid to do the work.

Fundamentally there's no way any professional would offer advice on this topic, due to legal libility and reputations reasons. Hence any response must be from amateurs.

When the shit hits the fan (which I believe you have encountered before in your treez guise), will or won't you be legally liable and how will your defence fare if you admit that you relied on responses from amateurs?

Alternatively,

• given your reputation on this forum - as briefly illustrated by the preceding responses - I suggest you post questions like this on other forums
• post this in the correct part of the forum, i.e. "Jobs ... Post your job ad here".

[Faringdon]

Thanks, I know what you mean, though I think as a rule, without further detail being needed, if you have an offline SMPS with transformer isolation, and an earth wire connection, and one common mode choke, and you are only allowed to use 2 Y caps for EMC, then you would never put one from earth to downstream common mode choke node..but always as follows…

A…1 Y cap across transformer pri to sec. (quiet node to quiet node, not switching node)
B…1 Ycap to HVDC or pri gnd.

..this way , you don’t so much unbalance the common mode filter, because the y cap to HVDC or pri gnd alternately connects to line and neutral, so gives a bit more balance to the common mode filter.
Obviously here I am speaking of being constrained to only using 2 y caps...which ayk, is an unwelcome constraint, but it "is what it is". in this case.

[mtwieg]

you would never put one from earth to downstream common mode choke node..but always as follows…

A…1 Y cap across transformer pri to sec. (quiet node to quiet node, not switching node)
B…1 Ycap to HVDC or pri gnd.

Y caps are not usually connected from HVDC to earth (unless the secondary is connected to earth, in which case the Y cap across the isolation barrier effectively becomes a capacitor from HVDC to earth). The most common implementation is to connect the Y capacitors from earth to each line, in between the common mode choke and bridge rectifier.

If the secondary is connected to earth, then the Y cap across the isolation barrier might provide significant reduction in EMI on its own.

[Zero999]

If it doesn't have a Y capacitor, then what else is missing? X capacitors? Adequate separation between the primary and secondary sides?

The correct answer to the question is. The product is defective by design. Return it to the supplier who should either remedy it, or give you a refund.

[jonpaul]

Just another FTSS  troll.

IF FTSS was a real engineer he would know about conducted  CM vs DM noise and the topology/design and construction of conducted EMI mains  filter.

But   FTTS (Farrington/treeze/time/sink) NEVER actually buys/study/reads/learns ANYTHING)

For  others interested  in EMI filters/compliance:

Suggest the classic old texts (NOT videos! ) on EMI compliance and filter design by great engineers like Nave, White, and our old friend Henry OTT.

These and  other books as well as Compliance Design magazine were my references and education for  many EMI designs, filters  and problems since 1980s

HAVE AN ABSOLUTELY FANTASTIC DAY!

Jon

[floobydust]

But would you help if it was for charity? Please help with failed EMC scan for charity system?

This thread cross-posted, zero replies so far: https://www.edaboard.com/threads/where-to-best-out-y-caps-in-offline-flyback.409024

OP did you learn from this? https://www.electro-tech-online.com/threads/our-smps-is-non-isolated-so-we-dont-need-y-capacitors-in-the-common-mode-filter.158739

Is this the SMPS the one with with triple-insulated mag wire, and oddball heatsinking board? Parasitic capacitance determines where the best spot is for the cap.
Power Integrations typically uses the +HV Bus node for Y-cap placement but the transformer winding is particular.

EMC is at the end of the product development cycle, and you're going to need to redo the PCB.
If you want to play in the Big League, you'll need a LISN and spectrum analyzer, rather than throw your design over the fence and spend $$$$ at the EMC lab.
I go down to the EMC lab as well, with soldering iron and a bushel of parts in case there's an issue or experiment I want to try. Great teacher.

[Faringdon]

The most common implementation is to connect the Y capacitors from earth to each line, in between the common mode choke and bridge rectifier.

Thanks, yes, i tend to agree. In this case though, there isnt space on the PCB to be able to connect to both legs of the CMC.
I do have the 1  Y cap across the transformer.
However, there is only room for 1 more y cap, and it has to go to either the one leg of the cmc, or the HVDC bus...there is no room for it to go anywhere else.

So, with this dreaded constraint, i deign to put it to the HVDC bus...at least then it kind of indirectly connects to each leg of the (downstream nodes of the) cmc in turn as the  bridge rect switches....and in that way give a kind of balance to the CM filter...or  "bit" of a CM filter.

[Zero999]

If the capacitor bridges the rectified mains and secondary side, then it must be a Y1 capacitor, or two Y2 capacitors of the same value in series. A single Y2 capacitor can be used to bridge the mains and protective earth.

I still don't get why you've not returned it as faulty. It's not your job to sort it out and isn't fair that manufactures should be allowed to get away with shipping faulty products.

[floobydust]

It's his own design failing EMC, the one that has been "20W" forever and zillions of questions about it on Internet forums.
No company would accept a product, some other's design that failed EMC- and offer to fix it.
Why didn't said company deal with TreezCo in the first place? Why didn't said company go back to the PSU original design house? Uh because it is TreezCo, now in a pickle.

You can also make a few special PSU's with the Y-cap(s) in various positions and take that to the EMC lab, do a quick pre-scan with them to see the difference, what might work.
If the transformer fab/design, snubber etc. is wrong, or that kooky heatsink happened, you'll need more than luck.

Do the added labour costs to make up the bodge offset the millions of dollars saved having no Y-cap?
If you think no Y-cap is necessary, provision for it on the PC board layout and don't populate it if EMC can still be met.

Nobody goes full bore into production quantities before passing EMC and safety, right? Common sense, right?