Planar PCB Coils As An Alternative To Winding Transformers

Those readers who have experimented with winding their own inductors will know that it’s not an easy task, and when those inductors are handling high voltages it can be especially tricky to maintain adequate insulation between layers of windings. [Open Frime TV] has a video addressing this in a novel way, by creating the windings for a switch-mode power supply transformer using stacked PCB coils instead of wire (Russian language; you’ll have to enable YouTube’s subtitle auto-translation).

The video below the break makes for a handy primer on PCB coil construction, reminding the viewer that the turns need all to lie in the same direction as well as the importance of insulation between windings. There’s a discussion of the properties of a PCB coil in relation to the switching frequency, and once the transformer has been assembled, we see it hooked up to a power supply board for a test. What happens next may be familiar to seasoned transformer-winders; nothing works, and the transformer gets hot. In making the PCB he’s left some copper on each board which amounts to a shorted turn — cutting these allows the transformer to work perfectly.

This technique might not be the solution to all transformer woes, but makes for an interesting option if your work takes you in the direction of winding transformers. If PCB coils take your interest, how about a Tesla coil using them?

35 thoughts on “Planar PCB Coils As An Alternative To Winding Transformers

  1. There is an company in israel (just can’t think of their name of the top of my head) that produces planar transformers to order. The windings are stamped/laser cut out of sheet metal. They make some great high power very low profile transformers.

  2. This is a neat idea, but once you have the idea to ise PCBs, wouldn’t the obvious next step be to… not use PCBs? What I mean is, the fiberglass takes up most of your vertical space and you don’t need its rigidity here, so why not just etch a plain sheet of copper with kapton tape on the back for insulation? Or if you’re isolation milling the boards, like here, you could use a larger end mill or a fly cutter to mill away most of the fiberglass.
    You could probably fit at least four times as many turns in the same space without changing anything else.

    1. It seems lie you have a few options if you want to shave non-copper out of the stack: you could do copper/kapton(either manually or by ordering a flex PCB from one of the prototype PCB outfits); and you could also implement it as a multilayer PCB if having it fabricated. I expect that would start to add up if you specified really high layer counts; but having the coils joined by vias rather than solder joints would be nice.

      It would presumably be the worst of all in terms of cost; but unstiffened multilayer flex PCB would probably make for the cleanest assembly and best copper to not-copper ratio.

  3. For small transformers, the main problem is that the insulation (the PCB laminate) takes up more volume than the conductors. I think he said at one point that he was using 0.4mm PCB laminate, so from the pictures this must be a pretty small transformer. There are 10 layers, which comes to about 5mm of total thickness for the windings, allowing for a little extra thickness from the copper.

    Another issue is the solder joints between layers, as these will push the layers apart.

    Whenever I’ve wound my own power transformers, it has always been a trade-off between core losses and copper losses – you can increase the number of turns/volt to reduce the core magnetization loss, but that takes up more space, forcing the use of smaller wire, which increases copper loss. And obviously vice-versa. Even if you increase the frequency, losses are still a balancing act between core and copper loss. So given a specific core, the main objective is to get as much copper (and waste as little volume on insulation) as possible.

    1. I wonder if a guy could extrude some sort of conductive ink onto a surface into a coil then spray on a thin layer of lacquer or some other non conductive material, then extrude the next layer of coils. Even some sort f screen printing process could work.

      1. Technical viewership is like this, for some reason. entitled people who “can do it better” and ppl who just want to disprove others. and don’t forget the “why don’t you just buy it” folks. sometimes it can get a bit annoying

    1. I once disassembled a switch mode power supply for an x-ray tube. Big beefy 3 phase unit submerged in mineral oil. Secondary coils were thin almost flexible PCBs stacked in layers. This is already a design used in obscure commercial products, saying it “doesn’t work” is just being ignorant.

  4. They have done this for isolated power supply modules (PUPS) ~200W range e.g. 1/4 brick, 1/8 brick. usually used for telecom market where the main multilayer PCB have both the primary and secondary winding. The FR4 dielectrics could be 10mil or thinner. The frequency is in the 200-300kHz range so you won’t need a whole lot of turns.

    There are also configurable transformers from the usual inductor companies that you can wire the windings in parallel/serial to get the ratio you want.

  5. I have been working on a PCB with 2 planar cores – it’s supposed to be an SWR power meter. Prior to Kicad 6, curved traces were a pain.
    Modelling performance before pulling the trigger and ordering the board has also been… Educational. I discovered Sonnet, but (the free version) doesn’t accept the boards produced by Kicad.

  6. Sorry, any serious transformer design can’t rely on this. As a thought experiment it’s interesting only. The ‘winding density’ is so low you’re off the bottom of the BH curve for any transformer core material you’d want to use. When I think of all the transformers I worked on in the 80’s 90’s, including airport runway, (50/60 Hz, w/ series primaries) and switchmode converters from 50Khz to 500Khz, this idea would not be a workable solution.

    1. In general, saying “it can’t be done”, or even “it’s not practical” when presented with examples of successful applications of an idea, is not useful. You might as well be saying that it will never be practical for heavier-than-air machines to fly.

    2. Lol, you haven’t dealt with high density energy systems. As mentioned by others, there are TONS of DC/DC converters that do this. Most are under 300w, but there are some pushing 600w in 1/2 in^3.

      Also this isn’t new, I’ve got modules from over 20 years ago kicking around my parts bins.

    1. Certainly. Of course, you can also just make glass-epoxy dielectric variable capacitors with a stack of interleaved single-sided PCBs. Or if you don’t want to have to deal with making contact to moving plates, you can use double-sided PCBs for the plates, and interleave these with thin sheets of styrene or other sheet plastic, that slides between the PCBs, changing the overall dielectric constant as they rotate in and out.

  7. I think if this was paired with a kapton-like substrate it might be viable for small to medium scale applications but I’m personally having a difficult time seeing the benefits. The tooling required to make something like this viable medium to large scale would be prohibitively expensive and not inherently better than coil winding. It’s a lot easier to draw copper wire than it is to etch plates.

  8. The transformer in the picture is a basic concept – 10 metal layers can easily fit inside 1mm FR4 stack – and high magnetic coupling 98% + transfer can be improved by adjusting the stack with other materials – Also current and voltage sense van be integrated with accurate sensor voltages for system power management so there is a lot of flexibility- The results with high coupling efficiency works to above 2GHz. So not such a low frequency requirement and line width can handle very high current capability in multilayer. It’s been done for some special applications since the early 80’s and a lot of passive manufacture construct the dielectric using Hi-K ceramic – which also helps thermal dissipation. However, another tip, the FR4 can be improved in 3D via structures so the winding does not have to be planar.

    1. Hi Mike! You have a point! If you have time to experiment, go ahead and enjoy. But it’s a 1% solution to a 1/2% problem, for me, in years of jobs, hi to low power, 60Hz to 500Khz it was not a concept to bet a whole contract on. For high voltage stuff there’s nomax, mylar and kapton insulation materials + vacuum epoxy impregnation. With oil bath magnetics we’d use vacuum to pull air out of windings and sudden air pressure at the top to ‘slam’ oil all the way thru!

Leave a Reply

Please be kind and respectful to help make the comments section excellent. (Comment Policy)

This site uses Akismet to reduce spam. Learn how your comment data is processed.