What's new
What's new

Need ideas for in-process measurement of spring.

KilrB

Stainless
Joined
Feb 18, 2007
Location
Angleton, Texas
Where I work we manufacture dozens of different sizes of these:

IMG_0042.jpg IMG_0043.jpg

They are done in progressive dies and should continuously coil-to-coil but ...

Current inspection procedure requires us to stop every 100', cut a section out, and inspect it in a video inspection system.

We are looking for an in-process solution that will allow us to achieve 100%, or near 100%, inspection and eliminate the current down-time and waste.

This means obtaining measurements of the height, width, slot width, leg width, web width, slot progression, inside radius, and any difference in leg lengths ... while running @ 60-90 strokes per minute.

This system needs to tie into the press and shut it down upon detecting any out-of-tolerance readings or mis-feeds.

So far we have had tried a system from Keyence and spoken with a couple of other companies, who's names I was not told, with no success.

One of the major problems (I am told) is that the material is too reflective, scattering the light/laser beam and causing erroneous readings and shutting the press down constantly.

I'm hoping someone here knows of a company that has a viable solution to this.
 
Did you try searching "Vision Inspection Systems"? I looked at the Cognex offering years ago and was impressed with what it could do under white light. But your product speed could be a big difference.

Perhaps the camera needs a FILTER to supress the noise.

Give Omega Optical a call 802 254 2690 for filter assistance. They do not do inspection systems.
 
"material is too reflective, scattering the light/laser beam and causing erroneous readings and shutting the press down constantly."


Would it be possible to surface coat it in the inspection area and then clean after inspection if cleaning is needed.
If you want to inspect it on the fly(moving) you should be able to coat it on the fly also.
Perhaps simple spray paint? or water soluble paint for easy cleaning? ammonia base for very fast drying?
argon or carbon dioxide based for fast drying? perhaps even the pigment could evaporate or not be a pigment but a powder like soot..
 
Thank you gentlemen.

Not sure if the Keyence system is properly termed a "Vision Inspection System" but I would call it that.

Will have to ask about the filtering ...

Coating the part is a no-go, at this point at least.

Not sure what the extra length required for the coating and cleaning operations would be, but would likely make "pushing" the lighter/thinner product on through near impossible ... and you can't pull it after it's slotted.

The samples in the photographs are representative and not our smallest or largest sizes.

I will be contacting Mr. Guru for sure.

Any other suggestions/insights are certainly welcome.
 
Dozens of sizes to deal with and 8+ dimensions to handle, oh boy, what fun.
Not a off the shelf system.
The reflection/glare/hot spots can be handled in the optics end but this does not come cheap.
Dare I ask a budget for solving this?
Speed is somewhat fast. Not sure how long the part is actually stopped for.

Unless you have somebody experienced you are going to need to go to an outside vision house.
There are a few problems that will not be easy here.

Even on a good day with bright spots not a problem the Keyence system could give you all these dimensions on a 100% check in real time?

If using a actual camera this place is a good starting spot for information on polarizers and filters.
http://www.edmundoptics.com/technical-resources-center/illumination/choose-the-correct-illumination/
In the machine vision world it's the lighting and lens that makes or breaks the system in production.

Bob
 
"material is too reflective, scattering the light/laser beam and causing erroneous readings and shutting the press down constantly."


Would it be possible to surface coat it in the inspection area and then clean after inspection if cleaning is needed.
If you want to inspect it on the fly(moving) you should be able to coat it on the fly also.
Perhaps simple spray paint? or water soluble paint for easy cleaning? ammonia base for very fast drying?
argon or carbon dioxide based for fast drying? perhaps even the pigment could evaporate or not be a pigment but a powder like soot..

Actually, the antireflective film that would suppress the scatter could as well be a temporary OIL film. Refractive index controlled oils are readily obtained. Easily applied by spray or wipe, and could be reclainmed and reused.

A thin liquid film of just about anything could change the scatter into specular reflection and be sent away from the recording camera and out of the system.

Might be an option,

"It's all signal to noise boys, it's all signal to noise"
 
There's probably a Pareto-like range of defects that you commonly see. As a cheaper alternative to a custom vision or laser system -- is it possible that the top two or three failure modes could be fool-proofed at the source (at the die, between dies, etc.)?? Start with an analysis of the cause of rejects, not the inspection system?

Even if you 100% inspect them, you still don't want to be making scrap.

It also looks like a collimated beam shot through the serrations and then picked up on the other side (think square wave display on a scope as the coils pass by at constant speed) might catch most of the common defectives in process??
 
In the old days, inverting the video (dark becomes light, and light+glare becomes just black) would take care of challenging lighting conditions, etc. Lasers may be overkill. Doing the majority of QC while items are flat should handle most different sizes, etc. on a nearly continuous basis. QC after the final folding/roundover step gets more challenging... but you can still do leg-length comparisons along with item height and width... perhaps on a contact/wiper basis rather than optically.

Based on history, where in the process are the most problems?

Chip
 
As a tool maker my first mode of attack would be careful examination of your tooling and setup. I used to maintain progressive dies making small leaf springs that would make 300k-350k parts per run of hardened spring steel. Approx 300 SPM, they ran coil to coil, holding fairly close tolerances ( probably nowhere near your part)

All the sensors in the world won't help you make a good part, just let you know when you haven't.
 
There are no defects in the flat state.

Unless a punch or die breaks, or slugs, any problems occur during forming.

There are not really all that many defects, but with our current manner of inspection it's still possible to run 100' of scrap between checks.

This, the down time during inspection, and customer demand are what have led to this push for 100% in-process inspection.
 
. . . it's still possible to run 100' of scrap between checks. . .

I still think that a collimated beam shone through the side in the folded state might immediately pick up many common forming defects within a foot or so? The interrupted light would generate a sort of square wave, which could continuously monitored for width, slot width, slot progression, etc. Any dangling slugs, missed punches, defects in spacing, crimps between die punches, etc. would likely show up. It wouldn't pick up folded height (but another cheap sensor might do that).

Point is you may only need to check a couple dimensions in-process to catch the vast majority of problems. "Inside radius" seems the toughest dimension to check cheaply; but I'd guess if the radius is off, the folded height might be as well?
 
I still think that a collimated beam shone through the side in the folded state might immediately pick up many common forming defects within a foot or so? The interrupted light would generate a sort of square wave, which could continuously monitored for width, slot width, slot progression, etc. Any dangling slugs, missed punches, defects in spacing, crimps between die punches, etc. would likely show up. It wouldn't pick up folded height (but another cheap sensor might do that).

Point is you may only need to check a couple dimensions in-process to catch the vast majority of problems. "Inside radius" seems the toughest dimension to check cheaply; but I'd guess if the radius is off, the folded height might be as well?

Exactly what the current system we are testing is supposed to do.

It uses a pair of beams, one horizontal and one vertical. If not for the recurring erroneous readings shutting the press off it would be acceptable at this point.

I'm still not sold on this system precisely because it cannot measure the inside radius.

It also has to be carefully timed to the press stroke/material feed cycle.

Supposedly the manufacturer of this system is working on the problem(s) and I am told we are talking to another manufacturer too.

I have also been told that we talked to one maker of a laser system that was supposed to be able to take continuous reading even as the material was progressing, but were told most of our parts were too small!?!? C'mon ... it's a LASER!

At least that's what I was told.

I'm only a bit player in this project, and I (obviously) don't know a lot about it, but I know some guys on the internet ...

Thank you gentlemen, I really appreciate you sharing what you know with me.
 
We have integrated all manner of Cognex and laser displacement sensors for process measurement and experienced our share of failures and enough success to keep us in the game. For this application I wonder if a simple mechanical measurement approach might be better.

I would make a precision instrument roll stand at the output of the press with a series of 2-3 radiused bearing mounted wheels all aligned along the path of the progressing spring. Smallest radius wheel first and largest radius wheel last. The spring being stamped would ride on the wheels with spring loaded cam followers on top (or something like a bellofram cylinder), each with a Heidenhain plunger scale that you could allow to drop onto the cam follower mounting boss whenever you wanted to take a measurement.

The radiused wheels would be machined to be slightly smaller and slightly larger than spec. Measure the plunger scale positions with an in-spec spring to get your baseline measurements and check them periodically while running by extending the plungers when a reading is desired.

This would be more than adequate for verifying radius as the mechanical gain of an out of spec radius is more than 1:1 and should be easily detected.

Two more plunger mics coming in from the sides adjacent to one of the wheels and two more coming up from the bottom. You can combine the data from all three sets of Micrometers easily enough and with small THK linear bearings for guiding the actual spring contact surfaces, it would be relatively easy to build a robust solution. Plunger mics are good to 1/20th of a micron and as I recall are relatively cheap at about $400 a piece.

Should be able to put a decent system together for under $15k-$20k
 
wow ... I'm going to have to think on that one just to wrap my mind around it.

I've read your reply several times today, as time permitted, and I'm finally starting to "see" it.

It sounds perfectly feasible to me and I can see it being the route we will have to go, if we cannot get an optical system of some kind to work.

I say this because I do not believe our CEO wants to invest the time it would take to make an inspection set-up like this for 12 presses, or the several dozen sets of radiused wheels it would require for the different sizes of spring.

This is getting way out of my league here but, how easy would it be to get a signal from this set-up to shut off the press in the event of an out-of-tolerance condition?

At any rate, something has got to be done sooner rather than later and your suggestion Mr. Guru is going to be presented to the people in charge of this project.

Thank you sir.

Sorry it took so long to respond but it's been a very busy day here.
 
One of the beauties of this kind of inline checking is that you can collect data with which to create X-Bar and R charts. Using these, you can chart how the process is running real time and detect trends before problems occur. This is one of tools used in 6-sigma quality control.

Tom
 
The bearing mounted radiused wheels can be dropped into precision sockets . . . just change them when you set up for another run. The rider cams on top can be the same.

As far as monitoring to shut down the system - this can be done a number of ways - what is your plant standard for automation? Rockwell? Siemens? No standard?

A few examples of measuring systems using this plunger scale from the Heidenhain website.

Anwendung_Welle-hi.jpg


image001.jpg


Generally, the plungers are spring applied with an internal spring and pneumatically retracted in applications like this. We have done fixtures where a cam follower or precision V-block is mounted to a small linear bearing and this is brought into contact measurement with the product being measured using a low friction actuator - then the plunger mic is extended to a precision landing point on the same block for the measurement function and then both are retracted.

In your case, if the contact points with your stamped product are in constant contact with a rolling element (cam follower for instance) and the cam follower is held in position with a light controllable force . . . and the plunger micrometer was extended to measure the cam follower mounting block position on a regular basis - this would allow spot measurement every XX stamping progressions.

I don't know how much the product moves around or if you could dampen out the shock loading from each progressive die stroke - but something like this could work.
 
Last edited:
One of the beauties of this kind of inline checking is that you can collect data with which to create X-Bar and R charts. Using these, you can chart how the process is running real time and detect trends before problems occur.
That's the trick.

Recognizing the fact that a problem is developing allows you to correct it before you start making bad parts.

- Leigh
 
Ah ha! I begin to see more clearly now.

As far as part movement and die impact goes ...

On the thin materials (.003"-.006") you would hardly know there was anything happening. There's no "impact". But the material is so flimsy that just looking at it hard will cause it to deform and move.

On the "heavy" side (.025"-.036") there can be quite an impact. However, though you can still move the spring fairly easily it doesn't deform without a lot more effort.

Thank you once again gentlemen.

Edit:

Oh, our plant standard for automation is ... what automation?

The servo feeders for the presses are the only "automation" I can think of around here, and I don't know what they or the presses have inside them.
 








 
Back
Top