squeezing the last bit of accuracy on a BP V2XT cast Iron - Page 2
Close
Login to Your Account
Likes Likes:  0
Page 2 of 3 FirstFirst 123 LastLast
Results 21 to 40 of 44
  1. #21
    Join Date
    Jan 2007
    Location
    Northern, NJ
    Posts
    682
    Post Thanks / Like
    Likes (Given)
    320
    Likes (Received)
    27

    Default

    Quote Originally Posted by madmachinst View Post
    Different points on screw giving different numbers with variations as much as 250 steps at different locations.
    250 steps would be 2.5 thou. That would be FAR worse than even inexpensive rolled ballscrews. You need to find out if each of these different spots are repeatable. If each spot is random, then it certainly sounds like motor tuning.

  2. #22
    Join Date
    Aug 2014
    Location
    bainbridge island
    Posts
    1,455
    Post Thanks / Like
    Likes (Given)
    318
    Likes (Received)
    361

    Default

    a fine scribe line marked on the end of the cog pulley.. will show you something.

    1/100th of a turn is about 3 degrees, or 2.5 thousanths of an inch for a .25" pitch ball screw. you should easily be able to see such a discrepancy.

  3. #23
    Join Date
    Mar 2006
    Location
    Vershire, Vermont
    Posts
    2,816
    Post Thanks / Like
    Likes (Given)
    1978
    Likes (Received)
    932

    Default

    Quote Originally Posted by madmachinst View Post
    Haven't addressed oil metering problem cause I don' have the time to tear it apart....
    I don't know enough to say if lack of lube is causing your problems or not, but the long term health of the machine is at risk.

  4. #24
    Join Date
    Jan 2007
    Location
    Central FL
    Posts
    908
    Post Thanks / Like
    Likes (Given)
    8
    Likes (Received)
    22

    Default

    well I was thinking about moving the table over to one end pulling the cover and spraying some lube on the screw and then do the other side. I can attack the Y axis from underneath the knee. I was thinking about doing that like every 2 hrs. running it like that till I can afford to have the thing redone like I was planning. But no one gave a reply on what kind of spray I should use and if every two hrs is a good idea. Or does the whole idea stink?

  5. #25
    Join Date
    Nov 2002
    Location
    Burlington, NJ
    Posts
    1,440
    Post Thanks / Like
    Likes (Given)
    42
    Likes (Received)
    176

    Default

    Replace all the metering units is the first thing you need to do. Lube USA makes lower price meters if your strapped for cash. Also check that the lube lines are still seated in the ballscrew flange.
    If the ballscrew still has preload through it's range than it should be accurate enough.
    A handy tool for this kind of work is a precision step gauge but they are stupid expensive. Keep an eye out for a Cadillac Height Gage, the one with the micrometer on top and rings on the spindle every inch. If you can get one cheap, pull the spindle out, it makes a sweet step gauge. I have one about 18" long. I use it on a set of precision V blocks and carefully square it to the axis I'm checking.

    Mr Bridgeport

  6. #26
    Join Date
    Jan 2007
    Location
    Northern, NJ
    Posts
    682
    Post Thanks / Like
    Likes (Given)
    320
    Likes (Received)
    27

    Default

    This has to be a servo/control issue. 750W servos will move that machine effortlessly. My Centroid would fault long before it settled on being off by 250 counts (unless it was WAY out of tune). I don't understand why this machine is not faulting if it is "closed loop". Maybe it is not really closed loop back to the controller. This sounds like something that would happen with a step-direction control.

    Get some way oil on all sliding surfaces and use a can or brush to get it on the ballscrews, but this will not solve the servo issue. The sliding surfaces are much more likely to cause you problems if they are dry and you are getting a lot of friction. I drive my Series II knee with a 40 in-lb servo and as heavy as it is, the Centroid puts it to within an encoder count or two and it has a high friction acme screw to elevate it.

    I have fully rebuilt 3 Bridgeports, including 2 BOSS machines. I had the ballscrews re-balled on the Series II, but it was only to remove backlash. Overall accuracy was not at issue.

  7. #27
    Join Date
    Jan 2007
    Location
    Central FL
    Posts
    908
    Post Thanks / Like
    Likes (Given)
    8
    Likes (Received)
    22

    Default

    Quote Originally Posted by Polaraligned View Post
    This has to be a servo/control issue. 750W servos will move that machine effortlessly. My Centroid would fault long before it settled on being off by 250 counts (unless it was WAY out of tune). I don't understand why this machine is not faulting if it is "closed loop". Maybe it is not really closed loop back to the controller. This sounds like something that would happen with a step-direction control.
    Thanks for the suggestion of brushing way oil on the screws. Will do that now regarding closed loop. it is true closed loop to controller, not the driver. I had ones that were closed to driver in the past. All they did was fault when steps weren't fully made. This one actually makes corrections. Then again I am not inside the system so don't know what is going on. If I had table locks like on a manual machine I would actually see a better picture what is going on. It's a Hiconn controller. You can actually see a screen in here that counts commanded and actual position and tells you error. like this

    https://www.vitalsystem.com/portal/m..._PID_Guide.pdf

    I strongly believe its mechanical either belts or ballscrews maybe even Yoke moving?

  8. #28
    Join Date
    Jan 2007
    Location
    Flushing/Flint, Michigan
    Posts
    10,914
    Post Thanks / Like
    Likes (Given)
    645
    Likes (Received)
    8760

    Default

    This drive configuration can lead to all sorts of strange errors.
    Z axis are normally gravity loaded.
    Testing here is best done in all upward moves if worried about the screw itself.
    One a test I'd do a basic block an down and then up to zero.
    Make the move at low speed and then stick in the larger gauge blocks at different sizes being careful to not disturb the indicator.
    End of test go back low and then up, stick in the "zero" block and make sure it still is zero.
    Do not use high speeds. At this count pulse level there can be other electrical problems.

    You have talked to the drives and they have settled at no error?
    Bob

  9. #29
    Join Date
    Jan 2007
    Location
    Central FL
    Posts
    908
    Post Thanks / Like
    Likes (Given)
    8
    Likes (Received)
    22

    Default

    Thanks BOB.
    So this is on my to do list
    A.) Check Z axis with only upward motion. Zero DTI on vise then move up an increment and check with gauge blocks, say 20 IPM?
    B.)check for excessive force needed to move ballscrews with torque wrench. 1st with belts tensioned ad then without belt tension.
    C.) check commanded moves by verifying on controller screen and servo drives that the correct number of steps have been made and recognized by controller.
    D.) take it all apart so I can check those lube lines and verify oil coming out of ball screw lines. I am really not too worried about the ways cause those are soaking wet with oil as seen in pics. Anything I can run through the lube system that may clean out any clogs?
    Any thing else I should add to this list?
    Meanwhile have parts I am behind on and am running em manually, then have this state test I have to prepare for. That is why I am taking care of this so pathetically slow. THANKS for all the help and suggestions.

  10. #30
    Join Date
    Jan 2007
    Location
    Flushing/Flint, Michigan
    Posts
    10,914
    Post Thanks / Like
    Likes (Given)
    645
    Likes (Received)
    8760

    Default

    The oiling would be the last very thing I'd chase in a non-loaded test.

    One of the real oh-mys I encountered was a similar setup that was unpredictable.
    Turned out that under the accel ramps set too high the amp would error out and shut down.
    To stop this bad behavior the user routed the error shutdown signal from the amp to the amp reset. Meanwhile Mach is just fine putting out steps as it has no way of knowing.
    Any steps during this reset cycle lost but the amp side happy also.

    Did not dig in but it looks like your drive software should allow you to see running torque or power in a move. This is much better than a torque wrench check.
    Are the pulleys factory or did you have to drill your own holes for this retro? Since it seems inch screws and only doing even inch tests this should not be a biggie.

    When the end error happens what if you grab the pulley and try to move it to zero. Does it fight back and go to the bad DTI reading?

    What are you expecting or wanting in 1 and 4 inch accuracy and any random position repeatability?
    One can not be very accurate but repeatable to very small numbers. Then one just simply lies or tweaks the cnc program.
    This a mess on a 5 or more axis but super easy on a 3.
    Bob

  11. #31
    Join Date
    Jan 2007
    Location
    Central FL
    Posts
    908
    Post Thanks / Like
    Likes (Given)
    8
    Likes (Received)
    22

    Default

    Quote Originally Posted by CarbideBob View Post
    The oiling would be the last very thing I'd chase in a non-loaded test.

    One of the real oh-mys I encountered was a similar setup that was unpredictable.
    Turned out that under the accel ramps set too high the amp would error out and shut down.
    To stop this bad behavior the user routed the error shutdown signal from the amp to the amp reset. Meanwhile Mach is just fine putting out steps as it has no way of knowing.
    Any steps during this reset cycle lost but the amp side happy also.
    Doubt I have that situation. But, will take this into consideration and test distance moved both in 5IPM and 30 IPM to see if there is any difference. So, add that to the super list of what I have to do.



    Quote Originally Posted by CarbideBob View Post
    Did not dig in but it looks like your drive software should allow you to see running torque or power in a move. This is much better than a torque wrench check.
    Are the pulleys factory or did you have to drill your own holes for this retro? Since it seems inch screws and only doing even inch tests this should not be a biggie.
    Pulleys are factory, their collet is for 5/8" shaft which is what the servo provides. I am not sure they are aligned. I also had to make servo adapters as teh old SEM servos seemed to have a diffeent hole pattern than these, so maybe the adapters are such that they require more ( excessive) stretch from the belts. Maybe when the ballscrew pulleys sat there for like 5 years some rust might have accumulated on them making them slightly bigger? GD knows.

    I have no info back feed on torque and such back to the controller. Machs newest controller doesn't use step and direction, but ethercat communication, so there is real 2 way comm with lots of data going back to the controller. I don't have that. Only info I have going back to the controller is encoder pulses. That's it, no load meters. I bet my hicon controller might have enough analog inputs to accommodate this, but would need some new fancy wiring and maybe even digital filtering which is beyond me.

    Quote Originally Posted by CarbideBob View Post
    When the end error happens what if you grab the pulley and try to move it to zero. Does it fight back and go to the bad DTI reading?
    Haven't looked at what happens on DTI. I just saw that the pulley seems to go back to original position. So, will add that check to my list above. Check to see DTI returns to position after fighting servo. Also, I have Readouts updating when in E stop mode. If that helps any.

    Quote Originally Posted by CarbideBob View Post
    What are you expecting or wanting in 1 and 4 inch accuracy and any random position repeatability?
    One can not be very accurate but repeatable to very small numbers.
    Kinda confused on Only thing I can check is repeatability and am not getting it. I do expect at least .0005" accuracy.

    Quote Originally Posted by CarbideBob View Post
    Then one just simply lies or tweaks the cnc program. This a mess on a 5 or more axis but super easy on a 3.
    Bob
    Totally lost.
    Carbide thank you for all this help.

  12. #32
    Join Date
    Jan 2007
    Location
    Central FL
    Posts
    908
    Post Thanks / Like
    Likes (Given)
    8
    Likes (Received)
    22

    Default

    Thanks BOB.
    So this is on my to do list
    A.) Check Z axis with only upward motion. Zero DTI on vise then move up an increment and check with gauge blocks. I will check at both 5 and 30 IPM to see if accel ramp has anything to do with this problem.

    B.)check for excessive force needed to move ballscrews with torque wrench. 1st with belts tensioned and then without belt tension.

    C.) check commanded moves by verifying on controller screen and servo drives that the correct number of steps have been made and recognized by controller.

    D.) Check to see that when I fight the servo while it is powered the DTI returns to the same position it was before.

    E.) take it all apart so I can check those lube lines and verify oil coming out of ball screw lines. I am really not too worried about the ways cause those are soaking wet with oil as seen in pics. Anything I can run through the lube system that may clean out any clogs?

    Any thing else I should add to this list?
    Meanwhile have parts I am behind on and am running em manually, then have this state test I have to prepare for. That is why I am taking care of this so pathetically slow. THANKS for all the help and suggestions.

  13. #33
    Join Date
    Jan 2007
    Location
    Flushing/Flint, Michigan
    Posts
    10,914
    Post Thanks / Like
    Likes (Given)
    645
    Likes (Received)
    8760

    Default

    Quote Originally Posted by madmachinst View Post

    Totally lost.
    .
    So this tweak in part.
    So one has a program and lets make it simple in Z levels not checking right.
    If the machine/control is repeatable you change to 0.9940 or 1.0025 rather than the 1.0000 on the print. You lie to the numbers.
    Error is who cares as long as the machine makes the part good.
    Repeat position is so different than accuracy.

    Bad here is first part cut good as the error unknown. That is a pain since you have to go outside, check, offset and recut.

    Given this crap talk I think your error is outside of any normal expected so other stuff somewhere.
    Bob

  14. #34
    Join Date
    Jan 2019
    Country
    UNITED STATES
    State/Province
    Tennessee
    Posts
    766
    Post Thanks / Like
    Likes (Given)
    228
    Likes (Received)
    229

    Default

    I wouldn't worry about encoder counts being different than expected. high res Rotary encoders rarely are actually what name plate states. Allen Bradleys are plus/minus 4 counts per rev. Can you ping encoder value before and after movement and see if it is within tolerance? At 100 hz polling rate the motors are spending a lot of time flying blind, and with your current PIDs there is no correction to flight path between pings. If you are running ethercat then mach should have the ability to read amps straight from the drives without going thru the controller. I am a big fan of modbus (ethercat and devicenet) for systems in a machine, but it would add un-needed complexity to a 3 axis machine. Can you run your controller via a standard ethernet or usb protocol? Set your counts from a long run (24 inchs in x, 10 in y, spit the difference for z as a starting point). Calibrating over an inch or two is calibrating over futile and frustrating.

  15. #35
    Join Date
    Nov 2002
    Location
    Burlington, NJ
    Posts
    1,440
    Post Thanks / Like
    Likes (Given)
    42
    Likes (Received)
    176

    Default

    "I wouldn't worry about encoder counts being different than expected. high res Rotary encoders rarely are actually what name plate states."

    I have never heard such a thing. If true than manufactures of machines would have all kinds of variables to contend with. None of it fun.
    If AB really states that they can't manufacture a rotary encoder with a known quantity of pulses per revolution I'd be very surprised. Many companies have been doing it for decades.

    Mr Bridgeport

  16. #36
    Join Date
    Jan 2019
    Country
    UNITED STATES
    State/Province
    Tennessee
    Posts
    766
    Post Thanks / Like
    Likes (Given)
    228
    Likes (Received)
    229

    Default

    Quote Originally Posted by Mr Bridgeport View Post
    "I wouldn't worry about encoder counts being different than expected. high res Rotary encoders rarely are actually what name plate states."

    I have never heard such a thing. If true than manufactures of machines would have all kinds of variables to contend with. None of it fun.
    If AB really states that they can't manufacture a rotary encoder with a known quantity of pulses per revolution I'd be very surprised. Many companies have been doing it for decades.

    Mr Bridgeport
    Most of our encoder are AB, whenever one gets replaced I have to go thru a new calibration counts routine. Scales are always right, rotary encoders are different. We have begenhoff branded encoders on another machine, those very by 100 counts from one another. They are super high res (6000 counts). The techs (young fellows) for that machine swore encoder counts are set in stone, replaced gear boxes, electric eyes, and pinch wheels before grasping encoder counts are ultimately found in calibration/imperical methods.

  17. #37
    Join Date
    Jan 2007
    Location
    Central FL
    Posts
    908
    Post Thanks / Like
    Likes (Given)
    8
    Likes (Received)
    22

    Default

    Quote Originally Posted by CarbideBob View Post
    So this tweak in part.
    So one has a program and lets make it simple in Z levels not checking right.
    If the machine/control is repeatable you change to 0.9940 or 1.0025 rather than the 1.0000 on the print. You lie to the numbers.
    Error is who cares as long as the machine makes the part good.
    Repeat position is so different than accuracy.

    Bad here is first part cut good as the error unknown. That is a pain since you have to go outside, check, offset and recut.

    Given this crap talk I think your error is outside of any normal expected so other stuff somewhere.
    Bob
    That's just it, the error is not consistent. If it was consistent I wouldn't even consider it error, I would just change steps per inch and all is great.

    Now on another note, you all think I can just apply like a 20" dro to my X axis and calibrate the machine off of that? Got one of these I gaging DRO's on my lathe, it seems accurate just really susceptible to oil and coolant. I just need to find way to mount it temporarily and measure. Quill is only like 4.5" so 4" block should suffice.

  18. #38
    Join Date
    Jan 2007
    Location
    Central FL
    Posts
    908
    Post Thanks / Like
    Likes (Given)
    8
    Likes (Received)
    22

    Default

    Regarding
    "B.)check for excessive force needed to move ballscrews with torque wrench. 1st with belts tensioned and then without belt tension."
    I took off X axis cover which also hold the servo on it, so couldn't check belt tension when off. Before doing that, my 1/4" torque wrench didn't click on teh lowest setting when using it to rotate the pulley. Rotating pulley when tensioned needed just wrist force, rotating pulley when cover off ( untensioned only needed finger force) Servo magnets seemed to be causing that.

    here are pics of what I discovered. Belt tear only between .045-.050"
    img_20210729_203028714-1-.jpg

    pulley misalignment has to be .450"-.500" in the x axis direction. I measured this

    img_20210729_203045423-1-.jpg
    to see how deep pulley was by taking up thickness of a straight edge, then laying it across the edge of the mounting box and measuring depth to pulley face. It came out to .562"

    Now, as you can see this servo pulley is almost right on the face of the cover plate. which goes on the edge measured from in pic #2
    img_20210729_203104601-1-.jpg
    Pulley is say about .07" away from mounting plate/cover.

    The servo shaft due to the thickness of the mounting plate ( 1/2" thick 6061) is about 3/8" short of getting to end of pulley so I didn't want to have the pulley just barely hanging on the shaft. Hence the misalignment.

    should I make the mounting plate for this one out of steel and say about 1/4" thick? Or, just move the pulley further off the shaft? Or in my opine best option make a coupler so I can lengthen the servo shaft? Shaft has a key in it. bore out a bushing that is press fit to shaft and make a slot for the key and make an extra 1" length of shaft. Existing servo shaft also has tapped hole in it. I can rely on screw, key, and bushing to keep it together as one piece. Any suggestions regarding this?

    But all that aside, wish there was a way for me to test belt tension when this is all mounted in. Old BP instruction book says 1/4" play with 7lbs force. Just that there is no procedure for testing X axis.
    Attached Thumbnails Attached Thumbnails img_20210729_203056544-1-.jpg  

  19. #39
    Join Date
    Aug 2014
    Location
    bainbridge island
    Posts
    1,455
    Post Thanks / Like
    Likes (Given)
    318
    Likes (Received)
    361

    Default

    since you've got what looks like a 3:1 ratio, it should be very easy for you to hold a ruler against the servo pulley and scribe a line on it, anywhere. then move the machine one turn of the servo repeatedly (whatever that turns out to be)

    and you should easily be able to witness the error, if it is from the servo.

  20. #40
    Join Date
    Jan 2007
    Location
    Flushing/Flint, Michigan
    Posts
    10,914
    Post Thanks / Like
    Likes (Given)
    645
    Likes (Received)
    8760

    Default

    Quote Originally Posted by memphisjed View Post
    I wouldn't worry about encoder counts being different than expected. high res Rotary encoders rarely are actually what name plate states.
    I would have to be big big no on this with my limited experience.
    This is insane, Never have seen 100 to 288,000 be off by one count unless mounted badly. This sort of goes with how these things are made.
    No...... they are not engraved in stone they are engraved in glass. If this count off other problems.
    Now rotary to actual linear inches moved. Many things become involved here.


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
  •