[thunderdork]

Hi all,

I thought I would post a link to my current build here. I'm posting videos and pictures for every step.

ThunderDork » CNC

I basically bought the servo kit with the c11 BoB upgrade from automation technologies in (kelinginc). I am starting with sourcing the harder to find parts and building the control box. I have experience with my zenbot cnc but this is my first stab at building a machine. I plan on borrowing designs and buying the harder to fab items from cncrouterparts.com. I'm gonna be using all steel instead of 80/20 thought to keep down the price.

Feel free to subscribe to my youtube channel. I would love any feedback or corrections!

[CarveOne]

Steel is cool. It's my own preferred material to build with.

[thunderdork]

hehehe, CarveOne, i'm using your signature one of these days. that's awesome.

[CarveOne]

hehehe, CarveOne, i'm using your signature one of these days. that's awesome.

Go ahead. I wasn't clever enough to think of it myself anyway. I'll find something else given enough time.

I have two steel frame CNC machines (3'x4' and 5' x 12') and soon to be a second red oak frame machine. (3'x4' and 28"x 28"). All of them use steppers.

[thunderdork]

Very nice

What type of steel did you end up using for the chassis? I as thinking of using 2"x4" or 4"x4" for the main rail ways (all 1/8" walls) and 2"x2" for the minor stuff. You think that'll be beefy enough for something up to 5' x 9'?

I decided to go with a slightly larger bed for ATC in the future and clampling.

[CarveOne]

Very nice

What type of steel did you end up using for the chassis? I as thinking of using 2"x4" or 4"x4" for the main rail ways (all 1/8" walls) and 2"x2" for the minor stuff. You think that'll be beefy enough for something up to 5' x 9'?

I decided to go with a slightly larger bed for ATC in the future and clampling.

Don't use anything less than 1-1/2" x 3" x 3/16" wall (or bigger and thicker wall) standard steel construction rect-tube . 1" x 2" x 1/8" is too flexy. My big machine made of 1" x 2" x 1/8" works ok but it shakes too much with accelerations of 20+. It shows up in the circular borders as little ripples. It's not so bad but is just unnecessary sanding work.

I tried 1-1/2" x 3" x 3/16" C channel on the smaller machine and it was just not stiff enough the way I mitered the corners. Being a newbie steel builder has its challenges. I made a new frame with 1-1/2" x 3" x 3/16"box tube and it is very (very) stiff now. The C channel was also bowed from welding too easily.

[harryn]

My suggestion is to do a calculation on the steel tube beams before you buy. It is amazing just how much things will flex, especially when you are talking about reasonable precision.

The things we normally own actually are designed to flex, we just don't think about it that much. If your machine flexes, then it is not just a matter of part size being slightly off, the finish will be poor and chattered looking.

There are some on-line calculators, but the basic math is not complicated to calculate beam stiffness. The idea is that you want:
- stiff material - steel if possible
- wall thickness - as much as you can stand, but at least 1/4 inch for that size machine
- The wall thickness as far from the beam centerline as possible. (big beams)

I was kind of shocked when I ran the numbers for my own build of a machine that size. Roughly:
- 4x4 uprights, but 6x8 was better
- Minimum 6 x 8 in for the gantry, but 6x10 was better
- If you support the main X beams at the 1/2 way point, then you can also use 6x8 for the X support beam.

You can sort of see a 6x8 trend in the numbers, so that is what I have on paper so far. (not built yet for many reasons)

Mass - getting close to 1 ton if built as a table machine. I know, sounds crazy, but that was for typical wood type flexing goals.

These results drove me toward thinking about a machine with the wood cutting near the floor rather than up on a table, mostly to save weight and cost.

[thunderdork]

wow, thanks for the info harryn and carveone.

It seems that with this much weight and extra cost, would it make more sense to go back to 80/20 aluminum? I know that the mass of steel itself will help with vibrations though. And I was going this way to save money. Well, that and being able to mig weld it in sections and not have a billion bolts everywhere.

Any links to these calculators for beam stiffness?

[ger21]

It seems that with this much weight and extra cost, would it make more sense to go back to 80/20 aluminum?

To get comparable stiffness from aluminum might actually cost considerably more. Be sure that your comparing apples to apples in regards to stiffness.

[OCNC]

Any links to these calculators for beam stiffness?

I use a program called MDSolids3.5, an educational program developed by Prof. Timothy Philpot. Attached are a couple of loading and deflection diagrams for a steel and aluminum section, both measuring 2 x 4 inches with an 0.25 wall. The aluminum is deflecting about 2.5 times more than the steel for the same point load of 100 lbs. Let me know if you want a particular section, beam length and load analyzed.

Chris

[harryn]

Hi Chris,

If you have a chance, could you consider to run a deflection on:
- 72 inch beam
- 4 x 4 x 1/4 steel vs 6w x 8 deep x 1/4 steel tube
- 100 Lb point load
- That is close to the gantry needs of a 5 ft wide setup

Would either of these hit deflections of less than 0.001 inch ?

Thanks

Harry

edit, obviously, Chris's numbers are more graphical and visually helpful. The web site engineersedge.com has some useful calculators as well. BTW, while some parts of the formulas are linear, some are not. Longer sections need substantially more depth and cross section than a linear scaling would suggest.

[OCNC]

Hi Chris,

If you have a chance, could you consider to run a deflection on:
- 72 inch beam
- 4 x 4 x 1/4 steel vs 6w x 8 deep x 1/4 steel tube
- 100 Lb point load
- That is close to the gantry needs of a 5 ft wide setup

Would either of these hit deflections of less than 0.001 inch ?

Thanks

Harry

edit, obviously, Chris's numbers are more graphical and visually helpful. The web site engineersedge.com has some useful calculators as well. BTW, while some parts of the formulas are linear, some are not. Longer sections need substantially more depth and cross section than a linear scaling would suggest.

Working on it.

Chris

[OCNC]

If you use 6x10 with a 3/16" wall you'll have a little bit less deflection and about
17 lbs less weight (134.52-117.78=16.74).

These were all figured using A500 GradeA steel with a Yield Stress 39ksi and Ultimate Stress 45 ksi.

Chris

[harryn]

Chris, thank you very much for those calculations. That info will help a lot of people, including me.

Thunder, the calcs Chris posted are really good indications of what will be needed for a large machine, raised up on a table. You can quickly see why I am considering to build the table closer to the floor. The more supports to the floor, the less steel is needed in the beams. (less distance between posts = much less beam needed. )

There is nothing wrong with structural extrusions like 8020 - great stuff, and the design of it is quite efficient for stiffness. I am not sure about others, but when I ran the numbers using Bosch extrusions, for similar stiffness to steel, it was going to be more than 5X more costly, once you include the fasteners. It would be really nice to use that stuff though, because assembly and changes are easier.

[thunderdork]

Okay, yes, that does make sense. So what are the sizes we can come to a consensus too for 9' of y travel, 5' of x travel and 3' of z travel? Should it all be 6"x10" beams?

Also, with this amount of mass would 2 of these motors for the y, 1 for x and 1 for z be enough at 1125oz?

I have a video run down of all the main controller parts on my site.

Sorry for my ignorance. I am by no means or stretch of the imagination a physics expert

[harryn]

Okay, yes, that does make sense. So what are the sizes we can come to a consensus too for 9' of y travel, 5' of x travel and 3' of z travel? Should it all be 6"x10" beams?

This somewhat depends on your design. Imagine a kitchen table that is 10 x 10 ft, with just 4 legs at the corners. That makes for a 10 ft "span".

If you now add a few more legs along the sides and one in the middle, you can make it so that this "span" is reduced to 5 ft, resulting in a need for much smaller horizontal support beams.

If the table were sitting on a very flat floor (completely supported) then the beam size is minimized, but loading it is less fun. This is one reason that routers with the gantry riding on top of fixed side rails are very stiff. Once again, somewhat harder to load material in though.

Added to all of this, is the reality that there are some economic and convenience benefits from using one size of steel tube for all major pieces, and whatever you use, you have to be able to lift it and move it around, at least some. Once you build a 1 - 2 000 lb frame, it is more challenging to move around.

Personally, I have not been able to make the mental jump to 6 x 10. I am playing around with making some plywood 6 x 8s to prototype a section of a frame to learn from. I don't have access to steel fab equipment, so I will need to job all of that out. Plywood is easier for me to proto from.

[thunderdork]

I was think of doing 6 legs if its the same tubing as the motion rails or 8 if its slightly less. Probably much like the mechmate except all my legs will go to the floor.

its probably going to be configured something like this:

I-----I
|xxxx|
IxxxxI
|xxxx|
I-----I

The 'I's are the rectangular leg beams. I plan on making the bed removable. So that I can have a z travel of 4" for sheet work, then remove the bed to do 3' of z travel for 3d foam block carving. I want to keep the front and back end open for sliding material in and out in that mode.

[harryn]

Hi Looking at your video of the power supply, your observation is dead on. Don't connect those sides in series or the voltage will be too high.

That transformer is wired with one primary, and two secondaries. Each of the secondary coils is center tapped. (the C) which can provide about 1/2 voltage in case you need it for something.

I am pretty sure you can also tie both sides together in parallel if you want so that it behaves like one large supply, rather than two in parallel. This actually has some advantages, because when a motor goes into reverse, it creates a back EMF which will go back into the supply and raise the voltage. If the other motors can consume this power, it helps, but a protection circuit is helpful.

http://www.geckodrive.com/returned-energy-dump

PMDX also offers a board you can buy with similar protection already done for you.

http://www.pmdx.com/PowerModules

All things I learned reading CPF way too much, wikipedia, and info from linear power supply companies in the past 2 months.

[harryn]

Those are some very healthy servo motors you have there. 1150 oz inch.

I have so far, just one 1200 oz in stepper, and you most likely already know that one significant difference between them, is that my stepper will loose torque as the rpm increases, while servos mostly hold their torque as the rpm increases.

Those are real horses. Impressive.

If you couple those with some small pinions, you might get away with no gear reduction.

Example:

1200 oz in motor + 1 inch pinion, so 0.5 inch torque arm.

(1200 oz - in / 0.5 in ) = 2400 oz.

( 2400 oz / 16 oz per lb ) = 150 lb force at the pinion tooth from one motor.

Stand out of the way when you turn those things on. They will move about any home made system around.

With the stepper system, the challenge I am having is resolution. The steppers have 200 steps x 10 microsteps, so 2 000 micro steps / rev.

1 in Dia pinion x Pi = 3.14 inch movement, so 2 000 steps / 3.14 inch / 2000 steps = 0.0016 inch per micro step, but really 0.016 inch per step.

What is the resolution of your encoders ?

Which rack and pinion setup did you buy ? At some point, you might need hardened rack teeth to hold up to that much power.

[harryn]

The steel decison depends on exactly what you are building. Maybe you can sketch something by hand and pencil on a graph paper and post some concepts for us to look at.

Don't worry about the diagram quality, we are informal and here for fun. I have no idea how to use CAD, I just pencil stuff out free hand, and it shows.