Accurate plus Precise distance

[manie]

As I said in another post, this project could really mean the difference between surviving (economically) or not, so IN ADVANCE, a really BIG THANK YOU for any and all input to assist. My previous trials with various Picaxe circuits had to lead here inevitably.....

The Chook-House controller is nearing final completion and will be tested under real bad (HOSTILE) conditions soon. I will report on that to this forum.

Now for this distance meaurement project:

The idea is to measure vehicle wheel alignment through a "drive over" platform mechanism. Slow speed drive over a "floating" platform with the wheel/s on one side. The other side's wheels will act as the "anchor" thus forcing any displacement onto the floating plate. The movement/displacement could therefore be either positive(outward) or negative(inward). With the right sensor, even very small displacement can be measured and interpreted, possibly as degrees or as x_mm over 3 meter distance etc. One would want to measure sub-millimeter movement to be as precise as possible and also be able to repeat these readings and hence be accurate to within tolerances of measurement.

I have asked already in the thread regarding the sonic measurement and that developed into BB's light/shadow measurement. That has a LOT of promise and I will one day look at that in earnest again. Since then, "logic" (possible ?? ) has led me down the path of good quality digital Vernier caliper devices such as those from Starret/Mitutoyo and Fowler_Sylvac as well as HeidenHain. Almost all of these are capable (claimed) of repeat accuracy of 20 micron and resolutions of 10 micron. For this project, 10 micron resolution is excellent and repeatability to within 20 micron is really good. So I thought to use these devices as most of them already have the built-in communications output, either via RS232 (Fowler 7bit data-2 stop bits) or Mitutoyo's proprietory type "SPC". In addition there are stand-alone Inductive sensors capable of similar resolution and accuracy with Analogue output (this I understand well and therefore like....) BUT, they seem to be virtually unobtainable in South Africa..... (any volunteer prospective suppliers on the forum ????)

Some specifications:

1. It seems one will have to allow for a minimum displacement of 20mm from center_Zero_point = 40mm total distance.
2. This displacement will occur with the wheel running over a +-300mm long floating plate.
3. Slow drive over, 3Km/H = 0.8m/sec = +- 0.25sec measuring period
4. Get as many readings as possible during the drive over in order to form a "graph" of Rate of Displacement = severity of mis-alignment.
5. Floating plate mounted on 4 x commercial Load Cells (also floating and already mechanically sorted out), to also get weight on wheel reading.

Why do this:

1. To provide an Auditing device of alignment done (sometimes not done correctly but paid for...)
2. To keep an eye on alignment and predict suspension etc. wear rates....(S.A. roads are getting notorious...)
3. To keep me economically alive.......... My previous sponsorship has now ended, but this is viable here for me...
4. This will result in the sale of a few more Picaxe 28x1's or 28x2's to "make up" for effort from Rev-Ed and this forum.

I will post some datasheets from various Verniers or Displacement sensors etc as I get them....

Thanks for reading, hopefully I can complete this within a reasonably short time.

 

[manie]

Some datasheets that I could get.

1. Starret apparently uses RS232 with "normal" ascii sent, data bits and stop bits = unknown at present.

2. Fowler_Sylvac uses start/7 data/2 stop bits RS232.

Here is a link to a S.A. site of Switches International. Could work but the measuring range is a bit small: http://www.switches.co.za/proximity.htm

Attacments below:

 

[manie]

Stand-alone sensors from Epsilon_Micro. Seems very good IF they repond to e-mail requests for more info regarding prices/availabilty and shipping. They do have Inductive sensors also.

Add to specifications, SENSING MUST BE INDUCTIVE because capacitive sensing will be severely influenced by moisture (rainy weather etc)

 

[Dippy]

Quite a task Manie.
Quite a lot for people to read and absorb when they can't actually see what you intend.


This 'floating plate'; how is it located? e.g. is there a pivot point?

A good technical drawing could save a lot people wasting time trying to imagine what you mean.

So, after all the above text and links and "can't get in SA" are you going to buy a sensor or make one?
After all, you can get anything from anywhere if you have time or money.

Which bits are you doing and which bits are you buying?

Where do people have to put their effort in?

Its an interesting project which might make you some money, but it's confusing as to what you would like from others (free of charge).

I'm sure many people can suggest weird and wonderful home-brew sensors but without a firm specification and method of mounting etc. they they could be spending hours for nothing.

And obviously with moisture kicking around you won't want cardboard and sellotape.
I'd worry about the long-term aspect of any standard vernier when exposed long-term to high humidity. And obviously you will have designed/bought/imagined some robust mechanism to protect the delicate vernier movement from neanderthals driving cars over it

PS. I wonder if Stan will suggest a floating breadboard

 

[manuka]

Manie: Ahem! Aren't you concerned that these postings are rather beyond the normal "scope" of the Forum? They've shades of a standard business or technical assistance request, yet key infomation is missing. Hence sketches,BUDGET,time frame,installation,calibration & maintenance? What of the IP relating to the idea?

For US readers, 20 microns is ~the diameter of fine human hair (1 micron= 1 millioneth of a metre, = 1/1000 of a millimetre). For wheel alignment this seems very fine indeed! Is 20mm travel enough given rough road surfaces? I've travelled African roads with potholes of such depth that one almost expects to hear echos from them. Stan.

Dippy- trust you're not implying a magic weather stone type application? "If wet it's raining, if swaying it's windy, if you can't see it then the sun has set" etc. Visions of a breadboard being used as a packing shim for wheel alignments arise...

 

[manie]

Hope the attachment here provides more info. View is square-on from the top. The sensors etc will be protected environmentally as best possible and contained within the frame/box support structure. The bearings are really industrial sealed types with high load capability. The arms of the "parallelogram" will be made of hardened or "EN-18" type precision turned steel and will be able to support a lot of weight. The bearings at bottom takes the brunt of the weight.
Description:
As the floating plate gets forced into one direction or the other, this will cause the parallelogram to move and thus cause movement of the green target. Away from one sensor and towards the other. This movement could be very small (alignment = OK) to quite large (alignment = OUT). The displacement distance is measured. Here I'm starting to think stand-alone Induction (Eddy current) sensors. See attached datsheet for those.

the resolution needs to be fine, at least 0.3mm=300 micron, but since the sensors can offer better resolution and linearity, why nort use that ? Please ask for any other info you require, if I have it I'll give it or will search for it.

Thanks for your interest guys, appreciated !!!!!!

 

[Dippy]

Well, if you think it'll work and you've selected a sensor ... then what's the question?

I'm sure Beaniebots will enjoy helping you with this, it's beyond my time allocation.

 

[manie]

Dippy: I have not selected one YET.. I would like to know from the datasheet if you think its possible with one of those, probably the 16mm one. Whether you see any obvious problems as I have never used anything like this. And then in future....

there will be possible code problems. I would hate to place extra burden on you, but sometimes one requires reassurance from those more experienced.

As for weather-proofing, these sensors are sealed to IP-67 standard I think I saw, so that means Zero dirt ingress and watertight to 1M... Keep it reasonably boxed and it should be OK for a long while. Also they seem temperature compensated, as to price, I'm trying to get info from the States...

 

[Dippy]

I see.

Well, I've never used one of those sensors so can't say. I can't see, in principle, why it shouldn't work.
It senses distance to a target and it gives an output. Bingo!
It'll be up to you to ensure everything is within range mechanically.

I don't spend much time in garage workshops and the only tracking devices I've seen (apart from the older ones) have been a floating plate design but the vehicle has to be stationary.
With a moving one there are too many variables for my little brain to absorb, so can't help.


I hope the price doesn't shock you. I hope you don't end up with string and a pot
Anyway, good luck. Nice project. The code should be a piece of cake .

 

[RobertN]

A couple notes from antiquity. Mechanical drive over alignment (toe in) check stands were in use in the US decades ago, don't know if they still are.
Not sure why you would want to weight at the same time. Most scales use one or 2 load cells to minimize the cost. Small vehicle scales are used in the road racing arenas, they also have methods for quickly measuring alignment.
It will be difficult to obtain an accurate weight of a moving vehicle. The readings will be all over the place, even with averaging. Successive weighing will result in a good spread in readings also.
Strain gauges can measure deflection accurately and over most any distance depending on the mechanical structure. They have a minimum of mechanical joints and can be sealed environmentally very well. We used to put strain gauges on hack saw blades too measure deflection.

 

[manie]

RobertN: Now there is quite a nifty idea and probably worth looking into...

Dippy: Do know about the price but won't tell ?

 

[papaof2]

Measuring the weight of a moving vehicle may be a challenge, but it is possible - Georgia's commercial vehicle weigh stations have "Maintain 35MPH" signs on the lane(s) leading to the load cells.

Not every vendor gets it right initially: the "weigh in motion" scales in Virginia went through a long period of testing/redesign after they were installed a few years ago.

Measuring the toe in/out of a moving vehicle is also possible. Some cities (Memphis, Tennessee, among others) have a vehicle inspection station that every vehicle must go through at least yearly. Checks include headlight aim & brightness, signal operation (turn & brake), braking equality (front/rear & side/side), toe in/out (a slow speed driveover), etc.

John

 

[BeanieBots]

Well, if you think it'll work and you've selected a sensor ... then what's the question?

I'm sure Beaniebots will enjoy helping you with this, it's beyond my time allocation.

It's simply beyond me!

I'd be tempted to take a step back and look at what is the actual issue.
I get the impression, it's tracking.
If yes, why is tracking important? (it causes excessive wear).
What causes wear? Friction. What causes friction? Force.

Maybe measure the sideways forces??

 

[manie]

Papa: That is the general idea but only with alignment and weight on wheel indication. The weight for laden/unladen can maybe indicate worn suspension parts...

BB: Hope your time off/away was good... I have considered load cells (being calibrated) or strain gauges (must calibrate) but I do not know if the accuracy/resolution/repeatability will be good enough. The PV cell concept I like, but time constraints prevent proper development at this point. This leaves either load cells (force) or Non Contact Eddy Current/Inductive (displacement) sensors. The displacement sensors, even good quality Vernier (Starret/Fowler_Sylvac/Mitutoyo) all have IP_65 or IP_67 sealing against contamination ingress, so placing these inside a box will give good protection. In addition they all claim excellent resolution/accuracy etc.

Problem is getting a response from manufacturer/suppliers either from Germany or USA or UK. Times certainly cannot be hard there ?? I have e-mailed but as yet NO REPONSE.

Can anyone here maybe get me a price and shipping cost ? Look at my last attachment for the Eddy Current sensor from Epsilon Micro (UK/Ireland) if you can.

EDIT:
As for tracking, the steeper the gradient (rate of displacement) the worse the (mis)alignment. Displacement must be tracked to determine maximum point (drive-over system). So the rate can be incorporated easily then.

Thanks so far guys...

 

[manie]

BB: Its back to the lights show..... the inductive/Non contact sensors are just TOO EXPENSIVE ! Also the expected advantage of resolution/accuracy is not really there compared to your results using PV sensing. I've looked at the Quad/Doublet photo diodes you posted in the ultra-sonic thread. Now these and similar are available locally and at reasonable cost, including various external lenses that could be used for collimation/focussing for sharp light/shadow transitions on the sensors.

Although i explored the "other" options, I did think that I'll have to get back to the light sensors due to the cost vs accuracy issues (no real benefit or gain for a LOT more...)

 

[BeanieBots]

Naturally, the choice is yours.
Do be aware though, that the final accuracy of an analogue/optical sensor is down to components and construction.
Your flag might only expand by 1um with a 10C change in temp, but the optics could magnify that to 100um very easily.
Your op-amp might only have 1mV offset error, but a gain stage of 1000 will bring that up to a 1v error.
Even thermocouple effects at connectors/solder joints might only produce micro volts but very high gains (which you need) that can soon turn those errors into significance.

Just apply a good amount of common sense and all will be well.
Unlike digital where each stage cleans the signal again, with analogue, each stage introduces a new set of errors.

 

[manie]

BB: I'm sure we'll talk a lot on this still... in the meantime I've stumbled on these sensors from Hamamatsu, available locally and at reasonable price +- 35 Pounds(Sterling). I attach the datasheet. Please look if you have time and let me know, this one removes a lot of work that had to be done otherwise. I'll be looking to use the S3932-03 (12mm length), with +- 50uM error. The light mask will be Laser cut, the person there says he can cut down to .002mm slot width in plastic etc. Should be able to get down to the 0.2mm light spot required. Two of these, one for displacement left (R/H remains dark) and the the other for movement right (L/H remains dark), type of setup. Sounds OK ??

Please let me know what you think...

 

[BeanieBots]

The sensors look good and are of the type I had in mind.
Not sure I fully understand your arrangement though.
The design I had did not have a slot. It was a simple (blocking) flag as shown in my diagram.

One thing I noted on the datasheet was a rather long wavelength. Make sure you get the correct LED for illumination.

 

[manie]

BB: Thanks, it seems you require a 0.2mm light SPOT on the sensor. I mentioned the laser cutter says he can cut a slot as fine as .002mm, so cutting a HOLE to produce a small 0.2mm SPOT seems very possible. The LED looks like it should be a very warm white(reddish tint) or maybe even go straight for a red.

Also in one App.Note I read that both the Resolution and the Accuracy is inversely proportional to the amount of incident light. This means, contrary to normal instinct, one needs to go for dim light (softer illumination). Makes sense, as there should be less "spread" or bleeding of light around the spot periphery. The system uses the Op-amps so the OP-07's and 1% R's etc. will still be used. Will keep you posted on light-box construction etc.
Thanks for all the input.