[kis-mat]

Hey folks,

I need help with the alignment of my linear rails.

The planned system is illustrated below.
- 2000mm x 1360mm ( 40x80mm aluminum bars; connected with angles )
- Linear Rails: Two HGR15 linear rails ( yellow ) with two carriages per rail ( dark grey ), connected to 40 x 40 mm aluminum profiles using self-drilled 10mm thick plates
- this setup is for a pick and place robot with minimal operational forces beyond acceleration, utilizing a Core-XY rope drive.



My assembly plan is as follows:
1) Lay aluminum bars on a PVC floor to ensure a level base.
2) Attach the rails loosely.
3) Move the carriage to one end, tighten screws between the carriages slightly. Repeat on the other end.
4) Continue tightening screws in stages, moving back and forth, until all are secure.
5) Repeat 3) and 4) over and over, giving all screws an extra quarter turn, until they are tight.
6) Stand the construction upright.

I will be frequently laying down and standing up the construction after completion. My concerns are:
a) Will this process achieve proper alignment?
b) Can the alignment withstand the frequent repositioning?

Appreciate any advice or insights.

Thanks,
kis-mat

[peteeng]

Hi Kis-mat - Given your application I'd consider using supported round rail:
1) they are cheaper and have a large flat extrusion base making its connection easier. They are stiffer than square rail as well.
2) They are more forgiving in alignment. Given your structure is a bit floppy and it will move often I think they will give you less grief. I have used them on a machine and they were satisfactory for the job.

Peter

[kis-mat]

Hi Peter,

Thank you for your response. I would like to ask two additional questions for clarification:

1) It's commonly mentioned that rails offer superior precision, quieter operation, and smoother movement compared to rods, although the latter point is contingent on proper alignment of the rails. However, I'm struggling to quantify the impact of these differences on my upcoming project. Specifically, could you provide an estimate of the potential precision loss or increase in friction I might experience over a span of 1200mm when using rods? Are they significantly louder in comparison? Unfortunately, I haven't been able to find a direct comparison, for rails and rods of the same quality/price tag.

2) In the case of a construction primarily utilizing aluminum extrusions, would you consistently prefer rods, or do you believe that rails could be a viable alternative under certain conditions, such as a) employing thicker/more numerous extrusions or b) avoiding frequent repositioning (laying down and standing upright) of the structure?

Thank you in advance!

Best regards,
kis-mat

[peteeng]

Hi Kis-mat:
1) Rails can be more precise (if you use preloaded cars and precision grade components) but precision depends on the foundations of the rails or rods, they are not precise within themselves. In your case the foundation is quite compliant so this aspect is negated for both types. Square rail cars are noisy as they do not have a cage. You can get caged cars to reduce noise. Round rail cars are uncaged as well and I expect they make the same noise as uncaged square rail cars. I have run both and have not noticed any noise issues but then I make routers. They do "wizz" a bit when you set them up and push them around by hand.

I have made medical equipment in which noise is an issue and have had to run caged cars. This was because the service environment was very quite (an office type environment) and you could hear the balls going around the corner in the bearing! click click click....
2) Your structure is large with small members with no bracing. I can't predict how it will behave or perform after being moved up and down a few times. I expect it will be wobbly....

Your issue will be what accuracy do you require (+/-1.0mm ? +/- 0.1mm? +/-0.01mm?) and will this mode of operation achieve it? It maybe fine day 1 but over time and some lifts it may distort. I expect you will have to calibrate it every time you lift/lower. I still suggest you use round rails but square will be OK as well if you feel that's the way to go.

Consider a standard rectangular aluminium hollow section (40x80x6 or bigger) vs the construction extrusion. It will be cheaper and much stiffer then the 40x80mm construction extrusion and maybe lighter. The aluminium foot of the round rail can be directly screwed to it using self tappers or self threading screws or you can tap the holes... if you know someone who can AL weld then have the corners welded and it will be a more permanent solution. I presume its stored upright and used horizontal? Peter


https://medibath.com.au/wp-content/u...rth%20bath.mp4

Here's the machine I designed/built that required quiet bearings. Mainly because they operated in the vertical direction so the balls went around the corner then fell to the other side clik clik clikkkk This would have occurred with round or sq rails.

why the coreX drive? a simple XYZ belt is simpler???

Heres an article I wrote comparing construction to std extrusions....

[kis-mat]

Dear Peter,

I am very grateful for your detailed explanation. This is exactly the exchange I was hoping for here.

I've reconsidered the router's design and decided against a design that allows transitions from vertical to horizontal. This approach seems to complicate rather than simplify matters. Instead, I am now focusing on designing a robust structure that will be securely attached to the wall.

Concerning precision, I aim for a tolerance of +/- 0.1mm. Could rods meet this stringent requirement? Initially, my preference leaned towards rails, considering their theoretical advantage as linear guides. Yet, if achieving the requirements of linear rails proves unfeasible with an aluminum framework, pursuing this option might be impractical.

The system is engineered for light-duty tasks, specifically picking and placing lightweight items, yet it must accommodate high accelerations of 10-20g. Hence, it's essential to minimize the weight of the moving components, prompting me to consider a coreXY, or more precisely, a coreXYZ motor arrangement.

I would like to work while the pick and place work is being done next to the construction, but I think that cased carriage will be too expensive...at least more expensive than noise canceling headphones. However, the medibath seems like a great product!

It is very unintuitive to me that standard rectangular aluminum hollow section is more stable than construction extrusion but the math doesn't lie. Thank you for opening my eyes!

Best regards
Kis-Mat

[peteeng]

Hi Kis-mat - 10g-20g is a big challenge. A vertical structure saves space and will be stiff if attached to a wall. I would also consider a laminated timber ring beam. A solid section is more stable than a hollow and since it now does not move it would be easy to build. Then you can glue and screw the corners and an 80x80 solid section would be very stiff and stable. I use a lot of F17 plywood for gantries and machines and its proven to be stiff and light. Since ply is 700kg/m3 vs al at 2700kg/m3 you can use a lot of it to equal the aluminium weight. But geometry wins so big is good...

Rods will easily meet 0.1mm tolerance so will square rail. Since your aiming at a relatively precise machine you will need medium preload bearings. So I'd go square rail so you can get the preloaded cars. This is because std cars have hysteresis. ie they will arrive at a slightly different spot from different directions and they will wobble a little when they change direction due to the internal clearance. A pick and place machine needs to be able to arrive and stop at the correct position. Good luck - keep the thread going, like to see where you end up Peter

A 2mx1.3m corexyz machine will have really long belts. I have built 2mx1.2m routers with belts and they need to be very tight to work properly. The longer the belt the tighter it needs to be even though yours is a P&P machine I suspect a shorter belt system is the go or a rack and pinion.... Hmmmm What width belt were you thinking? I'd vote fopr R&P. I've just been thru a development and costing program for a large router, belts vs R&P and helical r&P wins out on cost and scalability. They come in 1.4m lengths Peter

what Z do you need?

[kis-mat]

Hi Peter,

I had also thought about plywood, but I'm worried that the wood won't run straight and might bend the rails when you screw them on.

Thank you again for the valuable exchange!

Best
Kis-mat

[peteeng]

Hi Kis- Its up to the carpenter to build straight. The material is fine. Here's a plywood router build called Frankie that cuts aluminium easily and has round rail. It now is owned by a guy that makes surfboard fins with it.... Peter

[kis-mat]

Hi Peter,

Interesting. I may be overly cautious when it comes to the alignment of the guides. Anyways, Frankie looks amazing.

As I can't see a shoulder in the pictures where you have aligned the guides: Did you proceed as I described in the first post, moving the carriages back and forth on the guide and tightening the screws gradually?

Best
Kis-mat.

[peteeng]

Hi Kis- pick one side as a maser and use a straightedge to check its straight. Then using the gantry work it down the track and snug down the secondary side as you go. Move the gantry back and forth checking its smooth and removing tight spots. Your hands are sensitive, once it feels free all the way snug everything up and recheck... Peter

[kis-mat]

Hi Peter,

Thanks a lot!

Ill post some pics when I got everything to work.

Greetings
Kis-Mat

[kis-mat]

... to see where you end up Peter

A 2mx1.3m corexyz machine will have really long belts. I have built 2mx1.2m routers with belts and they need to be very tight to work properly. The longer the belt the tighter it needs to be even though yours is a P&P machine I suspect a shorter belt system is the go or a rack and pinion.... Hmmmm What width belt were you thinking? I'd vote fopr R&P. I've just been thru a development and costing program for a large router, belts vs R&P and helical r&P wins out on cost and scalability. They come in 1.4m lengths Peter

what Z do you need?

It's fascinating how the brain skips details sometimes; I overlooked the last paragraph after spotting your name in the farewell. However...

For my Core-XYZ, I'm opting for 1.0mm Dyneema ropes instead of belts, given the design requires twisting the drive cables. However, rope length is a concern. By crossing the ropes behind the workspace, as illustrated in the fourth image at https://corexy.com/theory.html, I can cut each motor's rope requirement from 6.4m to theoretically 5.5m, saving 14%. I dont know if this is a big help or not....

On the Rack and Pinion suggestion, employing this for all axes implies moving the drives too, doesn't it? This approach seems to significantly bulk up the moving mass, demanding a beefier gantry to cope. Achieving speeds of 10-20g seems far-fetched under these conditions. Plus, I'm limited to the power from a standard 16Amp, 220Volt outlet, which caps my current 3x1kW motors' scaling.

I need a travel range of maximum 10mm on the z axis.

Best regards
Kis-Mat

[peteeng]

Hi Kis - 1mm dyneema rope is not very stiff. Super strong but not stiff. Moving mass is usually not a problem just use suitable motors. Have you calculated the size of motor you need yet? Dyneema is very slippery so you will need a few wraps on the drive drum, will have to be careful managing the wraps at the accels you want. What top speed are you aiming at? Peter

I've used a lot of dyneema in yacht rigs and in movie stunts... prefer Vectran, it won't relax or creep like dyneema will....If Dyneema gets warm around the wraps then it will relax and stretch. Vecran won't as its a crystalline plastic ie its structure is set vs dyneema which has been heat stretched to achieve orientation. But this process is never perfect. Peter

1kW motors? are they 3000rpm AC servos? you'll move a house with those. Lets look at the motion requirement:
1) what distance do you want to get to max speed over?
2) What max speed do you want?
3) What is the rotational inertia spec of your 1kW motor? is it 3000rpm max
4) I presume you are trying to minimize some sort of production time with this high accel thing
5) what is your estimated mass of the gantry or target weight of the gantry assembly?
6) what is the average distance between pick and place?

Then we can see what's needed to drive this thing. Peter

[kis-mat]

Have you calculated the size of motor you need yet?
[...]
1kW motors? are they 3000rpm AC servos?

Yes, 3000rpm AC servos. To be honest, I bought the most powerful I could afford (considering price and power requirement). I ended up with these.

I've used a lot of dyneema in yacht rigs and in movie stunts... prefer Vectran, it won't relax or creep like dyneema will....If Dyneema gets warm around the wraps then it will relax and stretch. Vecran won't as its a crystalline plastic ie its structure is set vs dyneema which has been heat stretched to achieve orientation. But this process is never perfect.

Thanks for the advice...Ill definetly consider Vectran!

1) what distance do you want to get to max speed over?
2) What max speed do you want?
3) What is the rotational inertia spec of your 1kW motor? is it 3000rpm max
4) I presume you are trying to minimize some sort of production time with this high accel thing
5) what is your estimated mass of the gantry or target weight of the gantry assembly?

1 & 2) The planned machine is designed to perform extensive start-stop operations, with 10 to 200 million precise movements in the xy plane, ranging from 0.1 to 5 millimeters, followed by pick-and-place actions involving sub-millimeter adjustments in the z-axis. Speed is considered secondary, with a target of lets say 2 meters per second, but acceleration is the critical performance factor. To optimize for this, I bought the most powerful motors possible and aim for a lightweight gantry. Given the 12mm radius of the 3D-printed pulleys, achieving speeds of 2 ( even up to 3-4 ) meters per second is within reach. But I doubt that I will reach these speeds within these small start-stop operations.
3) rotational inertia spec: 2,0kg*m2*10-4
4) Purpose: Precisely, the goal is to reduce production times. I'm a scientific associate in robotics, experimenting with a concept during my personal time. This is a long-term passion project, driven more by interest in software development than by economic motivations.
5) Granty Weight: As you can imagine, I'm still in the planning phase. I'm still wavering between steel, aluminum or carbon rods or rails, constructing the tool-change mechanism and so on... But I think a weight of between 2 and 6 kg including cable routing and slide arm is realistic.

Thanks again!
Kis-Mat

[peteeng]

Hi Kis - Your aiming very very high for a precision machine to do that many ops on a shoe string (or a paraglider string actually )... But we must aim at something. So a quick calculation, say you want to move something at 10g over 5mm. V^2=U^2+2as so U=0 (2x9.81x10x0.005)^0.5 = 0.99m/s say 1m/s if it was 20g then 2m/s so your Ok there. I'll crunch the torque required...Peter

what torque is your motor rated at? 4Nm?

[kis-mat]

Hi Kis - Your aiming very very high for a precision machine to do that many ops on a shoe string (or a paraglider string actually )... But we must aim at something. So a quick calculation, say you want to move something at 10g over 5mm. V^2=U^2+2as so U=0 (2x9.81x10x0.005)^0.5 = 0.99m/s say 1m/s if it was 20g then 2m/s so your Ok there. I'll crunch the torque required...Peter

what torque is your motor rated at? 4Nm?

Hi Peter,
3,19Nm

If relevant, these are the motors

https://www.omc-stepperonline.com/de/t6-serie-1000w-ac-servomotor-kit-3000rpm-3-19-nm-17-bit-encoder-ip65-t6-rs1000h2a3-m17s

Thanks again!
Kis-mat

[peteeng]

Hi Kis - check my math - done quickly. Sorry motors not good for 10g maybe around 1g... and I have not included torque required to turn the motor yet or friction or other stuff.... Stopped at this point. But as the distances are small maybe this does not matter. Have to figure out a typical path and do more numbers. You will need a very smooth motion algorithm for this type of short start/stop. Ideally you need a controller with jerk control. Plus input shaping.... see my thread here...

https://www.cnczone.com/forums/mecha...30780-cam.html Peter

since your cycles are short and sharp you maybe able to sneak into the overdrive torque of the motor as its 9Nm. and since your motors are stationary you could water cool them. But this is a fine line between cooked motors and hot motors...

Is your project about motion optimisation ? path optimisation? or some sort of operational logic?

[pippin88]

2 metres per second = 120 metres per minute = very fast

20g is massive acceleration

Your frame is completely unrealistic for what you want

[peteeng]

Hi Kis - since the motor can move the system at 5m/s, and you need 1m/s if you use a 1:5 pulley or gear then you have enough torque from the 3Nm motor. But torque will be absorbed by the motor and the gearing so need to do more maths on that.... Peter

at 10g you need 1.3Nm to turn the motor so need 19+1.3 ~20Nm. If you use a 1:5 then you need 20/5=4Nm which is close to your 3Nm motor. So you could get close to 10g but not 20g... unless two motors are driving one axis.... Peter

now you have 60kgf plus pushing the machine around and my 250kg router dances at 0.1g so your frame will need to be nailed to the earth to stop it break dancing... Peter

[kis-mat]

2 metres per second = 120 metres per minute = very fast

20g is massive acceleration

Your frame is completely unrealistic for what you want

Hi Pippin88,

yes, that had already become clear during the conversation.
I will build the router much more stable, mount it in the wall and do without the possibility of reorientation.

Thanks!

---------------

Hi Kis - since the motor can move the system at 5m/s, and you need 1m/s if you use a 1:5 pulley or gear then you have enough torque from the 3Nm motor. But torque will be absorbed by the motor and the gearing so need to do more maths on that.... Peter

at 10g you need 1.3Nm to turn the motor so need 19+1.3 ~20Nm. If you use a 1:5 then you need 20/5=4Nm which is close to your 3Nm motor. So you could get close to 10g but not 20g... unless two motors are driving one axis.... Peter

now you have 60kgf plus pushing the machine around and my 250kg router dances at 0.1g so your frame will need to be nailed to the earth to stop it break dancing... Peter

Thanks Peter for the calculations!

There are many things to think about...A Core-XY arrangement acts like a very strange 1:sqrt(2) gear; depending on the direction of movement two motors pull on the granty; I am not sure if the z-motor helps for xy-motion or not; in some directions of movement gravity helps, in others the motors have to overcome gravity...

i have long days at work today and tomorrow, but i will try to make a detailed analysis of the dynamics and kneematics as soon as possible and share it here.

Thanks a million for your thoughts, I'll get back to you as soon as possible!
Kis-Mat