Hello Guest it is March 28, 2024, 06:44:45 AM

Author Topic: round circle with perfect backlash on straight moves  (Read 3709 times)

0 Members and 1 Guest are viewing this topic.

round circle with perfect backlash on straight moves
« on: September 06, 2014, 06:03:21 PM »

...sorry for the double post....

My backlash for 1" moves of 40" moves is within 0.0002" repeatability on x and Y axiz
(on a 28" x 48" working area)  slaved 900oz/in motors on the Y axis single on the X axis.
But whether it's a 1" diameter circle or a 8" circle... they are all out of round by 0.030"
What I think is over tightening the bearings/slides/screws/R&P parts....then a retest.... it's always 0.028-0.030" out of round on the same axis, (11 o'clock down 5 o'clock is narrower by 0.030)   regardless of climbing or normal routing directions.

This was tested with several circles inside each other, each running in the opposite directions.
ALL are out of round. so I even rotated the MDF test board just to validate it was not material stress related.
There is no (zero) racking (twin 900 oz/in motors) which might have made sense if backlash on one side were greater than the other
(trapazoidal resultant motion or reversal along the X axis is  the Y axis racked)

hhhmmmm... should my dual motor Y axis acceleration setting be twice the single motor X axis?
or perhaps half the X axis to compensate for power to weight of the gantry vs just the head on the X....
(or am I over thinking all this)

Now I know I read somewhere on these forums there is a difference possibly in absolute stop vs continuous speed....circle interp etc
An example visible to anyone is while making my dust shoe, one circle seemed to start and stop a lot, while others were a continuous motion/sound
This is visible at  https://www.youtube.com/watch?v=gr6VAfZCQlo
at the 1:54 mark you can here it starting and stopping cutting hard i nthe material, yet later circles are smooth and conitnuous.
I did not do anything different.... straight conversion from VCarve Pro to g-code, yet this one circle cut with the start/stop sound.

The MDF circle tests done a couple months ago now still bugs me as I've not been able to formulate any real logistics to hunt down, other than settings for the way MACH3 should/can be config'ed to cut circles.

for reference...the whole machine...  https://www.youtube.com/watch?v=KklafqyZpSY&list=UUNTCxkF_0pxnMa-g9HTVgxA

Any clues here as to where I should start testing to re-config this machine for clean round circles?

tia

ptegler

Offline Davek0974

*
  •  2,606 2,606
    • View Profile
Re: round circle with perfect backlash on straight moves
« Reply #1 on: September 07, 2014, 03:39:28 AM »
I would inclined to check / set my steps per with a dial indicator or vernier gauge on both axes.

One step out might do it.

Check over as long a distance as possible.
Re: round circle with perfect backlash on straight moves
« Reply #2 on: September 07, 2014, 09:22:54 AM »
What results do you get if you try small moves (0.005")?  Your problem may be axis "sticktion.''
John Champlain

Offline Sam

*
  • *
  •  987 987
    • View Profile
    • hillbillyhilton.com
Re: round circle with perfect backlash on straight moves
« Reply #3 on: September 07, 2014, 11:17:57 AM »
Quote
My backlash for 1" moves of 40" moves is within 0.0002" repeatability
Not exactly sure what you mean, but I think you mean 1" moves, repeated 40 times. Repeatability and backlash are two separate things. Think of repeatability as positional accuracy over a number of cycles. Backlash is the free play, or gap, between drive components when there is a change in direction. There can be a massive amount of backlash, and you can still have superb repeatability.
Quote
whether it's a 1" diameter circle or a 8" circle... they are all out of round by 0.030"
Davek0974's advise is great, and you should do that, however, you state that the deviated amount is the same, regardless of circle diameter. If the motors steps per unit were off, the deviation would not be consistent on different sized circles. Sounds like backlash now. That does not mean that your steps per unit are correct, and you should still check it. As Dave stated, use as long of a distance as possible. If you have a set of 6" calipers, use those. If you have 12", use those instead. Remember, move enough distance to take the lash out, and then measure in that same direction.
Quote
one circle seemed to start and stop a lot, while others were a continuous motion/sound
Most likely, the machine is doing exactly as the code is telling it to. There is nothing indicating a problem with the machine by this. The start/stop is segmented code (less than desirable in my opinion) being executed (G1), and the other pass is using circular interpolation (G2,G3). You can take a look at the code to verify. The reason behind the code being generated as segments, can be vast. There might be a setting in your CAM program that can have an impact on this.
Quote
should my dual motor Y axis acceleration setting be.....
When your trying to remedy a problem such as this, I would turn the accel down, to eliminate it from the equation. Turning it up may just introduce a new problem into the mix and make confusion/aggravation shoot through the roof. You can always tweak the accel later on, when the problem is fixed.

"CONFIDENCE: it's the feeling you experience before you fully understand the situation."

Offline Davek0974

*
  •  2,606 2,606
    • View Profile
Re: round circle with perfect backlash on straight moves
« Reply #4 on: September 07, 2014, 11:35:37 AM »
Could this be something as simple as a pulley or gear working loose on a shaft - this can create backlash which would cause oval circles.

The area to look at is the axis that shows the issue - if the circles are correct in the x-x direction but short in the y-y direction then look at the components for the y axis drive. A loose pulley can be a bugger to find until it falls off ;)
Re: round circle with perfect backlash on straight moves
« Reply #5 on: September 07, 2014, 01:58:56 PM »
both on moves of 1" and moves of 40" as tests moves....was referencing pulses per unit accuracies as well as repeatability of returning to zero.

long and short moves used to set/validate move accuracies and backlash calipers, rulers, lasers used... straight line moves on a single axis are dead on accurate (better than less than .0006") repeatability 

will go back and look at that code...was quite awhile ago  (Jan/Feb 2014 build date)  the G1 g2/3 makes sense

will minimize accell and run my circle test again.

I just spent the last two hours milling parts to minimize all free play in gear trains, bearing spacings, etc etc
will let you know how it goes.

thanks all.

ptegler