Machsupport Forum
Mach Discussion => General Mach Discussion => Topic started by: JustinTime on August 06, 2011, 05:16:02 PM
-
I'm sure that the problem was answered before but I didn't know what catch-wore to use to find it.
If I give an input, lets say x2, it will only go to 1.989 on the DRO. If I than give the input of x0 it will go to perfect 0.000. This error accumulates and make the circles be not round. I can't compensate with backlash since it's not the machine that is the problem, it's Mach3. The same problem is on the y axis too. Did I forget to make a setting somewhere?
-
Mach will ALWAYS go the closest position it can go to the commanded point based on step size AND it does NOT accumilate error. IF it did then you would NOT have gone back to zero(;-) IF MACH has to hold short based on step size THEN at the next point it CAN it will add the hold short value back in.
I would guess that your steps per unit are not exactly perfect. When that is so then when you make a LONG move you see the error and when you move BACK to the start it will always be exactly where you started.
Just a thought, (;-)TP
-
I did not set the steps per unit, it was done by Mach when I calibrated the axis' with the axis calibration feature. When I made a test-run of 10" it went to 9.986, 14thou short. Went back to 0 and I measured the 'backlash' and it was 28thou (coincidentally exactly 2x14). I adjusted the backlash 0.028 and cut a 1" circle and I got two halves, offset by about 28thou. You could see the tick on the top and on the bottom. The ticks were toward the -x. If I run without backlash the circles are not round and are all over.
-
Like I said your steps per unit where not perfect to begin with(;-) DID you fix that problem?
How did you measure the backlash?
(;-) TP
-
Like I've said, the steps per unit were calculated by Mach3 when I used the axis calibration feature. The backlash I calculated, like I've said, by measuring the difference between the starting point and the endpoint when I went 10" on x and returned to 0.
-
The way To calculate back lash is to set up a dial indicator against the Spindle or tool in the spindle using step jog set for.001 steps back the axis up until you see the dial indicator move a full .001"( Take the slack out of the movement). Then single step it in the opposite direction dirction unitl you see it move .001" again in the opposite direction. the amount of .001" steps is your backlash.
There are also some settings for backlash you may want to read up on such as Shuttlespeed tha can effect your backlash corrections.
(;-) TP
-
Your way, TP, is for sure the better way of doing it but if you lack a dial indicator the next best thing is to use a digital caliper and measure it the way I described it. Theoretically they both should show the same results all be it the caliper way will, most likely, not be as precise since there is the human factor involved but that should be only a thou or two and I'm not worried about that right now. First I'd like to solve the general problem and then I'll teak it.
-
First you need to recalculate the steps per unit to get that part correct so it moves the correct distance per command.
(;-) TP
-
Oh man, if I had a buck for every time I did that! :) But again, I used a digital caliper for the lack of dial indicator.
-
OK do you have any info on the Axis drives such as steppers?, micro stepping drives? , gear reduction?, turns per inch on the screws, is it rack and pinion?
What are the settings you are using now?
The Calibration routine will work but you have to keep refining the settings until it is perfect.Most times it is NOT always a one shot deal.
Run the routine let it setup the settings then measure the error and adjust the settings by the % of error. Then measure it again repeat if needed.
ALWAYS move in the same direction to measure AFTER you take up any slack(backlash in the axis) That will eliminate backlash in your measurement.
Calipers will work just fine. Just be consistant on how you measure it. NOTE: some low end digital calipers are only accurate to +/- .005.
(;-) TP
-
Oh man, if I had a buck for every time I did that! :) But again, I used a digital caliper for the lack of dial indicator.
I think you may have misunderstood TP’s comment (recalculate the steps per unit). Calculating the steps per unit and using Mach’s axis calibration feature are two totally different procedures. The former will result in an exact correct value while the latter will be subject to measurement inaccuracies (and I would guess errors approaching the magnitude that you describe).
You will need to know the details of your system as TP mentioned in a later post, i.e., stepper deg/step, lead screw pitch, etc.
There is a spreadsheet at:
http://www.machsupport.com/forum/index.php/topic,16315.0.html
That will do the calcs for you.
If you don’t know some of your system parameters, then give us as much info as possible and I’m sure we can work it out with you.
Al
-
Thanks guys for the answers. I'll go through the procedures again and will come back if I don't get it solved.
-
Hi Yoram,
I expect you have already seen this post http://www.phlatforum.com/viewtopic.php?f=264&t=2984 but that type of backlash would perhaps be difficult to spot and may not occur at every move. Curious because I have never seen it before.
Tweakie.
-
Yes, I've seen it, Tweakie, and it peeked my interest. I'll have to look into it since the x axis has a belt and it is very high tensioned. I may try to loosen it up a bit and see what will happen.
BTW, did you happen to see my post here? http://www.machsupport.com/forum/index.php/topic,19205.0.html
-
has a belt and it is very high tensioned
Is it a timing belt or a v belt?
I would recomend that when you check the backlash you base it on one complete revolution of the belt.
You can set the steps per based on a distance of one rev of the belt and then find that it will be out if you check a distance where the belt
is say only half way thru it's complete travel. Steps can be lost or gained with timing belts depending on tension, but, but, but..... you will be hard pressed to find those steps.
Even a cheap indicator will be of value in the shop and can be found new for say $15.00 on sale. Splurge my friend. ;)
Similar problem as yours can be found when a screw profile is not linear, yes it will return to zero, but depending on how you rationalize the
steps per based on a distance you will have varing results. Let me remark that what I am talking about here are small variances.
You problem is not Mach, as i see it, but rather you system.
Been Wrong Before, :D
RICH
-
It's a timing belt, Rich.
I never thought of making sure the belt made a full rev and I will check it out, and, BTW, I did splurge and bought a dial indicator, at Harbor Freight, on Sunday. ;D
I found an oddity after I bought the indicator which I detailed in my thread 'coming out of jog error'. I wonder what causes that problem.