In process grinding measurement - Page 2
Close
Login to Your Account
Page 2 of 3 FirstFirst 123 LastLast
Results 21 to 40 of 50
  1. #21
    Join Date
    Jan 2007
    Location
    Flushing/Flint, Michigan
    Posts
    7,225
    Post Thanks / Like
    Likes (Given)
    345
    Likes (Received)
    6086

    Default

    Quote Originally Posted by Robert R View Post
    Motion Guru has gone off and fixed the problem I guess.
    There is not much in the world of machine building Motion and his crew can't fix.
    It's nice when he throws us a bone as to what he is up to.

    One can do resin bond "porosity" wheels in CBN or diamond and they have a place.
    These wheels have lots of visible "holes" in the working surface. Not the same as a white wheel but not the same flat finish of a standard resin bond.
    Bob

  2. #22
    Join Date
    Dec 2003
    Country
    UNITED STATES
    State/Province
    Washington
    Posts
    4,575
    Post Thanks / Like
    Likes (Given)
    2070
    Likes (Received)
    3815

    Default

    “They are used when a mirror surface finish is needed. The goal in this instance is to retain the dull abrasive.”

    Robert has surfaced the fly in the ointment . . . it is a part that in addition to being tight tolerance, has certain aesthetic requirements.

    Marposs system is enroute and I have been working on counting data drops of CAT6 cable in the new building . . . if you told me we would need roughy 150 drops I would have told you were crazy, but it is what it is.

    We will have an update when the hardware shows up and we get it commissioned.

  3. #23
    Join Date
    Sep 2011
    Country
    GERMANY
    Posts
    2,437
    Post Thanks / Like
    Likes (Given)
    1438
    Likes (Received)
    816

    Default

    Quote Originally Posted by motion guru View Post
    I have been working on counting data drops of CAT6 cable in the new building.
    Off thread but I hope you are using CAT7 or CAT7a rather than CAT6.

  4. #24
    Join Date
    Aug 2005
    Location
    Raymond , CA
    Posts
    197
    Post Thanks / Like
    Likes (Given)
    0
    Likes (Received)
    130

    Default

    Quote Originally Posted by motion guru View Post
    it is a part that in addition to being tight tolerance, has certain aesthetic requirements.
    The CBN and the seeded gel ceramic grinding wheels recommended in the earlier posts are used for removing metal quickly and would be poor choices for producing polished surfaces.

    The old school approach for this kind of work is to use a conventional glass bonded aluminum oxide wheel with a coarse dress for grinding the part to size and then do a very slow wheel dress to dull the abrasive for producing the final finish. This approach might have a faster throughput than using a polishing wheel with a coarse dress to act as a grinding wheel.

    On the other hand a grinding wheel will produce large chips. If the filtration system of the grinder coolant or the wash down of the table and parts is less than perfect the polishing step will be damaged by loose abrasive and metal chips. I can see why the resin bonded wheel might be a better choice even if it means a lower throughput.

    The disk drive industry solved this problem a long time ago. The aluminum disks are ground to size with a mirror surface finish.

  5. Likes cash, michiganbuck liked this post
  6. #25
    Join Date
    Apr 2018
    Country
    UNITED KINGDOM
    Posts
    1,326
    Post Thanks / Like
    Likes (Given)
    0
    Likes (Received)
    628

    Default

    Quote Originally Posted by Robert R View Post
    The disk drive industry solved this problem a long time ago. The aluminum disks are ground to size with a mirror surface finish.
    Those are ground ? That's cool to know. I have seen machines for turning floppy disks - honkin' big blocks of granite, a diamond, and a whole lot of air bearings - so just assumed that hard disks were done the same way.

  7. #26
    Join Date
    Aug 2005
    Location
    Raymond , CA
    Posts
    197
    Post Thanks / Like
    Likes (Given)
    0
    Likes (Received)
    130

    Default

    Quote Originally Posted by EmanuelGoldstein View Post
    Those are ground ? That's cool to know. I have seen machines for turning floppy disks - honkin' big blocks of granite, a diamond, and a whole lot of air bearings - so just assumed that hard disks were done the same way.
    I chose a very poor example.

    I should have written that for material removal of less than .005 inches, a surface finish in the range of .05 microns and surface flatness in the range of 0.3 microns, a single side lapping process would be a better choice than a Blanchard grinding process. Reference volume 16 , Machining, ASM metals Handbook page 492., chapter on lapping. The point being that lapping can compete with grinding on a cost basis even when .005 inches of material are being removed and produce surface finishes and accuracy that cannot be obtained by grinding.

    My understanding is that the hard drive disc platters are punched from an annealed aluminum alloy sheet unspooled from a large roll. The blanks are then annealed again while being pressed flat to remove the cold work stress. There may or may not be a double sided lapping process as the next step depending on the depth of the surface defects. There then follows a cleaning-etching-chemical plating process that deposits a non magnetic tightly adhering hard film. The disks are then run through the double sided lapping process again, cleaned. and sent off to the disc drive manufacturers. The disc surfaces may then be textured and finally sputter coated with the magnetic media layers. A top layer of hard film and maybe a lubricant will also be applied.

  8. Likes michiganbuck liked this post
  9. #27
    Join Date
    Dec 2003
    Country
    UNITED STATES
    State/Province
    Washington
    Posts
    4,575
    Post Thanks / Like
    Likes (Given)
    2070
    Likes (Received)
    3815

    Default

    Ok, Marposs finally received, mounted, commissioned, and calibrated . . . first two batches of parts off the machine gave us the best accuracy and repeatability we have experienced to date with spot checking of parts revealing +/- 0.00015 in of target.

    More testing tomorrow and perhaps we can get this grinder back into useful production before the end of March.

    Major kudos to Frank from Marposs for spending two days with us to ensure everything was as close to perfect as it could be before we ever turned on the spindle.

  10. Likes michiganbuck liked this post
  11. #28
    Join Date
    Feb 2012
    Location
    California
    Posts
    1,288
    Post Thanks / Like
    Likes (Given)
    851
    Likes (Received)
    1376

    Default

    Quote Originally Posted by motion guru View Post
    Ok, Marposs finally received, mounted, commissioned, and calibrated . . . first two batches of parts off the machine gave us the best accuracy and repeatability we have experienced to date with spot checking of parts revealing +/- 0.00015 in of target.

    More testing tomorrow and perhaps we can get this grinder back into useful production before the end of March.

    Major kudos to Frank from Marposs for spending two days with us to ensure everything was as close to perfect as it could be before we ever turned on the spindle.
    Looks like something we might be interested in.

    Is this system integrated into your control or standalone? How does it address thermal growth?

    Glad to hear it's working out.

  12. #29
    Join Date
    Jun 2012
    Country
    UNITED STATES
    State/Province
    Michigan
    Posts
    9,523
    Post Thanks / Like
    Likes (Given)
    3242
    Likes (Received)
    3403

    Default

    Good that Marposs seems to have solve problem . .00015 is good IMHO


    Qt:[These wheels have lots of visible "holes" in the working surface. Not the same as a white wheel.... ] old school method to make visible "holes" in vitrified wheels was adding ground walnut shells to the mix before baking the wheels.

    Always wanted to make a highly porous clay wheel of about 220 grit for polish finish SG hard steel work.

    We had a high porous AO 46m we called the Popcorn wheel for SG chuck grinding..Think I used my last one..

  13. #30
    Join Date
    Dec 2003
    Country
    UNITED STATES
    State/Province
    Washington
    Posts
    4,575
    Post Thanks / Like
    Likes (Given)
    2070
    Likes (Received)
    3815

    Default

    The machine is on a plant wide coolant circulation system with stable temperature. Coolant on the wheel and chucks runs at around 30gpm so temperature stability is good once in production.

    A couple of setbacks.

    1.) The probe on the part leaves an ever so slight scratch as it drags across the part. The engineer on the project was able to query part thickness from the Marposs unit over Profibus and then wrote software to predict when to lift the sensor probe and then complete one more revolution of the chuck under the wheel before lifting off the parts. We will test tomorrow and see how this works.

    2.) The chuck reference probe also scratches the chuck. We designed a Delrin scrubbing block and are going to test that tomorrow as well. The probe would dig a trench in the chuck in no time without increasing the contact area. Now to make sure this doesn’t introduce yet another point of inconsistency.

    The Marposs system is completely embedded into the grinder control system. This 30 year old grinder has no shortage of technology deployed on it.

  14. Likes cash, johansen liked this post
  15. #31
    Join Date
    Aug 2005
    Location
    Raymond , CA
    Posts
    197
    Post Thanks / Like
    Likes (Given)
    0
    Likes (Received)
    130

    Default

    Quote Originally Posted by motion guru View Post

    2.) The chuck reference probe also scratches the chuck. We designed a Delrin scrubbing block and are going to test that tomorrow as well. The probe would dig a trench in the chuck in no time without increasing the contact area. Now to make sure this doesn’t introduce yet another point of inconsistency.
    Does Marposs have a LVDT sensor probe with a built in air sled?

    If not would it be possible to supply a metered air blast to the underside of the probe to allow it to float above the surface of the chuck. The idea is similar to what is used on the floating hard disc drive read/write heads. In this instance the air would be supplied externally rather than being a boundary layer dragged in by a disc spinning at 8000 rpm.

    It might also be realistic to replace the Marposs gage with a air gage head mounted on a servo driven stage. The servo control would be gated to monitor the air pressure when each work piece passed under the gage head. The servo would control the stage position once per revolution of the grinder turntable to maintain a constant air pressure. The servo position encoder then provides the part height .

    This would only work with polished part surfaces and a slowly rotating table. A air gage requires a surface roughness of less that 15 microinches to give accurate readings in the .0002 inch tolerance range and there would be a minimum part contact time to insure a stable pressure measurement .

    Another possibility is to use a capacitance sensor head to maintain the servo driven stage height. This would be a faster measurement capable of a much higher accuracy than a air gage but would still require a low work piece surface roughness.

    Something to think about for the next project.

  16. Likes johansen liked this post
  17. #32
    Join Date
    Jun 2012
    Country
    UNITED STATES
    State/Province
    Michigan
    Posts
    9,523
    Post Thanks / Like
    Likes (Given)
    3242
    Likes (Received)
    3403

    Default

    Qt: [The probe would dig a trench in the chuck in no time without increasing the contact area. Now to make sure this doesn’t introduce yet another p]

    agree no matter what the probe, the presence of abrasive makes any physical contact a problem.

    I once gauged a grinding job with sensing light reflection but never perfected it for other work. it was simple grinding a spot by plunge grinding, then withdraw and reciprocal grind to witness to reflection stop. That gave me +.001 /+.0015 to gauge spec, so x numbers of continual passes could give + .0005 (in spec). Yes, just mechanical and eyeball so never perfected a system that could be used in automation.

    Yes just an old grinding trick an old timer taught me..Wait now I am an old timer, Darn

    Sensing to <.0002 is tough.

  18. #33
    Join Date
    Dec 2003
    Country
    UNITED STATES
    State/Province
    Washington
    Posts
    4,575
    Post Thanks / Like
    Likes (Given)
    2070
    Likes (Received)
    3815

    Default

    Problem #1 solved simply by lifting the sensor and going one more rev under the wheel.

    The idea of an air gauge mounted on a servo stage . . . not going to happen. After $30k plus in the Marposs unit plus engineering, mount, software development . . . and the aforementioned 30 GPM of coolant mixed with grinding swarf, no air gauge would stand a chance.

    Monday we will run a dozen batches and see how things look. The Marposs seems fairly robust with a resolution of about 0.00005 inches (yes, I got the decimal place in the right spot - 1/2 of a tenth) . . . pretty amazing really.

  19. #34
    Join Date
    Jan 2007
    Location
    Flushing/Flint, Michigan
    Posts
    7,225
    Post Thanks / Like
    Likes (Given)
    345
    Likes (Received)
    6086

    Default

    One grind rev solved the scratch problem?
    Is that fact talking to you? Can it scream any louder?
    Why, why, why.. do I have to this five times?

    It works and all is good. Knowing when and if it will fail is also good.
    Wheel dress new, old and tired wheel?
    Many variables all hard to find in a machine build but one does not want to get with oops in the field a few thousand parts in.
    Will this one turn still work at a 5000 piece part count?

    Perhaps all wrong thinking but I've seen so many automation builders bit by the "works here" bug and one million cycle warranties that I'm a bit gun shy.
    Bob

  20. #35
    Join Date
    Aug 2005
    Location
    Raymond , CA
    Posts
    197
    Post Thanks / Like
    Likes (Given)
    0
    Likes (Received)
    130

    Default

    There is a missing piece of information.
    If the coolant is being effectively filtered and if the Marposs gage head is located at the far end of the rotating table away from the grinding wheel why is there grinding swarf under the gage probe?

    Is the probe area being flushed with filtered coolant or is the coolant flow only directed to the grinding wheel?
    Has the filtered coolant been checked for retained particle size?
    My wild guess is that a polished surface requirement for the work piece would imply a 5 micron or smaller filter .

    Another option would be to place a air knife device in front of the probe tip to remove a large portion of the coolant suspended swarf before it comes in contact with the probe tip.

    The chemical mechanical polishing machines used in the semiconductor industry do have height sensors. (They did 20 years ago. I do not know what the current standard is.) From what I remember they had the appearance of 60 inch diameter lapping machines holding 8 inch diameter wafers. The sensor did not contact the silicon wafer.

  21. #36
    Join Date
    Dec 2003
    Country
    UNITED STATES
    State/Province
    Washington
    Posts
    4,575
    Post Thanks / Like
    Likes (Given)
    2070
    Likes (Received)
    3815

    Default

    Quote Originally Posted by CarbideBob View Post
    One grind rev solved the scratch problem?
    Is that fact talking to you? Can it scream any louder?
    Why, why, why.. do I have to this five times?

    It works and all is good. Knowing when and if it will fail is also good.
    Wheel dress new, old and tired wheel?
    Many variables all hard to find in a machine build but one does not want to get with oops in the field a few thousand parts in.
    Will this one turn still work at a 5000 piece part count?

    Perhaps all wrong thinking but I've seen so many automation builders bit by the "works here" bug and one million cycle warranties that I'm a bit gun shy.
    Bob
    Bob, I honestly have no idea what you are driving at.

    When I say "scratch" - I am talking barely perceptible line on a component that becomes part of an $800 - $1200 tool. To say that aesthetics are important is like saying the sun is bright. The steels we are grinding are Rc 62 and harder, the stylus was selected by Marposs with full knowledge of the application including in person review of installation and assistance during commissioning.

    marpossinstall.jpg

    Coolant is filtered to 5 micron - this facility has no fewer than 30 grinders operating on this system at any given time and this company does a top notch job of coolant filtering and conditioning. The simple fact of the matter is that swarf covers the fixture, and magnetic chuck and when rotating at 25+ rpm, we simply cannot guarantee that there will not be swarf interfering with an air gauge. They are holding +/- 0.0006 inches now, on this same grinder design and our task is to get this design to +/- 0.0003 inches and I think we are very close at this point.

    Wheel is new, auto dressed based on evaluation of spindle torque, down force, and down feed rate. If down force goes up, spindle torque drops, and down feed drops - we auto dress, touch off with an acoustical sensor and then get back to grinding. At present, a completely automated process.

    One more week of testing / refining and we will know a lot more (grinding hundreds of parts). Goal is FAT during the week of the 8th, ship the week of the 15th, and CAT the week of the 22nd - then we push into production and see how it goes. The machine has two chucks - process is generally defined as grind first side using only the acoustical sensor, flip parts and put on the finishing side with the Marposs and rough grind (defined as down feed to 70% spindle load) to within 0.001 of target, transition to finish grind (reduce down feed to 40% spindle load), then spark-out till Marposs says you are good, then one additional rev of the chuck and lift off the parts. (I don't know what the specific spindle loads are that trigger the transitions at this point, it has all been optimized by empirical testing over 1000's of parts and I haven't kept up with the settings.)

  22. #37
    Join Date
    Jun 2012
    Country
    UNITED STATES
    State/Province
    Michigan
    Posts
    9,523
    Post Thanks / Like
    Likes (Given)
    3242
    Likes (Received)
    3403

    Default

    Any chance that spotters made of carbide might be ground in to be the Marposs gauge off locators. perhaps something harder than carbide like diamond coated. Agree abrasive will even score carbide and diamond in time. Perhaps the spotter 2” wide so able to have a number of strike zones across the spotters to move over as spotter goes bad. Yes, the spotters hand fitted/gaged in to close. Perhaps a go/no go spotter step of perhaps .0004

    Still .0003 is tough on any grinder for a production part (any part).

    Perhaps a dress at + something and then evaluate the grind/take per cycle/rotation with the dress at a constant, perhaps spec + .001. Yes this might require a stop and dress at a spot and pass, so time consuming.

    *I know these ideas are likely old-hat to you and your crew, and Marposs has been around the block many times so little they don't know...but still worth my giving them.
    I wonder if you are using the roller dreseors common to Blanchards and the like?

  23. #38
    Join Date
    Aug 2005
    Location
    Raymond , CA
    Posts
    197
    Post Thanks / Like
    Likes (Given)
    0
    Likes (Received)
    130

    Default

    The photograph shows the chuck magnetic field pattern displayed on the top of the work piece fixture plate. The magnetic field is not being completely shunted by the steel part and is extending beyond the work piece surface.
    The leaking magnetic flux lines between the poles will trap the metallic swarf. Would there be a reduction in probe wear if the sensor probe was moved over 1/8 inch so that it rides directly over the steel ring pole piece rather than the brass spacer ring?

  24. Likes adama liked this post
  25. #39
    Join Date
    Jun 2012
    Country
    UNITED STATES
    State/Province
    Michigan
    Posts
    9,523
    Post Thanks / Like
    Likes (Given)
    3242
    Likes (Received)
    3403

    Default

    Spotters might be non-magnetic on top sides so to avoid holding swarff and having an air or coolant blast, wiper or brush. to clear gauging.
    Last edited by michiganbuck; 03-24-2019 at 03:22 PM.

  26. #40
    Join Date
    Jan 2007
    Location
    Flushing/Flint, Michigan
    Posts
    7,225
    Post Thanks / Like
    Likes (Given)
    345
    Likes (Received)
    6086

    Default

    Fully understand why the air gage is a non-starter.
    To remove the "mark" left by the stylus you have to move some material or sort of brush it so you only see the wheel marks.
    Sometimes that mark will be a little deeper so you need worse case.
    As this is a grinding wheel and not a brush one rev seems oh-so little to me and walking the edge.
    I know it eats cycle time but I'd go at least three.

    The fact that one rev removes it tells you the wheel is still cutting stock.
    Which is all fine but this action changes with parts run.
    Do you dress every part cycle? If so my worries are wrong.

    With these controls I would fully expect a tenth or less ..... but you have to load the part well within that tenth off the bottom side.
    Being single side grinding that error can be a real problem.

    I can hold a +/-.0003 six-sigma on a Blanchard with one micron readouts.
    Flipping the part is the huge error.
    Even the one micron coolant filter does not help as what has been left over from grinding side one has not seen the filter yet.

    Plenty of room to work but don't mix chasing the second op size with loading for the second op.
    Side 2 is so full of possible added in errors

    In reality given these controls why are you not in the couple of microns in test? The machine and the feedback are.
    Bob


Tags for this Thread

Bookmarks

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •