Expansive 2kW SMPS acts like big antenna?

[Faringdon]

Hi,
The attached is a prototype of a 2kW offline PSU (240VAC) . Its 2 interleaved Boost PFCs (each 60kHz) into a single Half Bridge LLC converter at 50kHz.
The LLC FETs have series diode and anti-parallel SiC FET so as to avoid severe reverse recovery in output short cct etc.
Switching nodes are as small as poss, but still end up being quite large, due to the size of the thing. Power switching current loops also larger than would like. Do you think that this will be a problem and cause any malfunction due to antenna effects etc?

Both poured and un-poured views are shown, should you wish.

[tggzzz]

It won't be a problem if you put a large enough fuse across the output.

You owe the Oracle some ideas for a Father's Day gift.

(Only greybeards will understand that reference)

[Faringdon]

Thanks, yes, and if you have an SMPS with chunking big  hardware and heatsinks, then nice super-tight layout like you can  get with low power SMD designs just isnt possible....so do you find yourself compensating by damping the switching transitions and adding shielding copper pours over the switching node copper?.....i mean , if you have buckets of current, then you simply cant reduce the area of the switching node that much.

[CJay]

https://www.magic8.co.uk/q/iNiPxoJogyQ#a

[Faringdon]

Thanks..
....Must admit, in  laying out a 2kW Offline SMPS…due to the stipulation that  all components must be offtheshelf (transformers are custom wound but on offtheshelf gapped cores), its not terribly small.
It means that there is a constant robbing of Peter to pay Paul over…
1….Not having so little thermal reliefing that you cant solder stuff on (…without hot air gun/hot plate etc).
2…Keeping switching node copper small in area but somehow thick enough to pass the current.
3….Trying to keep power switching current  loops low in area, when you’ve got to dodge the tracking round heatsinks and other big chunky components etc.

Is this why high power SMPS is said to be so difficult?...ie, all the noise issues that exist  because  of the above layout difficulties?

[mariush]

I think that the pcb layout could be improved and you could also work on your pcb layout skills...
there's lots of places where things could be improved starting from simple things like not using 90 degree angle traces,
rotating components where suitable to have better traces or copper fills  (ex the big capacitors in top right corner,
do you need to have them positioned like that or you could have rotated them 90 or 180 degrees to have more continuous wide traces

lots of unused space in the bottom left area, maybe you could shift things to the left and move the LLC transformer to have more space between it and those 18-20 diodes (have more copper around them to dissipate some heat - also maybe consider adding some plastic rings or something to keep those diodes lifted off the pcb)
If the llc transformer can be moved a bit to the left it may be a good idea because it makes the isolation area straighter ... if that's what it is.  If that's the case could even make it extra better by making a vertical cut under the transformer for even more separation and a small help with air movement around it.
The diode bridges on the left could also potentially be rotated 90 degrees and you could have the fan above them, moving air in the same direction as the other fans (assume from top to bottom in the pictures)
the fan on the right seems to have air flow blocked by the big capacitors above...

[Faringdon]

Thanks,
Those are great points, also, i would like to minimse the HV switching nodes which are highlighted in the attached....but it will be difficult with all those chunky bits of hardware that must be connected to switching nodes.

[JohnG]

Don't forget that minimizing your switch node area is only one part of the solution to CM EMI.

Think about what else is close to switch node. Some possibilities:

• Heat sink capacitively coupled to switch nodes via FET drain where mounted. Tie it back to the closest low impedance node (ground, ideally), even if you need to use a cap to do it.
• Switch node connected to large magnetic component. Ferrite has fairly good electrical conductivity compared to most things and winding turns couple capactively to core, giving a big block to couple CM to the surroundings. Try to keep first few turns away from core, and also a grounded metal shield on the core can help. Sometimes this is part of the mounting, anyway. You don't need a lot of CM current to fail EMI. NiZn ferrite is a lot better in this regard, but has a lot of other issues.

John

[jonpaul]

Suggest you find and read one of the many texts on ELI compliance design regarding  PCB layout and EMI shields and filters.

both CM and DM mode noise can be minimal by design.

Jon

[Faringdon]

Suggest you find and read one of the many texts on ELI compliance design regarding  PCB layout and EMI shields and filters.

both CM and DM mode noise can be minimal by design.

Thanks and i agree....though i confess, for this particular prototype, i dont care how bad EMC is.....as long as the noise doesnt make it malfunction.

Here is a true story of malfunction due to noise from an SMPS...

At one company, they had had a PCB designed comprising three one hundred Watt Buck converters (48Vin, ~36Vout) on a PCB.
When the Buck converters were turned ON, the lab radio immediately went  OFF, when the Buck converters were turned OFF, the lab radio immediately came back ON….and so on.
The lab radio was plugged into the mains, the Bucks were powered from a 600W OffTheShelf PSU (from a well  known, global  PSU company) also plugged into the mains.
I  changed the  series resistors in the FET gate drives from 1R to 4R7  to solve  this “problem”.

....BTW, tne Meanwell supply, in the above story, had of course, passed EMC.

[pardo-bsso]

(Only greybeards will understand that reference)

It's still red but I got it.
(Can we play here?)

What does your Father like?