My CNC Building Adventure

[JayneV]

Totally agree with you Jim. I learned something new today regarding short belt drives as used in CNCs to transfer motion from the motor to a screw. The tooth profile is not as important as having correct tension and pulleys that are not too small. The idea is to have enough tension in the belt so that the leading edge of the first 4 or 5 teeth on the belt are positively engaged with the pulley teeth, then as the belt wraps around the back of the pulley, those teeth don’t really do much, then continuing around the pulley to the other side, the trailing edge of the last 4 or 5 teeth of the belt are engaged with the pulley teeth. That’s probably not a very good explanation, but if done correctly, it doesn’t matter which direction the belt rotates, there are always teeth positively engaged with the pulley with basically zero backlash. This also means pulleys with around 25 teeth are needed. Any perceived backlash then comes from the belt stretching under load which is minimal for a very short belt. Anyway, that is how it was explained to me which seems to make sense.

I got the quotes back today from the emails I sent out. One was around $150 and the other about $240 for three belts and six pulleys. I think the $150 quote was for the wrong parts so I sent them another email to confirm. The price is a little expensive but not unexpected from local suppliers. Tonight I sent a quote request to Powge at AliExpress. If they come back with a good price I will just order from them. It’s not like I have an urgent need for those parts, Carlos is still a long way for needing any drive system components.

Out of interest, I have heard that it is possible to compensate for backlash through the controller software. Thinking about the mathematics involved to accomplish this hurts my brain. Has anybody had any experience with compensating for drive system backlash through the software? That sounds like a handy feature to have to fine tune the machine.

Jayne

 

[Jim_in_PA]

Out of interest, I have heard that it is possible to compensate for backlash through the controller software. Thinking about the mathematics involved to accomplish this hurts my brain. Has anybody had any experience with compensating for drive system backlash through the software? That sounds like a handy feature to have to fine tune the machine.

I have no idea if that's possible...but in my mind, a properly adjusted machine really shouldn't have any "measurable" backlash. But maybe I'm wrong about that.

 

[JayneV]

I have no idea if that's possible...but in my mind, a properly adjusted machine really shouldn't have any "measurable" backlash. But maybe I'm wrong about that.

I was thinking of the complete drive system as a whole regarding backlash. Items like ballscrews and linear guides will all have some tolerance depending on how much one is willing to spend on these items. In my case these will be parts aimed at diy home builder market, which will be more than enough to suit my needs. My question was just out of general interest and if the software is capable of compensating for backlash, it would be an interesting exercise, more like a challenge really, to see how finely tuned the machine can get. Also, I suspect any backlash would only really be measurable when cutting metal. If it can be measured in wood then I’d be looking for an actual fault rather than a tolerance issue in the components.

 

[nickhofen]

Totally agree with you Jim. I learned something new today regarding short belt drives as used in CNCs to transfer motion from the motor to a screw. The tooth profile is not as important as having correct tension and pulleys that are not too small. The idea is to have enough tension in the belt so that the leading edge of the first 4 or 5 teeth on the belt are positively engaged with the pulley teeth, then as the belt wraps around the back of the pulley, those teeth don’t really do much, then continuing around the pulley to the other side, the trailing edge of the last 4 or 5 teeth of the belt are engaged with the pulley teeth. That’s probably not a very good explanation, but if done correctly, it doesn’t matter which direction the belt rotates, there are always teeth positively engaged with the pulley with basically zero backlash. This also means pulleys with around 25 teeth are needed. Any perceived backlash then comes from the belt stretching under load which is minimal for a very short belt. Anyway, that is how it was explained to me which seems to make sense.

I got the quotes back today from the emails I sent out. One was around $150 and the other about $240 for three belts and six pulleys. I think the $150 quote was for the wrong parts so I sent them another email to confirm. The price is a little expensive but not unexpected from local suppliers. Tonight I sent a quote request to Powge at AliExpress. If they come back with a good price I will just order from them. It’s not like I have an urgent need for those parts, Carlos is still a long way for needing any drive system components.

Out of interest, I have heard that it is possible to compensate for backlash through the controller software. Thinking about the mathematics involved to accomplish this hurts my brain. Has anybody had any experience with compensating for drive system backlash through the software? That sounds like a handy feature to have to fine tune the machine.

Jayne

Why did you request a price for six pulleys and three belts?
As I can understand from you photos you have four motors ,two for X axis, one for Y and another one for Z, right?

 

[JayneV]

Why did you request a price for six pulleys and three belts?
As I can understand from you photos you have four motors ,two for X axis, one for Y and another one for Z, right?

The long axis is Y on my machine, with two motors.
The X axis is on the gantry with one motor.
And the Z axis is direct drive. So only X and Y will be belt driven.

 

[nickhofen]

Ops right, don't listen to me!

 

[ghostchord]

Just wanted to also chime in with belts/pulleys should be fine. I worked on commercial precision machinery where a stepper motor was coupled to a ballscrew via a belt. Backlash something like 1 micron. Any backlash you see at the pulley is going to translate some minute offset on the axis since it's some really small fraction of a rotation which is only so much displacement (depends on your screw pitch). I don't recall what specific type was being used, we also used them for gearing (i.e. wasn't 1:1) and it's nice that you can actually swap different pulleys in if you need to change the gear ratio (let's say you find you want to go a little bit faster, or you put a more powerful stepper on, or you want a little more force/torque).

Another btw is that I don't think software backlash compensation will help much in these applications. There's a lot of direction changes when cutting and whenever you do that there's no avoiding the backlash. I think with the right belt/pulley this isn't really an issue anyways.

EDIT: found some pulleys in my old parts bin:

Any backlash is gonna be some tiny fraction of a tooth, the belts are very stiff... This pulley is rather large, you can go smaller and it'll still be fine...

Those flex couplings you'd use for direct drive can also be a pain, if they're not perfectly aligned they can fail relatively quickly and some types also have backlash, the belt is generally more tolerant...

 

[nickhofen]

Just wanted to also chime in with belts/pulleys should be fine. I worked on commercial precision machinery where a stepper motor was coupled to a ballscrew via a belt. Backlash something like 1 micron. Any backlash you see at the pulley is going to translate some minute offset on the axis since it's some really small fraction of a rotation which is only so much displacement (depends on your screw pitch). I don't recall what specific type was being used, we also used them for gearing (i.e. wasn't 1:1) and it's nice that you can actually swap different pulleys in if you need to change the gear ratio (let's say you find you want to go a little bit faster, or you put a more powerful stepper on, or you want a little more force/torque).

Another btw is that I don't think software backlash compensation will help much in these applications. There's a lot of direction changes when cutting and whenever you do that there's no avoiding the backlash. I think with the right belt/pulley this isn't really an issue anyways.

EDIT: found some pulleys in my old parts bin:
View attachment 914895

Any backlash is gonna be some tiny fraction of a tooth, the belts are very stiff... This pulley is rather large, you can go smaller and it'll still be fine...

Those flex couplings you'd use for direct drive can also be a pain, if they're not perfectly aligned they can fail relatively quickly and some types also have backlash, the belt is generally more tolerant...

Good info,thanks for sharing.
Is there a rule of thumb of the tension on the belts?
I mean too loose will cause many problems also too much tension on them will have some disadvantages, right?
I have read is that after you put your new belts on a machine you have to retention them after 1-2 hours of working on a project, this must be done only once.
But I can't find info about how much stretched they must be.

 

[ghostchord]

Good info,thanks for sharing.
Is there a rule of thumb of the tension on the belts?
I mean too loose will cause many problems also too much tension on them will have some disadvantages, right?
I have read is that after you put your new belts on a machine you have to retention them after 1-2 hours of working on a project, this must be done only once.
But I can't find info about how much stretched they must be.

The belts we were using (unfortunately this was 20 years ago and I probably didn't know the type or vendor even then, I was doing software) did not require very much tension. They were super stiff. Basically just pull on the motor bracket with very moderate force and tighten the bolts. I'm sure the manufacturer has specific deflection/force specs.

EDIT: Too much tension could damage the bearing in the stepper. There's really not a lot of tension at all.

EDIT2: random link I just found that seems to have some useful info: http://benchtopmachineshop.blogspot.com/2019/04/printer-belt-tension.html

EDIT3: this has a lot of details: https://www.gates.com/content/dam/g...ogs/powergripdrivedesignmanual_17195_2014.pdf

I don't recall the belts we were using getting re-tensioned but this could have been done without me seeing it The main thing I'd worry about is resonance, that's sort of a generic stepper issue, and not over-tensioning. I'd tension pretty lightly and as long as you're getting smooth moves just leave it there.

 

[JayneV]

Thank you @ghostchord for all that excellent information. I like the idea of plucking the belt like a guitar string and measuring the frequency it oscillates to check the tension, it seems very appropriate for this forum and the machines primary purpose.

Figuring out what that tension should be is not as easy as checking the manufacturer’s data. Manufacturers have lots of information for power transmission applications where the belt is used to transfer power from one shaft to another and rotated in one direction, but they don’t go into much detail when it comes to using the belts for precise positioning, rotating in either direction. It was explained to me that in the positioning application, the belt tension will need to be higher than what would be used in a power drive application. The increased tension is intended to seat the teeth if the belt against the pulley teeth on both sides of the pulley so no matter which way the belt is turning, there is never any gap between the belt and pulley teeth. Not having any hands on experience with belt drives except for looking at the tiny belts used on my 3D printer, I have to admit that I m overwhelmed with theory and formulas from manufacturers and discussion in my build thread at CNCZone.

The steppers I will be using have a 10mm shaft and are rated for 75N dynamic radial load which isn’t very much. However I suspect that is for infinite bearing life so it probably wouldn’t hurt much to go above 75N if necessary.

The ballscrews will have 10mm pitch (1610) except for the z axis which will have 5mm pitch (1605) but that will be direct drive.

I will probably order the belts and pulleys tomorrow just so I can commit to something and save myself from going crazy thinking about belts 24/7 I’ve decided to give HTD belts a try because they are very cheap and readily available pretty much everywhere. My first choice was going to be AT5 belts but they cost more and are not as readily available.

 

[ghostchord]

I agree it can be hard to grok those data sheets. I think the vendors are quite aware of positioning applications, the usage of these belts for positioning goes back many decades, like computer printer applications... Those teeth *really* sit very nicely in the pulley just naturally with practically zero tension and the belt length doesn't really change much under tension.. I thought I had a belt sitting somewhere but I can't find it right now.

I literally used to just loosen the bolts, change something, pull on it so the belt is "tight", tighten the bolts (no deflection measures other than just maybe pressing it with my finger to see it's not over tight). This is on imaging machines where micron level inaccuracies would pop out pretty immediately. There was a pretty fine pitched ballscrew there which obviously helps but personally I wouldn't really be super concerned here. That's me as a software person mimicking what our mechanical engineers would do I was working on the stepper control so I had to mess with it quite a bit, switch motors, switch pulleys etc.

EDIT: I think you'll be fine with those HTDs ... easy to also change belt type and pulleys down the road. I would say there's some chance you could get free samples from the local Gates rep if you drop some hints that more machines might be built at some point Looking forward to see the build come together!

Thank you @ghostchord for all that excellent information. I like the idea of plucking the belt like a guitar string and measuring the frequency it oscillates to check the tension, it seems very appropriate for this forum and the machines primary purpose.

Figuring out what that tension should be is not as easy as checking the manufacturer’s data. Manufacturers have lots of information for power transmission applications where the belt is used to transfer power from one shaft to another and rotated in one direction, but they don’t go into much detail when it comes to using the belts for precise positioning, rotating in either direction. It was explained to me that in the positioning application, the belt tension will need to be higher than what would be used in a power drive application. The increased tension is intended to seat the teeth if the belt against the pulley teeth on both sides of the pulley so no matter which way the belt is turning, there is never any gap between the belt and pulley teeth. Not having any hands on experience with belt drives except for looking at the tiny belts used on my 3D printer, I have to admit that I m overwhelmed with theory and formulas from manufacturers and discussion in my build thread at CNCZone.

The steppers I will be using have a 10mm shaft and are rated for 75N dynamic radial load which isn’t very much. However I suspect that is for infinite bearing life so it probably wouldn’t hurt much to go above 75N if necessary.

The ballscrews will have 10mm pitch (1610) except for the z axis which will have 5mm pitch (1605) but that will be direct drive.

I will probably order the belts and pulleys tomorrow just so I can commit to something and save myself from going crazy thinking about belts 24/7 I’ve decided to give HTD belts a try because they are very cheap and readily available pretty much everywhere. My first choice was going to be AT5 belts but they cost more and are not as readily available.

 

[JayneV]

Hi everyone,

Well after going back and forth between belt drive and direct drive for the ballscrews, I finally settled on direct drive. There are pros and cons for each system but it was the simplicity of direct drive that swayed me in the end. Direct drive also doesn't have the added cost of belts and pulleys. To make sure my decision is final with no more indecision, I placed an order for linear guides and ballscrews today, all sized for direct drive.

This is Carlos in its current (and final ) form.



Now to start sourcing the aluminium to start building the frame.

 

[John Nicholas]

Hi everyone,

Well after going back and forth between belt drive and direct drive for the ballscrews, I finally settled on direct drive. There are pros and cons for each system but it was the simplicity of direct drive that swayed me in the end. Direct drive also doesn't have the added cost of belts and pulleys. To make sure my decision is final with no more indecision, I placed an order for linear guides and ballscrews today, all sized for direct drive.

This is Carlos in its current (and final ) form.

View attachment 918720 View attachment 918721

Now to start sourcing the aluminium to start building the frame.

Congratulations on finishing the design phase! That's a huge accomplishment!

As for the belt vs. direct drive, my training and experience (which is minimal. LOL) is to follow the axiom K.I.S.S. for those who don't know, that means to Keep It Simple Stupid. Direct drive is simpler, hence it would be my choice unless there was an excellent reason not to.

Wow, you've now truly started the build now that you've ordered the main means of the Axis' moving!

Again Congratulations!

 

[Jim_in_PA]

This is Carlos in its current (and final ) form.

No battle plan survives the first shot from the enemy...

 

[JayneV]

Congratulations on finishing the design phase! That's a huge accomplishment!

As for the belt vs. direct drive, my training and experience (which is minimal. LOL) is to follow the axiom K.I.S.S. for those who don't know, that means to Keep It Simple Stupid. Direct drive is simpler, hence it would be my choice unless there was an excellent reason not to.

Wow, you've now truly started the build now that you've ordered the main means of the Axis' moving!

Again Congratulations!

Thank you John. the KISS principle is a very good principle to follow. But sometimes you get your mind set on doing things a certain way for no other reason than just wanting to. I guess that is where the second "S" in KISS comes in to play. BTW, I'm not often know for doing things the easy way.

No battle plan survives the first shot from the enemy...

Very true Jim! Those who know me know that I can be pretty stubborn, so with Carlos, it took a few shots before the battle plans changed, and then the plans took on a life of their own and changed many, many times.
Keep the enemy guessing.

 

[JayneV]

Hi everyone,

Just received a shipping update. My order from BST Motion has arrived in Sydney and cleared customs overnight. The estimated delivery date to my door is Thursday. Woohooo!!!

 

[JayneV]

As promised, the delivery arrived today. Yay!!!!! There were two boxes on the floor waiting for me when I got home from work. It was late and I resisted the temptation to tear into the boxes to see my new toys. I will also have to resist the temptation tomorrow because it will be late again by the time I get home from work, so I will patiently wait until Saturday. Well.....I will wait, can’t guarantee that it will be patiently.

 

[JayneV]

I opened the two boxes of goodies today to find some nice new linear rails and ballscrews. There is an extra pair of rails in the box which I didn't order, not sure how they got in there but everything else is exactly what I requested. On initial quick inspection, all the parts seem to be good quality for the price. Now to start building so these parts have a place to live.

 

[nickhofen]

Nice!
I see you have already started gathering parts for your next cnc mill machine also!

 

[guitarbuilder]

You can make a binding jig out of the extra pair of rails....