Ok now that we got that out of the way.... My name is Brian, I work in a factory part time making knife and scissor-blades. That is the origin of my screen-name, I run various types of grinding machines and I can be a real... um... sweetheart... to deal with when things go wrong.
I was in the hospital with dishydrosis recently and someone gifted me an arduino-kit to keep me amused while the sores on my arm healed.
Being of an Engineering background and of an Engineering mind -- despite being stuck in Production -- I decided that this was the best gift ever.
While laying there whacked out on various meds, I imagined a device... driven by an arduino-board... that takes the input from a contact sensor, converts said input into a number and displays it on the small attached screen. The numbers... indicate the curve of the blade. Those numbers could be interpreted by an utter b*stard like myself and used to correct the process.
Well I have a long ways to go here. I've done all I can with the kit I was given, I'm acquiring more pieces and parts as I get the money.
Long boring story short... looks like I need a linear position sensor of some kind, that can read the input from a probe. From there I need to convert that probe movement into distance, then crunch it to a display-able value.
Has anyone worked with sensors of this nature in conjunction with arduino-board? From there it's a matter of learning how to make everything work together to get me useable data.
Hmmm, there are a few criteria you need to decide on before you can get into a design.
The ones that come to mind are:
How automated do you want this to be? D o you want to just stick the blade in a jig, and have the gadget do all the moving around of the sensor? Does it do a mix of automated and manual operations to gather the measurements?
How precise do the measurements need to be? Tenths of an inch? Thousandths?
How many measurements do you need to take? And how fast?
A friend of mine has a CNC router. Router bits vary in length, and may not get chucked in exactly the same position every time. He puts a grounded aluminum block on top of the workpiece, and clips a wire going to a controller input to the router bit. His software slowly lowers the router until the circuit is closed, and then knows that the tip of the bit is exactly the height of the block above the workpiece.
You could do something similar, with a leadscrew-driven carriage travelling the lenght of the blade, and another leadscrew extending the conductive probe until it contacts the blade.
There are lots of projects on the web for doing (fairly) precise positioning from which you can get examples of how to build the leadscrew mechanism and get position feedback.
The device that immediately comes to my mind (and the name of which currently escapes me), is an obsolete engineering machine that my university recently sold. It looked kind of like a pen plotter, but it had two pens attached to eachother with a bunch of pivot points and arms. It allowed them to clamp a prototype into one side, and trace around it. The movements were duplicated on the other side, letting them see how close their production units were to the original. The entire thing was bolted to the floor, and had a 6" thick concrete table. Accuracy was a fraction of a millimeter.
Maybe someone here will know what i'm talking about.
That arm design could be used with something like this:
Currently we use a digital height-guage with a DTI. I zero at one part of the blade, than take multiple measurements along the curve... then I plot those numbers.
I want to zero at the same point, move along the part and take those same readings electronically and display them on a small screen.
First step is figuring out how to measure the part. I'll post a sketch this evening or late tomorrow.
Once I get that figured out and built I will mount the whole thing on an old inkjet printer mechanism, then figure out how to automate it.
@ devices come to mind - Potentiometers and encoders. Potentiometers come in linear and rotary versions.
Depending on resolution and repeatability a couple of linear potentiometers might be able to do the job, only testing would reveal the true resolution and repeatability. A couple of analog inputs and a push button to tell the Arduino when to make the measurement. Easy to do if you can find a linear pot with enough length. Accuracy - probably not so good.
Encoders - would work real nice with 2 Arduinos - one for each encoder, and a common push button. Encoders rigged to some kind of X-Y slide on the fixture - fine tooth gear drive. Resolution could be quite high - depends on the counts per rev of the encoder. Higher lower counts don't affect price too much for a given size encoder. Omron has some nice, small ones that are towards the low end of the price spectrum. Would be more work, but if built carefully it would be very accurate and repeatable. Harder to program, but good accuracy and repeatability.
Another option - various instrument manufacturers make indicators and other measuring instruments with a communication connection. Interface with that to the arduino and from the arduino to a PC.
Well being tight and not liking to pay for anything, i would make something similar to a coin sensor, a linear pot as mentioned before with a spring loaded arm to keep tension against the blade as it is guided through a slot
Sounds sort of like a two-dimensional coordinate measuring machine.
Grind_Bastard:
Those numbers could be interpreted by an utter b*stard like myself and used to correct the process.
Well actually, no. Google up Dr. Deming's "Funnel Experiment". Feedback mechanisms that perform automatic process adjustments increase variation. Now, if we want to discuss process improvement as opposed to correction, then certainly, but that is not something that has to (or should) happen on the production floor.
In the lingo we use in here... we take the information from the current measurement-method and correct the curve at the machine with a wrench and lead-hammer. In order to do that I need a measurement that doesn't take a half-hour in the lab to perform. Most of that is setup and teardown of the measurement equipment.
So... I use put my blade in the hypothetical Measurematic, and either by automation or manually positioning the head I get my numbers quicker, and I can correct the machine quicker.
Result, less tweaking and rework downstream.
The grinder dates from 1985, aside from the wheel-wear-comp PLC everything is adjusted by hand.
It's simply a matter of getting the data quicker so I can correct the machine quicker.
Assuming you can accept measurements taken along a straight line:
Mount a digital dial indicator, with a serial/USB output, on a carriage that runs parallel to the blade.
You'll probably want to drive the carriage with a lead screw, to simplify anti-backlash. It's a short travel, and a light, so it won't be terribly expensive.
It's going to be extremely important to hold the blade precisely parallel to the track. I'm thinking two V-blocks under the edge, and a single spring-loaded arm with a V-notch at the top. That'll need to be designed with zero slop.
The V-block material will need to be selected for a trade-off between not nicking the blades and the blocks getting worn out of precision. I'm thinking Delrin, but I'm an amateur, so don't give my suggestion a whole lot of weight.
Delrin is too soft. Currently we use 1,2,3 blocks... once we get to the point that we're automating my device I'll work out a holding arrangement out of tool-steel.
I am looking at digital dial indicators right now... We may have an extra at work that I can experiment with.