The latest fad in SMPS's....Planar transformers?

[Faringdon]

Hi,
Planar transformers look good for when high power with low profile is needed, and even then only when f(sw) is high enough to make the low turn counts possible.
Planars suffer  "edge effect" w.r.t.  core gapping with eg flybacks.
For high currents, the only way to do planar transformers is literally to have the turns interleaved Pri/sec/pri/sec...etc etc
As a result, you end up with loads of unwanted interwinding capacitance......this  can give really high di/dt at fet turn on...not to mention hideous common mode EMC problems.

The mountings for high power planars also require custom  tooling set-ups, which just arent needed for custom wound "normal" transformers, eg ETD, PQ, etc etc.

Planars look pretty un-appealing, would you agree? I think they  really need an application which can handle the high interwinding capacitance.....preferably a resonant design aswell, because hard switched in combination with the high interwinding capacitance is going to give a nightmare common mode EMC problem.

I would say only use a planar txformer if you absolutely have to, wouldn't  you agree?

Also, for a high power planar design,  the planar windings cant be implemented in a normal PCB layout program, unless you are able to write eg python or C scripts for them.
There is also a need for custom mech tooling for lead frames when windings are high current.
Good luck to anyone trying to get quotes on planar transformers.
And good luck to anyone trying to get lead times for manufacture of a few pieces (or even large qtys) of planar transformers.
Then if you need to get it slightly modified, good luck on the time and money needed to achieve that.
All this isnt a great problem with "normal" conventional transformers.

I reckon planar transformers and hard-switched SMPS just dont go together (due to high interwinding capacitance and the resulting high di/dt noise, and the  common mode EMC problem that goes with it.)
Vicorpower modules use planars...but they are all resonant converters, and desperately need to be v v low profile.

The last sentence of this says it all...
https://www.pulseelectronics.com/wp-content/uploads/2021/01/Pulse-Power-BU-Planar-vs-Conventional-Transformer.pdf

Another point with planars, is that they can only really support low turn counts.....and so if you need a particular turns ratio with eg a full bridge, so that you can maximise the duty cycle, then you are not likely to be able to achieve this with a planar. With a planar  you simply cant make a small fractional increase in turns ratio like you can with a "normal" transformer.

And as the following says.....
Planar magnetics take up a great deal more circuit-board footprint than traditional transformers.
https://passive-components.eu/whats-the-difference-between-conventional-and-planar-switching-power-transformers/
Also..
Planar devices require a printed-circuit layout and tooling for the magnetic core materials.
and...
Were it not for the demand for higher frequencies, it’s somewhat doubtful that planar transformers would be considered as an alternative to traditional wound-wire magnetics.

The following deals with the "edge effect" of planar transformers...
https://www.ti.com/download/trng/docs/seminar/Topic4LD.pdf
...this is one of the ills from which the Planar transformer suffers....and can require custom cores to be tooled up for to  mitigate it....it is mitigateable, but you very often walk into jobs where they havent done the planar transformer correctly and are suffering it. The desire to shoehorn a planar into a design so as to augment the CV, is often so great, that one finds a badly implemented planar...often an offtheshelf planar, which is totally unsuitable due to the intracacies of planar.

So planars are disadvantageous with DCM flybacks...
As stated in page 23 of the following...
https://www.ti.com/download/trng/docs/seminar/Topic4LD.pdf

Planars are said to have the advantage of easly facilitating multiple interleave winding, giving extremely low levels of leakage inductance...the problem with this, is that a certain amount of leakage inductance in eg a Full or half bridge is actually beneficial, as it acts as a turn on snubber. In fact, quite large leakage inductance does no harm in Bridge converters, since the bridge current is AC, and balanced, and so the current  in the leakage inductor does not staircase upwards....it just shunts backward and forwards, harmlessly.
Then there's the active clamp forward converter, where a decent amount of leakage inductance is actually helpful, as it will facilitate ZVS in high load.
This is a big downpoint for planar transformers.......they actually need to be multiple interleave wound...which then results in the really low levels of leakage inductance......which really is not wanted at all.
(perhpas with the exception of flyback converters...but even CCM flybacks benefit from a bit of leakage L  for the turn-on-snubber effect.)

[Gyro]

I don't think I would call it a 'latest fad', they've been around for years.

[fourtytwo42]

Clickbait as usual 

[jonpaul]

Easily available from TDK, Siemens, Philips 35 yrs at laset.

Many papers and app notes on design and PCB layout.

Vicor among others use them


FARRINGTON/TREEZES: PLEASE SEARCH AND DO YOUR RESEARCHJ BEFORE YOU TROLL US!!!

WE HAVE SO LITTLE TIME!!!

Many thanks for your kind consideration


Jon

[Faringdon]

Thanks, yes i have read huge numbers of planar literature. All of it is lauding the planar as the new messiah.
When you go into a job....you then find that you are working on a design that a designer has done where he/she has selected a planar txformer when it is not at all suitable.
They have simply selected it for their CV.
It is this area, where literature really doesnt exist.
Seek literature on "disadvantages of planar transformers" (of which there are a great many) and you find almost nothing.

It is realised that some people on this forum are genuinely concerned about the transfer of proprietary SMPS knowledge to places outside of the "western world", and therefore repeatedly seek to shut down any discussions such as this one...but please may i assure you, that..
1.....The "non western world" is miles ahead of the western world in all areas of SMPS design anyway. (so this isnt giving anything away)
2....The mainstay of missiles and military hardware etc ( the secret bits) is not in the power supply.

(AYK, point 1 above applies because the "western world" has outsourced virtually its entire power supply design and build industry to the Pacific Rim.)

So please feel free to discuss here-with, you are quite safe to do so, and you will not jeopardise anything.
(there are already 149 views and the post is only some 10 hours old)
__________________
Another point is that you often find yourself working on a SMPS with a planar transformer, but it has been implemented totally incorrectly.......for example, there are ways to do it which would end up with ridiculously high proximity losses.....and suffering edge effects........and often people dont get the full data on the transformer, since the design consultancy withholds this info......this is less than ideal when you need to work on it. The designer may also have forgotten to multiple interleave wind where it was necessary to do so, etc etc

[jonpaul]

Planar is used were the device thickenss must be minimal.

The cores are cstly and PCB layout needs rather tricky multilayer to get full benefoit of a PCB coil.

In high vol mfg this eliminats the copper wire wind and term.

For MHz converters, the flat PCB coils may have low skin effect, so litz is not needed

j

[Marco]

AFAICS the TI complaint about edge effects aren't really relevant for designs using modern design rules. I think KoRba88 is using that too in his microinverter.

[Faringdon]

Thanks,
The following is one of the ways to reduce CM noise from planar transformers
https://people.ece.ubc.ca/alisaket/Transformer_Common_Mode_Noise_Elimination.html

..however, it cant always be fully done, since there is not always a convenient other winding, with exactly the same dv/dt,  that can be paired with a winding, in order to make the overall dv/dt equal to zero.

But yes, you should always try and pair windings in this way......and you should insist on the characteristics spec of any planar, so that you can note to see if such interwinding capacitance reduction measures have been taken.

Its always necessary to recieve the full spec of any planar that you  are working on...since if the pri:sec is 4:5.....then you cant just put 4 primary turns then 5 secondary turns...it would cause too much AC losses....they have to be interleaved S/P/S/P/S/P/S/P/S/.
And it will be dawning on us that power supply engineers cannot make out a full manufacture spec for a planar transformer. Instead they are limited to simply giving the Np/Ns spec and voltage and isolation specs etc, to a consultancy...the consultancy  then uses their highly expensive software to come up with a design for you. As you can imagine..startups with little money, can forget planar transformers.
Many places actually do a SMPS with a conventional transformer, then send it to a planar company and ask for the "planar version".......and tell them not to increase the size profile, but to increase the high temperature withstand.

Regarding isolated Low voltage DCDC SMPS demo boards for sale, there is not one single one which has a Planar transformer...this says it all about planars. All their so-called massive advantages, and not one single Eval Board!....all Eval boards on the market use conventional transformers.

[dmills]

I seem to remember much the same whining about toroidal transformers shortly before they ate the line frequency EI lamination market.

In the right design planar transformers are just fine, I have used them in an isolated supply on a 16 layer sonar processing board, couple of slots for the core, and a mess of tracking (Which I largely scripted), worked fine, was thin, and since I needed 16 layers anyway (Big FPGA and loads of wide DDR3).

Would I go there for a flyback or hard switched design? Probably not, but there are more things out there then just that crap, and in something like a PSFB they do just fine (Used them there as well, Coilcraft will do you custom parts up to a few hundred VA or so if you order a hundred at a time).

[DavidAlfa]

Planar transformers pack some serious power density in a small size.
For example:
https://www.chargerlab.com/huawei-40w-supercharge-scp-charger-hw-100400c00-teardown-review/

That the charger my previous phone had (Not me doing the teardown lol), I was curious to see how it made 40W in such small size.
Nowadays they're getting even better!

[Faringdon]

Thanks, to be honest i dont think planars can be done in a  PCB layotu program...you need an adapted version of solidworks, due to the lead frames etc.
I also dont think engieners can do a planar design...it has to be done in the proper software which is only held by planar consultancies.....you have to do the "dv/dt pairing", the Edge problem management, the  full calculation of strays, the full interleaving, the exact calculation of RAC, etc etc.

I'd say its probably best to do a conventional design, then if youve got the need.....send that to a planar consultancy and get them to make an "equivalent",  (but obviosuly smaller and with higher  temperature capability). Probably tell them to give you a low price quote, otherwise youll just use your conventional transformer instead...otherwise, i can well imagaine the planar consultancy would be charging the earth. They likely have very big glass front windows for the unwary.

[dmills]

Depends on the performance you need, and you absolutely can do a planar as part of a PCB with NO lead frame as such, just a couple of slots to get the core halves thru, been there, done that.

Is a 2.5 or 3D field solver useful? Sure but unless you are going all out it is not necessary, a combination of a 2D FEM solver (Which is available open source), a half remembered electromagnetism course at uni, and some creativity will get it done, near enough.

I am not even sure I need your much vaunted conslutants even if going for ultimate power density, better to be talking to my vendors application engineers, most any transformer manufacturer in this space will tell you what they can and cannot achieve. The transformer manufacturers are generally quite good about helping you to specify this sort of thing.

[Faringdon]

Thanks,
And reference leakage inductance.......AWAK, Planar transformers allow us to get vanishingly small leakage inductance.....right!.....so what!.......try running the attached two tran forward with moderate L(leak) and with very low leakage inductance in the main transformer. (its an LTspice sim)......its horrendous when leakage inductance is low...massive current spikes......so AWAK, we actually dont want low leakage inductance anyway! (in hard switched SMPS).... In hard switched converters, the leakage inductance levels that we get with conventional wound transformers are actually beneficial.
Another BIG downer for planars. Planars give you high interwinding cap too....which is very unwanted....i think it would be tricky to do one with low IW Cap...there is a way to do it, but not quick to do.

Depends on the performance you need, and you absolutely can do a planar as part of a PCB with NO lead frame as such, just a couple of slots to get the core halves thru, been there, done that.

Thanks, may i please ask what power level that was?...ive done 4 layer PCB planars myself for very low powers, but when its 100's or 1000's of Watts i believe it gets more critical. Thanks  Ill look  into the field solver side of life.

[dmills]

I was running maybe 60-100W or thereabouts, NOT hard switched.

[Faringdon]

NOT hard switched.

Thanks, thought so.
We are waiting for the planar fraternity to come forward and declare that the two of the main salient features of planar txformers (low leakage L and high interwinding C)....are not AT ALL useful to hard switched SMPS's.
Even flybacks benefit from a bit of L(LEAK)....certainly  the L(leak) level you get with conventional txformer design is not bad for flybacks generally.....if you want higher power you would use 2TF, bridge anyway.

[jonpaul]

Far/Treeze:

There is no planar "fraternity" /conspirance/community.

Real  power electronics engineers and consultants are open to whatever tools/topologies and magnetics are best for the job.

eg  flat SMPS  heatsinked to a panel eg Vicor .

We have used  flat normal EE/EER/EI +bobbin

j

[dmills]

Yep, planar are just another tool in the designers kit, use them where appropriate, don't where not, understanding the tradeoffs, not trusting the marketing drool, and making those calls is what we get paid for.

Like all these things having them available increases the range of shit you can do, but it doesn't mean you HAVE to use them when something else is a better fit for the job.

I like having EI/Toroidal/Planar and everything else available, options are good, same way you want Bipolar/Mosfet/SiC/GaN/IGBT available in semiconductors, lets you optimise for a solution.

There is no Planar conspiracy any more then there is a GaN or SiC conspiracy, they are just options in your toolkit. 

[mag_therm]

I was designing "Planar" transformers 30+ years ago. But they were 50 ~ 200 kVA, 8000 ~ 15000 Amp.
That term was not used, the secondaries were "fins" and the primaries were "pancakes"
For automobile assembly welding, ratings and outputs were standardised worldwide by RWMA .
(Resistance Welding Manufacturer's Association)

[T3sl4co1l]

Looking at a 12V 80A project right now, did one proto with all litz windings: performed very well but it's practically unmanufacturable, soldering piles of winding pairs to a common bus.  Still need some refinement with the planar version but it's looking promising.  Will run hotter but still within power budget, and most importantly, will be compact.  Interestingly, have seen some commercial offerings that appear to have changed from planar (PCB based) windings, to punched sheet.  Much less freedom on what you can "wind" with sheet, but maybe that's still good enough.

Have done a couple other small ones before, turned out fine.  Losses tend to be high from the low copper density (low winding factor), but package is very compact so that can be worthwhile.

Have also seen some that probably shouldn't have been.  One, I think 1/8 brick 24V to 12V 10A isolated converter, something like that -- looked to be active clamp flyback (or was it forward?), ridiculous idle dissipation (like 5 or 10W).  No idea if that's from circulating currents (should've used burst mode at light load?) or eddy currents (planar tends to be bad at this, it's easy to "wind" wrong) or what, but it sucked.  Presumably a more conventional design would've been able to deal with eddy currents at least, if not the other possibilities.  Probably wouldn't fit in the envelope though (they do make wire bobbins for planar cores, but surface mounting one on top of the board is a bit taller than clamping the core around the board itself, eh?).

Tim

[T3sl4co1l]

I was designing "Planar" transformers 30+ years ago. But they were 50 ~ 200 kVA, 8000 ~ 15000 Amp.
That term was not used, the secondaries were "fins" and the primaries were "pancakes"
For automobile assembly welding, ratings and outputs were standardised worldwide by RWMA .
(Resistance Welding Manufacturer's Association)

We had some neat ones at a PPoE -- induction heating.  Mostly came from Jackson Transformer, massive bastards, core was stacks of ferrite blocks glued up.  Windings were copper tube/plate brazed together -- water cooled.  Interconnection from inverter modules (IGBTs) was dick-sized litz cable from NEWT.  (Yes, well, it just happens to be the nearest sized piece of anatomy. And given the soft flexible vinyl jacket, about as flexible as one -- well, take that whichever way you wish to take it...  )

I think those transformers were usually by flat layers like that, though I'm not real clear.  It's hard to see inside the things, let alone the potted ones(!).

Celem also claims a line of stacked-plate transformers, or something like that: https://www.celem.com/Matching_Transformers I don't recall we ever quoted them seriously, like as an alternative or replacement for Jackson.  Used a lot of their capacitors though.

Tim

[jonpaul]

most modern designed SMPS are soft not hard switching

Teslaco, instead of litz try foil, have used in radar HV pulse and high current magnetics

Jon

[Faringdon]

Yes, i'm unsucessfully seeking articles depicting a hard switched SMPS of 100W+ that would benefit from planar txformer design..not having any luck.

[ogden]

Interconnection from inverter modules (IGBTs) was dick-sized litz cable from NEWT.  (Yes, well, it just happens to be the nearest sized piece of anatomy.

Engineering is exact science. Next time please kindly pick more dependable units of measure

[T3sl4co1l]

Interconnection from inverter modules (IGBTs) was dick-sized litz cable from NEWT.  (Yes, well, it just happens to be the nearest sized piece of anatomy.

Engineering is exact science. Next time please kindly pick more dependable units of measure

Come to think of it, even 7/46 litz is bigger than mine... I meant human median earlier, of course!

Tim

[xavier60]

This transformer is from a Castolin Eutectic welder. Both primary and secondary windings are flat, the secondary being 1 turn per side.
The topology is phase shifting full bridge at 100KHz.
I'm guessing that the ring cores on the secondary outer legs are to soften the reverse recovery of the rectifier diodes.
Also, there is no output inductor. It must be making use of the welding leads' inductance for smoothing.