Ball Screw Accuracy DIY Improvement
Vložit
- čas přidán 24. 08. 2024
- In this video I find out the error that the ball screw of my CNC Router has and apply a pitch compensation in the software to improve the accuracy. The first step is to calibrate the linear glass scale to the DRO. For this I use a 100mm gauge block. Next I drive the axis a known distance, 10mm to start with and compare the DRO values. Unfortunately my C7 Ball Screw is not within the manufacturers tolerance. Next I use the EDING CNC Software to apply a pitch compensation to the Ball Screw Error. I am happy with the result, as I am able to improve the error from 0.15mm down to 0.02mm.
If you like check out my website for some projects:
www.JbWorxStudio.com
Damn! I wondered now for a few days if such a software compensation is possible. Will surely come back many times to your channel while building my CNC… Thank you so much!
Great! Good luck with your
built !!!
you are doing a very rare quality work here, so thank you for sharing these videos and i was wondering what software are you using to compensate for the BS errors and can mach3 do that?
Thank you! I am using EdingCNC and I am really happy with that software. So I know Mach3 can do backlash however I think it cannot correct the ball screw tracking error. But best is to look in the manual I am not 100% confident.
@@JBWorx Any Chance you could do a video about EdingCnc? I can certainly see the benefit of having both the control software and controller being made by the same company.
It is on my idea list but I have not really put anything together for it. The new card they have is pretty sweet lots of selections and there is also a new software version out but I have not installed that yet.
@@JBWorx From everything I'm reading Eding cnc seems like a fantastic solution. I'm 90% sure it's going on my next machine that I'm currently building. So far from ATC, Vac control, to Z axis and XY compensation, it is ready for all of it all. But if you ever decide to do a video on it and your wiring too, it would be most appreciated.
I think you will like all the bells and whistles Eding offers and yet it is relatively simple. The macro programming is more involved as I ever wanted to get into the controller but for z height and 3D probing there are some things to know.
Also, when you first used the dial indictor without the gauge block, wheren't you opening yourself up to cosine errors (or whatever the name is)? I mean, the dial indicator has to be 100% parallel with the glass scale for you to use it like that, no? When you switched to the gauge block, then that becomes the reference and the dial indicator can be at an angle as it no longer measures distance but just works as a "zero". Not sure, but I think that's the case?
Yes you are correct. And a small angle makes quite a difference once you start measuring in microns … so does temperature
Great video
Thanks for the visit
You are making a great job with these videos, thank you. Two questions: do you have the servo on z axis with a brake? What do you think about gas spring to compensate the z weight?
Thank you Marson, yes I am using the servo with the break. Less chance to ruin a bit. The gas spring is not a bad design it depends on the weight of your z axis. A counter weight will work too. The servo allows for a directional torque boost as long as the positioning is good in both directions I would not add anything to the axis that increases the inertia. Have a look at the servo tuning video on my channel
5:51 the missed tolerance could be dro calibration, not the ball screw.
1:06 i have the feeling, your micrometer is not perfectly aligned to the direction of travel, so it shows more than the axis actually moved. just like the dro in the end, which is also showing too much travel.
The calibration was done using a 100mm gauge block. This is the most accurate thing I own in that distance. Checking parts made after the software compensation I can tell the error on my machine has greatly improved. Actually the biggest error for this set up would be that the glass scale is not parallel to the axis travel.
@@JBWorx i was referring to when you matched to dial indicator and the dro in 1:15. if the dial indicator is not linear to the axis, it shows to much. since you matched the dro to the maybe false dial indicator, it might also show too big values. that would mean, your backlash is more like 48um and your ball screw might be precise.
Ok I see what you are saying. No 1:06 I did that all over using the 100mm gauge block. And what you see as not aligned is the round part / display can be freely rotated so it is readable from any position. The true calibration of the DRO calibration starts at 1:20 to 4:47. And the dial indicator is not used for any measurement it is just used to get to the same point reference Of 0 the dimension comes from the gauge block itself. So even if there is a misalignment of the dial indicator in this set up it is not relevant.
I'd feel better about this if I trusted that DRO. I know you calibrated it over 100mm, but the fact it was off at all makes me distrust it over any other interval. I'd want to check it over the full travel and with shorter lengths at different locations on the scale. You've got me shopping for bigger gauge blocks now 😞
Lol 😂 the glass scale in these is rather accurate. All you have to do is mount it dead parallel to your axis and find a standard in your shop that is of known accuracy. I talked to a metrology guy at Zeiss and he assured me the glass scale is good just by the means they are made they are accurate. However most people make a mistake mounting them. If you want to measure down to 0,001 mm you first need to set your environment to 20C. That is how sensitive of a measurement we are now looking at.
@@JBWorx I daresay if that was a Zeiss DRO and not a Chinese import you wouldn't hear me complaining! 😀I get that the lithography that is used to make the glass itself is pretty accurate and reliable, but there are more steps in the system before you are getting readouts. It is likely correct, I know I sound paranoid, just it's the sort of dependency chain (cheap DRO correcting cheap ball screw) that would keep me awake at night. Or even more likely, to get me to buy a bunch of long Mitutoyo gauge blocks that I really can trust to confirm things.
Lol not a bad view. So let’s see on the next cut what my Mahr digital dial indicator has to say.
@@JBWorx Mahr I trust! I have 1 micron and 0.5 micron Milliness units. Very nice stuff.
Yes I like them as well. However using something like this around a CNC router can drive you up a d down the wall. 0.01mm is just fine. Or for the woodworking guys 0.1mm will even do. However it’s nice to work with precision instruments and it makes me happy to improve my machine.
Coul you tell me about wiring diagram system into control box. Thanks you
This glas scale is a stand alone unit it is not wired into the machine itself. It comes with a plug and simply gets plugged into the DRO box. Very easy to set up.
Looks like an error that got transferred from the machine that rolled the screw or a very slight bend ? Seems to go over and under, similar to the leadscrew issue I'm trying to fix in my lathe cross slide.
Yes, rather typical for a rolled ball screw in that accuracy. Interesting however that it was out of spec. I guess most people will never measure it so the defect will never appear.
@@JBWorx if you're turning precision tapers, you'll discover it very quickly 😂 rather any linear surface that's turned with dual axis moves will have artefacts of the inconsistent positioning.
Hmm your comment gives me an idea … I could check the screw on my cnc lathe … I think my glass scale might just be a little long for that but something I should do.
@@JBWorx easy test would be to turn a 45deg taper on stock like 25mm dia or more. Check against a surface plate, gauge pin or a precision ground parallel.
Thanks I will give that a whirl next time I make a batch of z touch Probe housings.
I am using Eding software but I cannot get the pitch compensation feature to work. The difference is showing up in the Eding DRO but there is no correction at all in the axis travel. Eding have stopped replying to me😂
Lol 😂. So you have turned on the feature itself on the .ini file under say joint_0 pitchcompensationOn=1 ? Also try -1 mine is set to -1 don’t remember why I did that I think it applies the values different. I am sure you have that though if it shows up in the DRO. Sorry no idea why it would not physically move oh yes what is your backlash compensation ? Is that possibly playing with you ?
@@JBWorx I have manually taken up the backlash, I have even preloaded the ball screw nuts to further eliminate backlash. Mine is also set to -1 in the ini file. Its very frustrating indeed. 👍
Ok understand but there is also no value for backlash in the set up table of the axis correct ?
@@JBWorx Yes correct. When you set up your pitch compensation table, did you alter the number of positions as shown in the table? I altered the number of positions in my table. Could this be the reason why my pitch compensation is not working?
I did as well. I have only 3 line items in there.
Does your machine use servo motors or stepper motors?
I used to have steppers you will see them in my early videos but changed over to servos later on. No regrets I like the performance / dynamic so much better.
@@JBWorx Ahh I see. When you had the steppers, did you have the linear pitch compensation active with the steppers?👍
No I did that at a later time. I think there is a step by step process to get the machine more accurate. Using good quality bearings to hold the ball screw in pace is one of the first changes I made. Then I re-balled the guide blocks next I stiffened the Z-Axis and spindle holder and also used the linear compensation. I have one item open. The x-Axis has a slight bend in it, I could try to mechanically improve that but my controller can compensate for that as well. I will see if I can get that implemented.
U need to upgrade to a C5 grade ballscrew or better yet a C3 ground ballscrew to get better acuracy.
Well I did it on a C7 😀
@@JBWorx As a experiment this is a great resource for learning.
If U have time please post a link with the DRO unit.
Looking forward to the next video.i also have some isue with the limita switch on my router.
@@DizzyWood_shop you can buy the DRO with different length glass scales on Amazon as well as EBay. Check ToAuto DRO.
I plan to go through 4 different types of switches and will post the results.
@@JBWorx I think this subject (end stop switch comparison) is a winner, and as much as I want to see it soon, my guess is that it has a good chance to go viral as it applies across a lot of popular domains (3D printing comes to mind) and testing more than 4 switches might be in order. I'm sure you know that there are some manufacturers that have high precision switches that are specifically rated for this use, not cheap, but maybe worth putting up against the cheaper commodity units I assume you will be working with.
Well we shall see. Usually my videos are send into the CZcams abyss after one week lol. Yes there will be one precision switch in the game as well. I think it’s 3 microns repeatability and 3 million switching cycles guaranteed.