[eddie3000]

Hi there fellows!
It's been over a year since I built my first cnc out of mdf, threaded rods and normal ball bearings. It works ok, but it's time to get something a bit more sturdy and less "wobbly". I was thinking of using iron/steel tubing for the structure and the rails.
My questions: Is it is a good idea to weld the structure and rails or bolt it all together? If I were to weld it, can I use an machined mdf supports to hold the rails in place before welding?
And while I'm at it, would you recommend acme lead screws instead of ballscrews for a hobby machine for ocasional machining? Price is important, and as it's just a hobby I not sure whether getting chinese rails and ballscrews are worth it.
Thanks.

[davidmb]

Hi Eddie,

It depends on many factors, like what resolution/ tolerance you desire, what you are going to cut wood / metal. How big a machine you are planning and what you intend for the drive mechanism ( stepper motors or servo ).

[OCNC]

Hi there fellows!
It's been over a year since I built my first cnc out of mdf, threaded rods and normal ball bearings. It works ok, but it's time to get something a bit more sturdy and less "wobbly". I was thinking of using iron/steel tubing for the structure and the rails.
My questions: Is it is a good idea to weld the structure and rails or bolt it all together? If I were to weld it, can I use an machined mdf supports to hold the rails in place before welding?
And while I'm at it, would you recommend acme lead screws instead of ballscrews for a hobby machine for ocasional machining? Price is important, and as it's just a hobby I not sure whether getting chinese rails and ballscrews are worth it.
Thanks.

You can certainly weld the structure but I don't think that welding the rails in place is such a good idea. You usually want to have some adjustability in the rail system to tweak the alignment. Unless you have some way of providing for this with your welded design I wouldn't do it. If you're only machining wood and aluminum and price is a factor then ballscrews wouldn't be necessary although commercial grade rails might be a benefit.

Also I came across this yesterday. It explains very clearly the process of choosing a lead screw. Also this page has a nice article on the merits and faults of the standard drive systems.

Chris

[eddie3000]

Sorry for the lack of information!!
I always have wanted to be able to work on wood and aluminium, but until today I have only done wood work. The toughest material I've cut is olive tree and holm oak. My current router works ok and I'm proud of it as it is my first, but it has it's problems, and one of them is that the entire structure is starting to get more flexible as time goes by, losing sturdinness. And I'm getting some serious backlash problems due to wear. I've learnt from my mistakes, and that's great.
At the moment my router is 3axis with 200step steppers and a g540 and EMC2, with no encoders, yet.
I've seen some projects around here that are great looking that are just bolted together, and others are welded. Welding the entire machine might be a problem for future disassemblies so maybe a combination of both is best? But if I go the welding way, how to keep the rails perfectly parallel is a worry.

[Benonymous]

Have a look at my thread Eddie.

http://www.cnczone.com/forums/diy-cn...l_1_x_0_a.html

The main reasons I went with my no-weld design were:

1. lack of any real welding skillz

2. familiarity with bolted construction.

3. ability to tweak and fit components to achieve squareness.

4. ability to break machine down to components.

Any or all of these could speak to you. My feeling is that to do a proper welded job, I'd need some sort of reference surface and that just wasn't available at the time (might be now :-) ) The drilling and tapping can be a pain but if you check the last post on my thread, I've assembled what you see there in a day and have already begun tweaking it to be flat/square and level.

[eddie3000]

You have a nice looking machine Benonymous. Tough looking angles too. Yes, I had seen it some time ago and liked it.
It has just struck me that I could bolt everything together, adjust as necesary, and once everything is tight and right I could weld the parts I want welded and leave the screws as decoration. I can leave other parts with the screws only to make the machine easier to move. I am no expert welder myself, but I have done it in the past and making a good weld isn't all that tricky, it's making it good and look nice at the same time what's difficult. And another thing that's tricky is welding things in place and not a mm more or less here or there.
And thanks OCNC for those links. Good stuff.

[TechGuy5002]

You have a nice looking machine Benonymous. Tough looking angles too. Yes, I had seen it some time ago and liked it.
It has just struck me that I could bolt everything together, adjust as necesary, and once everything is tight and right I could weld the parts I want welded and leave the screws as decoration. I can leave other parts with the screws only to make the machine easier to move. I am no expert welder myself, but I have done it in the past and making a good weld isn't all that tricky, it's making it good and look nice at the same time what's difficult. And another thing that's tricky is welding things in place and not a mm more or less here or there.
And thanks OCNC for those links. Good stuff.

The issue with welding is distortion. Unless all of your weld points have no gaps (can't fit in a razor blade edge), the welds will pull on the work piece causing distortion. Bolting also provides the options to shim work pieces so they align or are square. This will permit you to construct a more accurate machine, especially if you cant machine workpieces in a mill to cut them to precision length or flat and square cuts.

If you go with bolts, there really is no reason to follow up with welding, unless you really undersized your bolts for the forces applied to the structure. My only suggestion with bolting is to apply loctitle or some thread adhesive so that machine use (vibration, dynamic forces, etc) does not work the bolts loose over time.

I believe the majority of people choose to weld instead of bolting is because it takes a lot more time to align, drill, and tap. Its usually much quicker to just weld the workpieces together.

My suggestion is to go with bolts and take your time building your machine. Don't buy all of the parts at once if cashflow is an issue. Purchase the material and parts you need to assemble one assembly. When you completed the assembly than purchase the next batch materials and parts for the next assembly. This may permit you to spend more on better parts and materials since you distributing the costs over a longer period and you can apply materials and part changes if you discover an issue. For instance you may discover an issue with bearings or rails or the material is warped\uneven\not square. Take your time and do it right. Its much more expensive to rush and take shortcuts.

[jsheerin]

The issue with welding is distortion. Unless all of your weld points have no gaps (can't fit in a razor blade edge), the welds will pull on the work piece causing distortion.

Why does the fit up of the weld make a difference? My understanding (not being a very experienced welder) is that as you weld, you locally heat the material which makes it expand. I suppose you're saying that if you have tight fit up, there's no where for things to expand? Anyway, as you fuse the joint together, it gets joined in that expanded state, but as it cools, it tries to contract. This causes further warping. I'd think if you had tight gaps, then you would just get different distortions and induced stresses. The insidious part is that you can then machine everything flat and square if you like, but over time the stressed state that the material is in will work itself out, causing further warping. This is why it's recommended to stress relieve the part after welding but before machining. You put the material back to a low stress state before machining so everything stays flat afterwards. I don't think most people do this though.

[eddie3000]

I would like as much resolution as practical. My current cnc has 200step motors, 10 microsteps and 1.5mm/rev threaded rods. It's totally ridiculous. Even though the theoretic resolution is 0.00075mm (insane), at the beginning I had about 0.05mm real resolution, and after a bit more than a year I have about 0.5mm errors on occasions, mainly due to backlash I think and flex, and I can still achieve 0.1mm resolution with slow routing and careful gcode programming.
For my new cnc I was thinking in getting a leadscrew of at least 5mm/rev, 10mm would be great, depends on what's available. For what I do, I guess 0.05mm res is fine, but if I can get more for less, well better.
I think you have just about convinced me to go the bolted way. I very much like benonymous' machine. I think I may go that way. I can now start to look for materials in my area to start designing.
Thank you all for your feedback very much.

[TechGuy5002]

Why does the fit up of the weld make a difference? My understanding (not being a very experienced welder) is that as you weld, you locally heat the material which makes it expand. I suppose you're saying that if you have tight fit up, there's no where for things to expand? Anyway, as you fuse the joint together, it gets joined in that expanded state, but as it cools, it tries to contract. This causes further warping. I'd think if you had tight gaps, then you would just get different distortions and induced stresses. The insidious part is that you can then machine everything flat and square if you like, but over time the stressed state that the material is in will work itself out, causing further warping. This is why it's recommended to stress relieve the part after welding but before machining. You put the material back to a low stress state before machining so everything stays flat afterwards. I don't think most people do this though.

Yes, you pretty much got it. When there are gaps, the weld pool will take some metal from the workpieces and it will liquified. The gap will close as the work pieices expand. When the workpieces cool they will contract, the gap will not re-open causing the workpieces to be shorter and pulling on the workpieces.

Stress reliving has more to do with the crystaline structure of the metal. When you heat a workpiece the crystaline structure expands and becomes relaxed, when it cools back down (especiallly if it cools fast) it will cause the crystal structure to become be distorted since different regions of the workpiece will be in different crystaline structures depending on temperature and how fast it cooled. By heat treating the work pieces, the crystaline structures re-align remove the distortion quickly. The hotter the pieces is held the faster the structure is restored (to a point of course). The workpieces must be cooled evenly to avoid distortion after heat treatment. The crystaline structures will slowly normalize, even at room temperature, but it may take years or decades for the structure to normalize at room temperture. Pre-heating helps because it permits the workpeices to cool more evenly. Post-heating after with pre-heating will helps more since it will slow the cooling rate so the workpiece also cools more evenly.

If you can avoid gaps in welding joints, distortion will be minimized. You still have to apply good welding techniques to avoid distortion, but leaving gaps makes it almost impossible to avoid distortion and heat treatment will not correct distortion cause by gaps in the joint. You can chamfer the joints, but the workpieces should sit flat against each other below the chamfer. A triangle gap shape will cause the pieces to warp, pulled in the direct of the gap when you weld them together. You should also chamfer and weld equally on both sides so that any distortion (pull from the chamfer gap) is equal on both sides of the workpiece. Ideally you want pieces to fit tight, but not so tight that the workpeices are pushing on each other so than when the metal softens under torch heat, that it warps under the force being applied. If it warps because of force being applied, it will stay warped when it cools.

[eddie3000]

When you talk about distortion, do you mean that even if you clamp the pieces downs hard and you leave gaps when welding the pieces won't maintain the exact position they were clamped in?
When you talk about preheating and postheating, what kind of temperatures are we talking about? Something one can achieve at home?

[TechGuy5002]

I would like as much resolution as practical. My current cnc has 200step motors, 10 microsteps and 1.5mm/rev threaded rods. It's totally ridiculous. Even though the theoretic resolution is 0.00075mm (insane), at the beginning I had about 0.05mm real resolution, and after a bit more than a year I have about 0.5mm errors on occasions, mainly due to backlash I think and flex, and I can still achieve 0.1mm resolution with slow routing and careful gcode programming.
For my new cnc I was thinking in getting a leadscrew of at least 5mm/rev, 10mm would be great, depends on what's available. For what I do, I guess 0.05mm res is fine, but if I can get more for less, well better.
I think you have just about convinced me to go the bolted way. I very much like benonymous' machine. I think I may go that way. I can now start to look for materials in my area to start designing.
Thank you all for your feedback very much.

If your goal is precision, you will need a small mill to mill the parts flat and to provide good length tolerances. This will provide the means to keep your machine square. You can add shims but this can end up like wack a mole. When you shim (or file) one spot square, its likely to unsquare another joint somewhere else, unless your able to cut the workpieces with some precision and cut flat with close 90-degree angles.

Also consider that leadscrews can introduce considerable error even without backlash. Consider that the threading on the leadscrew may not be very accurate. For instance in one small spot the pitch might be 5.001 mm, in another small area, it might be 5.002 mm, etc. these can add up over time. There are two possible options: One use a precision leadscrew, or two, use a linear encoder that can provide feedback to the controller so it can make constant adjustments to correct the inaccuracies of the leadscrew and backlash.

[eddie3000]

When you talk about distortion, do you mean that even if you clamp the pieces down hard and you leave gaps when welding the pieces won't maintain the exact position they were clamped in?
When you talk about preheating and postheating, what kind of temperatures are we talking about? Something one can achieve at home?

[jsheerin]

For pre and post heating, I haven't seen it recommended for steel so I don't know what temps you'd be shooting for, but for aluminum it's not super hot. You can use an oxyacetylene torch to heat parts up, for example. For smaller ones, you could use an oven. However the thermal characteristics of steel and aluminum are quite a bit different - different coefficients of thermal expansion and different thermal diffusivities (if I'm remembering the right term - how easily heat gets distributed through the part). So I've seen aluminum preheated because otherwise it will take forever to get enough heat in the part to get the metal to puddle. On the other hand, steel doesn't conduct the heat away from the joint quite as quickly.

For stress relieving steel, that is quite a bit hotter, like 1100-1200F, and you need to keep it at that temp for an hour per inch of thickness, and heat and cool the part at between 100F-400F per hour. Personally I have some little heat treat ovens in my garage, so I can do this 'at home' for small parts, but all of these things are definitely leaning more towards industrial processes.

[TechGuy5002]

When you talk about distortion, do you mean that even if you clamp the pieces down hard and you leave gaps when welding the pieces won't maintain the exact position they were clamped in?
When you talk about preheating and postheating, what kind of temperatures are we talking about? Something one can achieve at home?

Yes, because Clamping will not prevent expansion. The workpiece will expand from the Weld area to the clamp. When the workpiece cools either the clamp will slip or the workpiece will deflect (aka like a spring). When you remove the clamp the workpiece will spring back to its relaxed position.

Preheating does not necessary have to be super hot. The purpose is to reduce the temperature gradient in the workpiece. Typically you heat the work piece evenly with a gas torch, or even a propane torch (to a lesser degree will work). When you weld you need to avoid excessive dwell time with the torch. use the torch to create a small weld pool and join a small area, then remove the torch and let the work peice cool down. Then repeat. A practical method is to apply stitch welding where you well a small stitch of the joint, the move over to another area some distance from the place you just welded and weld there. This allows the region you just welded to cool so the temperature gradent decreases, Then go back and fill in the stitch gaps until the weld is continuous. The less heat you dump and the more distributed you apply heat into the workpiece will reduce distortion.

[eddie3000]

Thank you guys for that info. I've been thinking a bit and reading a bit here and there, and with your replies I have come to the conclusion that I will weld some parts like the legs and structure it'll be resting on, I will bolt the all moving parts. I guess that's a good idea.
Now come's the fun part of designing the machine. I also think I'm going to have fun trying to tap holes in steel/iron tubing with my mdf cnc. Will it be up to it? Hmmm....we'll see.
Thanks for the tips.

[OCNC]

I also think I'm going to have fun trying to tap holes in steel/iron tubing with my mdf cnc. Will it be up to it? Hmmm....we'll see.
Thanks for the tips.

I don't see how you would be able to tap holes with a cnc machine. What do you have in mind?

Chris

[jsheerin]

You can definitely tap holes with cnc - typical methods are using a tapping head, rigid tapping or thread milling. I don't think you'll accomplish any of those in steel with an mdf machine if you have a router for a spindle though.

[adprinter]

The issue with welding is distortion. Unless all of your weld points have no gaps (can't fit in a razor blade edge), the welds will pull on the work piece causing distortion. Bolting also provides the options to shim work pieces so they align or are square. This will permit you to construct a more accurate machine, especially if you cant machine workpieces in a mill to cut them to precision length or flat and square cuts.

If you go with bolts, there really is no reason to follow up with welding, unless you really undersized your bolts for the forces applied to the structure. My only suggestion with bolting is to apply loctitle or some thread adhesive so that machine use (vibration, dynamic forces, etc) does not work the bolts loose over time.

I believe the majority of people choose to weld instead of bolting is because it takes a lot more time to align, drill, and tap. Its usually much quicker to just weld the workpieces together.

My suggestion is to go with bolts and take your time building your machine. Don't buy all of the parts at once if cashflow is an issue. Purchase the material and parts you need to assemble one assembly. When you completed the assembly than purchase the next batch materials and parts for the next assembly. This may permit you to spend more on better parts and materials since you distributing the costs over a longer period and you can apply materials and part changes if you discover an issue. For instance you may discover an issue with bearings or rails or the material is warped\uneven\not square. Take your time and do it right. Its much more expensive to rush and take shortcuts.

I agree with Techguy, the bolt-together option allows for adjustments in machine acuracy which would only be possible on a welded-together machine by first CUTTING it apart, and attempting to clamp, warp, and twist things into a position for re-welding (usually resulting in even more distortions). By thinking about these things during the designing phase, you can incorporate adjustment mechanisms in your design to aid you in making fine-tune adjustments to your machine (once complete). Things like jack screws, for shifting rails, or table position. Yes, it does require a LOT of drilling, and hole tapping. But it does make for a more acurate machine in the end.

[TechGuy5002]

Thank you guys for that info. I've been thinking a bit and reading a bit here and there, and with your replies I have come to the conclusion that I will weld some parts like the legs and structure it'll be resting on, I will bolt the all moving parts. I guess that's a good idea.
Now come's the fun part of designing the machine. I also think I'm going to have fun trying to tap holes in steel/iron tubing with my mdf cnc. Will it be up to it? Hmmm....we'll see.
Thanks for the tips.

I would recommend either solid steel or alum. over tubing. There is more material for tapping. with tubing its possible to strip the thread or crush the tubing. If you're going to put the effort into making a quality CNC, might as well start out with good materials.

I suppose a compromise is to weld on solid steel to the ends of the tubing, and drill tap the solid ends. but that would increase your labor. And you run the risk of distortion. You probably would need to machine the welded ends because it will be extermly difficult to weld them dead square to the tubing.

When you do the table/legs that hold your CNC, be sure to include machine leveling feet so you can make sure you table sits level for your machine. If the floor is uneven it might cause a wobble. If each leg has its own adjustable foot you can make sure it not going to wobble and properly support the machine.