Accuracy I can expect from a RF-45 or similar cnc conversion? Bang for $$ point?


Page 1 of 2 12 LastLast
Results 1 to 20 of 27

Thread: Accuracy I can expect from a RF-45 or similar cnc conversion? Bang for $$ point?

  1. #1
    Member
    Join Date
    Mar 2017
    Location
    Australia
    Posts
    98
    Downloads
    0
    Uploads
    0

    Default Accuracy I can expect from a RF-45 or similar cnc conversion? Bang for $$ point?

    I'm debating whether to go down the CNC path with my RF-45, but I don't want to invest into something that gives inaccurate results. I am aware it wont be a metal hog, but can it be similar in accuracy to a skyfire which boasts <0.02mm accuracy? Within a reasonable budget.

    I have done some reading but don't know exactly what to buy. This is probably what I'm looking at:
    - C7 ballscrews, though I have seen some rolled C5 at a not too extravagant price on Alibaba.
    - Hiwin or similar rails (for z minimum)
    - Upgraded spindle bearings + new motor
    - Closed loop steppers seem like good bang for buck - https://www.ebay.com.au/itm/8-5NM-Ne...oAAOSwqd1Zv4pM

    Can such a configuration achieve the accuracy I'm looking for, or is the RF-45, as others say, just a poor base to start with?

    Could anyone recommend good bang for buck components and give me an idea of what accuracy they can achieve? Or good threads in this direction. (I am in Aus so access to some products is limited)

    Thanks.

    Similar Threads:


  2. #2
    Member
    Join Date
    Nov 2011
    Location
    United States
    Posts
    205
    Downloads
    1
    Uploads
    0

    Default Re: Accuracy I can expect from a RF-45 or similar cnc conversion? Bang for $$ point?

    Mostly it is YOU that determines the accuracy, not the starting machine.

    Choices, Choices, Choices

    Don



  3. #3
    Member
    Join Date
    May 2011
    Location
    Canada
    Posts
    790
    Downloads
    0
    Uploads
    0

    Default Re: Accuracy I can expect from a RF-45 or similar cnc conversion? Bang for $$ point?

    Quote Originally Posted by j3dprints View Post
    I'm debating whether to go down the CNC path with my RF-45, but I don't want to invest into something that gives inaccurate results. I am aware it wont be a metal hog, but can it be similar in accuracy to a skyfire which boasts <0.02mm accuracy? Within a reasonable budget.
    0.02mm = 0.00078"

    My guess is that this specification is simply derived by looking at what the electronics combined with screw lead and microstepping are capable of.

    I very very much doubt that their machines are capable of this in practice.

    Perhaps one of the Skyfire owners can chime in to the conversation. I am assuming that you are talking about a stepper driven machine, but it looks like skyfire also has closed loop options? I'm not too familiar with them other than they are a made in China CNC small mill that has a big thread here about people waiting years for their orders? Is that right, are we on the same page or am I thinking of something else?

    Even a small amount of spindle runout will cause more error than that with a 2" or longer bit, so for the accuracies you're talking about, everything has to come into play. A Hass mill may give you those kinds of accuracies or better, but a Skyfire, I am doubtful (but not certain).

    If your goal is to make or buy a machine that is very accurate, my advice would be to


    1. Ask the people who own Skyfire CNCs what kind of accuracies they are getting (don't trust the advertisement).
    2. If building your own, consider the preload of the ballscrews and the runout of the spindle.
    3. Ask Jim Dawson about his setup using linear encoders.

    EDIT:

    Looking through the skyfire website, it looks like they are using closed loop steppers on their entry models, and AC servos on everything else.

    Last edited by NIC 77; 11-17-2017 at 11:49 AM. Reason: More Info


  4. #4
    Member
    Join Date
    Mar 2017
    Location
    Australia
    Posts
    98
    Downloads
    0
    Uploads
    0

    Default Re: Accuracy I can expect from a RF-45 or similar cnc conversion? Bang for $$ point?

    The reason I mention skyfire is that I have read from people who have performed a conversion that they wish they'd just bought one of those (or a tormach, I could have just said that). What I mean is I don't want to invest a year or so and $3000 (or however much) and end up with a machine that isn't very useful as it can only make parts for aesthetic purposes, due to holes being out of round etc.

    So if I am concerned about accuracy then spindle runout and ballscrew backlash are the main enemies? This makes sense to me, although of course I am aware there are other factors.
    Backlash induced from stiction should be ok if I put linear rails on all axis.
    I have also looked into ballscrew accuracy, although I believe this should be able to be overcome if one was to perform a recording of the true positions of the ballscrews along their travel and have software perform a mapping to compensate. Ballscrew backlash seems to be a bigger issue.

    Thanks, & any more advice is welcome! Pulling the trigger for 3k investment is intimidating.



  5. #5
    Member
    Join Date
    May 2011
    Location
    Canada
    Posts
    790
    Downloads
    0
    Uploads
    0

    Default Re: Accuracy I can expect from a RF-45 or similar cnc conversion? Bang for $$ point?

    Yes, I would say that spindle runout is the #1 enemy, followed by ballscrew backlash.

    The third enemy is what you can achieve with microstepping, stepper motors, but that is usually less than 1000th anyway.

    For example, 10mm lead, 200 steps per rev (1.8 deg), 10x microstep

    10 / (200 x 10) = 0.005mm however that's a theoretical number and in reality it won't be that accurate.

    The ballscrew error can be in terms of over the length, so you may be able to program some of that out without a mapping, for example, 10mm lead, well, perhaps it's 10.05mm in reality, but a consistent 10.05.

    The kinds of accuracies you are talking about are at the extreme good end, IMO almost no DIY machines can do it or come close. What you are talking about might be suitable for press fitting bearings with half a thousandth of difference between the bearing and the pocket.

    Whether to buy a machine or do a conversion I don't know.

    Jim Dawson is the guy to ask about the linear encoders, because he has made that work and I believe he has made it work super accurately, and even made his own software.

    If I was looking to setup a really accurate machine, he's the first guy I would ask. I'm afraid it's a bit outside of my knowledge. I only know it's possible.



  6. #6
    Member ninefinger's Avatar
    Join Date
    Sep 2006
    Location
    Canada
    Posts
    509
    Downloads
    0
    Uploads
    0

    Default Re: Accuracy I can expect from a RF-45 or similar cnc conversion? Bang for $$ point?

    The real question is what kind of dimensional tolerance do you need to achieve? Working with both engineers and machinists I can tell you that 95% of what the engineers provide is over toleranced, especially from the inexperienced ones.
    Most DIY machines will make parts +/-0.005 thou easily which is still quite good, and with learning your machines behavior you can improve on that.

    As mentioned there are many sources of error on a machine and just improving 1 thing won't necessarily let you get to the 0.02mm (less than 1 thou) that you are looking for.



  7. #7
    Community Moderator Jim Dawson's Avatar
    Join Date
    Dec 2013
    Posts
    5717
    Downloads
    0
    Uploads
    0

    Default Re: Accuracy I can expect from a RF-45 or similar cnc conversion? Bang for $$ point?

    Thank you for the kind words NIC 77.

    Yes, the best accuracy is achieved with linear encoders on the table, those could be glass or magnetic scales. It is possible to hold < +/- 0.001 mm. That also requires that the backlash in the leadscrews be near zero, but with the scales on the load (table) backlash and/or leadscrew error is pretty much automatically compensated.

    Even my old worn out router will typically hold 0.04 mm with stepper drives. The steppers are set up at ~0.001 mm/step. There is about 0.2 mm backlash in the Y lead screw and about 0.3 mm backlash in the Z, I forget what the X backlash is. With the magnetic scales and some software magic it auto compensates for the error. Not to be confused with what is normally called ''backlash compensation'' by some software vendors. I also close the loop at the controller rather than the drive.

    My mill will typically hold +/- 0.003 mm and will get closer if I am careful in the set up. For most work I don't even bother to measure the finished product, I know how the machine is going to cut and am confident that it will do what I tell it to do. I'm running DC servos on the X and Y, and a stepper on the Z. The last upgrade I did on a really tight machine had no detectable deviation (<0.001 mm) on a 100 mm circular pocket, independently verified by a metrology lab in an aerospace shop. I don't have anything here that will measure that close.

    The Ditron magnetic scales I use are 0.001 mm resolution and seem to be quite repeatable. About $120/axis. I also use Galil Motion Control products as the machine controllers. Normally available on Ebay at reasonable prices.

    Don't confuse accuracy with repeatability. If your machine is repeatable, then you can adjust the accuracy with offsets when compiling the G-code. Just requires some experience with the machine, and all machines are different.

    Jim Dawson
    Sandy, Oregon, USA


  8. #8
    Member
    Join Date
    May 2011
    Location
    Canada
    Posts
    790
    Downloads
    0
    Uploads
    0

    Default Re: Accuracy I can expect from a RF-45 or similar cnc conversion? Bang for $$ point?

    Quote Originally Posted by Jim Dawson View Post
    Thank you for the kind words NIC 77.
    I'm glad you said that. After I made the second post I was thinking that perhaps I was being rude by dropping your name in a thread that you weren't already participating in.

    Definitely you know what you are talking about.

    Once the machine I'm making is finished I may have to revisit this issue and upgrade to some magnetic scales.

    I'm not too worried about it since I think my machine will cut accurately enough for my purposes, but I do think that you are the only one here who could really give some good info on the subject.



  9. #9
    Member
    Join Date
    Mar 2017
    Location
    Australia
    Posts
    98
    Downloads
    0
    Uploads
    0

    Default Re: Accuracy I can expect from a RF-45 or similar cnc conversion? Bang for $$ point?

    Thank you, and thank you Jim.

    Linear encoders certainly seem to be the go then, that is amazing accuracy and a solution involving actual position feedback to overcome backlash is brilliant. I already have some glass 0.001mm scales on my RF-45, perhaps I could reuse them. It looks like I might be able to get by with cheaper physical hardware and proper closed loop compensation.

    Where could I start reading about the drives and what not to implement such a setup? If the loop is closed 'at the board' (I am not sure what this means exactly but have a picture) does this mean you don't use conventional leadshine type steppers? Is there a thread somewhere detailing where somebody has a build log with model numbers of such a setup?

    Thanks again.



  10. #10
    Community Moderator Jim Dawson's Avatar
    Join Date
    Dec 2013
    Posts
    5717
    Downloads
    0
    Uploads
    0

    Default Re: Accuracy I can expect from a RF-45 or similar cnc conversion? Bang for $$ point?

    Your linear scales should work fine. Most of the ones made in the last several years output a 4 channel differential quadrature signal, they look just like an encoder to the controller.

    Below is an example of modern servo system that I would use. I would think that a 400 W servo would be enough for your X & Y axis. Maybe 750 W for the Z

    https://www.ebay.com/itm/Delta-B2-20...h8BNyTV2dJJo_g

    If you can find brushed DC servos, they work fine also and may be less money. The main thing that you want to do when buying servos is to get a matched set, drive and motor. It's possible to mix drives and motors, but you really have to do your homework. I wouldn't do it.

    Closing the loop at the board means that the encoder input is connected to the motion controller and is used by the motion controller to position the table. As far as motion controllers, Galil, Kmotion, and Mesa all make systems that would work well. My favorite is Galil products, they are industrial class motion controllers. There are a lot of BoBs listed on Ebay as ''motion controllers'', they are not.

    Jim Dawson
    Sandy, Oregon, USA


  11. #11
    Member
    Join Date
    Mar 2017
    Location
    Australia
    Posts
    98
    Downloads
    0
    Uploads
    0

    Default Re: Accuracy I can expect from a RF-45 or similar cnc conversion? Bang for $$ point?

    Thanks, I have done -some- reading and have a couple of questions..

    Can you still use hybrid steppers in such a setup? e.g. https://www.ebay.com.au/itm/8-5NM-Ne...oAAOSwqd1Zv4pM
    They are much cheaper in Aus than servos, $300 compared to $600+ per axis
    The Mesa boards are cheaper? If you don't include the PC cost.
    Does the output of the encoder plug into the MESA board, and is there anywhere this is documented or any search strings I can use to research such a setup? As well as implementation in software (LinuxCNC?). The enocder doesn't replace the position encoder on the servo, you are just now using two feedback systems?

    Thanks.



  12. #12
    Community Moderator Jim Dawson's Avatar
    Join Date
    Dec 2013
    Posts
    5717
    Downloads
    0
    Uploads
    0

    Default Re: Accuracy I can expect from a RF-45 or similar cnc conversion? Bang for $$ point?

    Yes, but they would not be quite like a full closed loop system. And I'm not sure how to handle the Mesa board input from linear encoders. What you would really be creating is sort of a dual loop system, but the motor encoders would not connect to the board, the error correction would have to be handled in the computer software at the end of each move. Those drives only accept a step and direction input and are really designed to be used with open loop programs like Mach3.

    I'm pretty sure the Mesa boards are compatible with LinuxCNC. Mesa Electronics I guess downloading the manuals might be the best way to understand how the system works. LinuxCNC

    On my machines, I'm using Galil boards and they do have dual encoder inputs for each axis and well as both analog or step & direction outputs for each axis. Very easy to integrate the entire system with either steppers or servos in any combination. Mach3 is compatible with Galil products, and I think LinuxCNC is also. Mach3 however still operates in open loop mode, but the Galil boards can close the loop at that level so it's possible to make it work pretty well.

    Jim Dawson
    Sandy, Oregon, USA


  13. #13
    Member
    Join Date
    Mar 2017
    Location
    Australia
    Posts
    98
    Downloads
    0
    Uploads
    0

    Default Re: Accuracy I can expect from a RF-45 or similar cnc conversion? Bang for $$ point?

    Thanks, can you recommend a budget Galil model capable of this? I have researched and some seem very expensive.

    This would be ideal if I could set up my mill in this fashion with my DRO with plain open loop steppers to start with, and later upgrade them to servos. All the while enjoying improved accuracy due to actual backlash / feedback correction.


    From Galil:
    "Dual-loop The dual-loop (DV) feature enables the controller to compensate for mechanical backlash. Typically, dual-loop systems use a rotary encoder on the motor and a linear encoder on the load (Galil controllers accept inputs from two encoders per axis as a standard feature). Dual-loop control changes the standard PID control and closes the position loop with the load encoder (“PI”) and derives the damping terms (“D”) from the motor encoder. This method provides smooth and accurate control along the motion path regardless of backlash."



  14. #14
    Community Moderator Jim Dawson's Avatar
    Join Date
    Dec 2013
    Posts
    5717
    Downloads
    0
    Uploads
    0

    Default Re: Accuracy I can expect from a RF-45 or similar cnc conversion? Bang for $$ point?

    To buy a new Galil board is very expensive, but there are a lot of them on Ebay. Some good deals, and some sellers are just crazy in their pricing. The ones you want are 18x0,18x2,18x6, 20x0, 21x0, 22x0, 41x3, 40x0. Where x = the number of axes. The newer firmware revisions have what is called Stepper Maintenance Mode (SM), that will error correct when using open loop steppers with encoders on the load. If you find some that you are interested in, you can post here or PM me for advice. I've been installing Galil products for over 20 years.

    Jim Dawson
    Sandy, Oregon, USA


  15. #15
    Member hanermo's Avatar
    Join Date
    Aug 2004
    Location
    barcelona
    Posts
    780
    Downloads
    0
    Uploads
    0

    Default Re: Accuracy I can expect from a RF-45 or similar cnc conversion? Bang for $$ point?

    I do not *know* what Jim Dawson is using .. nor doubt his numbers,

    But his 3 um accuracy is a very high nr for accuracy (not res. or feature/local accuracy), and has multiple caveats.
    Is the surface finish that good ?
    But all the advice re: servos, scales, etc. he made is spot on.

    As Jim intimated imho, accuracies of 0.02 - 0.01 mm in mechanical resolution are perfectly reasonable.
    That is resolution, not accuracy.
    And very much better is easily done, for little more money.
    Using a secondary measurement, like glass scales, easily allows one to gage to much better positioning//accuracies, 1 micron being quite possible.

    I use AC brushless servos on my 12" lathe with gibs (so far).
    750W, 220V, 10.000 counts, 1:3 belt drive via HTD8-30 mm belts.
    Theory - 0.2 microns step size. Approx 2000 kgf push for one step of 0.0002 mm. 32 mm screw on z, now, hiwin clone (TBI).

    It bounces, is not reliable.
    The belts and pulleys are not good enough - they are not linear.
    The gibs cause windup, sticktion causes tension, the bearings in the mounts are stock and are, in a word, crap.

    The x axis screw system I made is 18x more stiff than stock. Rigid-rigid mount in tension. 0.750" vs 15 mm or so.
    But it is loose, not linear.

    1.
    Yes. It is perfectly feasible to get about 0.01 mm or better in resolution and calibrate accuracies to better than that.
    As Jim said.
    2.
    The 400W servos are the sweet spot. As Jim said.

    All that I consider problems, are mostly less than 0.01 mm local errors in incremental positioning.
    You would not notice them, especially on a mill.

    Fwiw..
    I use an optical encoder DTI, electronic, external, to verify positioning.
    It is easy, and quite reliable, to position to 1 micron via manual MPG.

    This works well, but is useless for me as a business.
    I need the machine-co-ordinates to closely match the theoretical position.



  16. #16
    Registered MikeC8's Avatar
    Join Date
    Dec 2013
    Posts
    290
    Downloads
    0
    Uploads
    0

    Default Re: Accuracy I can expect from a RF-45 or similar cnc conversion? Bang for $$ point?

    This is just my opinion, but with todays market I would only look at doing your own conversion if taking on a project is something that you really enjoy. So the real question is do you want to start making CNC parts right away, or do you want to take on a project that might challenge you, and make it your own?

    I bought my machine already converted. However I have slowly been redoing just about everything on it. And if I did not have the machine I have today, I would probably buy a tormach 440, as that seems like the perfect size for the parts I make. I have yet to make use of my 19" x travel. And I have only used the 8" y travel once or twice.

    Work: Hurco VMX42/VMX50 - Shopsabre 4896 - HSMworks with Solidworks
    Home: RF45 with Ajax CNC Controller - Fusion 360


  17. #17
    Member CS900's Avatar
    Join Date
    Aug 2006
    Location
    USA
    Posts
    670
    Downloads
    0
    Uploads
    0

    Default Re: Accuracy I can expect from a RF-45 or similar cnc conversion? Bang for $$ point?

    Jim,
    Do you have a build thread or any more details on your linear scale set-up. I'd be very interested in learning a bit more about it.

    to the OP, for what it's worth I have a PM45 that I've converted over with some closed loop steppers (these ones specifically), have mapped my x and y ballscrews and I can hold +/-.001" without much effort at all. But like ninefinger said, it's all about what you can afford to get away with.



  18. #18
    Community Moderator Jim Dawson's Avatar
    Join Date
    Dec 2013
    Posts
    5717
    Downloads
    0
    Uploads
    0

    Default Re: Accuracy I can expect from a RF-45 or similar cnc conversion? Bang for $$ point?

    Quote Originally Posted by CS900 View Post
    Jim,
    Do you have a build thread or any more details on your linear scale set-up. I'd be very interested in learning a bit more about it.
    I really don't, but I can create a thread and go into some depth. Mechanically it's pretty simple, mount the sensor and slap on the magnetic tape. To the controller it looks just like any encoder.

    Here is a picture of the one on my mill X axis. Mounted on the front of the table, been there 3 years with no problems. This particular sensor is a Renishaw, but I have had excellent luck with the Ditron units also. The Ditron units are identical in size and performance to the Renishaw units, and considerably less expensive, about $120 / axis.



    Attached Thumbnails Attached Thumbnails Accuracy I can expect from a RF-45 or similar cnc conversion? Bang for $$ point?-img_0126-jpg  
    Jim Dawson
    Sandy, Oregon, USA


  19. #19
    Member
    Join Date
    Mar 2017
    Location
    Australia
    Posts
    98
    Downloads
    0
    Uploads
    0

    Default Re: Accuracy I can expect from a RF-45 or similar cnc conversion? Bang for $$ point?

    Quote Originally Posted by CS900 View Post
    Jim,
    Do you have a build thread or any more details on your linear scale set-up. I'd be very interested in learning a bit more about it.

    to the OP, for what it's worth I have a PM45 that I've converted over with some closed loop steppers (these ones specifically), have mapped my x and y ballscrews and I can hold +/-.001" without much effort at all. But like ninefinger said, it's all about what you can afford to get away with.
    Thanks CS900.

    I am very interested in what Jim recommended and I've been researching it since. It certainly takes a bit to get your head around, I am currently looking into Kflop (~$250) as it seems a more economical alternative to Galil ($ much more ?).
    Steppers in Aus are $200 / axis compared with servos which are $550/axis (starting). I would like an electronic system to start with steppers and linear encoder feedback, possibly upgrading to servos later.

    Mechanically what is necessary for good accuracy? I am thinking of C7 ballscrews with anti backlash mechanism + Hiwin rail CLONES. Genuine Hiwin I hope I can avoid, unless I win some lottery.
    That's just to get me up and running, later on I will probably go for a new spindle and accessories.



  20. #20
    Member
    Join Date
    Jan 2008
    Location
    Australia
    Posts
    1523
    Downloads
    2
    Uploads
    0

    Default Re: Accuracy I can expect from a RF-45 or similar cnc conversion? Bang for $$ point?

    You can get genuine HIWIN for pretty good prices (I use "BST automation" on aliexpress).



Page 1 of 2 12 LastLast

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
  •  


About CNCzone.com

    We are the largest and most active discussion forum for manufacturing industry. The site is 100% free to join and use, so join today!

Follow us on


Our Brands

Accuracy I can expect from a RF-45 or similar cnc conversion? Bang for $$ point?

Accuracy I can expect from a RF-45 or similar cnc conversion? Bang for $$ point?