Planar Inductor Questions

T Chap

I am working on some boards for this inductive switch product that Tony recommended.

http://www.ti.com/lit/ds/symlink/ldc0851.pdf

After including the 3v3 LDO and other parts, the board is getting a bit large to fit in tight spaces. I am wondering if anyone has any experience with this to know how badly it would affect the performance to break this up into 2 or 3 boards.

1. Main components with Regulator, IC, caps, etc.
2. Ref planar inductor
3. Sense planar inductor

If I use the same length wire( ie 2") for both the sense and ref coils, would this screw up the performance? It would nice to be able to break this up and park the main board slightly away from the parts to detect. I could get by with having the sense and ref on the same board, similar to how this is shown now, just divide the board in half with jumpers. These are 10mm diameter coils just like the examples in the PDF. The distance for detection should be less than the diameter.

I think I will not be using a reference metal for distance setting for the trigger, but rather the other mode where you can set the threshold with a resistor. I think it does not matter which side of the coil is used for detection of the metal.

kwinn

Based on the data sheets I think you should be able to put the coils on one board and the circuitry on a second board mounted a few mm behind it. A square board with the coils placed diagonally could be somewhat less than half the size of the board pictured.

Tracy Allen

The caution to keep it all balanced would apply to the jumpers that connect the boards, so if they are both the same length (2") and symmetrical It seems it should be okay. The change in inductance due to the dead bolt will be large, me thinks. With experimentation you can probably come up with balanced L configurations that depart widely from the data sheet examples.

There are also the more costly LDC1000 or LDC1101. They kick out quantitative values for L and Rp of a single coil, which you could use to measure how far the dead bolt has traveled. µP required though.

At the other extreme, it might work with a simple metal detector, like this one...
http://www.micro-examples.com/articles/index.php/PicoDetector

Beau Schwabe

Is it a necessary requirement for you to etch your coils on the PCB?

Reason I ask, is that I have been working on a pinpoint metal detector, and I'm using 100uH chip inductors and they seem to work great. It looks like with the IC you are using that the chip inductors would work as well and allow you to balance any component tolerances through the ADJ pin.

Video test of my pin-point metal detector using the chip inductors:

jmg

Beau Schwabe wrote: »

Reason I ask, is that I have been working on a pinpoint metal detector, and I'm using 100uH chip inductors and they seem to work great.

Yes, we have used chip inductors for magnet sense, and they worked well.

For detect of the wings on the dead-latch, maybe something vertically aligned, like this ?

Beau Schwabe

Oh I didn't put two and two together that this might be used for the latch detector. ...I would have gone a different direction, have you explored a capacitive approach? ... Same principle as having two inductors and using one as a ref, you do something similar with a capacitor where you also use one as a ref. Imagine 3 plates facing the lock mechanism, the center plate would be common. One of the outside plates would be the reference... the one that always "sees" the deadbolt, and the other outside plate would be used to sense the presence or non presence of the deadbolt.

Note: On Inductors, an air core might be better suited than an inductor with a metallic core. A metallic core will limit overall sensitivity due to saturation issues.

T Chap

Thanks for the ideas. This is partly to test for the lock but also just to get a good understanding of the inductive sensing concepts. I will make 2 boards, one with 100uH and one with PCB traces. The board can be cut up into 3 parts if needed and wired together. I could not find any 100uH in air, but I have these 100uH on hand for my power supplies. 1.5" x 1.25" board. NPN and Push Pull Output. Either adjustable sense with trim pot or reference metal option.

tonyp12

There is a TI inductive sensing forum:
https://e2e.ti.com/support/sensor/inductive-sensing/

T Chap

Major blunder on the pcb coil, I had in mind the smaller hole was the actual drill and the larger was the screw head as I always draw it that way on my CNC machine. I wiill try to hack it but luckily I did the SMT inductor version to test.

T Chap

Well the SMT inductor version is working, but not a good enough distance. I am going to try to find out how to make a coil for ref and sense, as the datasheet states that the diameter / 3 is the target distance to trigger at.

T Chap

I have been experimenting with routing some PCB traces since my board had a mistake. Some observations so far. The spiral with the wider air gap and narrower trace gives the best distance of approx .375". I tried less gap and wider traces and the distance dropped to about 1/8". The datasheet says the diameter/3 is the distance to try to trigger within. One strange accidental thing happened, when I put the PCB down on a piece of 2x4 the trigger distance goes to 1.25". I would love to use this greater distance for my project, so I am trying to find out if the effect will change over time as the wood dries out as well as finding out what the effect is coming from. Anyone know what would cause this?

Beau Schwabe

That goes with what I said earlier ... an air core will give you the best dynamic response... remember to have a core doesn't necessarily mean it's a ferrous core... the adjacent copper traces in a pancake coil will also act like a core. If you really want to dial it in, then some imperial testing is in order with some control coils you have as a baseline measurement.

T Chap

Thanks Beau, I am learning that inductance is a time vortex that you can get lost in. I just did some more tests after your notes. I observed that if I put the PCB on a 1/8" rubber mat that is on top of 5/8" plywood, at an angle with the nylon spacers holding one side up about 1/4" gives almost .5" trigger. If I put the rubber mat direct on the 2x4, the tape the PCB direct to the rubber mat which is direct on the 2x4, I get slightly over 1". If I put the PCB direct and flat held on the 2x4, I get a very short distance. If I hold the PCB direct on the 2x4 at an angle with one side held up at 1/4 with the nylon spacers, then it goes back up to the above 1" distance. Seems it does not want to sit direct on the 2x4 without the angle, but with the rubber mat it can sit direct on it. Strange stuff. My question is, if I can nail something down in my shop with one of these voodoo combinations of rubber and 2x4, should I expect the result to maintain over 10 years as the wood dries or is that of little effect? I don't have 10 years to imperically test it, although I suppose I could test with drying out some 2x4s in a an oven? Very fun stuff to play with.

T Chap

Forgot the latest pics. I am trying to understand your points on air core. As I understand it wood is the same as air in terms or a core. So why is it that if the PCB is above the 2x4 by 12" then the trigger is at 3/8", but if I put the PCB on the 2x4 as shown either tilted up on one side or with the rubber mat and I get 1" trigger distance, how is the wood causing the effect in this case?

jmg

Strange results, maybe you are at the limits of where the TI chip likes to work ?
ie it may like some parasitic C, to pull frequencies down, but not so much coupling that a lossy dielectric affects Q ?

T Chap

The TI chip wants a minimum of 33pf Minimum including parasitic, and since I don't know the board C I am using a 33pf. I tested 47pF, seems like slightly less distance, any higher it does not work. I just ordered 20pf and 25pf I will get tomorrow to see what happens. As I understand it, Beaus suggestion for air vs metal core results in lower inductance, which I take to mean that the goal is less inductance for better distance.

Beau Schwabe

I would guess parasitic C also ... BTW... two 33pF's in series would give you roughly 16.5pF ... likewise two 47pF's give you about 23.5pF ... or one 47 and one 33 would give you about 19.4pF

...It's not so much lower inductance that gives better results, it has to do with the permeability of a metallic core. in the absence of metal with the exception of the wire in the coil itself, less metal will provide a more dynamic response when there actually is metal present you are trying to detect.

T Chap

If it is parasitic, then that could only be additive to the total capacitance the inputs are seeing including the cap required on the com input. If it is adding capacitance then perhaps I could see the same result with a higher value cap(for a slower freq) . The total parasitic + cap will set the frequency, so that means I should be able to see the effect of frequency change on a scope by moving it towards or away from the 2x4 which is producing dramatic distance effect.

Beau Schwabe

The scope would be helpful, but keep in mind with capacitance that low, the scope will also add capacitive loading.

I think that there are slight differences between each coil that you are able to tune or detune by adding or subtracting parasitic capacitance <-- IOW, you are able to hold your tongue just right in this experiment.

I'm not sure exactly how the TI chip works, but I have an idea, since it uses a common cap, and it has subtile hints in the block diagram and similar to what I was doing in the video.

First imagine pushing one person in a swing at just the right interval or resonant frequency. you apply minimal force to maintain the oscillations. Now imagine pushing that same person but on every other interval. It still takes little force to maintain the oscillations. Now add another person on a swing. Keeping the same pace as the every other interval now you push one person, and then the next, alternating between the two. Both people in the swing however are synchronized and swinging together. Now substitute the people swinging for coils. Any perturbation on one coil or the other will cause an imbalance that will show up on the output.

Note: It looks like your tilting adjustments could be accomplished by adjusting the "balance" electrically using the ADJ pin on the IC... it looks like though that the symbol is drawn wrong for the 4-bit-ADC in that it should be facing inward, as I "think" this provides a bias to adjust for any coil/layout/environment discrepancies

T Chap

I put a 50K pot on the bottom and 49.9k on top of the divider, the result when I was testing the pot was that it went from 3/8" down. The data sheet says I should only get 3/8' with the diameter I am using. The 1.125" distance just happened as an accident tonight as I was moving it around. I am totally fine mounting the thing on a 2x4 with the rubber separation. Trying to design something and not be able to see the magnetic field, board capacitance, and inductance is tricky. Purely guess work.

Beau Schwabe

"Trying to design something and not be able to see the magnetic field, board capacitance, and inductance is tricky. Purely guess work." ... Sometimes, although there have been several Aha! moments in my 25 years of playing with coils.

T Chap

Wood as a lossy dialectric! jmg is suggesting the wood is affection the inductance:

The "Q" of a coil or inductive circuit is the measure of the inductive effects of that coil or circuit. Sometimes refered to as the (Q)uality rating of the coil or circuit. A coil or circuit that is highly inductive with little resistance can be said to have a high Q rating. A coil or circuit with alot of resistance to dampen the inductive effects is said to have a low Q rating

T Chap

TI tech told me based on my pics that there was likely capacitive coupling on my wires running to the sensors and the wood was affecting the sensitivity. I built a new version today and it is working perfectly with no effect from wood nearby. This is a neat gadget Tony referred me to, very simple to build and a fun learning project. I am easily getting a good .5" trigger which is more than enough for the project. I need to mount a solid copper section of the PCB at the trigger height above the REF spiral sensor to set the limit. The was routed on a CNC, two sided spiral copper connected in the middle.