Cheap way to measure linear positional displacement w/ high accuracy?

I'm working on a paid project for a professor, but this isn't for a grade so you aren't helping me with a school assignment. It's to test his composite bike forks. I have load cells to measure the force, but I need a really high accuracy way to measure displacement and capture it with windaq and a DAQ kit. A good target would be +/- 5 thousandths of an inch at 30 samples per second. here's some options I'm thinking of, if anyone can add to them or give me more options please do:

1.There are those graduated capacitive strips they use in digital calipers, I'm not sure how reasonable they are to buy. I was thinking, they have $20 digital calipers that offer the accuracy I need, but is there any reasonable hope of modifying one to be captured? I suppose the easy and dirty way would be to take the LCD lines and program a PIC to decode them back into numbers, but that would be slow.

2.Just get a really reliable slide pot, or set up a very reliable rotary pot with linear motion rigging.

3.Graduated optical tracking like they use in printers.

as a side note, I'm a mechanical engineering student who knows some about robotics (pic programming, motor control, PC programming), but I'm by no means an electronics expert. I've got the mechanical testing table and most of the electronics covered, but the positional part is going to require some stuff I haven't done before. That's why I'm seeking a bit of advice on where to start from some people who may have done electronic position measurement at high accuracy.

Some of the Chinese digital calipers have an IO port under the cover. A while back I posted a link with instructions on interfacing with it.

I think what 3v0 is saying is all you'd have to do is mechanically mount the caliper to the piston you're trying to measure and you're good to go. Out of curiosity what do you need that much accuracy and fast update rate for?

I think potentiometric is the way to go. How much travel do you need? What do you consider low cost? Digikey has a few linear position sensors. Here is one for $18.00
**broken link removed**

For a broader view what Digikey has look at this link.
**broken link removed**

Mike... For .005" resolution? For 20 bucks he can get a .001 (or better) resolution solution for the price of JUST the potentiometer there. All that has to be done is is the mechanical modifications to to couple the caliper to the shaft, and it'll work up to 6 inches.

I imagine that many of those cheap calipers would fall apart when you try to hack into them. Also reverse engineering one may not be so simple unless you can get a hold of one that has hacking instructions on the net. At any rate, the sensors I posted are just one option.

I'm not sure what range of travel I will need, it's going to be an oscillating load applied to a composite bike fork, but I haven't received all of the specs yet. I would imagine not much more than an inch one way or the other, the thing won't bend much more than that without breaking. For that matter I'm not 100% sure on the precision that will be needed. I want to study up on this technology before we meet to outline the project, I have a good idea of how to work with the other parts of this project, DAQ, control loops, pnumatics and so on, but I don't want to be at a loss when he brings up position monitoring.

Thanks for pointing out the tutorial 3v0, you're very helpful, as usual. Mike, It may turn out that the precision isn't as high as he speculated before, and a linear pot could be perfect. I do wonder if those caliper mechanisms and electronics can handle oscillating quickly without loosing accuracy, or lagging, but I'll have to get a good idea of the speeds were talking about here.

If there's one thing I've learned about contract work, it's to get all the specifics you can about everything that could come up. Preferably in writing. Most of my contract work is more mechanical and less electronic, but I'm working my way into robotics as I learn. If anyone is interested I'll keep you posted on the progress of this thing, or probably just stick it in my blog.

Triode said:

I'm not sure what range of travel I will need, it's going to be an oscillating load applied to a composite bike fork, but I haven't received all of the specs yet. I would imagine not much more than an inch one way or the other,

Click to expand...

Would you like to ride a bike with a one inch oscillating fork on rocky terrain? It's pretty dangerous.

I assume oscillations of maximum 1mm.

Boncuk

Mikebits said:

I imagine that many of those cheap calipers would fall apart when you try to hack into them. Also reverse engineering one may not be so simple unless you can get a hold of one that has hacking instructions on the net. At any rate, the sensors I posted are just one option.

Click to expand...

This is the link regarding communicating with the caliper. There is no physical hacking. The communication port already exists.

I have no idea how fast the caliper will respond of how long it would last. I was just addressing the OP's idea of using one.

I don't know if it's fast enough either, but I can investigate these things if they aren't known. My calipers actually have one of those ports on them, perhaps at the school electronics lab I can find a slip connector that will mate with it without soldering, otherwise I can make a small funding request and buy a cheap one for under $20 and try out the idea.

Boncuk, you have a good point, but you have to remember that this testing equipment will be used with many forks and in many conditions, in some cases he's probably going to want to use it to bend them until they break, which could be more than one inch. When they test airplane wings I've seen them flex them until the tip is 60 degrees off from the base. You would not want to see it do that when you were in the plane!

Wait, bend them till they break?! That's going to destroy the caliper from the shock on the rebound more than likely. Can you flesh out exactly what you're trying to do here with a lot more detail? You're generic question is unanswerable because it has some extreme qualifications which you failed to mention.

OK, I can give you a start as to how I would do it. First of all you mention WINDAQ and a kit. I will take a guess at a DATAQ Starter Kit? Let's assume I guessed right. I would further guess a DI 194 or 148 starter kit which gives you an input of +/- 10 volts with 10 bit resolution. Maybe a DI 158 which affords 12 bit resolution. Knowing the kit and the resolution is very important. The basic versions of both kits offer a sample rate of a maximum of 240 samples per second. There is a caveat there. The sample rate is 240 divided by the number of channels used.

You will be using 2 channels and plotting force against travel, therefore your best sample rate is 120 samples per second per channel. Unless you are using WINDAQ PRO. Personally I would run two channels at 120 samples per second.

Next, back to the bits. If you have a 10 bit ADC you have 1024 quantization levels so on a good day you have 10 volts FS / 1024 = 9.765 mV per step. If you have a 12 bit device you have 4096 quantization levels so 10 volts / 4096 = 2.44 mV per step. That is what will determine your resolution be it in pounds of force or inches of travel.

Now what you use to measure travel is your call, however, a linear pot or if it trips your trigger an LVDT (Linear Voltage Displacement Transducer) is how I would go. You really need to know the maximum travel to choose the pot. Here is what happens if you use a pot.

Lets say maximum travel is 0.5" you would set it up so 0 to 0.5 inch is equal to 0 to 10 volts. However, if max travel is 1.0" you set up for 0 to 1" is equal to 0 to 10 volts. Think about the resolution for your kit.

Again, I assume DATAQ WINDAQ basic software you will be using their charting software? I am also guessing the record software? You would then scale the channels in your engineering units. Using user annotations.

Now if you want to find the peaks just browse back.

The biggest problem you face is using accurate and cheap (inexpensive) in the same sentence. You want a pretty good linear pot.

Hope all this helps....

Ron

Sceadwian, I'm a mechanical engineer. I may have to ask dumb questions about electronics, but don't worry about common sense mechanics. I just meant that these will exceed the displacement one would expect in normal usage, if they actually crack the positional measurement won't matter, and the way it's going to be rigged it won't go shooting off if it breaks, so it shouldn't harm the sensor. As for specifics, the meeting today to get all the specs was moved to tomorrow, so I don't know all of the details. You do raise a good point about protecting the sensor. You are right that I need more detail. Even though I don't have enough detail to nail down a solution by any means yet, this is serving the purpose of letting me know what questions I should ask and what things I should consider when I meet up to get the specs.

Reloadron, those are some good points. As for cheap, I should have been more specific, I don't know the budget yet, but what I mean is that were looking at the hundreds or low thousands. when saying cheap there is a lot of play because you have middle school science fair expensive all the way up to air force expensive. But to be more to the point, I'll be sure to ask him what DAQ kit, load cells and software he has for me to use, and what amount of deflection, oscillation speed and accuracy were looking for.

I know it seems like I have no idea, but this does serve a purpose and you guys are helping me. Most of my contract work is things like frames or automated positioners (ones that aim and focus telescopes for example). I'm getting in to new territory here and while I'm sure I can figure out all the technology, I don't want to walk into the meeting without even knowing what to ask about. Thanks for the help so far, and keep giving me any comments about the technology, I am reading up on it where i can but there are things I might not even know about to look up. Now is the time to be pedantic!

What I was concerned about is the extreme forces unleashed if something under that much load releases, being a mechanical engineer you probably have a better idea of it than me. I was just concerned for the sensor itself if it was connected to a device that was even related to that kind of shock, the elasticity in the mounting material itself is going to put the caliper or other measuring device under incredible strains if a few thousand pounds of forced is released suddenly, massive G forces because of the velocities if it's a stiff mounting, nothing whatsoever to do with flying debris from the shock itself.

I was more curious about what you intend to measure with this device? Any of the oscillation frequencies should be easily calculated from the physical construction and materials themselves. You only want 30 FPS from your sensor, you could do better than that with a web cam (I've seen some that offer 60FPS rates) if you don't mind a loss in the specific resolution you have you can still calculate bulk velocities to incredible accuracies after calibration. If you're trying to figure the resonant frequency of a composite fork especially if it's extremely rigid 30FPS wouldn't cut it I wouldn't think. But I'm just being pedantic.

What I'm really trying to ask is EXACTLY what are you trying to measure? You're a mechanical engineer and don't know much about electronics so you from the get go don't know what you should be using yet you in your first post made a few suggestions of what you think you should use, the combination of large distances extreme accuracy and high resolution just doesn't fit to any specific measurement I can think of, then again I'm not thinking too hard =>

Just to give a little background, I'm a grad student making a living as a custom equipment fabricator, freelance stuff. Here is what I have to go on so far, there have only been some emails, I'll get a lot more specifics when we meet.

" The goal of the project is to create a servo-pneumatic test machine for testing bicycle forks. We will need to capture loads and displacements and record the data for many cycles. The loads are less than 200 lbs, but the parts experience significant deflection of nearly an inch or so."

I asked him about what equipment he already had and what kind of bicycle forks were talking about and got this:

"I have a bunch of load cells already and got a started DAQ kit from WinDAQ. Not necessarily the way to go but these are what are now available. We also need to measure displacement and I am looking for a cheap way to do that. Maybe a string pot. Regarding servo controls, I found the following company we may want to consider.

Enfield Technologies - Servo Pneumatic Proportional Control Systems
http://machinedesign.com/video/proportional-control-of-pneumatic-cylinders-0311"

I also know that he works with composites and that these will be layered up fiber materials. This is going to be the kind of project where they don't tell me what they want specifically, but what they need, and I have to figure out how. It's more nerve wracking because then you have the pressure of designing it, not just building it.

So to answer your question, it sounds as if it will be measuring "displacement" and it will move "about an inch". I can go back and meet with the guy again, so it's not like I have to think of every question now, but it's best to get as much out of a meeting as possible. I think you guys have already given me some good things to think about. As always with contract work, I'm going to try to get him to tell me exactly what he wants.

OK, where he seems to make refrence to WinDaq what he is really talking about is Dataq. When he refrences DAQ kit I think he is refrencing Dataq Starter Kits which you will see in the link I provided. I suggest you read the links for the starter kits and their associated PDF files. That will give you a handle on the specifications. Things will fall into place as to the DAQ device once you know exactly what he has.

Now as to the load cell. The most common flavor is frequently called a "Pancake Load Cell" used to measure tension and compression. These units typically use a bridge configuration and their output is specified as mV/Volt. What that means is if you have a 100 Lb load cell specified at 2 mV/Volt and you use (apply) 10 Volt excitation (very typical) the output of the bridge at full load of 100 Lbs will be 20 mV. Obviously this creates a small problem so the load cell output is fed to a signal conditioner typically to give you a nice linear signal of 0 to 10 Volts (or 4 to 20 mA). That gets you to 0 to 100 Lbs of force (tension or compression) = 0 to +/- 10 volts. Compression typically positive and tension typically a negative output.

With a short stroke like 1" or so a string pot really isn't practical. I would consider a simple LVDT (Linear Motion Pot) to measure displacement. something like this. They come in all sizes and flavors with or without sperical balls and bearings for mountion. They can be as cheap as $50 and as expensive as $1,000. I use many made by Honeywell. Again, for a short stroke a string pot really isn't the best choice. The Honeywell Short Longfellow series is nice but can be pricey. They do have some with a .1% linearity.

Overall, back to Dataq. Their recording software, even the basic lite version is nice stuff. If you can write in Visual Basic (or several other languages) you can write your own data acquisition software. Pretty much where you want to go with things and budget.

Ron

Thanks. I can write in visual basic, I'm pretty well versed in programming in general (C, VB, FORTRAN, java ect). The load cells might be a bit of a challenge, I haven't worked with them before, but I should be able to figure them out, there appear to be some schematics for using them online. I'm glad you mentioned them though, they're more complicated then I suspected, so it's good to be reading up on them. I am pretty well versed at working with PIC chips, I wonder if I could set up a decent converter using the DAC on one, it sounds like the right voltage range.

I'll have to see what budget and precision we're going for before I know if a lineal motion pot like the one you linked to is right for the job. It looks good to me, if the budget is only a few hundred I'll probably go with a slider like what Mikebits posted.

I'll post what I find out about the requirements and the equipment. It looks like the daq kits aren't too expensive and he seems flexible about it, so if he has one that is too low end I can probably get one of the better ones, the very top model is only $200, not that it would be needed necessarily.

I wonder if a flex sensor would be useful here, they seem to be under $20, but I won't bring that up unless it sounds like it's needed.

By the way, I'm pretty sure he won't be breaking them in this machine, it turns out he has a destructive tester already. This is specifically to test loaded oscillation. He also has a machine that tests oscillation, but it gives you a controlled distance, not force, and it doesn't measure load (or displacement). He seems to be building quite an array of testing tools for these things. It makes sense though, figuring out the very best way to wrap the carbon fiber on a complex shape like a bike fork could be pretty hard. I'm guessing it can be profitable too, but that's his business.

finally met up. The project sounds pretty reasonable. He wants to use existing pneumatic control kits and hes willing to buy them. The DAQ device is this DI-148 Data Acquisition Starter Kit - USB Data Acquisition system, but thats just the one he gave me, he said that they would be willing to buy one if I can recommend a better one. The pnumatic control kits seem to include position tracking and a servo valve with a PID device capable of adjusting to output a given force. The only things it lacks are feedback on the amount of force being generated, which will come from a load cell, and a wave function, which will come from anything I can make to generate it. He even gave me digital strain indicator (https://www.electro-tech-online.com/custompdfs/2010/05/M0450.pdf) to get the strain gauge data into a nice linear format. So the original question is null, the pnumatic/servo device already contains a suitable tracking pot. It sounds like the challenge for me will be properly setting up the load cell to provide feedback, and the other work involved will be making a user friendly data logging front end and function generator, and designing and building the whole apparatus.

Here's the specs for anyone interested:
accuracy +/- 0.05"
expected movement: 2" forward or back from center
maximum speed of oscillation: 4 HZ
maximum force applicable: 200 lbs push or pull

This thread was still worth while, I think I asked better questions because of the questions you guys asked me, thank you. I think I can do most of this without too much trouble, I've built many mechanical frames and it sounds like the electronics will consist of buying the right parts and possibly making some adapters and data converters, then there will be some programming. It's good to know that this guy is reasonable, sometimes as a custom equipment maker you run into stuff like a client who wants you to build a multiphase vibrational 1/4 scale earthquake simulator with 3D optical tracking that he can control with his cell phone built completely from scratch for $100.

Hi Paul

The DI 148 would work but the DI 158 series would be a better choice. The 148 being a 10 bit device and the 158 being a 12 bit device. They also make the 158 in scalable versions so it isn't a flat + / - 10 volt input. Additionally on the 148 all channels share common while the 158 offers differential inputs.

Nice you now know what he has and what is expected.

I am currently on a road trip but I still check the forums.

Ron