Trying to loosen the belt on my BP CNC.

madmachinst · Dec 16, 2022

OK here is the video of my worth a try hairbrained scheme If it don't work no time wasted.

Any suggestions?

CarbideBob · Dec 16, 2022

What is the problem???
I can not see what kind of belt/pulleys you are using. The sort of square teeth or the round teeth?
Trapezoidal or square looking is not so great at positioning and more for just drive apps.
Ratio and spacing distance looks good to me.
I am not understanding the question. We need more info.
The thing with this reduction is belts are not perfect tooth to tooth and the contact runout on a pulley is not the OD runout.
Bob
(Please take no offense from this post, so many questions and me still learning )

Jonathan Smith · Dec 16, 2022

Sorry, I can't answer your question directly but can advise the belt will not stretch. It might wear but won't stretch. Probably not even a cheap chinese knock off with a polyester carcass. Now my rant, why do cnc machine manufacturers insist on using motor mounted encoders instead of dro like scales directly mounted on the machine axis. Get rid of the fancy drive and belt problems. Use the traditional power feed systems. It would make converting manual machines much more straight forward too.

ChuckMTB · Dec 16, 2022

Did the accuracy issue start when you replaced the belt? Or have you just fired up the machine and trouble shooting accuracy. The step count can be adjusted in the control. Are you running Mach or another control? I didn't recognize the screen as it passed by in the video.
Stretching the belt won't change the pulley ratio.

madmachinst · Dec 17, 2022

ChuckMTB said:
Did the accuracy issue start when you replaced the belt? Or have you just fired up the machine and trouble shooting accuracy.

Its a retrofit. 1991 BP V2XT. I retrofitted with closed loop mach motion system. Never replaced the belt but did replace the servo. This meant adapter plate for servo and everything shifted a little to the tight end, IE away from the ball screw say .04-.07" . Originally thought cause belt was off those two pulleys for a long time. I guess not.

ChuckMTB said:
The step count can be adjusted in the control. Are you running Mach or another control? I didn't recognize the screen as it passed by in the video.

Mach 4. Step count can be adjusted, but throughout the travel of the screw, step count can be as different as 30 steps per inch. On 25000 steps per inch that comes out to like .0012

ChuckMTB said:
Stretching the belt won't change the pulley ratio.

I notice that with the belt stretched too tightly, beyond spec as I explained in the video, sometimes the encoder is fighting itself so to speak, and all kinds of other weird things happening. I don't expect stretching the belt to change pulley ratio, but would like to get one consistent steps per inch throughout the travel of the screw.
Thanks

madmachinst · Dec 17, 2022

Jonathan Smith said:
Sorry, I can't answer your question directly but can advise the belt will not stretch. It might wear but won't stretch. Probably not even a cheap chinese knock off with a polyester carcass. Now my rant, why do cnc machine manufacturers insist on using motor mounted encoders instead of dro like scales directly mounted on the machine axis. Get rid of the fancy drive and belt problems. Use the traditional power feed systems. It would make converting manual machines much more straight forward too.

Give it a break, that machine is from 1991. Saw pictures of a German servo from a V2, saw a similar looking beast with tachometer on it just like the old SEM servo this thing had.

madmachinst · Dec 17, 2022

CarbideBob said:
What is the problem???
I can not see what kind of belt/pulleys you are using. The sort of square teeth or the round teeth?
Trapezoidal or square looking is not so great at positioning and more for just drive apps.
Ratio and spacing distance looks good to me.
I am not understanding the question. We need more info.
The thing with this reduction is belts are not perfect tooth to tooth and the contact runout on a pulley is not the OD runout.
Bob
(Please take no offense from this post, so many questions and me still learning )

Spec says that this belt should deflect 1/4" with 17 lbs of force applied. This thing is guitar string tight and I suspect why I am getting inaccuracy through the travel of the table. When I get it within .00075" error throughout its travel, will be happy to rely on the screw mapping feature in the controller. Belts and pulley are original to machine.

madmachinst · Dec 18, 2022

Looked close and took cover off of Y axis servo drive and there seems to be no way to shift the adapter plate. Seems that the 750 W TECO drive is fatter than the old SEM servo so no way to expand the hole in the adapter plate a little to the left, servo flange will be over the hole then. Looks like I need a different belt say 1/4"-9/32" more on the circumference. Old one was Goodyear 270H100

memphisjed · Dec 18, 2022

Tooth belts do not slip. I think you are looking for hardware or servo tuning issue.
How much turn on lead screw is .01”? Mapping is for chasing tenths or really Really bad screws.

Did it work before you started changing parts?

madmachinst · Dec 18, 2022

memphisjed said:
Tooth belts do not slip. I think you are looking for hardware or servo tuning issue.
How much turn on lead screw is .01”? Mapping is for chasing tenths or really Really bad screws.

Did it work before you started changing parts?

Yes, it worked before changing servos. Old servo mounting plates had slots not holes for the mounting screws. I don't have them or any way to drive them anymore. I need to go with what I have. .01 is 1/20 of a leadscrew turn. I have a pulley from the servo driving a 2X bigger pulley on the leadscrew. Wish I could apply heat with heat gun on belt and that will loosen tension. If only life was that simple. I really hate to think might have to trim metal off the mounting flange on the TECO servos. I hate modifying expensive parts.

memphisjed · Dec 18, 2022

Servos need tuned.

madmachinst · Dec 18, 2022

Servos brand new and I figured out how to adjust the servo tension by modifying the adapter plate. shifted the servo mounting holes by 25 degrees so have enough meat to enlarge the mounting bolt holes .06" one direction.
went from

to

Now those 4 bigger holes can turn into slots. so the servo is sitting slanted 25 degrees . Lets get real you don't get shit like this right the first time.

memphisjed · Dec 18, 2022

Servos needs tuned. Each just needs a simple* 3 number code entered, nothing hard like 4 holes.
*simple is relative, ie simple relativity.

madmachinst · Dec 18, 2022

memphisjed said:
Servos needs tuned. Each just needs a simple* 3 number code entered, nothing hard like 4 holes.
*simple is relative, ie simple relativity.

3 number code for what? explain elaborate a lot. These are auto tuning.

madmachinst · Dec 20, 2022

after loosening the servo all I got was like 8 less steps per inch and max error in a few position was like .0006". when I test calibration @ 300 IPM I get everything dead on .0000 error but at 100 IPM more like .0005 max error and at 30 IPM I get like .0004 max error. Also notice that when the table was at 4" from limit switch towards me ( -Y) all the way to the limit switch error was like 0. Kind of like a used lathe near the chuck? I guess ballscrews can go bad. Anybody have anything to add about these observations?

CarbideBob · Dec 20, 2022

What? Steps per inch or degree do not change. This is pure math. Do not try to calibrate it. If it disagrees with the math look deeper.
Guessing what we have here is a step/direction into a servo drive. I do not like this for many reasons but it is very common and used by many.
This puts a loop into this drive. Different approach speeds give different check results although not sure how these checks are being done.
"I" term in the drive is bad or some others like a bad screw but that rare indeed.
Self tuning gets you in the ballpark but is not so great. More like a starting point for the settings.
Can you query the drive as to where it is sitting and error?
Error is less on one end? Way alignment and torque needed different? Backlash and acc/dec curve vs mass?
A confusing and interesting problem.
Bob

memphisjed · Dec 20, 2022

mr Carbide, how to you get I value over p value to start the tuning voodoo ratio? Honestly interested because you know more about this than most.

CarbideBob · Dec 21, 2022

memphisjed said:
mr Carbide, how to you get I value over p value to start the tuning voodoo ratio? Honestly interested because you know more about this than most.

This gets complicated fast and there are torque, velocity and positioning amp systems.
The P term is just proportional to any error and will happy sit all day off of by many counts when it is the right number.
It is gain matching. If you put low efficiency speakers on your stereo you will turn the knob higher than your old high efficiency speakers to get the same volume output.

The I term is the lower frequency side of the loop, is much slower to respond but builds power over time eliminating stopping error.

My first attempt to build a servo was on a surplus nine-track tape drive and was all P. Not good and very frustrating... :wall:

But we do not know if the amp is zero end error or not here. If it zero we get into mechanical stuff at the differing speeds.
Not sure any of this makes sense. Not sure I know more than others here so pennys from the peanut gallery on my side.
Bob

memphisjed · Dec 21, 2022

It makes logic when I read about it. When I have had to tune servos it lost logic and went to dark voodoo. I am not a master, or even amateur; I always have gotten them to spin right, eventually. Even a monkey can write Shakespeare with enough time.

If his are not correcting, or trying to correct, out of position wouldn’t this be the p value? I can see from your explanation the I value being why it stopped long or short.

Thank you.

madmachinst · Dec 21, 2022

CarbideBob said:
What? Steps per inch or degree do not change. This is pure math. Do not try to calibrate it. If it disagrees with the math look deeper.
Guessing what we have here is a step/direction into a servo drive. I do not like this for many reasons but it is very common and used by many.
This puts a loop into this drive. Different approach speeds give different check results although not sure how these checks are being done.
"I" term in the drive is bad or some others like a bad screw but that rare indeed.
Self tuning gets you in the ballpark but is not so great. More like a starting point for the settings.
Can you query the drive as to where it is sitting and error?
Error is less on one end? Way alignment and torque needed different? Backlash and acc/dec curve vs mass?
A confusing and interesting problem.
Bob

Yes it is a step and direction type system. The step pulse generator/ BOB gets feedback back from encoder to see if any corrections need to be made or if error too great and then servo gets disabled. Happened to me once before belt got loosened. When I have next session with tech support and machine hooked to network will ask him about query and see if too much error.
I understand P comes into play when the steps per are all consistently off. I needs to be tuned down when error oscillates between + and - equally and a lot. D needs to be increased when say I consistently get like .0002 more motion in the direction of motion?
currently at
9" I get +.00025
8" = +.0002
7" = dead on
6" = -.00015
5.25"=-.0004
5" = dead on
4" = -.0002
3.25"= -.0005
3" = -.0004
2" = dead on
1.5" = -.0004
1.25"= -.00025
1" = -.0004
.75" = dead on
.5" = dead on
As good as I can get it and happy with that or am I an underachiever? 1991 Machine keep in mind.

Trying to loosen the belt on my BP CNC.

Stainless

Diamond

Plastic

Plastic

Stainless

Stainless

Stainless

Stainless

Stainless

Stainless

Stainless

Stainless

Stainless

Stainless

Stainless

Diamond

Stainless

Diamond

Stainless

Stainless

Similar threads