Gantry design and FEA analysis

[Kitwn]

I know you are designing on the basis of 50 x 50 material but rather than the complexity of the braces in the designs above have you considered two pieces of 100 x 50 welded together to give a flat face 200 high? Your overall height looks to be about that anyway. I'm really asking because that's what I've built and I'd love to see the deflection calculations.

[eci22]

I know you are designing on the basis of 50 x 50 material but rather than the complexity of the braces in the designs above have you considered two pieces of 100 x 50 welded together to give a flat face 200 high? Your overall height looks to be about that anyway. I'm really asking because that's what I've built and I'd love to see the deflection calculations.

Hi Kitwin, do you mean like this ?


This seems like a very stout design and is better than the bracing I used. I have used it to generate some numbers below- hope they are of interest. But the analysis suggests the biggest component for deflection is the material of the Z plate, it seems to contribute the most. If you are interested let me know your exact materials ,including Z plate and I would be happy to run the numbers for you

I've changed all model calculations to use 100 Newtons of force. Also the Z plate is now 200x300x12mm Aluminium



My main question is where did you perform the welds for the two pieces of 50x100?, A seam weld covering the entire length would lead to significant warp id imagine and tack welds wouldn't hold up, bolts would work I suppose. How did you make yours ?

Agree with Voicecoil.
Analysis 2 is closer to worst case as this puts a twisting moment on the gantry as the force is applied at some distance away from the structure and will usually cause the biggest deflection.

Hi routercnc, can I just say the videos on your CNC build is really inspiring. I learnt so many subtle things the info is worth its weight in gold to me. It's something I come back to time and time again. Thank you for putting it out there !

I wrote a spreadsheet a long time ago just to play with the numbers and I looked at 3 load conditions.
Vertical load at the centre of the gantry due to the mass of the Y and Z axis and estimate the sag.

(All simulation diagrams use 100 newton force)

Vertical Load like this ?


Presumably the only force is gravity and the weight of your cutting tool, which I haven't included.

Lateral load applied at the tool tip as this will want to turn the gantry into a parallelogram- one reason why raised bed sides is stiffer than tall gantries.

lateral load


correct ?

Foreaft load at the tool tip to twist the gantry. This usually causes the most deflection and you need to make the gantry have a shape with material around the outside not 2 isolated beams. So yes plate them together is an option.

Is that the same as moment force around the Y axis of the Z plate ?

Remember this is all to see relative performance and get a feel for what aspects are important as it will assume perfect joints and you would need to model the whole machine and the cutter to get close to the actual deflection and even this is the steady cutting force not the deflection caused by vibration which will add some more.

Yes definitely, I just find this really interesting to do. It's a good way of comparing different designs. I'll use this for guidance only and for learning FEA which is something I have not done before. I also appreciate not having the Y axis, the feet, basically the rest of the machine modeled will skew the results somewhat

The stiffest gantries in terms of numbers are raised bed sides with a large box section as this gives good stiffness for all load cases. But practical considerations mean other options are good including the famous L shape gantry with 2 good sized rectangles joined in an L shape.

Would you mind directing me to a picture of this gantry design ?

Setup 2 is much more realistic since you want to design for decent performance when the cutter is hanging down a fair way below the gantry, 'coz that's where it will often be used. Having an equal horizontal force on each beam really isn't going to happen as that would mean the workpiece being considerably higher than the bottom of the gantry. 1000N cutting force is maybe a bit much unless you're using big cutters, I've heard figures of 200N tops bandied about before for typical machines such as people on this forum build though I haven't seen the sums behind that. I did find this reference a little while back though:

https://www.ctemag.com/news/articles...e-when-milling

Putting some numbers into the formula for a 1.5mm DOC on a 3 flute cutter at 0.1mm feed per tooth I get a tangential force around 90N for the best 6000 series ali.

It would also be interesting to do an analysis with a plate fixed on the back of your 2 bits of box section as that will make kind of a super box with the Z plate.

PS looking at your pics again, there might be a decimal point adrift in the text for the second set of results.....

Thanks for the link Voicecoil, much appreciated. I updated the the force to 100 newtons in my above simulations, I also made my Z plate thinner as I realised 20mm thick tool plate is out of my budget. If I have done the simulations correctly then the numbers still look promising. I will look to add the fixed plate at the back tomorrow. Thank you very much for your input

[Kitwn]

Hi Kitwin, do you mean like this ?


My main question is where did you perform the welds for the two pieces of 50x100?, A seam weld covering the entire length would lead to significant warp id imagine and tack welds wouldn't hold up, bolts would work I suppose. How did you make yours ?

That's about it. just two pieces of 50 x 100 welded together to give 200 x 50. The whole is then welded to the gantry uprights.

Bear in mind that this gantry was the first time I ever tried welding, I bought a cheap stick welder specifically for this job. To minimise the warping I clamped the pieces together and used a series of short welds alternating between the front and back seams. I use a welding technique I've called "Bird Poo" since that's what the result most resembles. Copious use of an angle grinder, gobs of car body filler and a layer of paint make it look much better than it is. The front surface was flattened (not very well) using epoxy.

[Voicecoil]

I also made my Z plate thinner as I realised 20mm thick tool plate is out of my budget. If I have done the simulations correctly then the numbers still look promising. I will look to add the fixed plate at the back tomorrow. Thank you very much for your input

Tip: if you can add some bits of bar to stiffen your Z-plate (i.e. turn it into a channel section) you will do better cost for cost than a flat plate. For example a 10mm plate 180mm wide with some 30 x 15 bar fixed down the edges is about 2.5x as stiff front to back as a piece of 20mm plate 180mm wide - go simulate it if you want to. I mention this because you generally end up with some space between the front and back z plates due to having to accommodate the carriages, ballscrew etc., so why not put something useful in there. The front (moving part) of the z-axis maybe isn't such an issue as the z rails will stiffen that.

PS If you want 0.1mm accuracy, you will likely need to design for static deflection figures rather better than that - remember that cutting metal produces a lot of vibration which can mess things up.

PPS what FEA package are you using? looks good.

[eci22]

Tip: if you can add some bits of bar to stiffen your Z-plate (i.e. turn it into a channel section) you will do better cost for cost than a flat plate. For example a 10mm plate 180mm wide with some 30 x 15 bar fixed down the edges is about 2.5x as stiff front to back as a piece of 20mm plate 180mm wide - go simulate it if you want to. I mention this because you generally end up with some space between the front and back z plates due to having to accommodate the carriages, ballscrew etc., so why not put something useful in there. The front (moving part) of the z-axis maybe isn't such an issue as the z rails will stiffen that.

Do you mean like this ?

PS If you want 0.1mm accuracy, you will likely need to design for static deflection figures rather better than that - remember that cutting metal produces a lot of vibration which can mess things up.
PPS what FEA package are you using? looks good.

The FEA software is Fusion 360, it is really nice once you get used to it. What kind of ballpark figures should I be looking at for in the deflection analysis ? At the moment I'm not actually taking figures as real world numbers, but more to compare the deltas between different designs.

Yes, vertical direction forces are mostly due to gravity but they will be much higher than 100 N as this only represents a Y and Z assembly of about 10 kg. More like 25-50 kg depending on the design, possibly more. This is to see how much the gantry will sag if you are surfacing a plate as in the extreme you would cut a dished shape. There is also a pull down force when using spiral fluted cutters, and there will be forces when drilling or plunging.

Great I will, add the additional forces for future analysis

The fore/aft analysis is in the wrong direction - it would be in the 'Z' direction using the coordinate system you have in the top right corner and would cause the part of the Z axis plate which is hanging down to all bend forward or rearward, and cause the gantry to twist about the 'X' axis as per your coordinates in the picture. This is usually the worst case of all the load conditions.

So rotating around the X axis would would be the following force, correct ?

Bear in mind that this gantry was the first time I ever tried welding, I bought a cheap stick welder specifically for this job. To minimise the warping I clamped the pieces together and used a series of short welds alternating between the front and back seams. I use a welding technique I've called "Bird Poo" since that's what the result most resembles. Copious use of an angle grinder, gobs of car body filler and a layer of paint make it look much better than it is. The front surface was flattened (not very well) using epoxy.

Yes I'm quite familiar with the bird poo welding technique : ). When you were welding, did you let the weld cool before removing the clamps and turn the piece around to clamp the other side or did you weld one side un-clamped, turn it over clamp it down and then weld the other side ? It you clamped down the first side then welded and turned over to weld the second side the clamps wouldn't have much effect on the first side at least ?- I'd be really interested to hear your workflow as I've alredy welded my Y axis and I am trying to improve my amount of warp in the rest of the build ( I'm using gaseless MIG, a very very basic machine), I've seen a bunch of you tube videos but haven't dialled in my sequencing yet.

Combine both and that's what you will have if you do not design properly the machine.

If you design your Z properly , use proper bearings and mount your spindle correctly, variant 2 will be non existent, so variant 1 is the real deal.
As if you clamp your spindl properly its body will strengthen the Z and only a couple of braces will make the Z equivalent to like a solid chiung of 100x100mm metal


Here is some design wisdom to you: it does not matter what profile you use. A well designed machine will have such design as to simulate at least 10cm wide x 3cm thick steel plate against all cutting directions. Against the 3cm no the 10cm. Or the equivalent. in whatever material/s you do it.

So go and make it 50x50mm but remember what i said if you want your machine to cut vibration free aluminum or even steel.

Check my build at page 7 for the gantry and page 16 and 17 for the Z, to grasp the idea of how serios a gantry and a Z have to be to machine fully extended at 200mm.

Its not only my machine i am bragging about. There at least a couple machines on forum that are seriously heavy duty and one way or another they are build like that.

Hi Boyan, thanks for your input- do you mean your build log from project 1 in your signature ? If so I just want to point out one thing from the first page of that build log:

2. Money is not an issue for the frame or the length of the supported rails, so i am not going to go cheap to save 10cm of rail or steel profile.

Unfortunately this is not the situation I am in, so I am trying to generate as much data and perform as much analysis on my design so I have enough information to make informed decisions about my build. Do you have images of your final build, it would be great to see an example of the Z axis you are describing. Thanks

[Kitwn]

Yes I'm quite familiar with the bird poo welding technique : ). When you were welding, did you let the weld cool before removing the clamps and turn the piece around to clamp the other side or did you weld one side un-clamped, turn it over clamp it down and then weld the other side ? It you clamped down the first side then welded and turned over to weld the second side the clamps wouldn't have much effect on the first side at least ?- I'd be really interested to hear your workflow as I've alredy welded my Y axis and I am trying to improve my amount of warp in the rest of the build ( I'm using gaseless MIG, a very very basic machine), I've seen a bunch of you tube videos but haven't dialled in my sequencing yet.

You are overestimating my skill and ability! I used a cheap stick welder I bought online which is the only welder I have ever actually used. The pieces were clamped together with C-clamps and vice-grips and if I remember correctly I used a couple of pieces of wood as braces to keep the pieces as flat as possible. I don't have a welding table so worked on the gravel path outside my shed. I clamped everything as tightly as I could before I began and didn't undo any of the clamps until it was finished. Having spent the last couple of days properly aligning this machine for the first time I can say how pleasantly surprised I am with the result of my first attempt at welding. It's needed some shims to get everything level and parallel but has turned out very well I think. The important thing is to think about how you will make your design adjustable, where you will need to have joints that can take shims for alignment and how you will access those joints when the machine is complete without taking half of it to bits again.

[eci22]

You are overestimating my skill and ability! I used a cheap stick welder I bought online which is the only welder I have ever actually used. The pieces were clamped together with C-clamps and vice-grips and if I remember correctly I used a couple of pieces of wood as braces to keep the pieces as flat as possible. I don't have a welding table so worked on the gravel path outside my shed. I clamped everything as tightly as I could before I began and didn't undo any of the clamps until it was finished. Having spent the last couple of days properly aligning this machine for the first time I can say how pleasantly surprised I am with the result of my first attempt at welding. It's needed some shims to get everything level and parallel but has turned out very well I think. The important thing is to think about how you will make your design adjustable, where you will need to have joints that can take shims for alignment and how you will access those joints when the machine is complete without taking half of it to bits again.

Thanks for the advice, yes I think the joints and adjustability becomes really important- so many choices!

Not really, I'm pretty sure it would be best to have the Hiwin rails fix directly onto the plate (which will add some rigidity already, especially with the extra stiffness of steel, so would be worth adding into you simulation) I was meaning maybe either side of the ballscrew (assuming there's enough clearance) or even on the front either side of the spindle mount if there's space. Just trying to help you add some cheap rigidity :-)

Like this ?



That does seem like a great idea to add rigidity

I think I've read a few times that it's better to have the guide blocks fixed and the linear rails moving down on the Z axis- I've also seen this quite a few times on builds, apologies if this is not what you mean by saying ('best to have the Hiwin rails fix directly onto the plate')

[Kitwn]

Thanks for the advice, yes I think the joints and adjustability becomes really important- so many choices!

You then have to choose what measurements you need to make and instruments to buy in order to set those adjustments correctly. There are a lot of worms in the can but you learn a lot too.

[Voicecoil]

Do you mean like this ?

Not really, I'm pretty sure it would be best to have the Hiwin rails fix directly onto the plate (which will add some rigidity already, especially with the extra stiffness of steel, so would be worth adding into you simulation) I was meaning maybe either side of the ballscrew (assuming there's enough clearance) or even on the front either side of the spindle mount if there's space. Just trying to help you add some cheap rigidity :-)

The FEA software is Fusion 360, it is really nice once you get used to it. What kind of ballpark figures should I be looking at for in the deflection analysis ? At the moment I'm not actually taking figures as real world numbers, but more to compare the deltas between different designs.

I'm no expert on dynamic loads whilst cutting, but as it's not exactly a smooth process (hence the noise!) I would expect peak loads to be may 1.5...2x the static???