584,829 active members*
4,944 visitors online*
Register for free
Login
IndustryArena Forum > MetalWorking Machines > Benchtop Machines > 5um vs. 1um linear encoders
Results 1 to 12 of 12
  1. #1
    Join Date
    Apr 2018
    Posts
    88

    5um vs. 1um linear encoders

    Hello,

    I am about to purchase linear encoders for the mill.

    They will be connected to a DRO for now, but in the (not so distant) future the mill will ve converted to CNC and the linear encoders will be used for closed loop control.

    I have the option to purchase encoders with 5um resolution or for a small price increase with 1um resolution.

    The end goal is offcourse the CNC conversion. Will the higher resolution feedback encoders improove the overall closed loop CNC accuracy? Or will they cause more trouble than added value? I am thinking here at motor hunting and the likes. I will be using servos to drive all axis.

    Thank you,
    Andrei

    Sent from my SM-A520F using Tapatalk

  2. #2
    Join Date
    Dec 2013
    Posts
    5717

    Re: 5um vs. 1um linear encoders

    I would use the 1um encoders. I have them on all of my machines, both for a DRO on my manual lathe and on my CNC mills and had them on my CNC router also. I will also be installing them on my CNC lathe and CNC Surface grinder, a couple of future projects. A properly tuned servo will not ''hunt''.
    Jim Dawson
    Sandy, Oregon, USA

  3. #3
    Join Date
    Aug 2015
    Posts
    142

    Re: 5um vs. 1um linear encoders

    How do u get ur servo fine tuned ?
    I mean what u want to achive with a properly tuned servo if u use velocity , or position control .
    Thanks .

  4. #4
    Join Date
    Dec 2013
    Posts
    5717

    Re: 5um vs. 1um linear encoders

    Quote Originally Posted by Andreiir View Post
    How do u get ur servo fine tuned ?
    I mean what u want to achive with a properly tuned servo if u use velocity , or position control .
    Thanks .
    In a perfect world a properly tuned servo will exactly follow the commanded motion profile, without any lead or lag error, and stop at the exact encoder pulse that was commanded. In the real world, you tune the servo to get as close to perfect as possible using the adjustments provided by the servo drive and/or the controller.
    Jim Dawson
    Sandy, Oregon, USA

  5. #5
    Join Date
    Jun 2011
    Posts
    692

    Re: 5um vs. 1um linear encoders

    From the research I've done (not that much,) adding linear scales as a second servo loop isn't really done much. I think most linear scales have slower response than servo encoders, so that will limit your dynamic gains. Also servo encoders typically have much higher effective resolution than linears. A 2500 line (10,000 pulse) encoder with a 10mm pitch screw will have the same resolution as a 1um resolution linear. Also the coupling is looser between the motor and the linear than it is between the motor and the servo encoder (screw acts as a spring, plus backlash and the mass of what you're moving can be a complicated system,) can lead to instability.
    Accurate linears can be good for tasks like mapping your ball screw, etc. though.

  6. #6
    Join Date
    Apr 2018
    Posts
    88

    Re: 5um vs. 1um linear encoders

    Quote Originally Posted by skrubol View Post
    From the research I've done (not that much,) adding linear scales as a second servo loop isn't really done much. I think most linear scales have slower response than servo encoders, so that will limit your dynamic gains. Also servo encoders typically have much higher effective resolution than linears. A 2500 line (10,000 pulse) encoder with a 10mm pitch screw will have the same resolution as a 1um resolution linear. Also the coupling is looser between the motor and the linear than it is between the motor and the servo encoder (screw acts as a spring, plus backlash and the mass of what you're moving can be a complicated system,) can lead to instability.
    Accurate linears can be good for tasks like mapping your ball screw, etc. though.
    Partially I agree with your statements.

    Adding a control loop might add instability to the system in the form of jitter or faults, depending on the setup. But only in case the control loop is set too tight.

    OTOH, adding feedback will help a lot with backlash compensation. The servo encoder does know the position of the leadscrew. Not the axis position. The linear encoder will know the axia real position and the motion controller can compensate or fault if the error is too big.

    Carefull parameter tuning will bring benetits. And in the end if it does not work properly I will simply disconnect the linear encoders. It is a hobby and it is all about trial and error

    Sent from my SM-A520F using Tapatalk

  7. #7
    Join Date
    Jan 2008
    Posts
    1523

    Re: 5um vs. 1um linear encoders

    I think it might be worthwhile for screw mapping alone.

    1um scales are cheaper than better accuracy ball screws. Map the screw error and compensate for it (LinuxCNC does screw compensation, though I am yet to use that function).
    7xCNC.com - CNC info for the minilathe (7x10, 7x12, 7x14, 7x16)

  8. #8
    Join Date
    Dec 2013
    Posts
    5717

    Re: 5um vs. 1um linear encoders

    I am using 1um magnetic linear encoders as my only feedback, not as a dual loop system. The linear encoders are tied back to the controller and the loop is closed at the controller, and am using analog control of the servos. Doing it this way compensates for any ball screw error, and will automatically compensate for some backlash. The controller ''knows'' where the table is. There is no latency in the encoder feedback, as I recall, these encoders are rated at 3 meters/second, and on a mill I'm operating at a small fraction of the rated max speed.
    Jim Dawson
    Sandy, Oregon, USA

  9. #9
    Join Date
    Feb 2009
    Posts
    6028

    Re: 5um vs. 1um linear encoders

    I also do a fair amount of scale installs, mostly on Fanuc, and mostly optical with Zerodur glass though, but I will throw my 2 cents in. A proper scale response is not slower at all once on an axis, its moving at a speed one to one with axis motion, not 5 to one etc like a motor encoder, and we usually move much faster than home machines. It will also help compensate for thermal growth on a given axis, so long as the hard mount point is in the center of the axis travel, and the floating points spread outwards from there. 1um/5um is the resolution, not accuracy of the scale, so while a scale can get you "pretty close" on positioning accuracy, there is always error in the scale, even the mid end ($2-3K scales) you can still be off 10um on position, and i've seen 3um on the $18K laser scales at 750mm. As Jim said, servo tuning is critical, and its almost an art if doing it by hand, software helps a lot if your supplier has the software capability. The only time i've seen hunting show up was with excessive backlash, since the control is trying to calculate the velocity and position loops at the same time, but this gets solved every time by reducing the backlash and occasional re tune of the servo loop.

  10. #10
    Join Date
    Oct 2018
    Posts
    79

    Re: 5um vs. 1um linear encoders

    Quote Originally Posted by skrubol View Post
    From the research I've done (not that much,) adding linear scales as a second servo loop isn't really done much. I think most linear scales have slower response than servo encoders, so that will limit your dynamic gains. Also servo encoders typically have much higher effective resolution than linears. A 2500 line (10,000 pulse) encoder with a 10mm pitch screw will have the same resolution as a 1um resolution linear. Also the coupling is looser between the motor and the linear than it is between the motor and the servo encoder (screw acts as a spring, plus backlash and the mass of what you're moving can be a complicated system,) can lead to instability.
    Accurate linears can be good for tasks like mapping your ball screw, etc. though.

    Only closed loop feedback can account for heating of the ballscrew though.

    The ballscrews will change temperature depending on a) ambient temperature and b) movement generating friction.

    So if you put some figures into:

    https://goodcalculators.com/thermal-...on-calculator/

    e.g. a 20deg c change in temperature over x axis e.g. 0.8m gives 0.11mm expansion (i.e. 0.11mm error) . Since ballscrew temperature is not measured maybe it's 50deg giving 0.3mm error? But it is uncontrolled, temperature will change with preload and travel speed. So you don't ever know what you are getting. Some industrial machines run coolant through the middle of the ballscrew to mitigate this.

    So I think worst case you can get in the order of 0.3mm error (depending on travel etc) So servo's being able to encode to 1/1000 or 1/4000 of 5mm (0.05mm) doesn't really matter.

    Of course this is (sort of) worst case - it depends what accuracy you need and you can mitigate this but I don't like the options:
    Reduce preload (increasing backlash, or at least backlash under load)
    Run coolant over the screws - needs coolant running all the time, also difficult for z axis
    Use higher tolerance screws and balls (reduces friction) - very expensive
    Run cooling fans on the screws - not a bad option but difficult to package, will not compensate for ambient temp changes.

    Of course it all depends on the accuracy you want and most people get by very nicely with open loop and steppers.

  11. #11
    Join Date
    Feb 2009
    Posts
    6028

    Re: 5um vs. 1um linear encoders

    Some use cooler oil through the ballscrews, the most common way to compensate on higher end machines is to pre-stretch the ballscrews using the thermal growth calculator. This is why they have thrust bearings on both ends of the ballscrews. Not cheap, so usually not for home machines, but then home machines aren't traveling at 2000+ IPM.

  12. #12
    Join Date
    Apr 2018
    Posts
    88

    Re: 5um vs. 1um linear encoders

    Thank you all for your advise. I went the 1um sensors route.

    I ordered them from a uk based supllier. The sensors are made by some company out of india: electronicaems. They seem to work very well and the price was good when bundled with an 7" LCD DRO.

    Sent from my SM-A520F using Tapatalk

Similar Threads

  1. Linear encoders
    By leix_99 in forum Mach Software (ArtSoft software)
    Replies: 8
    Last Post: 08-08-2012, 02:31 PM
  2. Low Cost Linear Encoders For Linear Servo's
    By protomate in forum Want To Buy...Need help!
    Replies: 0
    Last Post: 12-15-2010, 03:08 AM
  3. servo Vs. linear encoders
    By Yovav in forum Mechanical Calculations/Engineering Design
    Replies: 7
    Last Post: 08-10-2009, 01:04 AM
  4. Just saw a linear motor XY table with linear encoders today...nice
    By guru_florida in forum MetalWork Discussion
    Replies: 6
    Last Post: 04-29-2009, 01:39 AM
  5. How Should I Mount Linear Encoders?
    By lerman in forum Uncategorised MetalWorking Machines
    Replies: 6
    Last Post: 01-13-2005, 10:50 PM

Tags for this Thread

Posting Permissions

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