Making an LVDT

[Doctorandus_P]

I've been reading a bit about LVDT's and not really convinced this is a wonderful thing. It uses a complicated way to generate

an analog voltage, and then you have to put that into an ADC.

I also bumped into the AD598. This little chip costs EUR122 from mouser, which seems quite ridiculous.

Now you've built your sensor coils and are doing some experiments, I suggest you also do some experiments with a simpler method.
Ever since the '90-ies there are a bunch of circuits floating around the 'net for an LC meter. It's build as a free running LC oscillator and a microcontroller to measure the frequency of the oscillator though some reciprocal math. Years ago I did some experiments with this, and you can do quick measurements with about 0.1Hz resolution, (and could get more resolution by integrating over more oscillation periods). The hardware is very simple, and schematics and software is floating around the 'net.

Note:
Now re-posted in the right thread. I accidentally posted this in:
https://www.eevblog.com/forum/projects/diy-variable-ac-voltage-and-frequency-source-for-calibration/msg5553581

[Kleinstein]

For the solution with µC / ADC the drive signal should ideally use the same reference as the ADC. So more like square wave from the µC with a little low pass filtering.

Something ferromagnetic near the LVDT can interfere with the readings. It gets tricky to build a shielded LVDT.

[Conrad Hoffman]

some useful stuff here- https://www.researchgate.net/publication/221918107_Linear_Variable_Differential_Transformer_Design_and_Verification_Using_MATLAB_and_Finite_Element_Analysis

The strong point of an LVDT is almost unlimited resolution. If your needs are more modest, there are simpler ways as mentioned above. I've built simple LC oscillators with a longish coil and a ferrite that moved in and out, just measuring the resulting frequency. Worked fine but today everybody wants a digital solution with as little analog stuff as possible. Things like an https://ams-osram.com/products/sensors/position-sensors/ams-as5304-linear-sensor or it's many relatives are accurate and cheap.

[Doctorandus_P]

Those HALL based sensors don't give absolute position (i.e. the pattern repeats every 4mm).

And there are more options, for example EE-SX1070 is an analog optical sensor. It costs EUR 3 and can have a resolution of 1um over a range of 1mm.
I accidentally bumped into this sensor via:
https://hackaday.com/2024/05/18/flexures-keep-this-printed-displacement-sensor-in-line/

Edit / Addition: (not worth a second full post)
I had a look at TCUT1800X01 datasheet (See post below).
The sensors in these are staggered, and have a "linear range" of around 0.2mm These seem to be optimized for digital sensing, while the EE-SX1070 has a relatigely linear range of about 1mm and is desgned for linear position sensing.

[PCB.Wiz]

And there are more options, for example EE-SX1070 is an analog optical sensor. It costs EUR 3 and can have a resolution of 1um over a range of 1mm.
I accidentally bumped into this sensor via:
https://hackaday.com/2024/05/18/flexures-keep-this-printed-displacement-sensor-in-line/

That's just a diode-transistor photo interrupter.
The narrow slots give it a quasi linear transition with % of light transfer, but it is very temperature and unit to unit dependent.
A second sensor, mounted next to the working one, (or a dual sensor like TCUT1300X01) could give some tracking compensation if you wanted to push these.

Vishay have some interesting ones with parts like TCUT1800X01 having 4 slots.

[Muffins]

Hi guys,

Just a quick update on this one. It's been slow progress since the coil was built but now I can get some things after payday. I've ordered a function generator off ebay and I'll order a couple supplies this week. Frustrating not having stock of the basic components.

[Muffins]

For the start I would consider RC4558 or NE5532

I've picked up the RC4558 

Do you plan to use a MCU with a lookup table to correct for the S curve errors ?
 

Haven't quite planned that far ahead. But I have managed to get a function generator. This morning I hooked it up to the main coil and managed to see a waveform on one of the output coils. It was faint but there. Tomorrow I'll build the audio amplifier circuit to drive the coil properly.

Very nicely wound, very neat and even. Did you do it by hand or did you use some mechanical aids?

I mounted the 3D printed tube to a drill and wound the coils under power. I used a small amount of tape to secure each layer in place which also helped the next layer wind over it neatly. The inductance calculator is pretty handy.

A first test would be with function generator and scope, to get an idea of the signal level (depends on the core) and also suitable core length. The form with 3 coils in series depends quite a bit on the core length.  The simpler form would be with only 2 coils side be side and the 3rd coild wound on top in the center, possibly all the way over the others.

Well the first test today was close to this. I input a signal to the centre coil and had the scope connected to one of the side coils. For the core I used a 10mm x 30mm dowel pin that happened to be on my desk. Sliding the dowel pin back and forth did cause the amplitude of the output waveform to vary which I was happy about. A small screwdriver I used to push the pin into the coil seemed to be magnetized to the pin so I used that to move it. I might experiment with other coil configurations after seeing how well 2 seem to work.

For the solution with µC / ADC the drive signal should ideally use the same reference as the ADC.

I've picked up a triple output power supply. I was just looking for a dual rail one but then this one was really well priced. It has +-12v and a 5v supply. I thought I might use that to power the amplifier and a uC.


I've attached a picture of the testing, the oscilloscope shows a good sine wave on the output coil. I had to turn the resolution way up, reading a 37mV peak to peak at around 1kHz! The output voltage of the function generator dropped down a lot when connected to the coil, I think I won't use it to drive them directly haha.

[PCB.Wiz]

A first test would be with function generator and scope, to get an idea of the signal level (depends on the core) and also suitable core length. The form with 3 coils in series depends quite a bit on the core length. 
The simpler form would be with only 2 coils side be side and the 3rd coild wound on top in the center, possibly all the way over the others.

.... I might experiment with other coil configurations after seeing how well 2 seem to work.

It depends on your requirements.

3 coil designs have a natural centre-zero balance point, that does not vary with inductance.
They also have a balanced 's' curve that is somewhat linear near the centre, and they allow ratio metric (A-B)/(A+B) calculations.

That said, MCUs these days have cheap memory for tables and calibrate is not hard, so 2 coil designs and even 1 coil designs are cheaper to make, and cheaper to connect, and may be 'good enough' with a calibrate step, and maybe temperature compensation.

I see TI's inductance chips are getting cheaper,  the LDC1101 1ch 180ksps TI says is  1ku | $0.850  and the 2 ch LDC1312 1ku | $0.850  i2c with moderate 80-160sps

[Kleinstein]

37 mV is not such a small amplitude. AC is relatively easy to amplify before going to the phase sensitive rectifier or ADC.
It may be easier to amplify the detected signal than to use a larger drive level.

The connecting part to transfer the morement to the active core should be non metallic, like a plastic tube or glass tube.

[moffy]

I've attached a picture of the testing, the oscilloscope shows a good sine wave on the output coil. I had to turn the resolution way up, reading a 37mV peak to peak at around 1kHz! The output voltage of the function generator dropped down a lot when connected to the coil, I think I won't use it to drive them directly haha.

Try 10kHz, you should get greater amplitude as the coil impedance rises with frequency and signal generators typically have a 50R output impedance.

[Muffins]

I've made up the drive circuit with the OPA548T driving the coil. After a little fiddling to get it right I managed to get around a 4v peak to peak signal on the output coil driving with about 12v peak to peak at 20kHz.

I taped a pen tube to the dowel pin and used that to move it in the coils. It works fine while the dowel pin is in the coils but seems to shut down when it's moved out of them. I'm guessing the coils pull too much current when the steel core is removed. When it's placed back in the coils the thing starts up again and works fine.

I'm still just looking at one output coil. Time to work on the output circuit. And also add a few isolator switches so I don't have to keep unplugging the power supply.

37 mV is not such a small amplitude. AC is relatively easy to amplify before going to the phase sensitive rectifier or ADC.
It may be easier to amplify the detected signal than to use a larger drive level.

Wouldn't a low amplitude signal need sensitive circuits?

[Kleinstein]

The moderate excitation frequency implies a low to moderate bandwidth. Anyway reading the mechanical displacement is not fast by design.
With a noise of some 20 nV/sqrt(Hz) and maybe a 40 kHz bandwidth to the ADC with would be noise of some 4 µV or about 1/20000 of the range. The typical µC internal 12 bit ADC would still be more noisy. If really needed lower noise amplifiers are available.

[PCB.Wiz]

It works fine while the dowel pin is in the coils but seems to shut down when it's moved out of them. I'm guessing the coils pull too much current when the steel core is removed. When it's placed back in the coils the thing starts up again and works fine.

A series resistor reduces the effect of load inductor variations.