Help understanding planar transformer structure?

Thread Starter

SiCEngineer

Joined May 22, 2019
299
Hi All,

I am trying to get my head around planar transformers and how they are physically connected in the real world. I was originally under the assumption that one would create a single 4- or 6-layer board (for example) with each layer containing a planar coil which represents a transformer wire/inductor. However, I have been looking through Standex Mayer transformers, see page 10: https://standexelectronics.com/wp-content/uploads/SME-Planar-Magnetics-Design-Guide_web_en.pdf

And it looks like there are individual PCB boards, which are connected together individually through mechanical connectors. What is the benefit of having individual PCB's connected up in this fashion, rather than just having one single multi-layer board which performs the same functions?

Are these going to be multiple single layer boards (4 to 6 of them) all connected to each other via the mechanical fastenings, or alternatively, is each individual board its own multi-layer board, so that in effect we have a collection of multi-layer boards which each represents a transformer winding?

Hopefully by including the Standex transformer reference here some can understand my confusion and help me understand this physical configuration a little better?

TIA!
 

fourtytwo

Joined May 2, 2017
78
The reason for using a stack of two layer boards externally interconnected is foil thickness.
2 layer boards can have 2oz on both sides but multi-layer boards are usually limited to thinner foils on inner layers.
There is also a difference in thermal resistance.
Every planer trafo I have ever encountered is a stack up of 2 layer boards.
 

Thread Starter

SiCEngineer

Joined May 22, 2019
299
The reason for using a stack of two layer boards externally interconnected is foil thickness.
2 layer boards can have 2oz on both sides but multi-layer boards are usually limited to thinner foils on inner layers.
There is also a difference in thermal resistance.
Every planer trafo I have ever encountered is a stack up of 2 layer boards.
Perfect!! Thank you so much!!
Whats the best way to go about doing this on Altium? Design 2-4 individual 2 layer boards and dimension them the same? I see they are connected using PCB to PCB connectors?
Sorry, I am new to planar trafos and also PCB design in general so bit of a head bender!
 

fourtytwo

Joined May 2, 2017
78
I am very sorry I have not designed one either, I stick to big ole ETD's and hunks of copper!
Also these days being frugal I use Kicad...... Altium is one of those big expensive things I have not used for years.
If i was doing it I would do a step and repeat design so suppose you want 9 turns and you can fit 3 on a board then design it so 3 identical boards can be stacked. You also have to consider insulation so you may actually use single sided boards rather than double sided.
I dont know what currents your into but most I have seen just use wire pins soldered into PTH vias for interconnect.
 

shortbus

Joined Sep 30, 2009
8,169
Maybe a dumb question, but is this for one of your high voltage designs? If so how are you going to supply the distance needed between tracks and not make the PCB huge? By distance between tracks I mean to prevent flash over/shorting in the planar.
 

fourtytwo

Joined May 2, 2017
78
That is one of many disadvantages I see with planar, having to use either single sided pcb's or insert insulating washers between layers, thats why I stick to ETD's and traditional wire/litz/foil windings I have never been persuaded of any advantage in planar.
 

Thread Starter

SiCEngineer

Joined May 22, 2019
299
Maybe a dumb question, but is this for one of your high voltage designs? If so how are you going to supply the distance needed between tracks and not make the PCB huge? By distance between tracks I mean to prevent flash over/shorting in the planar.
Yes, it is for my high voltage designs. However the voltages that the transformer will step up to is 750V for the resonant design and 600V for the flyback design. I then plan to generate the high voltage using voltage multipliers (different configurations for each output). I understand what you are saying though, this is my biggest worry as well. At the moment I have just got to a point where I’ve chosen the cores and the required primary turns to avoid saturation of the core, etc. I will calculate trace width and distances in the coming weeks.
Does this change the required separation distance of the tracks, since they do not generate the full HV?
 

Thread Starter

SiCEngineer

Joined May 22, 2019
299
I'm no expert on this, but wouldn't the normal track separations used for a standard PCB apply? I know there are some calculators for tracks out there, for both voltage and amperage of them.
I believe so. I just wonder whether the tracks necessarily have to meet high voltage clearance and creepage rules despite not generating the entire full voltage, because of possible fault conditions.
 

fourtytwo

Joined May 2, 2017
78
Yes, it is for my high voltage designs. However the voltages that the transformer will step up to is 750V for the resonant design and 600V for the flyback design. I then plan to generate the high voltage using voltage multipliers (different configurations for each output). I understand what you are saying though, this is my biggest worry as well. At the moment I have just got to a point where I’ve chosen the cores and the required primary turns to avoid saturation of the core, etc. I will calculate trace width and distances in the coming weeks.
Does this change the required separation distance of the tracks, since they do not generate the full HV?
Track to track within the spiral is controlled by your volts per turn. I assume you would use spirals on both sides of a pcb with a PTH at the center interconnecting them. You would have to watch creepage around the edge of the pcb from the outside turns that of course will have the highest voltage between them. From pcb to pcb you would have to use an insulator of a suitable material & thickness, normal trafo tape would be difficult to cut to shape so maybe ptfe sheet or even FR4 with no copper. All planars I have seen use unusually thin pcb's for the windings.
 

Thread Starter

SiCEngineer

Joined May 22, 2019
299
Track to track within the spiral is controlled by your volts per turn. I assume you would use spirals on both sides of a pcb with a PTH at the center interconnecting them. You would have to watch creepage around the edge of the pcb from the outside turns that of course will have the highest voltage between them. From pcb to pcb you would have to use an insulator of a suitable material & thickness, normal trafo tape would be difficult to cut to shape so maybe ptfe sheet or even FR4 with no copper. All planars I have seen use unusually thin pcb's for the windings.
I believe I will use Kapton material as the insulator for the PCB, which will provide 15kV of insualtion for 3mil of material. I’m not sure whether I am able to put this into Altium or whether I will physically insert it between the transformer printed circuit windings I produce. I suppose doing it the second way we will also have the advantage that the windings are also separated by dielectric of air.
Issue remains of isolating the core. I believe the only way to do this is to pot the entire stack. Yes, the planar transformers I have seen are usually lower voltage so do not see any kind of engineering that we are discussing.
Plated through holes will indeed be used to connect the two layers of each board - but I am still wondering how best to connect the individual boards within the stack together. Standex use LCP material with standard pcb to pcb header connectors which are rated to around 400V. I don’t know whether I would need to isolate those to such a high voltage because I would expect they are kept clear from the high voltage potential, but there may be 600-750V placed across the pins for the secondary outputs... maybe a different solution is required.
 

fourtytwo

Joined May 2, 2017
78
I believe I will use Kapton material as the insulator for the PCB, which will provide 15kV of insualtion for 3mil of material. I’m not sure whether I am able to put this into Altium or whether I will physically insert it between the transformer printed circuit windings I produce. I suppose doing it the second way we will also have the advantage that the windings are also separated by dielectric of air.
I might be wrong but I thought this Altium was just a pcb cad program so you would have to define other materials elsewhere.

Issue remains of isolating the core. I believe the only way to do this is to pot the entire stack.
I may be wrong but I have always treated ceramic ferrites as insulators, but then the way I construct transformers the windings are encased in insulation anyway (bobbin + trafo tape) so the issue doesn't arise

I am still wondering how best to connect the individual boards within the stack together. Standex use LCP material with standard pcb to pcb header connectors which are rated to around 400V.
Surely each pin in a series connected stack is standalone and doesnt require any voltage rating as such if the pcb ensures suitable creepage and clearance, its not a multi-pin connector, just a single conductor ?
 

Thread Starter

SiCEngineer

Joined May 22, 2019
299
I might be wrong but I thought this Altium was just a pcb cad program so you would have to define other materials elsewhere.

I may be wrong but I have always treated ceramic ferrites as insulators, but then the way I construct transformers the windings are encased in insulation anyway (bobbin + trafo tape) so the issue doesn't arise


Surely each pin in a series connected stack is standalone and doesnt require any voltage rating as such if the pcb ensures suitable creepage and clearance, its not a multi-pin connector, just a single conductor ?
I’m unsure if there is any difference between normal and planar transformer wrt core to primary and secondary isolation. Like I mentioned I believe I must isolate the entire transformer to 12kV so must find a way to isolate not only the windings of the PCB but the core itself. In lies the gritty details.
You’re right about the connectors. I must have just been worried about pins and plastic with a dielectric breakdown of day 400V would not be suitable for board to board connections but Standex seem to use these as standard.

on page 10 of the PDF there is also some kind of beige looking holder for the planar transformer - am struggling to understand the purpose of this, since the same page shows that the transformer core is bonded to the PCB. I assume it is connected to ground with heat relief vias, and then possibly to a heat sink or chassis with heat bonding material. Seems strange they did not show this.
 

ronsimpson

Joined Oct 7, 2019
1,121
My planar transformers were low voltage and the core had very large area so the turn count was very low. I tried and failed making many turns work.
1608061452664.png
My High Voltage transformers (kV) used wire because of the 1000s of turns. With HV I am looking for long winding windows in the core.
1608061319401.png
 

Thread Starter

SiCEngineer

Joined May 22, 2019
299
My planar transformers were low voltage and the core had very large area so the turn count was very low. I tried and failed making many turns work.
View attachment 225000
My High Voltage transformers (kV) used wire because of the 1000s of turns. With HV I am looking for long winding windows in the core.
View attachment 224999
Thank you as ever Ron for the good high voltage insight. As mentioned I have dimensioned the cores to have only 4 primary turns for the primary planar coils. My secondaries then need to be about 26 turns, and for the flyback 2x13 turns centre tapped because it feeds a bipolar Cockcroft Walton multiplier.

I do not believe that this is a lot of primary or secondary turns for the high voltage transformer. I believe your 1000 turns transformers may be for lower frequency multi-kV flyback transformers with a single secondary winding.

Do you believe this is achievable, since the transformers do not generate themselves the full HV but instead feed HV multipliers such as quadruplers / multiple CW Multiplier chains?

the flyback transformer feeds a 5x bipolar Cockcroft Walton multiplier and the resonant transformer feeds a 4x voltage multiplier. The resonant has a single secondary 600W, and the flyback produces 60W in an identical centre tapped winding structure, so technically needs two identical secondaries.
 

shortbus

Joined Sep 30, 2009
8,169
Again just thinking out loud here. If this design is intended for a product, not just your own use, Wouldn't it need to meet what ever commonly used standards to be sold? With having never seen this high voltage planar being used, is there a standard? If no standard will you need to apply for an exception and pay for testing? Not trying to be nosy but just putting things out for consideration.

As mentioned I have dimensioned the cores to have only 4 primary turns for the primary planar coils. My secondaries then need to be about 26 turns, and for the flyback 2x13 turns centre tapped because it feeds a bipolar Cockcroft Walton multiplier.
I thought Cockroft waltons mulitpliers were pretty amerage hungry when used to multiply voltage. Meaning you get far less amps out than you put in. The number of turns you described seems to be low amperage to start off with?
 

Thread Starter

SiCEngineer

Joined May 22, 2019
299
Again just thinking out loud here. If this design is intended for a product, not just your own use, Wouldn't it need to meet what ever commonly used standards to be sold? With having never seen this high voltage planar being used, is there a standard? If no standard will you need to apply for an exception and pay for testing? Not trying to be nosy but just putting things out for consideration.



I thought Cockroft waltons mulitpliers were pretty amerage hungry when used to multiply voltage. Meaning you get far less amps out than you put in. The number of turns you described seems to be low amperage to start off with?
it is for a phd project but I do agree I will need to look for standards or possibly have the company I work for help with the extensive testing of the finished product.

In regards to the CW multiplier this is for the 60W output which only has 10mA output current, so I believe no Issue. With the resonant transformer it is a different configuration which does not hog the current in such a way.

The high current traces are the primary planar coils. I will need to think more about the ability of current delivery for the secondary coils. I think 13 turns in series (one centre tapped) may be able to deliver enough current rather than a 26 turn winding.
 

ronsimpson

Joined Oct 7, 2019
1,121
This transformer we called "car tire" has a single winding. Has a low self resonant frequency.
1608071308510.png
This type of transformer has multiple secondarys stacked up, each has its own diode. We made each secondary slightly different to help reduce the ringing. Resonant frequency is much higher. The AC losses can be very low.
1608071418107.png
 

shortbus

Joined Sep 30, 2009
8,169
it is for a phd project
Oh, if that's the case you do know your doing it wrong? Most times people come here saying, "my project is due next week, please help". :D:eek::)

I would think using a LOPT/flyback coil like Ron showed would be, in the end more compact than doing a planar type coil and adding a ferrite core to a PCB. But like I said earlier I'm no expert.
 

Thread Starter

SiCEngineer

Joined May 22, 2019
299
Oh, if that's the case you do know your doing it wrong? Most times people come here saying, "my project is due next week, please help". :D:eek::)

I would think using a LOPT/flyback coil like Ron showed would be, in the end more compact than doing a planar type coil and adding a ferrite core to a PCB. But like I said earlier I'm no expert.
I have about 1.5 years left of my project now, and have got all circuit layout schematics, and all components sorted and ready. Just need to design the PCB and the transformer then give it all a whirl.

I do believe the size of the solutions would be about the same. I am interested in planar for other reasons such as the uniform heat distribution in the planar magnetic and the increased efficiency. Repeatability is also very important for me and my company - having predictable component parasitics is definitely a plus side.
 
Top