Capture image at 1000 meters/min
Discussion
I have an industrial application where i need to capture a sharp image at up to 1000 meters per minute.
Think it's possible??
Seconds / mm travelled
0.1 / 1666.66667
0.01 / 166.66667
0.001 / 16.66667
0.0001 / 1.66667
0.00001 / 0.16667
0.000001 / 0.01667
0.0000001 / 0.00167
Think it's possible??
Seconds / mm travelled
0.1 / 1666.66667
0.01 / 166.66667
0.001 / 16.66667
0.0001 / 1.66667
0.00001 / 0.16667
0.000001 / 0.01667
0.0000001 / 0.00167
Edited by Glade on Thursday 20th December 09:14
Edited by Glade on Thursday 20th December 14:25
I think (Ok am speculating but what the hay) you can achieve the same effect as a massively short shutter speed with flash.
So shutter speed of 1/250th second but lit with flash, the flash is not going to last 1/250th but is more in the 1/4000th range. If the flash is the primary source of illumination then this will be all that is recorded and hence you get a sharp image. I recall this is how they do the bullet through a balloon type of image.
What you need to work out is;
a) the flash power that will give you the length of time of illumination you need,a nd
b) how the smeg you are going to fire that at the absolute instant to get the image framed correctly
I suspect you'll have to go with a darkened room, shutter open and then a carrying device cutting a laser beam and hence triggering your flash but am sure it will take some experimentation to work out exactly where the beam needs to be so that when the flash fires all is correctly aligned.
Cheers
Chris
So shutter speed of 1/250th second but lit with flash, the flash is not going to last 1/250th but is more in the 1/4000th range. If the flash is the primary source of illumination then this will be all that is recorded and hence you get a sharp image. I recall this is how they do the bullet through a balloon type of image.
What you need to work out is;
a) the flash power that will give you the length of time of illumination you need,a nd
b) how the smeg you are going to fire that at the absolute instant to get the image framed correctly
I suspect you'll have to go with a darkened room, shutter open and then a carrying device cutting a laser beam and hence triggering your flash but am sure it will take some experimentation to work out exactly where the beam needs to be so that when the flash fires all is correctly aligned.
Cheers
Chris
DIW35 said:
Or you could try panning! 
I'm sorry but that was class!ETA: I take it this isn't a piece of industrial plant that spins up less than instantly? I.e. you can't shoot as the machinery gets into its stride and is running at lower speeds? Other than that I've got no clue how to do this...
Edited by TheStoat on Thursday 20th December 11:25
Let me give more details of the application:
Wire is being manufactured on a production line. in this project sizes 0.5 to 10mm diameters. Currently gross faults are detected with laser gauges. Lumps and Necks in the insulation.
We currently cannot detect defects such as surface roughness when the machine is running. If we run too fast the surface of the wire can be "as rough as a bear's nether reigons" (to quote some official documentation we have here).
There are methods of shining laser or light at a surface and dtermining the surface roughness by analysing the light reflected back.
I want to develop a non contact method for wire at 1000mpm, but if cameras can't get a good image at that speed we might be buggered before we start.
Wire is being manufactured on a production line. in this project sizes 0.5 to 10mm diameters. Currently gross faults are detected with laser gauges. Lumps and Necks in the insulation.
We currently cannot detect defects such as surface roughness when the machine is running. If we run too fast the surface of the wire can be "as rough as a bear's nether reigons" (to quote some official documentation we have here).
There are methods of shining laser or light at a surface and dtermining the surface roughness by analysing the light reflected back.
I want to develop a non contact method for wire at 1000mpm, but if cameras can't get a good image at that speed we might be buggered before we start.
Yep, the equipment for high speed film will easily cope with this particular speed but it costs a bit for the initial camera and the film stacks up very quickly as you'd imagine. I'd want to see example film too, just to make sure the achievable quality provides sufficiently sharp images to examine the wire's surface. I've no idea of the running costs of such equipment though, maybe someone else here has prior experience because that could be prohibitively high to start with?
As a super lazy Google example here's a datasheet for a US-based offering that looks fit to rig up and use remotely (no price mentioned though) http://www.fastecimaging.com/pdf/PDF/Fastec_InLine... These guys have stuff going up to 16K frames per second (obviously dependant on proper lighting)...
As a super lazy Google example here's a datasheet for a US-based offering that looks fit to rig up and use remotely (no price mentioned though) http://www.fastecimaging.com/pdf/PDF/Fastec_InLine... These guys have stuff going up to 16K frames per second (obviously dependant on proper lighting)...
A flash based system can, as has been demonstrated on many occasions, capture a clear clean image of a bullet in flight. 1000meter/min is only 37mph so we're not talking extremes. Provided you can set up a point on the production line where ambient light can be kept to a miniumum (or preferably excluded), I believe a flash based system would work to an acceptable level when combined with a decent resolution/quality digi cam.
You need to remember that with such a system it is the duration of the flash which dictates the image - not the shutter speed, hence the need to keep other light out!.
Failing that, have you tried accoustics - I'd lay money on there being something to listen to as the wire is manufactured which can be directly related to the surface finish. This in turn can be hooked up to an alert system.
You need to remember that with such a system it is the duration of the flash which dictates the image - not the shutter speed, hence the need to keep other light out!.
Failing that, have you tried accoustics - I'd lay money on there being something to listen to as the wire is manufactured which can be directly related to the surface finish. This in turn can be hooked up to an alert system.
Analysed by Machine.
THis needs to be a live system, though refresh rate doesn't need to be high for this application e.g. capture image and check for fault every 10 seconds. (our laser gauges scan constantly for discrete faults, this type of fault would be continuous through the length of the wire)
An image is taken by camera or sensor, computer analyses image and quantifys surface roughness by some method, it then makes a go/no-go decision and outputs a signal to Computer Aided Manufacture system. If there is a defect the operator is alerted so he can take action. Benefit is we can push to run as fast as possible without faults.
I have found papers online for this reflected light method for flat surfaces, its could be another issue making it work on a curved surface. And wether you could calibrate a strobe to freeze the image and act as the correct light source.
Its interesting that its the flash speed that governs the image not the shutter speed. Let me check my understanding... the light freezes the image and thats all that gets recorded. hence no ambient light because that would reflect off the object and be recorded for the whole duration that the shutter is open and give ablurred image. (we actually use strobes to check inkjet marks at high speed)
THis needs to be a live system, though refresh rate doesn't need to be high for this application e.g. capture image and check for fault every 10 seconds. (our laser gauges scan constantly for discrete faults, this type of fault would be continuous through the length of the wire)
An image is taken by camera or sensor, computer analyses image and quantifys surface roughness by some method, it then makes a go/no-go decision and outputs a signal to Computer Aided Manufacture system. If there is a defect the operator is alerted so he can take action. Benefit is we can push to run as fast as possible without faults.
I have found papers online for this reflected light method for flat surfaces, its could be another issue making it work on a curved surface. And wether you could calibrate a strobe to freeze the image and act as the correct light source.
Its interesting that its the flash speed that governs the image not the shutter speed. Let me check my understanding... the light freezes the image and thats all that gets recorded. hence no ambient light because that would reflect off the object and be recorded for the whole duration that the shutter is open and give ablurred image. (we actually use strobes to check inkjet marks at high speed)
Edited by Glade on Thursday 20th December 13:26
How do other wire manufacturers measure / QC their output ? DO they use the kind of system you are talking about or do they simply run a batch of wire and hope all is OK?
As I thought and forgive me if you have already thought of this or if I'm heading off down a dead end (I'm not a manufacturing expert).
It sounds like the capital set up cost could be large, maybe a different way of coping would be to run a number of tests to determine at what speed the deformation in the surface layer happens and then run your machines at a bit under that speed, say at 90-95%. Whilst you might not be as efficient as possible in terms of speed and you will ahve the occasional batch that goes to waste you do save yourself the capital set up and constant monitoring costs going forward.
Chris
As I thought and forgive me if you have already thought of this or if I'm heading off down a dead end (I'm not a manufacturing expert
).It sounds like the capital set up cost could be large, maybe a different way of coping would be to run a number of tests to determine at what speed the deformation in the surface layer happens and then run your machines at a bit under that speed, say at 90-95%. Whilst you might not be as efficient as possible in terms of speed and you will ahve the occasional batch that goes to waste you do save yourself the capital set up and constant monitoring costs going forward.
Chris
Do you need to see the whole length of wire, or is it enough just to take samples? If it's just samples you need, the idea of panning it might be worth another look. I'm thinking in terms of a spinning mirror synced with the shutter with some post-processing to correct for the distortion.
CVP said:
How do other wire manufacturers measure / QC their output ? DO they use the kind of system you are talking about or do they simply run a batch of wire and hope all is OK?
As I thought and forgive me if you have already thought of this or if I'm heading off down a dead end (I'm not a manufacturing expert).
It sounds like the capital set up cost could be large, maybe a different way of coping would be to run a number of tests to determine at what speed the deformation in the surface layer happens and then run your machines at a bit under that speed, say at 90-95%. Whilst you might not be as efficient as possible in terms of speed and you will ahve the occasional batch that goes to waste you do save yourself the capital set up and constant monitoring costs going forward.
Chris
I don't know if anyone else actually checks for surface roughness in-line and i can only find commercially availible vision systems that look for defects by knowing where and what to look for - some are very advanced but don't do this particular job! I have also found shutter speed in microseconds but having no experience with vision systems so far can't tell if this is up to the job. As I thought and forgive me if you have already thought of this or if I'm heading off down a dead end (I'm not a manufacturing expert
).It sounds like the capital set up cost could be large, maybe a different way of coping would be to run a number of tests to determine at what speed the deformation in the surface layer happens and then run your machines at a bit under that speed, say at 90-95%. Whilst you might not be as efficient as possible in terms of speed and you will ahve the occasional batch that goes to waste you do save yourself the capital set up and constant monitoring costs going forward.
Chris
We currently feel the wire to check for surface finish but this is frankly a crap method. Operators tend to slow the line down just to be on the safe side, because they'll get a bollocking if they run 40km of rough wire. However to get cost down we need to run as fast as possible!!
What you have described, running at 90% is basically what we do now... but of course Its not quite as easy as setting a max speed and then running just below: variation in matieral, tooling etc means that one day it will run fine at high speed with high pressures and the next it won't... the REAL solution is to get all the inputs to the process under control and the output will be fine, unfortunately its not as easy as it sounds. The sceince of mixing the polymer when it is manufactured and the properties we give it in the barrel are very complex.
Because of the wide range of application for a non-contact, high speed surface roughness detection system in our factories worldwide there is a pot of money there to develop the technology. Its an interesting project!
GreenV8S said:
Do you need to see the whole length of wire, or is it enough just to take samples? If it's just samples you need, the idea of panning it might be worth another look. I'm thinking in terms of a spinning mirror synced with the shutter with some post-processing to correct for the distortion.
Yeah just samples. We don't need 100% inspection. Mirrors, might be getting a bit too complex! But i'll look into it.
I think We'll have to work on the calculation for analysing a standard image before we go that far!
Glade said:
I don't know if anyone else actually checks for surface roughness in-line and i can only find commercially availible vision systems that look for defects by knowing where and what to look for - some are very advanced but don't do this particular job! I have also found shutter speed in microseconds but having no experience with vision systems so far can't tell if this is up to the job.
We currently feel the wire to check for surface finish but this is frankly a crap method. Operators tend to slow the line down just to be on the safe side, because they'll get a bollocking if they run 40km of rough wire. However to get cost down we need to run as fast as possible!!
What you have described, running at 90% is basically what we do now... but of course Its not quite as easy as setting a max speed and then running just below: variation in matieral, tooling etc means that one day it will run fine at high speed with high pressures and the next it won't... the REAL solution is to get all the inputs to the process under control and the output will be fine, unfortunately its not as easy as it sounds. The sceince of mixing the polymer when it is manufactured and the properties we give it in the barrel are very complex.
Because of the wide range of application for a non-contact, high speed surface roughness detection system in our factories worldwide there is a pot of money there to develop the technology. Its an interesting project!
Good points, the science of manufacturing is so complex. I think you're absolutely right, getting the inputs right is the key to sorting out your issues. I hate to say this but it sounds like it's ripe for a Six Sigma type approach on the input side of your process. Yes I'm sure you'll need a detection system that detects errors when they happen i.e. rough wire is being produced but that is after the fact, get the inputs right an your error rate goes right down to start with.We currently feel the wire to check for surface finish but this is frankly a crap method. Operators tend to slow the line down just to be on the safe side, because they'll get a bollocking if they run 40km of rough wire. However to get cost down we need to run as fast as possible!!
What you have described, running at 90% is basically what we do now... but of course Its not quite as easy as setting a max speed and then running just below: variation in matieral, tooling etc means that one day it will run fine at high speed with high pressures and the next it won't... the REAL solution is to get all the inputs to the process under control and the output will be fine, unfortunately its not as easy as it sounds. The sceince of mixing the polymer when it is manufactured and the properties we give it in the barrel are very complex.
Because of the wide range of application for a non-contact, high speed surface roughness detection system in our factories worldwide there is a pot of money there to develop the technology. Its an interesting project!
Maybe instead of an optical system for QC for the output you could use some kind of pressure test for a device sitting on the wire as it comes out of the production process. A kind of mechanical substitute for the operators fingers. This could detect any roughness in the wire being produced. I would have thought this might be an easier approach.
Chris
CVP said:
Glade said:
...
Good points, the science of manufacturing is so complex. I think you're absolutely right, getting the inputs right is the key to sorting out your issues. I hate to say this but it sounds like it's ripe for a Six Sigma type approach on the input side of your process. Yes I'm sure you'll need a detection system that detects errors when they happen i.e. rough wire is being produced but that is after the fact, get the inputs right an your error rate goes right down to start with.Maybe instead of an optical system for QC for the output you could use some kind of pressure test for a device sitting on the wire as it comes out of the production process. A kind of mechanical substitute for the operators fingers. This could detect any roughness in the wire being produced. I would have thought this might be an easier approach.
Chris
I used to press metal parts and assemble them, thats EASY compared to getting a highly filled polymer compound mixed and uniform, then spreading it in a very thin layer and ensure that it passes Mil spec and space agency testing!!
Interesting project! Can't help but think that the biggest issue in an optical system is the image processing required (having worked with a load of image processing spods for the last decade!). You will help yourself considerably by lighting the cable properly to highlight any imperfections (ie cross lighting to put shadows into any depressions in the surface). Or could you pigment the other suraface so differences in colour would easily picked up?
Can help but think there must be an easier way though. Maybe pass an HT voltage through the cable and measure any leakage (or change in resistance) to a cathode somewhere down the line. Or maybe a system that induces a current in the cable (it is moving after all)and looks for any fluctuations in the inductance of the cable (thickness of insulation). Failing that the accustic system seems good - a similar approach is used for robot end effectors (hands). Basically a high sensitivity microphone is built into the grip of the finger to listen for the sound of the object slipping. A similar (contact based) system could be wrapped round the cable, with the added advantage is that you could listen for the specific frequency relating to the speed of flow.
Will keep thinking...
Can help but think there must be an easier way though. Maybe pass an HT voltage through the cable and measure any leakage (or change in resistance) to a cathode somewhere down the line. Or maybe a system that induces a current in the cable (it is moving after all)and looks for any fluctuations in the inductance of the cable (thickness of insulation). Failing that the accustic system seems good - a similar approach is used for robot end effectors (hands). Basically a high sensitivity microphone is built into the grip of the finger to listen for the sound of the object slipping. A similar (contact based) system could be wrapped round the cable, with the added advantage is that you could listen for the specific frequency relating to the speed of flow.
Will keep thinking...
There was a doctorate thesis written by someone I knew at Cardiff Uni, not even sure if he finished it because he got to caught up in trying to make a business of it.
He's what i remember, its a few years back, not more than 5.
The task was to inspect rubber grommets during manufacture for imperfections, as they were spun off the moulds, which ran at some incredible speed, he had a LED source which flickered at some rate or other and then detectors picked up the reflections, analysis was made on the results by microprocessor, the turnout rate was huge.
Nigel, last time I saw him was working for a company in the Bristol area, who bought him and his system.
If this of use I'll try to remember more.
He's what i remember, its a few years back, not more than 5.
The task was to inspect rubber grommets during manufacture for imperfections, as they were spun off the moulds, which ran at some incredible speed, he had a LED source which flickered at some rate or other and then detectors picked up the reflections, analysis was made on the results by microprocessor, the turnout rate was huge.
Nigel, last time I saw him was working for a company in the Bristol area, who bought him and his system.
If this of use I'll try to remember more.
Gassing Station | Photography & Video | Top of Page | What's New | My Stuff