belts vs screws?

[NIC 77]

Did everyone else just skip over the 5 or 6 axis part and go straight to the belt vs screw discussion?

From what I've seen, the OP didn't say leadscrew or ballscrew. He just said screw.

Also, there has been no mention from the OP about dimensions of the machine. IMO, an important detail to leave out.

No vote for rack and pinion? Depends on machine size, IMO.

I'll vote for the ballscrew. Belts, perhaps the least expensive, perhaps they have a place for hobby machines on a budget, I'm not going to slam that too much. But on a 5 or 6 axis CNC machine? No, they don't belong there.

Perhaps the OP can give more info.

[NIC 77]

Wow. Old thread. How did this find it's way into "Today's Posts?"

[ger21]

It's a poll. Any new votes bump it to the top.

[lalebarde]

Hi there, I wonder how compares in precision the different solutions : Belt vs Ball Screw vc Lead Screw vs Rack & Pinion vs Chain ?

[wizard]

Hi there, I wonder how compares in precision the different solutions : Belt vs Ball Screw vc Lead Screw vs Rack & Pinion vs Chain ?

The comments below are largely opinion based on years doing automation maintenance and may not reflect the experiences of others.

1. Don't even consider chain drives! They are a maintenance night mare.
2. When it comes to ball screws vs Trapezoidal lead screws ball screws are generally the better choice. However lead screws are likely more reliable in hostile environments. Lead screws and ball screws are the common solution for precision machine movements.
3. Rack and pinion is a good solution for very long axis. It has its own unique maintenance issues. The biggest problem with rack and pinion is backlash management. Precision can be good but may require the use of servos and gear boxes which end up being hidden expenses.
4. Belts (I'm assuming timing belts here) are very good with respect to precision but do suffer from belt related issues such as vibration in the belt and often a lack of stiffness. Sadly almost every router implementation of a belt drive solution I've seen has used a belt that is way too narrow and probably of a lighter series than would be ideal. Even on pick and place machines they regularly use belts 50mm wide, sometimes wider, for major axis motion.

Speaking of precision, often what you really want out of a CNC machine is solid repeatability. That is the first part comes out exactly like the last part. MOST low end routers are not precision machinery like you might expect form a milling machine. This is due to light frame construction, stepper drives in open loop designs, rolled lead screws (often the cheapest) that simply are not that accurate and a number of other factors. Of course in the opposite direction, where money isn't a problem, routers can be built that are nearly the equivalent of a bridge mill. Generally this forum focuses on lower end DIY machines with precision expectations in line with that.

[ramakarl]

I want to add a vote for chains.

Why:
- Chains are the only real cost effective option for high precision at larger sizes.
- Leadscrews just too costly at 8'
- Belts are hard to get precision at 8', by which I mean 1mm, due to the issues mentioned of vibration, slack and slip.
- Rack and pinion (ie. linear tooth rails) are the industrial soln for this, but also very costly at 8'. 8ft gear rack = $300. 8ft #35 chain = $15
- The maintenance of chains is low IF the system is well designed, by which I mean proper chain size, pathway and tensioning.

Really the best drive depends a lot on size, cost and precision goals.

[pippin88]

Show us a decent machine that uses chains

[ramakarl]

This is from 2015.

My latest machine has 1mm accuracy on XY axes, chain driven. Rebuilding Z now.

[ramakarl]

PS. I should have joined cnczone years ago. Making up for lost time I guess. In general I'm too busy to post much so I'll likely disappear for another 10 years until my next machine.. lol. When I started it wasn't even possible to find cheap, high current stepper drivers on-a-chip. Would have been a lot easier for me if I had found this community earlier.

[peteeng]

Hello Rama - For cnc machines such as routers, belts will outperform chains most times. Chains present issues with lubrication and stretching. The routers I have built with belts have achieved better then 0.1mm repeatability. Generally belts are undersized and under tensioned this leads to inaccuracy in use. Belts correctly designed do not have vibration, slack or do they skip teeth. "cost effective" is a relative statement and will mean different things to different people. Its fine to design a machine using chains if you understand what the limitations of them are and the design achieves your objectives. This is also true for any other drive system no matter the cost. Peter

[ramakarl]

Hi Peter,
I would be interested to see that machine. Do you have a video, plans, or links to it?
What is the travel/bed size of the 0.1mm machine?

> "Its fine to design a machine using chains if you understand what the limitations of them are and the design achieves your objectives. This is also true for any other drive system no matter the cost."

I think we are in agreement. I did say that the best drive depends on ones goals. I'm not suggesting that chains are better in all scenarios.
I've put in one vote for chains for having good precision and cost at larger bed sizes when designed properly. As you've said the quality of the design applies to any drive system.

[peteeng]

Hello Rama,
Here are some photos. It uses 16mm AT belts on all axes. It does not use "U" drives. I do not like U drives as the belt gets bent two ways which fatigues it faster and then the motors are on the gantry adding weight. The full loop drive is not as stiff as the belt is longer but then the motors are fixed so are not moving mass. One image shows it cutting a hydrofoil wing which had to be very accurate. Scoot was sold to a boatbuilder for making boat furniture. I'm now building one with ballscrews all round. See Brevis-HD thread. I wouldn't use a belt on Z in the future as Z needs much more force for plunging. But the belt allowed up to 25m/min feed which is really good when doing finishing cuts on things that have a small stepover... Peter

Valhalla was made for a lady that used it to float and burn her husbands ashes,,,, a nice little project, shame to burn it but so be it.

[ramakarl]

@peteeng
Very nice machine! Nice work. That ship is a really deep relief and impressive build.

I finally finished my Designer CNC Machine rebuild.



I also avoid belts or chains on Z-axis, and only use chains on X/Y.

[peteeng]

Hi Rama - I have just finished my latest machine yesterday. Shaking it down now for dust extraction... Peter

https://fb.watch/86KtxbwUUi/

[cncmakers001]

Screw Drive should be better.

[peteeng]

Hi All - There is no "better" depends on what your doing and what the requirement is and the budget. Chains, belts, leadscrews, ballscrews, rack & pinion. linear drives (have I missed any? trained mice in wheels with cogs, they wear little aluminum hats connected to the controllers and get little shocks that they count and take the correct number of steps) all have their place... Just need to understand them and get the best out of the solution you choose.... Got to go and feed my mice and clean their hats. Peter

[imbaine13]

trained mice in wheels with cogs, they wear little aluminum hats connected to the controllers and get little shocks that they count and take the correct number of steps) all have their place... Just need to understand them and get the best out of the solution you choose.... Got to go and feed my mice and clean their hats. Peter

Hahahahahahaaaa

[Rickey_H]

Awesome design! It's nice to see people thinking outside the box.

[Rickey_H]

I would like to add my 2 cents to this conversation. I would like to preface my response by weighing the pros and cons of each system.

I start with the ball screw transmission which is hands-down the most accurate system available. It is also the most expensive and does not have a path for upgradeability, as it requires replacing the expensive lead screw. Normal lead-screw systems use a lead nut that wears out quickly (or requires constant tightening) and is plagued by back lash.

A rack-and-pinion drive system is also a good solution, but it requires mounting two stepper motors on the gantry. This increased weight reduces the acceleration performance of the machine. I suspect that the little pinion gear needs to be replaced on a regular basis. Overall, this system is still more expensive than a belt drive, but the precision is going to be slightly better than a belt drive.

Belt drives come in different flavors. The most common configuration uses a drive system similar to the rack-and-pinion. Instead of pulleys, it anchors a belt to both ends of the table and uses a reduction drive stepper motor on each side of the gantry to push and pull the belt, hence moving the gantry back and forth. As others have pointed out, this system requires an aggressive belt-tensioning system that wraps the belt around the circumference of the drive pulley. The belt tensioning system places sharp radius bends on the back and front of the belt causing it to eventually stretch and fail. Does it really stretch enough to affect accuracy? I bet the belt will last twice as long as a pinion gear and will be much cheaper to replace.

I have a 1500mm x 1500mm DIY table. The main axis that moves the gantry on the table is done by two sets of belts. These belts are driven by two axles mounted on each end of the width of the table. On each axle are two pulleys that drive the belt. On one of the axles is mounted another pulley in the middle that is driven by yet another belt to a stepper motor mounted on the table, not the gantry. I am still evaluating this system working out all the minor issues. Although I spent a lot of money and time on this project, my net investment is still a fraction of what you would pay for a DIY kit.

One last thought- All inkjet and dot matrix printers used belts. CD players used lead screws. which lasted longer? (rhetorical question)

An off-topic question to this group, how do you quantify CNC router machine precision? What methods and metrics are used?

Regards,

Rick

[peteeng]

Hi Rick - see ASME B5.54 - Peter