529,951 active members*
2,454 visitors online*
IndustryArena Forum > Mechanical Engineering > Mechanical Calculations/Engineering Design > Moving Gantry column bearing spacing - open discussion

# Thread: Moving Gantry column bearing spacing - open discussion

1. ## Moving Gantry column bearing spacing - open discussion

Hi All - The spacing of the column bearings often comes up in many areas of the forum. Its an area that I have looked at intensely over the last 3 years. There's no guidelines that I know of other than "position them as far apart as possible". So I'd like to engage the forum to share its knowledge on these things. Firstly my history on the subject:
1) 3 years ago I built my first machine Scoot Alpha. Scoots design credo was "minimal function prototype" which meant the design issue was looked at and the minimum solution was implemented. This meant it used one car per axis and one rail for the gantry and Z axis. The bearing design manuals had lots of examples of single rail/single car solutions and the the router worked well within the cars load specified capacity.
2) Shortly into Scoots life it developed a wobble. The cars were zero clearance cars so after settling in actually had some clearance. This brought me to investigate the different preloads and what to use them for. Scoot had been designed so that two rails could be used across the gantry and I ordered a new rail and a high preload car for the Z axis. The Z axis car was then to be used on the gantry second rail
3) I also decided to up the cars from one to two on the gantry columns, so ordered 3 high preload cars... This became Scoot Beta. ScootB was much more stable and I even cut some aluminium composite panel that was impossible with Scoot-A.

So to discuss the bearing arrangement - from this experience its clear that 2 rails and 3 or 4 cars are needed for stiffness. The further apart along the rail they are the better. One guideline is that a group of bearings needs to be square. With square rails this gives very good stiffness in all directions as the bearings themselves support moments. If round rails are used the square pattern is good as well as this removes the rolling freedom a round rail has. But its not practical to have column bearings in a square pattern as they would have to be the table length apart so the gantry would not travel far or even not at all!! . This leads to the bearing group having a huge aspect ratio. So huge in fact that I think the LHS and RHS have to be considered as separate bearing groups. This means the gantry is considered as two cantilevers joined in the middle. Which sort of makes sense as gantries no matter how stiff "walk" so each side is behaving independently, one side catching up with the other. This is a structural issue vs a bearing arrangement issue I think.

Very large machines address this by using two rails each side and having 4 cars each side. So this has a square bearing group with a cantilever and each cantilever meeting in the middle. Cantilever loads have there own set of problems with bushes that is minimised or solved by using rolling element cars. This geometric problem is taken advantage of with F clamps and various other clamping systems allowing the clamp to bind when loaded. Exactly what we don't want to happen...

So our typical moving gantry is like a car in which its wheelbase is considerably shorter then its track ie it will steer badly! Some builders expect the drive system to correct this situation. Maybe it can but mostly it can't at a basic level. Using some sort of positional feedback is the general solution.

So my guideline is make the bearings at least one dimension apart prefer more, where the dimension is the along the rail bolt spacing. eg on a std 20mm flanged car this would be 40mm min between the middle bolt holes.

Question - Has anyone used one long car in this application vs two cars close to each other?

That's about it would like to hear anything anyone has to say on this.... Peter

2. ## Re: Moving Gantry column bearing spacing - open discussion

Interesting thread, need some more time to add my input on it. But I thought another question would be nice to add, with some visual aid to make more sense of what I'm about to write.

My thought/question;
How much impact does the placement of the guides in either X or Y axis have on the gantry? What I mean is, is there a co-relation between the placement in each axis.

For instance if you narrow the spacing in X-axis with XX amount, would it help by adding that decrease in to the spacing between the cars in the Y-axis to compensate it. Or vice versa.
My intuition says no, but I have not done any testing, to see how much or little this kind of change does to a gantry assembly. And to be frank, just got the thought when reading your initial post about placement.

Miniature images, with links to full size. and just a rough mockup of the idea behind the thought/question;
- Normal placement
- Longer Y, shorter X
- Shorter Y, Longer X

I thought this could be another good question to add on top of the future discussion of car placement.
I hope I wrote it so it made some sense, I'm not a native English speaker nor writer. So the grammar might be off.

3. ## Re: Moving Gantry column bearing spacing - open discussion

Hi Jaws - If the moving gantry is moving in the X axis (column bearings) and the transverse gantry axis (saddle bearings) is the Y axis then I think your intuition is correct. Irrespective of the axis names I don't think the rigidity of the two axis via the car spacing is related. The transverse gantry bearings (saddle bearings) are usually mounted in a near square geometry which is a stable condition. The column bearings have no chance of ever being square so the aspect ratio is huge and unstable. I'm about to design a mill and have decided the only way to have the column bearings geometrically stable is to have two rails and 4 cars each side of the gantry. regards Peter

4. ## Re: Moving Gantry column bearing spacing - open discussion

Then we are in the same boat, so to speak. In regards of what we see as stable conditions for bearing placement here.
Thanks for your input on the matter.

I'm going back and forth between two rails per side or just the one myself, going to start off with one per side and see how that setup performs. If I end up needing to double the cars and rails. I can do so at a later stage, in theory at least.
But based on the design on the machines at my current job and a former co-worker's workshop, 1 rail plus two cars per side should do fine, if my design is equally stable, or more so.
The one I am running at work is an "older" 2012 Kimla BPF-1531, while my former co-worker has a MultiCAM "X-X", both are routers and are mainly cutting Alu between 0.5 up to 50 mm thick. Ranging from 5000 alloys up to 7000 alloys. With the Kimla I run between 50-5000 mm/min, depending on type of operation and tooling.
Just some general info

And I would like to say thanks for all the valuable info and time you have put into this forum. I've read some of your post in another thread, regarding Gantry and Z-Axis design, and took your general design principles and added it into my own design.

5. ## Re: Moving Gantry column bearing spacing - open discussion

Hi Jawz - Happy to help its been two way, have picked up stuff here that's not commonly available. Most routers even huge commercial ones only have one rail per side and two cars per rail. Its very important to get medium or prefer heavy preload cars. They made a large difference on my first router when the normal ones developed clearance. For my mill I'm aiming at as stiff/damp as possible so I'm thinking of two round rails each side to make it economic, including two on each column as that's the same issue. I'm learning about round rail specs now. But two squares would be best. Maybe one direction 600mm so its half a sheet or block. I make moulds so need high Z's....Regards Peter

6. ## Re: Moving Gantry column bearing spacing - open discussion

Hi Jawz - If you have a CAD model of the gantry I'm happy to run it in FE. Publish here as step file and I'll compare the two rail to single rail to see what happens. I'm flat chat analysing a big yacht hull at the moment and need some diversion therapy sometimes. I have to sort and define over 4000 surfaces in Rhino3D. Been a while since I've done meshed FE work, been spoiled with meshless over the last year or so....Peter

Kimla interesting company. Linear drives are an interest at the mo as well. Hiwin sell them.... but where can I get hobby level linear drives> maybe in a few years then the stepper/hybrid/servo debate is finished...

7. ## Re: Moving Gantry column bearing spacing - open discussion

So our typical moving gantry is like a car in which its wheelbase is considerably shorter then its track ie it will steer badly! Some builders expect the drive system to correct this situation. Maybe it can but mostly it can't at a basic level. Using some sort of positional feedback is the general solution.
Driving the gantry from both sides solves this issue, at least at the DIY level.

As for the bearing spacing...
You can use the manufacturer's formulas and design for a given load. The Hiwin Catalog has all the formulas you need.
https://www.hiwin.com/pdf/linear_guideways.pdf

Or you can use the simple DIY method, and space them as far as you can within your particular design.

It doesn't make a lot of sense to stray from the "standard" 2 rails with 2 cars each for each axis. Trying to get away with less will almost always sacrifice rigidity.

I used to use an industrial router with a 50lb spindle mounted directly to 2 cars on a single rail, bit it was a large roller bearing rail, bolted to a steel assembly. I think the main reason for this was to keep it as compact as possible.
Roller bearings are considerably more expensive, so for a DIY machine 2 rails is probably cheaper, easier, and more rigid.

8. ## Re: Moving Gantry column bearing spacing - open discussion

Hi Gerry - If you design on load this will indicate that you only need one bearing on each side. And yes one rail one car systems work as well (from experience but not as well as two cars). That's what I did for my first machine on all axes. But it walked a bit so upgraded to two cars which made it better. The drive system does not solve the instability it just hides it. The guideline to make the cars as far apart as possible is the only valid way to do it at present that I can see. There's no math's to find the "ideal" spacing. I think this is the case because the aspect ratio is so big they are in fact independent ....Regards Peter

9. ## Re: Moving Gantry column bearing spacing - open discussion

Hi Peter,

That I can imagine, it being two ways. There seems to a lot of knowledgeable people around in this forum, from the brief glance I have had so far.
Yes that I have noticed a lot of them have, 1 rail + 2 cars per side is most likely the most economical while providing enough stiffness for router production. The Kimla is after all intended for wood working, not metal working officially.
That I did not not know, that it could have such an impact on rigidity, sure somewhat but not something that I thought would impact the end result that much. I was going to play around with the IGUS miniature rail system, with different preloads on my custom 3D printer, I'm thinking such a use would serve as a good testing setup, light weight and fast movement. That is easily visually comparable on the finished printed part, the difference the play makes between each car that is.
Ok well utilizing two round rails (Supported?) right next to each other should improve the nature of it wanting to "rotate" while being loaded. That was something I had an issue with, with my first CNC machine, but I also just grabbed what I had lying around. So the bars was none supported and too weak for my structure, but it made that issue quite visually obvious.
Molds, so much fun while waiting...., I make Ureol based moulds at work for vacuum forming tools. The female part is made in Alu 5083.

Sure I would appreciate any help in that regard, my simulation skills are not the most developed. Just done some basic stress and flow simulation throughout the years. To get some info more than my best guess.
A yacht hull, now that must provide a challenge! I can understand the need for variation and battling a more common issue. Having a somewhat non uniform pressure to consider on the underside, while also including the weight and it's distribution of the above structure.
Damn that is more than I would want to face right now, no pun intended. Meshless? Is it some sort of vertex based system then??

Yes they seem to have improved a lot on their machines and software since 2012, the machine we have has traditional rack and pinion for XY, and ballscrew with air cylinder attached to remove backlash even further.
The linear drive system was something I had missed, sound very interesting indeed, with the claimed specs!
Hopefully, it might end up becoming as "cheap" like servo motors and stepper motors have come to be in the past 10+ years.

Will upload a step later on by the way.

Hi Gerry,
If such a math formula exist, I would not want to use. Based on how much info one would have to add to make it even remotely viable as an option. Compared to using proven methods and then simulating it, to get a theoretical data point on each introduced change.
What I mean, based on my knowledge, is that you would have to incorporate mass of different assemblies, the center of mass of each of those assemblies, add moving mass estimates, velocities. And that would still be based on a theoretical assumption of each part. And then you still have the simulation process to go through to verify it is good enough.

Again my knowledge and insight on this matter is limited, and this is just my own assumption based on the knowledge I posses.
And i skimmed through the HIWIN document, and I did not see any math formula that covered anything but the linear guides themselves. Nothing about spacing, except for predefined values. But I did just skim the document right through, so I coould have missed it.

10. ## Re: Moving Gantry column bearing spacing - open discussion

Another thought of placement, with the round bar and linear bearings in mind this time.

Angled with 2 cars
Angled with 3 cars

Just an interesting idea I got, again while reading your comment Peter. Unsure of how much better it would prove to be compared to a 90º traditional mounting. I would suppose the square rail could benefit from such a mounting arrangement more, if the load could be carried like this without introducing any weakness. The weakness would come, I think, from the edges/corner of the rail, and could minimize the amount of load the car could carry. That would however not be the case with a round rail car. The load is evenly distributed in any rotation. Unless the linear bearing is designed with a specific load "target point".

EDIT;

A more optimal design would be removing the standard housing, and implement reamed holes to fit the linear bearings. To be used directly into the "car". Ensuring an easier and more proper placement of the rods to each other,

A great benefit from linear bearings, beyond their price point, is their end user customization possibilities of the cars themselves. With HIWIN cars or IGUS cars, you are restricted to use their housing for proper utilization of the rail and car system.

11. ## Re: Moving Gantry column bearing spacing - open discussion

Hi All - I have attached class notes from an MIT class on machine design, Bearing arrangement geometry is discussed briefly. The issue with square cars is that they can resist moments in any plane. This means the reaction can be in any direction and the second or "rear" rail cars are actually kinematically redundant. Its actually better to have 2 cars on the primary rail and one car on the "rear" rail. But we do like symmetric solutions and the car cost is small compared to the rail cost. If round rails are used the 4 cars are kinematically required to resist rotation and they will correctly share loads ie 4 round cars are required kinematically for a "rigid" connection. And yes supported rails are a must.... But that's a cost thing, square are better if cost is not a concern otherwise machine builders would use round rails. . Who worries about money anyway Maybe 2 round rails for the load side a square for the rear car? Then the stack heights of the two systems would be different and a new rabbit vortex would appear....

RE FE: many years ago I did quite a lot of FE for naval architects, they gave me the hard to do stuff. Every panel on the boat has a different pressure and being a composite it has all the plies and directions to cope with. Its quite a Chinese puzzle within a few rabbit holes. There are a few "meshless" FE systems. I use simsolid look it up its quite spectacular. It takes standard CAD solid geometry connects it all together including nuts and bolts if present and can solve massive problems in minutes. For instance the yacht would require millions maybe billions of equations to be solved yet Simsolid would do it with under 100,000. SS does not do composites so I have to do it old school. Will take 4 days to set the model up. If the CAD for SS existed I could do it in less then a day total with a report. Meshed FE is now outdated I feel. I read articles on "boundary value analysis" some 25 years ago and its taken that long for someone to code it vs looking at it academically. Means I can run your step file in 15mins vs in my mesh FE system say 6 hrs to set up the model, run it debug it then rerun and get answers maybe a day. cheers Peter

you can buy the round bearings san carriage...

12. ## Re: Moving Gantry column bearing spacing - open discussion

Hi All - Jawz has sent me a model and I have run it. Its a very stiff design but the object is to investigate the interaction of the gantry bearings and rails. So I used 1000N as my usual reference load on the tool and restrained one side of the gantry and allowed the other side to slide down the rail. It moved 1.1mm. I then deleted the rail and rerun the model. The free side now moved 2.6mm. So the rail constrained the movement quite a bit. Now the question is if we used two rails each side would it be better? or do we move the cars further apart until they have the same compliance? The 1.1 to 2.6mm compliance is the elasticity that the motion system has to control in this config. This compliance disappears in a fixed gantry design. The cars are 20mm in this case and the cars are just over one body spacing apart. Waiting on permission to show results here. Peter

13. ## Re: Moving Gantry column bearing spacing - open discussion

Hi All - Couldn't wait for Jawz so built a simple model to investigate and drum rollllllllllll. My twin rail theory is SHOT. I modelled a 150x150x8 SHS aluminium gantry 1400mm wide. The columns are 16mm al plate. A) has twin rails and slides on all rails B) has a free side and C) is sliding on single rails. A and C slide the same distance 3.72mm. The free displacement is 8.2mm. So now have to vary the bearing spacing to see what happens....

14. ## Re: Moving Gantry column bearing spacing - open discussion

Hi All - so I modelled the same gantry using a single rail each side. The base of the gantry is 200mm, 300mm and 400mm long. The 200mm moves 3.8mm the 300mm 3.3mm and the 400mm moves 3.2mm so there is a diminishing return. But this points to the gantry column. In this model its not very torsionally stiff, even though its 16mm AL plate with only 10kgf on the car. So seems the message is to make the column very torsionally stiff. Now, I do realise that the column can't move this far as its connected to the motion system. But this compliance contributes to vibration and poor performance when changing directions and contributes to walking as one side will clearly be different to the other.... So the rule of thumb, make the bearing spacing as long as possible holds and it seems 2 or 3 bearing spacings is as much as you need if possible and make the column torsionally stiff (round?). All of those single plate risers or columns take notes!! Peter
In this case

200mm is 3 bearing spacings
300mm is 5 and
400mm is 7 bearing spacings as the length of the car body is 50mm

What I'm trying to also say is that the structure should be made stiffer enough so as not to move "structural" functions to the motion parts. So my guess is that many of the flat plate type columns are relying on the motion parts to remain stiff. This has hidden issues like vibration, wear, movement hysteresis and poor accuracy.

This also means Milli does not need twin rails, excellent...that was part of the reason to start this thread. cheers and Keep Making Peter

15. ## Re: Moving Gantry column bearing spacing - open discussion

Originally Posted by peteeng
Hi Jawz - Happy to help its been two way, have picked up stuff here that's not commonly available. Most routers even huge commercial ones only have one rail per side and two cars per rail. Its very important to get medium or prefer heavy preload cars. They made a large difference on my first router when the normal ones developed clearance. For my mill I'm aiming at as stiff/damp as possible so I'm thinking of two round rails each side to make it economic, including two on each column as that's the same issue. I'm learning about round rail specs now. But two squares would be best. Maybe one direction 600mm so its half a sheet or block. I make moulds so need high Z's....Regards Peter
It's not uncommon to have 2 sets of rails and 4 bearings on a moving Gantry I have built some like this, there are a few companies that build them this way, I'm not sure but I think this hobby machine has 2 rails on each axis

16. ## Re: Moving Gantry column bearing spacing - open discussion

Originally Posted by --JawZ--
Another thought of placement, with the round bar and linear bearings in mind this time.

Angled with 2 cars
Angled with 3 cars

Just an interesting idea I got, again while reading your comment Peter. Unsure of how much better it would prove to be compared to a 90º traditional mounting. I would suppose the square rail could benefit from such a mounting arrangement more, if the load could be carried like this without introducing any weakness. The weakness would come, I think, from the edges/corner of the rail, and could minimize the amount of load the car could carry. That would however not be the case with a round rail car. The load is evenly distributed in any rotation. Unless the linear bearing is designed with a specific load "target point".

EDIT;

A more optimal design would be removing the standard housing, and implement reamed holes to fit the linear bearings. To be used directly into the "car". Ensuring an easier and more proper placement of the rods to each other,

A great benefit from linear bearings, beyond their price point, is their end user customization possibilities of the cars themselves. With HIWIN cars or IGUS cars, you are restricted to use their housing for proper utilization of the rail and car system.
What you are showing here is unsupported round rails which don't work well at all they must be supported if you are thinking of using round rails, IGUS don't have anything suitable for CNC machines as far as rails go

17. ## Re: Moving Gantry column bearing spacing - open discussion

Originally Posted by mactec54
It's not uncommon to have 2 sets of rails and 4 bearings on a moving Gantry I have built some like this, there are a few companies that build them this way, I'm not sure but I think this hobby machine has 2 rails on each axis
Hi Mactec - By above do you mean 2 rails and 4 cars per side? or 2 rails and 4 cars total? I've seen 2 rails and 4 cars each side on very large machines. I have always thought they do this to aid assembly. My modelling prior indicates this as well. I think a lot of machines do not have enough torsional stiffness in the columns. cheers Peter

18. ## Re: Moving Gantry column bearing spacing - open discussion

Originally Posted by mactec54
What you are showing here is unsupported round rails which don't work well at all they must be supported if you are thinking of using round rails, IGUS don't have anything suitable for CNC machines as far as rails go
That is two errors on my part there, taking the first bearing housing I could find in my library, and second. Not mentioning it was supported round rails that I was talking about.
And the IGUS system is possible for a cnc machine, it's just limited by the amount of weight that can be pushed back and forth safely, under a long period of time.

19. ## Re: Moving Gantry column bearing spacing - open discussion

Originally Posted by peteeng
Hi Mactec - By above do you mean 2 rails and 4 cars per side? or 2 rails and 4 cars total? I've seen 2 rails and 4 cars each side on very large machines. I have always thought they do this to aid assembly. My modelling prior indicates this as well. I think a lot of machines do not have enough torsional stiffness in the columns. cheers Peter
2 rails 4 Bearings each side the same for each axis

20. ## Re: Moving Gantry column bearing spacing - open discussion

Originally Posted by --JawZ--
That is two errors on my part there, taking the first bearing housing I could find in my library, and second. Not mentioning it was supported round rails that I was talking about.
And the IGUS system is possible for a cnc machine, it's just limited by the amount of weight that can be pushed back and forth safely, under a long period of time.
No I have tested the IGUS system and it is not suitable for CNC machine use, even at Hobby level, it would be a very bad choice if you where to build a new machine, IGUS can be used for other sliding systems that don't need any kind of accuracy

The design you have with the round rails would not work with supported round rails as drawn the Bearings would have to be open, round rail bearings also have large clearance between the rail and bearing some have small adjustments that can be made but are ok for hobby level machines, if you want any kind of accuracy then you would use pre-loaded profile linear rails

Page 1 of 2 12