LVDT design ratios

[CopperCone]

I want to build LVDT. Seems easy but what ratio. Should i use between core thickness, coil lenght and travel?

I.e. the LVDT sensor is 3 inches long, with one inch per coil.

How long should the magnetic device which travels should be? 3 inches? Slightly larger?

Also, how do i set the limits before nonlinearity due to leakage inductance sets in? I.e. for a three inch coil, whats the recommended displacement?

I.e. if the displacement is 0.8 inches, this leaves 20% of the original core lenght in the top inductor. Does anyone know the percentage of core required where displaced vs the nonlinearity incurred? (i want to know when i can avoid look up table

I assume it will lead to weird coupling if displaced too far.
I wanted to use 100KHz

Also the middle can be thinner right? So long the end bits are symetrical.

[retrolefty]

I've only worked on LVDT in industrial applications. They were pretty expensive and the winding used on the 12" travel I saw were quite complex. In talking to the supplier he said the winding was designed not only for the travel but also for some temperature compensation because of the ambient variations in outdoor service. I suspect that real world info will be difficult to find other then just basic theory.

[CopperCone]

Any hunches?

I found a website online, someone made one without knowing wtf they are doing and it came out to be like 2% linear.
http://www.mikesflightdeck.com/lvdts/lvdts.html

[Conrad Hoffman]

You might try simulating your design using the free magnetic simulator FEMM- http://www.femm.info/wiki/HomePage
There's a bit of a learning curve, but it's really pretty easy to use. Not sure if it will give you the voltage out, but it can at least show you what the fields from your coils look like.

edit- it's been done, but the article is behind a paywall- https://www.researchgate.net/figure/Use-of-FEMM-for-LVDT-modelling_fig2_221918107

[tecman]

The typical LVDT length (physical) is 4-6 times the measured stroke length.  If the core exceeds about 50% of the length of the secondary coils the linearity goes to hell.  As mentioned by retrolefty, very long stroke units use an altered scheme using multiple secondary coils over the length.  Difficult to make accurately.

paul

[CopperCone]

How long of a stroke length can I do at home before it gets weird?

[CopperCone]

how thick should it be in comparison to the stroke lenght too?

I am thinking about something thats like 4 inches

[joeqsmith]

I was talking with someone the other day about LVDTs.   Many years ago I was reading an article in a trade magazine, maybe Machine Design or EDN about LVDTs.  The article was about a design that they were investigating that used the phase information as well.  This allowed the sensor to go through multiple phase changes.   Basically, you could have a much longer working distance with a smaller package. 

I called up (we used phones back then to talk to people) and spoke with one of the developers.  The problem at the time was there was not a cheap and easy way to decode the information.  I think I saved that article thinking one day I may take a crack it it using modern parts. 

[Conrad Hoffman]

LVDTs excel at short stroke stuff where you need almost infinite resolution. IMO, for long stroke you're better off building a capacitance sensor like digital vernier calipers use. Or even a resistance slide wire if you can stand a bit of friction. What's the application?