[Jtepper]

Evening Ladies and gentlemen.

My Father and I have both taken the next week off to put this project together, have a look at it here.

JT engineering projects

There should be updates almost every day over the next week, my girlfriend loves taking photos, so there should be heaps of them.

Also Ill check back on this forum pretty regularly so feel free to post questions.

Enjoy
JT

[DonFrambach]

Looking good. I'm sure its great working on a project with your dad! Have you considered diagonal braces for increased stiffness?

[Jtepper]

Do you mean some triangulation in the legs? I did a quick FEA in solidworks, and that table will hold 12 tones before the legs buckle, and its just a table anyway. So I think it will be fine, haha.

As for the gantry, both the main beam and the legs/gantry support things, are 200x100x4 rhs, and my quick fea indicates about 0.00055 mm deflection with the router fixed in the centre of the table, and the appropriate forces applied at the bearings and ballscrew mounts ( most of that deflection is in the little bit of those arms that hangs over the edge of the table to hold the ballscrew nuts). I expect more inaccuracy than this just from the gantry being out of square and this strain calculation didn't include my aluminium z, which will probably flex the most anyway. (if this doesnt make since let me know and ill post some fea screenshots tonight)

(any suggestion how to get it better than the 0.1 degrees tolerance of the digital level are welcome?)

Also, as i mention briefly in the blog post, Im still tossing up whether to spend another $250 upgrading to 490 oz-in motors and a 48v power supply (currently 290 oz-in and 24v, was just going to get another 24v supply and series them) from ozxmods.
My gantry looks like it will weigh about 40kg, but even a 290 oz-in motor on a 5mm lead screw will make about 400kg of linear force so cutting force isn't going to be an issue.

So as I see it at the moment, Im going to try the 290 oz-in ones and expect fine operation but sluggish acceleration for rapids. And then the 490 oz-in ones will be bolt in replacements if the rapids are too slow.

Does this seem about right? Some feedback on this would be good, tempted to just bite the bullet and get the bigger motors, but if there isn't any real benefit then its just wasting money.

[PaulRowntree]

Do you mean some triangulation in the legs? I did a quick FEA in solidworks, and that table will hold 12 tones before the legs buckle, and its just a table anyway. So I think it will be fine, haha.

Buckle-proof should never be the design objective : you are trying to keep your frame square and in position to within a few thou of an inch subject to the static and dynamic forces involved. Check that out in the FEA. If it won't flex, it won't buckle either.
Don's suggestion of triangulation of the legs and table top is a really, really good idea.
Cheers!

[Jtepper]

The stiffness of the linear rails on the table top relative to eachother, and relative to the ball-screw supports is very very high, (only a very small contributor to the 0.00055 at the centre of the gantry). All the legs do is stop the table falling over, i don't see any need to make the table top stiff relative to the floor, or am i missing something?

Also the heaviest thing i will probably ever put on it will weigh about 20kg, so theres a 50 times safety factor in that 12 tones

[bushwakka]

It is not straight compressive loads that you are trying to deal with. It is the momentum of the gantry acting sideways on the leg.

I think you'll find that if the legs aren't stiff enough, that the unit will walk around your workshop. Especially with a 40kg gantry (will become 50kg+ with z-axis and chinese spindle) moving back and forth.
On my small machine (2x4, gantry z-axis and spindle around 30kg), it used to slide on the concrete floor around 40mm under heavy deceleration from rapids (admittedly with 400W servos), and that is with no legs. I'd hate to think how much movement you'd get with 600mm (2ft) legs or higher (even with small motors). The diagonal bracings are a must in my opinion. They are also a very cheap and easy to install early on, and it gets progressively harder to do as your construction gets further along.

Alternately, a lower level shelf tying the legs together or some big gussets will help.

Cheers
Paul

[dmalicky]

Very nice documentation of your build!

Don, Paul, and Paul have excellent advice. Every time the gantry accelerates, a horizontal force is applied to the lower frame. The magnitude is the (acceleration in m/s^2)*(gantry mass). Worse, Mach3 only has a linear accel profile, so there is a 'jerk' at the beginning and end of each accel -- this jerk induces vibrations, and the more horizontal deflection, the worse the vibration.

In your FEA model, try a horizontal force applied to the main rail of the frame top. Note the horizontal deflection. Then triangulate the lower frame, for example like this: Machine frame - MadVac CNC
(EDIT: better to make those diagonals meet at the center, or for your smaller overall length, a single diagonal per face (corner to corner) would probably be fine.)
Compare deflections -- they will probably be on the order of 100x less. Note the diagonals don't need to be particularly large in cross section -- it's the truss structure that gives the stiffness.

I'd suggest posting screenshots of your FEA, as it's easy to specify unrealistic constraints.

As for welding them on, use lots of small tacks, and/or lap-rosette welds to induce the least distortion. (Butt welds contract and pull the diagonal a little shorter, putting a kink in the frame [Edit: or pull the face out of square].)

[bushwakka]

It is more important to get the bracing between the legs in the x direction than the y direction, because of the mass of the gantry. The movement in the y axis will have a lot lower dynamic force, because you are only moving the z-axis and the spindle. (F=ma and the mass is smaller).

Also be careful with tack, stitch, stitch welding as you will get some deformation. You'll want to work your way around doing little tack, then little stitches on each side the small stitches to finish each side. My gantry sides (100x10mm) pulled close to 3mm over 125mm from a single stitch on one side only. Thankfully, it was the only unimportant dimension on the gantry.

I also found it easier to have it all assembled at the tack stage, so I could tell if it was warping as the slide became difficult to move.

The way you are heading though you should have a very usable machine... I was pretty happy with what can be achieved with a chop saw,a spirit level and a square.

Cheers
Paul

[Jtepper]

If you look at the cad there are leg ties at the bottom ( there 50x50 shs). Ive just welded these in, in the last hour, and its rock solid. Ill put photos up tonight. (as bush wakka suggested in his first post)

I do recognize exactly what the diagonal ties would do, but I cant get steel to do it today anyway, and i want to do the resin top overnight tonight. If i weld them in after i do the resin, the distortion is just going to ruin my resin surface.

Although you do all have me curious about how much its going to move so ill do an accelerating gantry load case on the table tonight to have a look.

As for the welding, we've got it all jigged up (admittedly in a wood jig), and i tack every edge before seaming it. Our legs were pulled about 0.2 degrees out of square by the first seam on each leg though (not mission critical dimensions), because they weren't jigged.

If Im going to get the bigger steppers though, i would like to order them tonight so they'll turn up Tuesday, if i could get some feedback on that it'd be good.

Also thanks for the feedback guys, anything that get me thinking about the design is positive I think.

[bushwakka]

Jtepper,

You can always put a turnbuckle style adjuster (with chain, cable or steel strap) in later (one each way) if you find that there is a bit of movement. It might not look pretty, but it will still achieve the desired result. Alot of what we are talking about depends on the acceleration rates of the gantry (since Force is mass x acceleration). If you only have little motors then your acceleration rates are limited and the problem may not even exist on your machine. You can also tune it down in Mach3 if needed.

I have no experience with stepper motors so can't be of assistance in that regard. You do appear to have 2 stepper motors working the x-axis, so you should have reasonable amounts of power. The 1605/2005 leadscrews may load them at rapid speeds though. Set your rapids to whatever suits. Not sure what sort of cutting forces you will expect either, obviously that depends a bit on the bit size.

With my 400W servo motors, in the early stages I was playing with 20m/min to stopped, and the base of the machine would slide across the floor (some law of physics there - for every action there is an equal and opposite reaction). So the force of stopping the gantry, was applying a force that moved the base, that was being absorbed as a frictional force of the steel frame on the concrete floor). I've since settled on slower acceleration curves, slower maximum rapid speeds, because it was brutal (even though it was awesome to watch). I had no legs on the machine, so there were no moments acting on legs to worry about.

The other thing I found is that a router for wood can handle 100thou (0.25mm) vertically without too many issues. The sheet timber (mdf) will normally conform to the bed. So if you surface the table with the machine, it does't need to be perfect across the whole table. This statement is probably defunct if you intend to cut non sheet product or metals.
Any more that 100thou though becomes the difference between a clean cut and an onion skin that you can't break (you then need a knife or a bearing guided flush trim bit in a router table.

You'll need to be pretty accurate with the alignment of linear rails. From experience, if they are out then you start putting forces in your bearing blocks which will cause the little plastic guides to crack and spit the balls out. I found that cutting forces exacerbated this, and I only spat ball bearings when I was in the middle of cutting something important.

The other area where you need to focus some effort is the axial alignment of the motor shaft and the ballscrew shaft. I didn't realise the importance of this, and had to resort to plywood interface plates with washer shims to get it in alignment. It is a PITA trying to work out where the motor mounting holes are, given that your reference point is normally the base of the bearing block (which is in a different plane, and on wrong side of the plate you want to mount the motor on.

Keep the pics coming. It is looking good.

Paul

[Jtepper]

Evening, a new post has been uploaded to the blog with todays photos JT engineering projects

As you will see, we had some really big issues trying to lay down our fiberglass resin , to try and make the rail supports level relative to each other. The resin didnt go flat....at all. As in long undulations, about 1mm in total height difference all over the surfaces? Any idea what went wrong?

Also after a few hours of deliberation and fea model making, we decided against adding ties, since the table moves at most 0.5mm as is before the feet slide. So it will be as simple as, if the feet slide around, well turn the acceleration down.

[DonFrambach]

I have never leveled anything with epoxy resin but I think I know what went wrong. As the resin cures it undergoes an exothermic (heat producing) reaction. The metal table acted as a uneven heat sink so there were a number of temperature gradients. This, I think, caused the ripples to form. I think the solution is to use a very slowly curing resin which begins with very low viscosity.

[bushwakka]

I'd say Don is probably correct.

The question is do you really need the epoxy? The only 2 surfaces that need to be levelled are the long SHS that will have the linear rails on, and they might not need much work done to them to get level.
Not sure what you are using for a base fill (cutting surface), but you can surface it once you are done.

If the jigging was good when you welded then you should find that it is pretty close anyway, there is not really that much deformation over 1200mm length.

Place a linear rail on each side and see what you get in gaps between the base of the rail and steel. You may be able to shim it a little if required.

Then use your spirit level or eye and see how they compare to each other horizontally.

From here, I would weld up your gantry and fit the x-axis bearings (or make a jig with a bearings each side). Then mount themlinear rail on one side, fit the jig and then mount the second rail, making sure you can slide it easily from x- to x+ (each time you put some mounting bolts in the linear rail). If you get it right, it should slide with about 50mm tilt on the table, or be easy to push with 1 or 2 fingers.

If you really need to use the epoxy, then you might need to mount the linear rails with standoff bolts and use the epoxy as a grout.

Cheers
Paul

[dmalicky]

Nice progress. That's a bummer on the resin, but what would a project be without hiccups? Some suggestions--

The CAD model shows the X ballscrews mounted lower than the X rails. You have the X rails smartly mounted at a similar height as the cutter. Can you raise the screws to be at the same height as the rails? Then everything is in 1 plane and the moments are reduced.

It's great to see the stiff gantry tube, but it looks like the CG of the gantry is pretty high off the table. If you can spread the X bearings out further (looks like they're inset from the leg ends?), then gantry accels will result in less tipping due to bearing deflection.

The accel may need to be turned down a lot to keep vibration under control and feet from slipping. That's a good check on the 0.5mm deflection, but also think about the jerks at accel start/stop and emergency stops. Machine tools always run better with the base both stiff and heavy (much heavier than the moving parts). A usual solution for sliding feet is to add mass to the lower frame--sand bags, concrete, steel, gold, etc. With the resin at least on hold, I'd diagonalize while it's easy.

Yes, those tube caps make a big difference. In my FEA models of 1500mm gantry tubes (200x200x6mm alum), placing about 6 bulkheads evenly spaced down the tube increased stiffness at the tool by 200 - 300%. Thin walled tubes are great for I and J, but the cross-section can't take concentrated loads very well, even on the edges. (Your steel gantry tube is probably stiff enough for its length already, but it'd be interesting to model a few bulkheads to see the difference.)

When welding the gantry, I'd avoid continuous beads. Welded steel will stress relieve over time, causing distortion. Strength isn't an issue, so I'd suggest just enough tacks and stitches for stiffness. Rosettes welds are useful because they tend to self-equilibrate.

[ger21]

So as I see it at the moment, Im going to try the 290 oz-in ones and expect fine operation but sluggish acceleration for rapids. And then the 490 oz-in ones will be bolt in replacements if the rapids are too slow.

Odds are that the 490oz motors may give even slower rapids, due to the high rpm's required to spin your 5mm pitch ballscrews.
In most cases, as the motors get bigger, their maximum rpm's decrease.
In order to go faster with a bigger motor, you need to make sure you get a very low inductance motor, which will usually be very high current and require more expensive drives, or use a belt reduction to spin the screw faster than the motor.
This keeps the motor rpm's lower, where it has more torque.
When using steppers, properly matching the motors to the drive system can be critical to good performance.

[Jtepper]

Hey guys heres another new update,
JT engineering projects
Im sure youll be pleased to see there's added leg braces


@ dmalicky
I put the ballcrews down there for two reasons, firstly to try and protect them a little from swarf from the cutter, and secondly, I would have needed a 200x100 tube to fit them both on the top, instead of a 100x100, which just seemed excessively large to me.

Also when i welded the gantry, I didnt want to just tack it, because i wanted the tubes sealed to avoid corrosion, and just general aesthetics, so i ended up doing a few tacks on each part, cycling between the different welds, just adding tacks until the seams were all filled, and it seems to have come out pretty square.

[DimONN]

I like , good job:cheers:

[Jtepper]

more photos
JT engineering projects

[dodger889]

You might want to keep you smaller table. You never know when you want to do something on a small table when the big one is busy. Clean it up and set it aside for a little while. Or make it just for engraving little things.

[dmalicky]

It's looking great! And kudos to your gf for the excellent pics.

If you want, it looks like the ballscrews could be raised a bit higher and still stay on the side of the tube.

I'm curious--did the FEA show the wood y-car/z-plate (with rails) was stiff enough for longitudinal loads, especially with the router halfway or all the way up?