Can backlash give a cumulative error?

[Gerald D]

. . . . . on my friends machine there was so much back lash that when he cut a sign you could see the letters get smaller as it went down the sign, it was like it was missing steps, and when it was done cutting and returned to zero it was off by more than .5 of an inch, (in the defense of the geared motors they were really large files) I believe they were hundreds of thousands of lines of codes long, but it caused errors in the machining from the back lash . . . . . .

That is NOT backlash - that is lost steps from overloading the stepper motor. (or an under-powered stepper motor)

[chopper]

Gerald,
If you do the math on this you will see what I mean, if I have a file that is say 200 thousand lines of code long and my machine has a back lash of .005 of an inch, with every direction change there is a possibility of gaining or loosing .005 of an inch, so worse case,... 200,000 x .005 = 1000.inches,( my friends machine had at least ten thousands backlash in each motor) so you see it could be off by a long ways, the cuts were smooth no lost steps,no jagged moves,...
just backlash ....I repeat they were not the oriental 7.2's
the steppers that were used were setup correctly, not overworked or underpowered...they were switched out to kelling 450 ozin motors, and the belt drives, the motors were the same size as the geared motors no configuration change was required, other than the steps since He changed out the pinions for smaller ones and the ratio changed...and after the changes were made (new motors and drives) this never happened again...
//chopper

[bradm]

Chopper, backlash is not cumulative. if you gain .005 one direction, you lose it back the other direction. It's just the slop. Only lost steps (or shaft slippage) will accumulate.

[Gerald D]

There is no maths to be done. You can not accumulate backlash (it does add to itself).

.005" of backlash will stay .005" anywhere on the table, even after a year of running 200 000 line files every hour. That is backlash.

If you are getting an error that accumulates with time, that is not backlash, it is something else. In a stepper motor system it is most likely to be "lost steps" due to magnetic field slippage. Or it could be a slipping grubscrew.

Would you agree that a slipping grubscrew and backlash are two different things?

[Gerald D]

Brad types faster than me! :)

[chopper]

Brad,
I disagree it accumulates, I have seen it happen, just because you loose in one direction you may not gain in the other, if the machine is cutting in one direction and jogging in the other this type of gain or loss can happen quite easily and if this occurs over and over it will accumulate in one direction or another, or in several directions, the slop is a constant as in how much of it exists, but there is no rule of physics that says you will gain and loose in equal amounts, I also agree that you will have gain and loss just not of equal values that is why when it went back to zero it was only off by .5 of an inch not a 100 inches
and there was no shaft slippage on the machine either..
//chopper

[domino11]

Here (http://en.wikipedia.org/wiki/Backlash_%28engineering%29) is a good description of Backlash on Wikipedia. From the description is does not sound to me like it is accumulative. I thought that was one of the benefits of using the spring loading on the rack system, was that it pre-loaded the rack and pinion contact point and minimized the backlash there.

[bradm]

Chopper, imagine a gear system where the stepper motor shaft is welded to the casing and cannot move. That gear system would still exhibit backlash (in response to external pressures), but clearly cannot accumulate, since the shaft is welded in place. It can still rock back and forth, and that's backlash.

A great real world example of this is the "Park" position on an automatic transmission. You can get out and push the car back and forth from one extreme of the backlash to the other, but you can't get it to move down the road unless you get the tires to slide.

Whatever the effect is that you observed must have another name. From your description, it sounds like slippage of one kind or another.

[chopper]

Brad,
I think you are not following what I am saying.....
I know that when you have backlash on the machine it is a given amount which may grow over time but it does not accumulate in that manner, not on the machine, but it accumulates on the part being machined,... to use your analogy of the parked car, if the car is in park and your rock it back and forth and it slides slightly it is not in the same spot and there will be an error in the part being machined...and then multiply that by 200,000 moves
//chopper

[bradm]

Chopper-
It's been clear that a lot of us have been talking past each other on this. I think I now get what you are trying to say, which is that the momentum from extra backlash carries through into slippage, and slippage can accumulate. I can agree with that, although that effect is definitely *not* called backlash, it's slippage of some kind. Going back to the car analogy, the term backlash *cannot* apply to the sliding of the tires against the road. Heath's link is pretty clear on that.

Getting back to the question at hand though, which is if you can have motors on a MechMate that have enough torque to handle normal situations, but which experience slippage caused by accumulated momentum from backlash? I think you may be on to something with that thought, and apparently it's borne out by your experience. Thanks for hanging with us to get it clarified. Some of us can get a bit pedantic around our terminology. :)

[chopper]

Brad,
now that you are starting to see what I am talking about I will try to clear this up a little better, if you have a known amount of backlash, say ten thousands, in all drives, and you are going to cut a part, you zero out your machine on the corner, and auto zero the bit you hit run on the file the bit moves over to where it is to start, at this point it doesn't have a load against it, so it is unknown where you are in the range of the .010 backlash
in either X or Y your bit drops in and now forces are applied against the bit...
the bit executes a square in the material being cut say 1 inch square, but becouse of the backlash the square is off by .010 in the Y direction
and .005 in the X, then the bit lifts out of the material .010 off from where it should be in the Y and .005 in the X, when it is out of the material the forces are no longer against the bit, so it may move back a little or not, the bit moves over 2 inches to the left and cuts a circle when it drops in it is .002 off from where it should be but that is multiplied from it being off .010 in the Y and .005 in the X, when it finished the square, now after cutting the circle it is out of round by .005 in the Y direction and .010 in the X, and that is multiplied by the fact that it was off .010 when it finished the square and off .002 when it started cutting the circle, then it picks the bit up and moves up 3 inches to cut a triangle, when it drops in it is off by .009 which is multiplied by all the previous errors etc.etc..as you can see this has an accumulative effect,and remember this is going on with all the axises... and on and on it goes till it is done cutting, and it goes back to zero, but as you can see there is no way it will hit the exact zero again,this is because of the accumulative action of backlash, and the larger the file the more apparent this will become,
//chopper

[bradm]

Chopper, I think the point of disagreement here is that after the stressed bit lifts from the material; it should go back to a neutral position similar to its initial state. As soon as it does that, there is no possibility for accumulating error. If there is accumulating error, it occurs while the bit is under stress, not after it has been relieved (or reapplied). If it occurs under stress, it is due to slippage, because backlash is the absence of force, not the application of it.

Going back to the car analogy, you can't make the tires slide while the car is in the middle of rocking from one extreme to the other; only when you are all the way forward or backward in the slosh. The slosh is the backlash. The tire slide (even if causd by the backlash) is slippage; and is the only thing that can accumulate.

[sprayhead]

To me what matters if there is "play" in any drive mechanism is....

All the steps "lost" with the backlash will accumulate...

If your motor puts out 10 steps to "get to the other end" of the backlash, and your Gcode does 100 direction inversions, you end up loosing 1000 steps...

1000 steps gone, it will show in the part very well

F.

[Gerald D]

Francis, you also have it wrong.

[Greolt]

Backlash itself will not accumulate. It is simple physics.

However if using Mach3's backlash compensation there is another factor that MAY come into play.

Backlash compensation does not honour the acceleration settings in Motor Tuning.

So if you were to have too aggressive settings for backlash compensation, that in itself can cause lost steps.

So I guess it would be feasibly possible for those lost steps to accumulate. Although I would have thought they would even them selves out.

And this would be an accumulation of lost steps not backlash itself.

Greg

[sprayhead]

it hurts, just back from kneeling on corn.

With a sketch on paper I realized that it indeed doesn't accumulate.

f.

[Gerald D]

Thanks Francis. A simple sketch does indeed show the simple physics. Driving an old car or truck with backlash (slop) in the steering wheel also shows it - the wheel spokes alway go back to the same position when driving straight ahead.

Here is a sketch.

8099

Red is the driver, blue is the follower. Red needs to move left or right to drive blue to the left or right. The red "finger" is a very loose fit in the hole of blue - the gaps either side are called backlash.

The size of each gap is 0.003" (the max "error" between red and blue will be +/- 0.003" even though the "total" backlash is 0.006")

Red is driven with a stepper motor, it needs 10 steps to move 0.003"

If red moves backwards and forward by less than 10 steps to a side, and does this 200 000 times, and blue be more than 0.003" offset from red?

8101
8100


Does is matter if red moves more than 10 steps at a time.....can blue ever be more than 0.003" offset to red? Of course not.

(The 0.003" and 10 steps values are approximately what the guys with OM 7.2 geared motors will be seeing, if the gearboxes are in good condition)

[chopper]

I am sorry but Francis was correct,
I do not disagree that there is a constant in the machine, and that the backlash dose not grow on the machine,
where the issue is is in the part and the program because of the backlash the part will not be machined exactly to the program,( the backlash will and does cause deviation from the program in an accumulative manor) and with each error it will grow, so when the program thinks it is back at zero it is not it may be off a little or a lot it may be so little you cannot see it
but it is there, I tested this when I had the 7.2's on my machine, when I was driving 20 tooth pinions the amount that it was off was minute, but when the 30 tooth pinions were installed it was visible, it was not as bad as my friends machine, but it was there, if I was cutting wood I probably would even notice it with the 20 tooth pinions but with the 30's I would, the way I found this out was I wrote a program to cut a part out in .250 aluminum, but when I went to the shop to cut it the metal was slightly thicker than .250, I ran the file and cut the part, but it didn't cut all the way through
so I rest the height of my auto z plate to cut the remaining .010 off the part and re ran the program and when it dropped into the first hole it was off by a few thousands, not a big deal but I noticed it, I checked the part to make sure it was clamped down correctly checked for loose pinions etc, and nothing was loose. this is what caused me to install the 30 tooth pinions, I was thinking that they were going to run smoother on the rack because of the bigger dia. (read that here on the site) after putting them on the problem was more evident than ever, I thought this was a fluke, checked the whole machine to make sure nothing was loose,( went through all the math for the steps and all my settings in mach no errors) ran the file again, it was off again, after repeating this several times without change, I decided to go back to the 20 tooth pinions, and I ran the file again, it was off again but not near as much as the 30's were, this is when I decided to build the belt drives to correct this, after the belt drives were installed I could cut the same file in the same material and re run the program and never touch the sides of the hole that it dropped the bit into, I repeated this many times to make sure that I was correct, and I was, I can still do it now, after I had this success,I told my buddy about what I did and I agreed to build him a set of drives, after he put them on he was astounded as to how well they worked he also tested like I did, no more backlash, no more problems,
the backlash issue was corrected on two different machines built by to different people in two different states, who had an accumulative issue with backlash, there were no other issues with either machine no loose pinions, mounts bolts etc... the only thing that was removed was the backlash,
I don't know how to explain this any better, you can say it isn't so, heck you can call me a liar, but it doesn't change the fact that the backlash
accumulates, I have proven this to myself and to my friend, and others that were there when this all took place and quite frankly you cannot convince me other wise I have seen this same effect in mills and other machines,
just because you wont except it doesn't mean it isn't there..
//chopper

[bradm]

Chopper, no one is saying that you haven't observed _something_ accumulating.

We are pointing out that calling whatever is happening "backlash" is wrong.

Because we are getting hung up on that point, it's impossible to have a productive conversation about what effect is actually occurring.

Can I convince you to simply stop using the term "backlash", and substitute something else, like "accumulating error", or "factor x", or "unknown"? At that point, a discussion might take place.

So I hear you loud and clear that you have observed an effect with geared drives that causes a machine to behave in a non-repeatable manner. Further, you have observed that those same machines behave in a repeatable manner when changed to using a belt drive. No dispute with your observation. If all you want to do is get better performance, we can stop here - you have a recipe for this, and it have been shown to work well.

However, if we want to sort out what is actually happening we need to go farther. Since we all know that the 7.2s cannot be rebuilt into belt drives, the gears versus the belt are not the only change here, at the very least there is a motor change as well. There is more than one variable here. It's also true that there is no way to directly observe or measure lost or slipped steps, so we cannot simply rule them out.
So attributing the observed effect to "backlash" is jumping to a conclusion. It so happens that the definition of "backlash" does not permit accumulating effects, so we're still searching for what the observed effect is, and what to call it.

Personally, I'm convinced from the evidence thus far that there are steps being lost (or gained) here, and I'm willing to believe that the backlash may be a factor in those steps getting lost. Testing for this would involve cutting materials with various amounts of resistance to see if the repeatability problem exists in all of them. If the repeatability problem does not occur in say, air and foam, but it does occur in acrylic and aluminum, then is caused by some effect other than backlash, because the amount of backlash should not vary in response to the amount of resistance.

[chopper]

Brad,
I am not hung up on the term backlash, but to deny that this is not an accumulative effect of it is ubserd, ( if the backlash is removed and the "effect" stops, and you put the backlash back in and the "effect" comes back that would have a direct relationship with the backlash ) you can call it what you like it doesn't change the fact of what it is, it is not slippage, missed steps, backlash does not accumulate on the machine that is accepted, and understood...
that is not what I am saying......
and of course backlash will vary on the amount of resistance, if you have forces pushing in one direction your backlash will move to the other but it depends on how much force is applied, it will take a given amount of force to overcome the weight of the gantry or Y car to change where the backlash is, or what state it is in... all the way to the left or all the way to the right or somewhere in between
I guess we can agree to call it "accumulative backlash error" and yes I am jumping to a conclusion, based on what has happened in a repeated test
on two machines, with the same results, if you want run the tests yourself on your machine see what happens,
I am not arguing here, for me it is cut and dried, I am not the one who put this question into play, I am just stating the facts...
//chopper

[domino11]

Chopper,
In your case though, since you changed from the OM geared motors to new steppers and belt drives, I do not think you just merely removed the backlash. Torque curves are different (2 different motors), gearing is different, step signal frequency from the computer is different. There seems to be many factors that have changed when you changed drive systems. Not just the backlash. Would you not agree?

[Greolt]

Unless you lose steps there is nothing to accumulate. (besides some mechanical slippage)

However those steps got lost can be argued to the end of days.

Backlash in a round about way may cause lost steps but backlash itself can not accumulate.

Mach neither knows nor cares about mechanical issues. All it knows is electrical pulses which a driver converts into motor steps.

The gcode tells it to move 1" that way, so Mach puts out the correct number of pulses to be translated into steps and move that distance.

What Mach does NOT do is say to itself,

"Oh, because of some mechanical issue the axis did not quite get there so I will add some steps"

Mach neither knows nor cares.

Greg

[riesvantwisk]

Chopper,

out of curiosity, if you air surface the table, does your spindle come back at 0,0 ?

Ries

[chopper]

Chopper,
In your case though, since you changed from the OM geared motors to new steppers and belt drives, I do not think you just merely removed the backlash. Torque curves are different (2 different motors), gearing is different, step signal frequency from the computer is different. There seems to be many factors that have changed when you changed drive systems. Not just the backlash. Would you not agree?

I agree with this to a point, the motors are a different brand, but about the same power, I do not know the difference in the torque curve but I am sure they are similar unless the 7.2's motors are complete junk there shouldn't be any difference, a stepper is a stepper, in fact the motors I am using are about the cheapest steppers you can get, the frequency may be different but I wouldn't say that many factors have changed, one could believe that there is more than one change in the machine but in reality the only real change,that is the removal of the ghost in the machine "backlash"
I really don't know why this is so hard to accept? in fact it is quite simple...
I also do not understand why some are trying to be politically correct
with the term backlash, anyway once you get down to the facts it is what it is I didn't invent it, it is a byproduct of using gears or at least cheap gears, and you are also correct that the gear ratio has changed from7.2 to 4 to 1,
if anything the cuts should have been better with the higher ratio since the resolution would be tighter, so I guess what I am saying is the ratio change and motor change are insignificant, and the logical deduction would be,
that the backlash was removed..
//chopper

[chopper]

ries,
when my machine goes back to zero it is right on no matter what I cut,
air, wood , steel, aluminum, plastic, stainless steel, I cannot go back to the 7.2's to check they are long gone,

you see I do not have a problem, I am trying to help others with their problems, and it turns into me defending my claims,
quite frankly, I want all of you to use the 7.2's. and when you cannot hold your tolerances on your jobs you can hire me, to cut them for you, because you have "skittles" in your machine heaven forbid we use the term dare I say it ............backlash.......
//chopper

[chopper]

greg,

Backlash in a round about way may cause lost steps but backlash itself can not accumulate.
how many times must I repeat myself?
I agree that backlash on the machine dose not accumulate, but in the part it causes an accumulative effect which is caused by backlash, it may not be backlash but it is caused by it...that is why I called it "accumulative backlash error"

Unless you lose steps there is nothing to accumulate. (besides some mechanical slippage)
greg,
you are absolutely incorrect here, if you go back and read you will understand what I am saying, the errors caused by the backlash do accumulate and you cannot prove that they do not, on the other hand I proved that they do exist, and if you test it you will see that it does ( if you are honest with yourself )
//chopper

[domino11]

Chopper,
I dont think anyone is trying to get you to defend your claims, rather just get the correct mechanism and cause for the error that we are talking about. Your machine does function well with the belt drives and the motors you are using. Nobody can argue that. :)

[Gerald D]

Chopper, I have a big problem with your participation at this forum because you only make posts about: geared motors have backlash (true) that cause cumulative errors (false) and that this is cured by replacing them with belt drives (false). Your tone in your choice of words is out of line and you have resorted to shouting by increasing the size of your text to factor 5 (I have reduced it again).

So, I am sending you to the cooler for a few days.

This thread will now also be closed because it is a waste of time.

[Gerald D]

Folk that want to pursue this argument can get it out their system at Need help with backlash...please (http://www.cnczone.com/forums/showthread.php?t=97003)on the CNCzone.

[chopper]

post moved from elsewhere


This will be my last post and I am sure Gerald will delete as soon as he sees it,
it is kinda hard to stay here with things like this : Chopper, I have a big problem with your participation at this forum,
and when I come back all my pms are deleted, and in the meantime he is harassing me at the cnc zone, the problem here is Geralds ego cannot handle it when some one proves him wrong,
I would like to see all my posts removed since I am such a detriment here, along with all of my attachments there really is no need for them since according to gerald the belt drives don't work anyway.....
//chopper

[Gerald D]

That post will stay - it makes a good epitaph

[jessyjames]

For those with limited vocabulary, like myself...

EPITAPH: a phrase or statement written in memory of a person who has died, esp. as an inscription on a tombstone.

[Kobus_Joubert]

I am glad...now we can have Rest in Piece.:rolleyes:

[jimbo]

Hi,
I just joined this site. I have been on the cnc zone for awhile, I read some of this before it was deleted, I have no comment on the backlash issue. I am just really getting started in learning about cnc, I hope to learn a lot here,..it seems like a great place!!
But it seems that weather you liked the guy or not he sure could build a machine.
I do not want to overstep my bounds here with my first post so if I am out of line I apologize, but I don't see anyone else here able to make the cut this chopper guy could.
or in the verity of materials he worked with.
I never seen anyone cut stainless steel, steel, and aluminum with this type of machine before and get that kind of quality, when I get ready to build I hope that the machine I build will work as well as his does.

Jimbo

[Castone]

Jimbo welcome to the forum. Most everyones machine here could cut steel or alumn if they wanted to , but the machine is designed to cut wood. I have cut both alumn and steel on mine.With equal results to choppers. But it is much cheaper to run it on a machine that is designed to do so. I have never seen anyone banned for saying what they wanted to on the forum. But the guy seemed to have a one track mind and the only help he was willing to share was the same old story and get mad if no one agreed with him. He did build a good machine , he had good plans to show him how to build it , but with the plans provided at this site you can build a machine just as good as his and the machine will run better if you paint it blue. Good luck with your build.

[riesvantwisk]

When I was young an naive, I made a software tool that did 3D rotations of wire frame objects and due to the way I solved some problems within the software my 3D model on the screen got smaller smaller and smaller (or bigger) depending the kind of calculations I was doing.

The effect I was seeing was simply floating point math errors that got accumulated during each rotation I did on the screen.

This let me to believe the effect that was seeing might have been coming from software.

Now I don't know what software this person but it aswell be a software error rather then a hardware error.

Could it be??

Ries

[jimbo]

Castone,
if you have cut steel and stainless steel, and aluminum would you mind showing pictures of your cuts? and tell me what settings and type of bits you use to do this? I am excited to plan out my build, the information here is overwhelming though, it will take awhile to understand all of this stuff...
thanks for the help..
jimbo

[Castone]

Here you go Jimbo. These were cut with and .500" shred mill @ .030" depth of cut , at 90 IPM feed . I just cut these to try it out . The material is one inch thick grade 50 carbon steel. Of coarse like I said before it is not the smart thing to do. As a machine that was made to machine steel could take a .500" wide and .500" deep cut at over a 100 IMP , so I dont know why you would want to use a wood router to cut steel that is just crazy. But never less the MechMate will cut it. If you want to cut steel you might look at building a milling machine. I might put up a vid of the Mechmate cutting some of these blocks , but this is getting kind old now. We will see. I hope that this helps you out.

[sailfl]

Leo,

Nice cuts and those look like nice plates.

[domino11]

Leo, thanks for the pics. You certainly have your machine in top order!

It looks like Jimbo is actually Chopper in disguise. Funny how he took to Choppers cause. :)

[Gerald D]

Jimbo, and at least one more, shared chopper's IP address. They have all been shut down.

[Charlie]

On page one it was mentioned that if your using backlash compensation in Mach it could perhaps lead to these errors. chopper never cooled down long enough to see the post or say if he used the backlash setting in Mach or not. Mach surely couldnt get it right every time, different cut paths and so forth. Just guessing here but it does seem like he is pushing things to hard/fast on his table. Perhaps the torque curves of his new motors/belts come closer to "his" ideal cut force/speed.
Coming from someone who has the 7.2's/30pinion (all still new not in action yet) I am once again concerned about my motor choice. Because they are still new Im thinking of getting 20 pinion now. Gerald wondered if maybe the 30's were putting to large of a load on the gearboxes with the weight of the MM gantry.
So maybe.... *crossing fingers* if I break them in with 20's they might last longer? Who knows we shall see.
BTW backlash does not accumulate. *had too...*

[cab. guy]

Hay now, lets beat this dead horse.:D I think there's a easy to understand explanation using our virtual machines as an example that illustrates how backlash works in reality.If instead of .0005 backlash, lets just use a whole number as the example,say 1cm for the metric people.
In this model, we home the machine at 0,0, we then send the x-axis to 30cm.
The dro reads 30,but the bit indicates 29 because of the loss due to 1cm BL.
Now we send the x-axis back to 10 -- the dro reads 10, and now the bit is at 10.
The action back and forth of the x axis is backlash, followed by correction each and every time. It is the same for all three axis.Chopper, I know you got booted:( but If your reading this,in order for your cumulative error theory to be correct you would have to debunk this simple principle first (not easy).I'm not piling on,nor am I a political person,In fact I was impressed with your build job--excellent.You have some good ideas, but the debate is won by weight of substance not by personality.

[Gerald D]

Ron, there is no merit in pursuing this "debate" here. The days of arguing whether the earth is flat or round are long over.

Thread closed