Multilayer PCB coils viable?

[Gunghouk]

Hi all, here's a notion that's been rattling around inside my head but I know nothing about detector coils other than the basics.
Would it be possible to design a viable, multilayer pcb coil, utilizing vias and possibly using an interconnected ground plane as shielding?

Seems to me a multilayer mylar, flexible type would keep the weight down and improve coupling over a fibreglass substrate board but what do I know.

Also how about a spider coil in pcb form? Seems an ideal candidate.

Looking forward to your thoughts.

Regards
Neil

[pito]

[Gunghouk]

Thanks for that. Looks cool.
I never said the idea was original

So how do they perform?

How about a dual layer PI board say 12 turns per side, 26cm average diameter made from standard weight, conformal coated, FR6 board, 0.5mm track width & 0.07 track spacing?

[Gunghouk]

Actually, after I checked out some pcb track calculators, you could use 0.25 mm width track and use whatever spacing gets the inter track capacitance down to an acceptable value (PI) or gives you resonance with a single, selected, standard value (surface mounted?) capacitor. All this on a standard 1oz board with 0.018mm copper thickness.

[GeoMax]

Hi Pito,

nice PCB job!!! Question. Why are you using two connecting cables for your PCB coil? So far I have seen this only at Lorenz detectors with the latest generation of PI coils.

[lucifer]

Thanks for that. Looks cool.
I never said the idea was original

So how do they perform?

How about a dual layer PI board say 12 turns per side, 26cm average diameter made from standard weight, conformal coated, FR6 board, 0.5mm track width & 0.07 track spacing?

It depends what you want to achieve. Why FR6?
Placing turns above each other (top and bottom layer) will increase the coil's parasitic capacitance. Also IMO the flat spiral coils are not as deep. Their advantage is lower capacitance if designed properly so you can sample earlier in a PI device.
PCB coils generally have a higher resistance too. For your example, 24 turns with average diameter of 26cm and 0.5mm trace width 1oz copper weight makes ~19Ohm at temperature of 25C. This makes the peak current in the coil much lower and further reduces your detection depth.
I would either increase the trace width or use 2oz copper weight (higher PCB price) or both.

[pito]

Hi Pito,

nice PCB job!!! Question. Why are you using two connecting cables for your PCB coil? So far I have seen this only at Lorenz detectors with the latest generation of PI coils.

This is not my design, I have just pictures, one cable for Tx second for Rx, as far I remember author was looking for cable with two separate shielded pairs.

[Gunghouk]

It depends what you want to achieve. Why FR6?
Placing turns above each other (top and bottom layer) will increase the coil's parasitic capacitance. Also IMO the flat spiral coils are not as deep. Their advantage is lower capacitance if designed properly so you can sample earlier in a PI device.
PCB coils generally have a higher resistance too. For your example, 24 turns with average diameter of 26cm and 0.5mm trace width 1oz copper weight makes ~19Ohm at temperature of 25C. This makes the peak current in the coil much lower and further reduces your detection depth.
I would either increase the trace width or use 2oz copper weight (higher PCB price) or both.

Exactly. I'm not sure of the dielectric constant of fibreglass, solder mask or conformal coating combined but I'm sure someone here can get the numbers for the inter track capacitance on 2oz board (0.036 mm thick tracks?). The spider design transferred to pcb should also help to reduce the capacitance.

[lucifer]

Exactly. I'm not sure of the dielectric constant of fibreglass, solder mask or conformal coating combined but I'm sure someone here can get the numbers for the inter track capacitance on 2oz board (0.036 mm thick tracks?). The spider design transferred to pcb should also help to reduce the capacitance.

Dielectric constant of the PCB substrate depends on the manufacturer and is typically between 3.5 and 4.5. The solder mask has a dielectric constant usually in the range 3 - 4. You can use tin plated traces instead of a solder mask.
It's hard to tell what the coil capacitance would be in advance. Spider web seems to be the best in terms of low capacitance IMO. I had a Python script to generate the coil PCB footprint for KiCAD.

[Gunghouk]

How about laying the upper side tracks over the gaps in the lower trace to reduce the capacitance?

[Teleno]

Exactly. I'm not sure of the dielectric constant of fibreglass, solder mask or conformal coating combined but I'm sure someone here can get the numbers for the inter track capacitance on 2oz board (0.036 mm thick tracks?). The spider design transferred to pcb should also help to reduce the capacitance.

Not a problem with thick copper PCBs.



https://www.pcbway.com/pcb_prototype...pper_PCBs.html

[Teleno]

I've made these double sided with interleaved spirals.

[lucifer]

How about laying the upper side tracks over the gaps in the lower trace to reduce the capacitance?

Yep, that's how the spiderweb coil I posted is done.
Note that this way if trace width is bigger so is the gap between adjacent traces. That lowers the capacitance but also lowers coil inductance.

[ahermida]

I've made a few PCB type coils, 8" and 10" in 1 oz and 2 oz copper.

[Carl-NC]

A while back ahermida sent me some samples of his PCB coils to try out. All of his are single-sided. Here are the parametrics:



where
N = number of turns
w, s = width and spacing of the tracks in mils; 39.37 mils = 1mm
L, R, and f0 are measured; f0 is the SRF
C is calculated from L & f0
RD is the calculated damping resistor for critical damping

Note the 2 oz copper resistance is only slightly lower, which makes me question whether the PCB maker really does what they say.

I also have some double-sided PCB coils that I made:



Note the double-sided coils have much higher C. However, these were not laid out in a good way; the top-side coil spirals in, vias to the bottom, which then spirals back out. Teleno's method is much better and something I would like to try next.

Finally, ahermida sent me two spiral-wound wire coils for comparison:



Parametrics are almost identical to his PCB coils except, of coarse, for the R.

Finally, here is a pic of my coils:

They are nested for cost efficiency. Also, JLCPCB won't route circular boards so I have to hand-route them.