PCB coil to replace hand wound coil.

[rickford]

I have an application in which I am transferring power from a stationary device to a device that is rotating. The device that is stationary outputs an alternating magnetic field at 26kHz (sine wave) using a U shaped ferrite. The device that is rotating has a large, 7" diameter, coil wound in air with 24 turns of 26 gage wire. The device that is being powered by this coil has capacitors across the input such that the coil and capacitors form a resonant circuit. This aids in the power coupling across the gap. Power transfer works quite well. I have since designed a PCB with 4 layers of 6 turns each, one ounce copper, with traces .01" wide with .01" spacing between coils on the same layer. The board house tells me that the inter layer spacing is .04". The coil windings on the PCB are all in series, so it is also 24 turns. In the same way, capacitance is added to the device under power to bring the system to resonance. Problem is, this PCB coil does not pick up power nearly as efficiently as the hand wound coil does. It also doesn't seem to resonate nearly as well. I need to make changes to the PCB to improve this but changes are very expensive, so I need to understand what's going wrong. First off, the DC resistance of the hand wound coil is about 3 ohms, where as the DC resistance of the PCB coil is about 35 ohms. I tested the hand wound coil with a 30 ohm resistor in series with the coil, but still, ample performance was found. The resistor created loss of power transfer, but not nearly enough to cause failure. There are 2 other issues that I'm thinking of, but I can find very little about either on. First is skin effect. Second is the air gaps between the coil windings in the PCB are greater than those of the hand wound coil. I'm wondering if there is anyone that has a better understanding in this area that could shed light on what I should do to improve the coil design. Thanks.

 

[VEBill]

Are the original hand-wound coil and the new PCB coil both approximately the same dimensions and orientation?

If not, then perhaps a hand-wound version of the PCB coil might be a good experiment to distinguish between the geometry changes and material/coil-Q/resonance issues.

 

[rickford]

They are the same diameter. The hand wound coil is wound "tightly" with normal (round) 26 gage magnetic wire, so each winding is very close to the next. With the PCB, there are air gaps between the copper that are a fair bit bigger than the spacing between the copper of the hand wound coil. I'm not sure what to do with the PCB... pay extra to have the coils places extra close to each other? Pay extra to up the copper weight? Or, both, or something I haven't thought of?

 

[MikeHalloran]

I don't understand why the DC resistances should be quite so different. ... but that's probably not the source of your problem, as indicated by the experiment with resistor.

I think the problem lies in the magnetic circuit.

Draw a cross section of the source coil, the ferrite cup, and the target coil, and I think you'll find that the printed target coil's windings are in the farther fringes of the target coil's magnetic circuit, because of the spacing.

It may be possible to add some passive soft iron to tighten up the magnetic circuit a bit, or maybe add a complementary ferrite.



Mike Halloran
Pembroke Pines, FL, USA

 

[rickford]

The DC resistance is high because the PCB traces don't have anywhere near the cross sectional area that the 26 gage wire has.

I can put the PCB coil in almost exactly the same location as the hand wound coil. Compared to the size of the u-shaped ferrite, the coil width/depth size isn't so large. An entire portion of the coil can sit between the 2 sides of the ferrite on the primary.

I wonder if the Q of the PCB coil is much lower than that of the hand wound coil, and if so, if this could be causing the problem.

 

[MikeHalloran]

The 26 gage wire has just under twice the cross sectional area of the stated traces, so the resistance should differ by a factor of ~2, not ~10.

The measure of an electromagnet is ampere-turns.
Since the turns are the same and you are relying on resonance to push the amperes, the Q does matter.





Mike Halloran
Pembroke Pines, FL, USA

 

[rickford]

Diameter of .026 gage wire, .015". Cross sectional area, .0002sq".

Thickness of 1oz copper, .0014". Width of copper trace, .010". Cross sectional area, .000014sq".

.0002/.000014 = 14.3, which is about right.

 

[Skogsgurra]

What power levels are talking about?

The diameter hints at at least ten watts. I cannot imagine that a ten-fold increase in resistance wouldn't be the difference between sucess and failure in such an arrangement. It is so obvious that I need to question how you verified operation of the coil+resistor. Same input and roughly same output?

Gunnar Englund

http://www.gke.org

--------------------------------------
Half full - Half empty? I don't mind. It's what in it that counts.

 

[MikeHalloran]

Thanks, rickford; I had the thickness wrong.

Something still ain't right.


Mike Halloran
Pembroke Pines, FL, USA

 

[rickford]

I have a small circuit that runs on the secondary coil. It rectifies the power from the coil and filters it to get a clean DC power supply. The power supply working range is anywhere from 6.5V to 25V. With the hand wound coil, I can tune the power circuit up to provide as much as 25VDC (at about 30mA) to the circuit. Putting the series resistor in drops the voltage at the same tuning to about 17VDC. The load runs off of a linear regulator, so the current draw is roughly constant over the input voltage. Putting the PCB in place, the highest I can tune the supply to is about 7V. It works, but it's way too marginal. I would like to see from 12V to 15V available from the supply so I can have confidence in it.

 

[IRstuff]

Have you looked at whether the lossiness of the FR4 might be impacting the resonant circuit?

TTFN
faq731-376

Need help writing a question or understanding a reply? forum1529

 

[rickford]

No, I didn't think there would be any losses there. It doesn't seem to be magnetic in any way, and it's not conductive, so I don't have to worry about eddy current losses.

 

[rickford]

BTW, the relative permeability of FR4 is 1, so I think this supports the idea that FR4 is not creating any more losses than the air wound coil.

 

[itsmoked]

The two big issues are the resistance of the traces and the gaps between them. Magnetic flux that cuts between the traces is lost from exciting the other coils. It's a multiplicative loss.

I'd suggest re-doing the board with the traces stacked up as much as possible so if it were all clear you wouldn't be able to shine a light past adjacent coils. I'd also have the coils about 8 mils apart (just before a board fab premium kicks in) if you can pull that off. Set them up with all trace spacings symmetrically covered by a centered trace on a different layer.


For the heck of it I'd also do a 2 layer board with the same overlap scheme with the same total winds with "4 mill copper" which really means 2 mill copper with 2 mills of tin on them. This would dump the high resistance and reduce the cost on the board by being only 2 layer.

Keith Cress
kcress -

http://www.flaminsystems.com

 

[MacGyverS2000]

"4 mill copper" which really means 2 mill copper with 2 mills of tin on them.

I certainly hope not (though I'm assuming you mean 4 oz[/u] copper, not mil). Layer thickness should not include anything other than bare copper... it could be copper electro-dsposited onto a thinner starter layer, but in the end the 4 oz should be all copper.

Dan - Owner

http://www.Hi-TecDesigns.com

 

[rickford]

Interesting that you suggest staggering the coils like that. That's exactly what I don't have. Right now, all 4 layers lay right on top of one another so that you can see right through the gaps.

What happens to the magnetic flux as it passes through one of the copper layers? Is it redirected? If so, how does it then interact with layers behind it? I would think it has to be redirected to some degree because the copper trace will have a current induced and that will produce a magnetic field of its own. This stuff fries my brain. If the windings of a coil are far apart, the magnetic field produced by any one winding circles the winding itself, and not the group. When the windings are close together, the bulk of the magnetic field engulfs the entire group of wires... So what happens to the magnetic field that induced the current? Yup, brain in frying.

 

[OperaHouse]

I am reminded of that Govt. warning.....Your resistance may vary. You may not have enough skin in the game. Hate to calculate so this what I use:

http://www.powerstream.com/Wire_Size.htm

Look down and at 20 gauge there begins a penalty over 27khz. Given the thickness of the trace, your resistance could be a lot higher than 30 ohms.

 

[rickford]

I found a calculation of AC resistance due to skin effect, and it was minimal, assuming I did it right. Only a couple ohms.

 

[VEBill]

It might be something that we can't see over the Internet. Triple check that the PCB turns are wired in series and in the correct phase. Triple check that you haven't got a shorted turn (eating power). Triple check the absolute basics, because the higher order potential problems aren't leading anywhere so far.

 

[itsmoked]

Yeah MacGyverS2000 I slipped up on the units there, I meant "oz" not mills.

I think you may find a common way board houses increase the "oz" it's purely by running the board thru a wave solderer before solder masking.

What happens to the magnetic flux as it passes through one of the copper layers?

Doesn't pass thru the copper - around it.

Keith Cress
kcress -

http://www.flaminsystems.com