Poll: Acceptable lost steps?

Hi all, I was just wondering what you consider an acceptable amount of lost steps when running your STEPPER based machines?

I have a cheepo X2 with a common cnc stepper based conversion ( 270 oz-in steppers) on ballscrews.

I am pretty sure I loose about .0005" to .0008". That is after about 340" of movement, about 11,000 lines of G-Code, running roughly 3-4 hours.
While that is not a lot of loss I will have to re-zero after each run.

SO HOW ABOUT YOU GUYS?
How much do you loose and after how much running of the machine?

Please, lets keep this thread on track.

Quote:

---

Originally Posted by cncuser1

I am pretty sure I loose about .0005" to .0008".

---

What are you basing that measurement on?

No. IMO, you haven't lost any steps. That kind of repeatability is what you might get from a servo turning a ground ballscrew. Perhaps that is your backlash. Not the same as lost steps. Could even be a variance in your limit switch behavior.

Loosing steps would result in a noticeable shift in one or multiple axis, often part way through the cut, caused by a crappy driver, improper mounting, or exceeding the performance of the stepper motor by trying to push it at faster speeds and accelerations than it can handle.

Quote:

---

Originally Posted by cncuser1

Hi all, I was just wondering what you consider an acceptable amount of lost steps when running your STEPPER based machines?

I have a cheepo X2 with a common cnc stepper based conversion ( 270 oz-in steppers) on ballscrews.

I am pretty sure I loose about .0005" to .0008". That is after about 340" of movement, about 11,000 lines of G-Code, running roughly 3-4 hours.
While that is not a lot of loss I will have to re-zero after each run.

SO HOW ABOUT YOU GUYS?
How much do you loose and after how much running of the machine?

Please, lets keep this thread on track.

---

If that is the change of dimension you are finding after a long period of running,have you considered the possibility of thermal expansion of the machine?I would think that to improve on your consistency would require a laser calibrated machine,running compensation software.

Zero lost steps.

If you are losing steps then at some point you will have a major loss of steps and break a bit or waste an important part.

I used to be in the design it well and open loop steppers are fine camp. Then I had a stall during a rapid on an important part, broke the bit and caused a lot of extra work. Now if I was building a new machine I would use closed loop steppers at minimum.

Yes I was running close to the limit of the machine, but I'd run it with those settings for years without an issue.

.0005" to .0008" seems like it might be in the range of the machine tolerance. Pretty good for a light machine, I'm pretty sure my Haas is not better than that over an extended run without very careful setup and a lot of attention to the finish passes.

To All: I used an edge finder to locate off the fixed edge of my vice. Took the average of four touches in X and Y. Ran the program ( 11K lines of Gcode, 3-4 hours), then redid the edge finding ( 4 times each axis). difference in before and after POSITION is @.0005"-.0008". These numbers aren't part tolerance, they are POSITION.

So out of 4 answers so far ONE has no lost steps. ANYONE ELSE keep track of LOST STEPS?

any lost steps is unacceptable in my opinion , and , if you were losing steps then it would likely be more noticeable with that amount of run time .
.0005-.0008 can easily be the difference between a cold machine and a warm one . A low resolution setting for your steppers may also accumulate a few increments out over time

All in all , that kind of repeatability in a converted mill/drill is likely the best a guy can expect . Even top of the line machinery is to a slight degree susceptible to heat expansion , etc

What sort of dimensional accuracy are you holding on parts? If I understand things, this is a hobby grade machine, with a hobby grade conversion, using steppers and I assume ball screws of a good but not ultra high precision tolerance. Driven open loop. And after three hours of machining you are having a positional drift of under a thou? That's an almost amazing repeatability over time for such a machine. Sounds like you are seeing thermal drift, normal wee bits of backlash stacking up, software interpolation in the step generation, all sorts of things adding up or perhaps luckily cancelling out to see only half a thou "error" in a low end machine. Heck, most home conversions don't reliably hold half a thou in accuracy, some may hold half a thou in repeatability for a little while, If I can run my G0704 all over the place, then go home, seeing the indicators all pointing to within a thou of the staring point makes me very happy. That's with double nut ballscrews of hobby grade import variety, nothing sexy.

Machines I've seen that are loosing steps don't do it in some subtle part of a thou way, they tend to make nasty sounds as they loose the steps and usually you could get an idea of how far the machine really is from the commanded position with a tape measure!

Metalmayhem: I haven't really kept track of temperatures but seeing as how this is a small machine the only thing that would really change temperature is the steppers. I agree that my loss is not alot, I think it comes out to 16 lost steps, not much, but when I started I was losing .02"-.06"! But that lead me to ask what is a reasonable loss? As it is now I'll have to re-indicate my zero for every part. I am using Gecko G540 so I have no control over microstepping. So you don't really lose position on your mill?

Stocker: On a good day I can hold a couple thou tolerances( small steel parts). I hope to be able to keep things the way they are now with losing less than a thou on spindle position over a run. I never heard the steppers knocking when I lost steps. What I did was move my cheap linear power supply 6" away from my control electronics. I changed computers, accel paramters, ect. I never actually found the cause. I did a bunch of stuff and the problem went away. If I keep things the way they are I'll be happy, I was just wondering what is normal or standard in these circumstances.

I've only tried it for about 20mins atm on my PM25 clone conversion.
Left to right X and forward to back Y at a speed of 2000mm/min (78.7 inch/min).
I work in mm and hit either side of the 0 mark by half a digit on my 0.01mm increment dial gauge (so 0.005mm either side).
So it's holding at 0.01mm which is 0.00039" in inches in this test.

But..... I have found that If I hit my soft limits at speed it does get knocked out (over-travels) by 0.2mm. So I'll have to make sure I don't do that.

Quote:

---

Originally Posted by cncuser1

Metalmayhem: I haven't really kept track of temperatures but seeing as how this is a small machine the only thing that would really change temperature is the steppers. .

---

most of your heat will be on the spindle head , this has an effect on large and small machines . I've owned numerous x2 mills that have thousands of hrs on each . I used to not shut them down for weeks on end and I've never had any problems with creeping .
They are turd machines and a guy can only expect so much out of a mill/drill that isn't meant to be doing what we make them do . All things considered I think what your getting are pretty good overall numbers

With that said , I didn't find any of them to be all that precise but they did what I needed them to do . The only times I've put dials to them were for adjustment , and part precision of a couple thou with good surfacing finishes were all that I needed

Quote:

---

Hi all, I was just wondering what you consider an acceptable amount of lost steps when running your STEPPER based machines?

---

Zero.
If you are losing steps, the machine is unusable.

Hello.

I agree with ger21. No step loses are acceptable.

As I mentioned in another post of yours that amount of error looks more like an accuracy or repeatability limitation..

You can improve your precision by setting your drivers to smaller microsteps adjusting your control to their new travel distances per step values but take into consideration that the overall machine speed will slow down.

Also some controls allow to set a TOLERANCE value which sets how far the actual trajectory is allowed to depart from the trajectory generator. Its value has to be carefully defined as a too small value might lead to trajectory overshooting, that is not locating the end of a segment. Too big a value leads to geometry distortion.

That tolerance value leads to an effect I know as ripple. There is no CNC or robotic machine without some degree of it.

I remember attending some IMTS show where the advertising of a machining center mentioned that if the Earth had been made on their machine the biggest mountain would be 3 mts high. Just imagine what the precision of that machine was. Yet the ripple still led to certain inaccuracy.

If your machine has an auomatic reposition function you can make the following test:

1. Move the machine a certain distance and then stop. Mark that position. Let´s call it point 1.
2. Move to the right say some 4 inches, stop and then return to point 1 using the repositioning. Mark that position as point 2. You will notice that while pretty close to point 1 this one isn´t precisely there.
3. Now move the machine to the left. Again certain distance. It is not important how much but again about 4 inches will suffice. Now againg using the auto reposition return to point 1. Again you will notice that it is now another point next to position 1. This is position 3.
4. For positions 4 and 5 you´ll have to do about the same in the front and backward directions. Label the new two points as point 4 and 5.

You should normally end with points 2, 3, 4 and 5 around point 1.

Just consider this. Your microstep setting defines certain amount of travel per step pulse. It is the smallest distance that the motor can give you. If required by the control program to move to a position that happens to be in between of two mechanically reachable positions the control system will usually decide on the closest one. Then again, it will not likely be the actual position you requested.

When designing products we usually have to take into account the possible dimension deviations due to machine characteristics as the parts that make those products require among other things to be exchangeable. That´s what sets the tolerances of any part that is to be manufactured as you can see in any mechanical blueprint or production drawing.

The link is of one of our retrofitted machines. Its tolerance is about 0.1 mm or 0.0039" with the arguable advantage that the motors are analog servomotors which leads to accuracy and repeatability to be nearly the same. Resolvers are used as position sensors with 24 bits of accuracy effectively giving the possibility of simulating an encoder with 16,777,216 lines of resolution. With some hardware and software tricks that amount can be quadrupled but in fact we only use 16 bit resolution and internally the control program reduces that a little bit more.

https://www.youtube.com/watch?v=B6FiInwcxh8

I might be wrong as I don´t know your particular requirements but again given your numbers I believe you are all set and ready to work your machine to its fullest.

Regards.

On a 5 TPI ball screw and a 200 count stepper motor with direct drive I think that is .001" per step.

I would say your not loosing steps.

Not only that but if your loosing steps they will add up over time so you will be out way more than .0008". Its rare to loose just one step.

After 8 years of making parts with steppers I still thank for small parts runs they are fine. I would call you system working correctly.

I would run it for two hours, set a zero and re run it for two hours. If you under .001" your good.

Arizona:20320 steps per inch.

My parts run for 3-4 hours, @11, lines. I can in the end live with my missed steps. I was just wondering what is normal?

Quote:

---

I was just wondering what is normal?

---

Normal should be zero lost steps.

ger21: A properly working stepper setup shouldn't lose steps, got it.

A back of the envelope calculation: if a 10" long piece of steel raises temperature by 5F, it expands by .0008".

Metalmayhem: It is summer time here and I sometimes have the AC on, sometimes off, so that might explain it.

Thanks all, I think I have my answer,and hopefully this thread may help some in the future.