Why won’t this work…or will it?

This topic has 29 replies, 14 voices, and was last updated 14 August 2013 at 18:10 by Mike Poole.

Viewing 25 posts - 1 through 25 (of 30 total)

1 2 →

Author

Posts
28 May 2013 at 14:34 #120846
John Hinkley
Participant
@johnhinkley26699
While making slow progress on my Harold Hall grinding rest and making extensive use of the DROs on my mill, I got to wondering about a less cumbersome means of measuring cutting tool measurement in three axes.

Now, it occurs to me that modern video game controllers and even my iPad must contain some form of motion sensors and/or accelerometers probably in some form of chip, in order to detect movement of the device and even my (cheap) digital angle gauge can measure to a knat's of a degree. So, why couldn't these chips be attached to a mill table to measure displacement in 3-dimensions, with optional remote readout by wire or wirelessly?

Can someone with more electronic knowledge than I have please explain why this couldn't be acheived? Are the sensors not sufficiently accurate? I can't think that they'd be too expensive.

If it wouldn't work, I'm happy to be told why…If it would, how brilliant that would be!

Over to you boffins…….

John
Advert

28 May 2013 at 14:34 #6728
John Hinkley
Participant
@johnhinkley26699
28 May 2013 at 14:40 #120848
Russell Eberhardt
Participant
@russelleberhardt48058
Unfortunately they use acceleration sensors rather than distance sensors. In principle, if you could measure the acceleration vs. time accurately enough and performed a double integral you could get a measure of distance but in practice the errors would be enormous.

Russell.
28 May 2013 at 15:59 #120850
John Hinkley
Participant
@johnhinkley26699
Russell,

Thanks for your prompt reply. I should have known it wouldn't work – or somebody would have done it by now! Oh well, back to the tried and tested method.

regards,

John
28 May 2013 at 16:11 #120851
jason udall
Participant
@jasonudall57142
Just think , want to resolve say a thou on 12 ".. thats one part in 12 000 …not imposible ( initial guidance platfoms achive that)….16 bit arithmatic resolves 1 in 64 000 ( yes I know 64 *1024 -1) and 16 bit is old hat ..

will have a nose and see what my android phone is capable off…

Shame you posted ..here you might have done your self out of a patent..
28 May 2013 at 16:57 #120856
Michael Horner
Participant
@michaelhorner54327
Hi John

Arn't you describing CNC?

In Mach3 You can type in your coordiates from the keyboard and the cutter will move to those points. The accuracy depends on how much you want to spend. A basic CNC machine is more accuate than I am on a manual machine.

Cheers Michael
28 May 2013 at 18:09 #120857
Another JohnS
Participant
@anotherjohns
I had a game-controller on my (LinuxCNC controlled) mill.

It did not "work" with me – I'm so used to turning knobs to machine that the game controller was confusing! I got a "mpg" rotary knob controller and all is fine for those times when I want to manually control machine movement.

Michael – As with Mach3, LinuxCNC can take keyboard entries and move the machine to specific locations. I'd think that any CNC system would also do that?

Another JohnS.
28 May 2013 at 18:40 #120859
jason udall
Participant
@jasonudall57142
I understood the op was suggesting the sensors as a alternative to scales /encoders
28 May 2013 at 19:54 #120864
Russell Eberhardt
Participant
@russelleberhardt48058
Inertial navigation systems are used on fighter aircraft and missiles they use acceleration sensors, integrating the acceleration to get velocity and integrating the velocity to get position. These systems respond quickly but need input from GPS to correct the position frequently.

It would only take a tiny error in the measurement and maths for the system to think it's moving when in fact it is not. A similar error will occur when moving at a constant velocity. The error would then accumulate over time. Even with frequent homing, the errors would be much too great.

A typical chip (ADXL362) has a resolution of 1 mg on the 2g scale. That's one part in 2000 and the error will be much more as the nonlinearity is up to 0.5%. Remember that's just the acceleration not the position.

Good luck if you're going to try it!

Russell.
28 May 2013 at 20:03 #120866
jason udall
Participant
@jasonudall57142
I think that thats the crux…in this application noise in the sensor channel would be of the same order of magnitude as the rate of change of position. .and after two integrations errors would rapidly exceed the signal..
Another reason why ..I doubt that this technique would not have been explored if plausible
28 May 2013 at 20:23 #120867
Michael Horner
Participant
@michaelhorner54327
Hi Jason

He does mention a digital angle gauge. Not sure why he would find a DRO cumbersome! Might be a pig to fit but then that's it. Surely? Is he after something like a mini GPS? If so I think to get the accuracy it would be out of the price league of normal model engineers.

Cheers Michael.
28 May 2013 at 20:45 #120871
John Hinkley
Participant
@johnhinkley26699
Oh dear! This is a bit like Topsy – it just growed and growed. I have no intention of making a device of the type which I described in my original post – I have insufficient time and zero ability in that direction! It was just a random thought that occurred to me while I was winding the mill table backwards and forwards for what seemed like hours.

Had it been a "goer" I imagine the result would have been very small and needed one or two screws to attach not to mention hundreds of pounds/euros/dollars cheaper than CNC, nice though that would be.

Thanks for all the replies – like I said – back to the drawing board.

John
29 May 2013 at 09:58 #120898
MICHAEL WILLIAMS
Participant
@michaelwilliams41215
Hi John ,

(1) Several ' backwards ' versions of more or less what you suggest have been devised .

These involve tracking devices on fixed parts of machine remotely detecting positions of a target on moving parts of machine .

(2) Most commonly these are laser devices but in more recent times digital cameras have been used to literally watch the target and calculate positions from changes in successive images .

Neither are commonly used on machine tools as such but are certainly used in scientific and experimental equipment .

(3) A different system uses an unpowered transponder on the moving parts and radio frequency detection from one or more fixed transmitter / detectors . Something like GPS but not quite the same .

(4) A problem with these systems is ' line of site ' – there is always a chance of something blocking the detection signals . Usually overcome by using multiple redundancy .

Regards ,

MikeW
29 May 2013 at 11:58 #120907
Anonymous
A simple calculation may be illuminating. Let us suppose that we want to move, and track, the milling table at a moderate 300mm/min, or 5mm/sec. Let us assume that we have a 5mm pitch leadscrew, and that we accelerate to 300mm/min in one turn of the handwheel over 1 second. So, over 5mm and 1 second we've gone from zero velocity to 300mm/min, or 5mm/sec. Acceleration is the change in velocity divided by the time. So, using metres, we get 0.005m/s/s. But 'g' is about 9.81m/s/s. So our acceleration is about half a milli-g. Which according to Russell is less than we can resolve anyway.

Regards,

Andrew
29 May 2013 at 12:26 #120909
jason udall
Participant
@jasonudall57142
Andrew…"So our acceleration is about half a milli-g. Which according to Russell is less than we can resolve anyway."…

or as we used to say "down in the grass"
12 August 2013 at 09:40 #126772
John Hinkley
Participant
@johnhinkley26699
I've had a few more thoughts on this topic, so I thought that I would resurrect it! (It would appear that I don't have enough to do!)

After extensive research over a period of minutes, I discovered that computer laser mice – as opposed to optical – can resolve very small movements, typically 2000 dpi or better. What's to stop one using the laser gubbins in these to detect the movement of lathe/mill slides? An additional benefit would be the ability to connect wirelessly to the display computer. I know that a chap on HMEM has done a similar arrangement with an obselete Android tablet for the display, but he still uses caliper-type scales, hard-wired to an Arduino card and thence via a bluetooth transmitter to the display. If it would work, my system could be significantly more compact and, I think, accurate enough for the sort of work I intend to carry out. An added advantage of using the laser would be that it detects movement by comparing successive pulses reflected from a surface – virtually any surface – so no graduated scales required?

Any thoughts, all you electronics wizards?

John
12 August 2013 at 09:55 #126774
jason udall
Participant
@jasonudall57142
Hmmm
12 August 2013 at 09:55 #126775
jason udall
Participant
@jasonudall57142
“Laser” mice do on occasion “stick” that said I have seen print scanners using calibration grids ” watched ” by something simular looking down into base from the scan head…
“

Edited By jason udall on 12/08/2013 09:59:50
12 August 2013 at 13:09 #126790
Trevor Wright
Participant
@trevorwright62541
The standard length of a metre is measured by so many wavelengths of light…….

Trevor
12 August 2013 at 13:17 #126792
blowlamp
Participant
@blowlamp
Maybe something like a mini GPS setup might work?

Martin.
12 August 2013 at 15:54 #126803
John Hinkley
Participant
@johnhinkley26699
Trevor – When I was at school – admittedly a long time ago when rods, poles and perches were still on the back of our exercise books – the wavelength of light varied, depending on the colour of the light. Good job, too, otherwise we wouldn't have had the wireless to listen to!

I did think that I might get shot down in flames with this idea, but it looks like it could be worth pursuing. I feel an ebay mouse purchase coming on!

John
12 August 2013 at 16:01 #126804
Michael Gilligan
Participant
@michaelgilligan61133
Posted by Trevor Wright on 12/08/2013 13:09:03:

The standard length of a metre is measured by so many wavelengths of light…….

Trevor

.

Not quite true, Tevor

… see here

MichaelG.
12 August 2013 at 16:56 #126808
Billy Mills
Participant
@billymills
Method 2 IS optical wavelenths Michael. For a very long time the metre WAS defined by optical wavelenths. Time measurements present the problem of the variation in velocity of the media, they only work without secondary effects in a vacuum. However there is no need for national standards accuracy on a home machine tool. For a home machine glass scale DRO's work very well, are not that expensive and cope with life quite well.

A cheap way of getting distance readings is to use a modified digital caliper or the ready made digital scales. Or you could do what people did BIC ( before integrated circuits) and use a dial gauge or two.

Billy.
12 August 2013 at 17:30 #126812
Michael Gilligan
Participant
@michaelgilligan61133
Billy,

No argument with the practicalities, nor with the history … but the fact remains that the metre is now effectively defined in terms of time.

MichaelG.

P.S. My apologies; Trevor said measured, he did not say defined.
13 August 2013 at 20:55 #126892
Sub Mandrel
Participant
@submandrel
Fascinating choice to use the time it takes light to cover a known distance. I'm surprised physicists would choose a standard they could never measure with absolute accuracy (due to the Heisenberg uncertainty principle), although I would concede it will have higher resolution than using optical methods.

Neil
Author

Posts