Suggestions for linear displacement sensors, durable, underwater, low rez.

[Triode]

I'm working on a project to make a collision sensing bumper for an underwater robot. The bumper will be a bar across the front, the bar will be about 4 feet wide and have a stiff shock absorber (650 lb/in) at each end. It will sit about 6-8 inches in front of the frame and we want to put linear sensors at each side so it can detect how far the bumper is pressed in, and by using the difference at the two sides, roughly where the impact occurred along the bumper. There are a lot of trade offs to consider, the LVDTs I've found are too expensive, I'm worried a linear pot wouldn't be durable enough, I've considered capacitive and optical encoders but there are a lot to choose from so I'd like any input you guys have. Here are the basic requirements:

Can be put in waterproof casing, being inherently water proof is even better
Allows 3-5 inches of travel
Costs under $120 each
Not very high resolution, within 0.10 in is fine (+/- 0.5)
Requiring complex data interface and support circuitry is acceptable

If you know of a type of sensor, or even better a manufacturer or store that sells them, please let me know.

 

[jpanhalt]

How about an accelerometer? There are lots of manufacturers for them. I am currently working with a Memsic unit. It is sealed, but I don't know its pressure rating off hand. The Memsic site has several other applications.

Sourcing in the USA is a bit of a problem. One of the major parts vendors is being considered as a distributor. In the meantime, you can buy from Memsic directly in single lot quantities for <$5.00 each plus a huge amount for shipping from China.

John

Edit: The Memsic site seems designed to discourage use of its products. Here is a summary: https://www.electro-tech-online.com/custompdfs/2012/02/Product_selection_guide_components.pdf

I am using the MXD2020; another attractive device is the MXC6232.

 

[duffy]

4' wide? 650lb/in? Sounds like a pretty big robot! What's it for?

I made an displacement sensor for a throttle a couple of years ago using an HMC1501 and a magnet. Works great, cheap, we have thousands in the field. You mount the magnet on the moving part, put the sensor nearby. Sensor has a Wheatstone bridge output, send that to an op-amp, send that to an A/D on a μC. Needs a little scaling and linearizing in software, but it is as rugged as you can make the mounting for the magnet and board. You can use cheap alnico or ceramic magnets if you want.

 

[Triode]

jpanhalt, I think an acceleration may work. Since the springs can be assumed to return the bumper to the full forward position under normal conditions. I wonder how precise of displacement I can get, I know I would need to integrate the signal twice, but that isn't a big deal. It might even give me some useful vibration data.

duffy, that is a very interesting option. I'd looked at hall effect sensors but this seems like it is better documented than what I looked at before. Actually a combination of sensors may be the best thing and I think being so inexpensive and durable this would be a good one to include.

Since you asked, the robot is designed to drive on the bed of lake Michigan for various scientific experiments including sampling and measurements, it's named Lake Michigan Amphibious Robot (L'MAR). It was developed at UW Milwaukee's school of fresh water science. It has a huge mounting plate on top for whatever instrument the robot may be carrying. There's a picture and more info**broken link removed**. Since it's driving in the surf it's basically blind, the water is too murky for vision, too turbulent for sonar or lazer range finding, and too deep to just scan through the water, so I am developing these tactile bumpers for it. The robot weighs about 500 lbs and has horsepower about equal to an ATV.

 

[duffy]

That's a pretty nifty project, a littoral zone robot. Sounds like it needs tactile bumpers alright - maybe "feelers" like a crab or something.

 

[alec_t]

How about using a simple piezo disc? Bumper moves lever to apply pressure to disc. Couldn't get much cheaper. Easily waterproofed with elastomer. Robust. Piezo voltage output v displacement should be roughly linear.

 

[Triode]

Actually duffy, we do hope to move on to active tactile feelers after we have basic sensing bumpers working.

That's a good idea alec, and in general the use of some kind of mechanical system to scale down the movement to a smaller sensor may be in order. It could isolate the sensor from impact and put it on a more fitting resolution scale, since devices tend to be much higher rez than this requires, and smaller. Almost any sensor that can inherently sense over a 4 inch range of motion has thousands of degrees of resolution, but a device with closer to 100, which is all I need is usually tiny. I'll see what I can come up with with this, maybe a pressure sensor and a small spring, the spring could be size so that 4 inches of movement was equivalent to the right range of pressure change on the sensor.

 

[alec_t]

the spring could be size so that 4 inches of movement was equivalent to the right range of pressure change on the sensor.

That's what I had in mind if the lever is rigid. Another way would be to use a length of spring steel/brass etc strip as the lever. The bumper pushes on one end. The other end is fixed and acts as a pivot. A short distance from the pivot (depending on the movement scaling you want) is a small protrusion which bears on the piezo.

 

[duffy]

Don't forget these guys -
https://www.sparkfun.com/products/8606

Absolute output resistive bend sensors. Probably be perfect for "feelers".

 

[Triode]

That's a good idea too duffy, I wonder if they make ones that are a bit longer, I suppose when I get to the feelers I could just get the longest ones they have and put one after another. Or just cover a springy bar in a series of strain gauges and create a similar effect.

For now I'm ordering some force sensing resistor pads, I'll need to see how well the work and determine if they can give me a nice consistent output with a spring.

 

[duffy]

What, those little thin film ones? I can tell you how they work: LIKE CRAP.

 

[Triode]

I've seen some that give a pretty consistent output, they have them at several levels, but even the nicer ones that cost $30 a piece are way cheaper than a linear transducer that can cover 4" of almost any type. I know what you mean, I have a sample kit at my lab of the thin film kind, they take a lot of force to change then they span their whole range with a tiny bit more.

 

[duffy]

Yeah, those are the kind I have. You can get an improvement with a compliant backing - but then they wear out quick. Color me unimpressed with THOSE things.

 

[Triode]

well hopefully I can make it work. It has the advantage of being easy to water seal and being pretty tough. I've had all kinds of ideas to feed linear motion into a rotary encoder, by way of a cable, a lever or a scissor lift, but rotary encoders are very hard to water seal. I'll try to find a force resistor with an actual data sheet available which gives its approximate force to resistance curve. If this doesn't work out I could go with a very stiff spring and a load cell, but that makes the cost quite a bit higher.

 

[duffy]

Yeah, I wouldn't use a rotary encoder for this even if it wasn't under water.

What's wrong with that magnetic gizmo? Once you get into the mechanics I think you will find attaching a magnet somewhere on the moving part, and attaching a board in a waterproof box somewhere nearby on the stationary part is pretty simple compared to your other options.

 

[Triode]

I'm actually ordering parts for the force resistor option, the accelerometer option and the magnetic option. They're all inexpensive and some combination of them may work well too.

 

[alec_t]

I've just been doing a rough-and-ready experiment with a piezo disc. Voltage out does increase with applied force (could't tell if it's a linear response), but the voltage decays in a fraction of a second even when the force is maintained. Not sure if that's due to charge leakage internally of the disc or via my meter. So the disc is behaving something like an accelerometer.

 

[Pommie]

A slight variation on the hall effect sensor. There are many examples of using a Hall sensor with a gear wheel to measure RPM. If you use a toothed rack and two sensors you could make a very robust and reliable linear encoder.

BTW, the magnet is mounted behind the sensors and the teeth increase the field strength.

Mike.

 

[Triode]

That's an interesting idea Pommie. it would certainly be water safe, it is a bit complicated to make, though we do have a machine shop. I've ordered the accelerometer, the magnetic sensor and the force sensors, so I'm trying several of these ideas. I have some doubts about the magnetic sensor, it requires a lot of support components and it may get interference from the motors which are only 13" behind it. I guess we'll see.

 

[duffy]

The problems we had were more with the mounting brackets themselves. We had permeable steel mounted near the sensor in the first prototype, and the flux would "jump".



But even this crappy arrangement gave useable results. Solution was to use non magnetic steel. We also went with a magnet that we could screw down to the end of the rod.



I don't know what the "suspended load" of particles is like in the surf zone of Lake Michigan, but one problem you might have is if there is iron or something being churned up in the water that could stick to the magnet.