1
\$\begingroup\$

I am thinking of using the LDC1612 by TI to sense position over a 0 to 10mm range with an accuracy better than 100nm. I have been told this is possible by a TI engineer.

However, my question concerns relative dimensions of coil diameter and target. I am especially concerned with linearity of response, or the ability to linearize it either in circuit or mathematically.

Diameters - Is it best to have a coil>>target?

Does anyone here have any experience with this type of sensing?

\$\endgroup\$
  • \$\begingroup\$ I would think that both accuracy and linearity depend on shape, material and direction of an object more than size of a coil. SNOA930A and SNOA957A referenced from datasheet have a lot of relevant info. I don't think you'd achieve required precision without specifically designed target. \$\endgroup\$ – Maple Jul 25 '18 at 9:44
  • \$\begingroup\$ @Maple Well, we are quite prepared to specifically design the target, and if necessary make it out of Silver. There is quite extensive information available but I cannot find much on linearizing or what happens when the sense coil is much larger than the target \$\endgroup\$ – Dirk Bruere Jul 25 '18 at 9:55
  • 1
    \$\begingroup\$ From those documents it seems the size of a coil is primarily constrained by required inductance/resistance, which in turn dictated by frequency. I'd recommend looking at design of Zettlex sensors, working on the same principle. They have amazing precision, although it comes with huge price tag. \$\endgroup\$ – Maple Jul 25 '18 at 10:03
  • 2
    \$\begingroup\$ Did the TI guy say "accuracy" or "resoluton"? \$\endgroup\$ – Andy aka Jul 25 '18 at 10:06
  • 1
    \$\begingroup\$ Re: linearity, the design notes are quite clear in that rectangular or trapezoidal coils are much better than round (implied by "diameter" in your question). Zettlex sensors I mentioned above have exactly these kinds of traces. See "stretched coils" in SNOA931 for design examples. \$\endgroup\$ – Maple Jul 25 '18 at 10:16
1
\$\begingroup\$

If your primary concern is linearity, here is an idea not mentioned in TI design guides that might work better for you. (or might not work at all, I am not an electrical engineer after all)

Use two coils, one at the bottom one at the top of your "metallic container". Wire them into one in opposite directions. If the coils are geometrically identical then their non-linearity should somewhat compensated each other, although with the price of reduced sensitivity.

Alternatively use two channels of the chip for two separate coils and combine (subtract one from the other) them in software, basically making differential sensor. This should actually double the sensitivity while also reducing non-linearity.

\$\endgroup\$
1
\$\begingroup\$

Page 1 of the DS seems to hint at what you want to know: -

enter image description here

So if you want better than 0.1 um precision and 10 mm range then the target distance divided by sensor diameter needs to be less than 55%. This points to a minimum diameter of 18 mm.

However, making the diameter much bigger is also going to be a problem because the relative disturbance caused by the target to the magnetic field becomes smaller and hence signal levels drop. You will also start to get potential interference problems when using a larger diameter because the magnetic field depth becomes bigger and other (possibly moving) objects may create an error.

To obtain maximum sensitivity you might also consider attaching a shorted coil (or tuned resonant coil) to the bottom of the target. This will create a bigger signal than just having a metal target and induced eddy currents.

\$\endgroup\$
  • \$\begingroup\$ It also seems to answer linearity question. Under 40% it looks very good. \$\endgroup\$ – Maple Jul 25 '18 at 12:01

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.