[elitechicken]

Hello Chaps,

I've lurked here a long time, and have tried my best to soak up the knowledge which is freely shared. I don't normally post on web forums but this forum comes across as very supportive and well meaning.

Since late last year, having read the excellent MadVac webpage and being thoroughly inspired, I have set out trying to acquire the skills needed to build a capable 8x4 machine for cutting MDF and ply sheet. I had a 6 month false start where I tried to do it on the cheap, using 10' scaffold rails as my linear rails and a frame made from scaffold, rolling on small bearings. I got as close as moving the various axes with Mach3 before realising that I would never be satisfied with such a machine and immediately went back to the drawing board.

Attached are some work-in-progress images of my design so far. The steel tube is 120x60x4mm on the long axis, 60x60x3 on the "end assemblies" and 80x40x3 on the gantry. The Zaxis is two sheets of 15mm aluminium. The X and Y linear bearings are 25mm profile type rails, and the Z axis uses unsupported 25mm linear rails. The CAD model uses parts modelled from the dimensions in the datasheets.

The motors used will Nema34 frame stepper motors: 12Nm on the long axis and 8Nm on the other two.

The ballscrews will be 25mm diameter 10mm lead on the two longer axes and 16mm diameter 5mm lead on the Z.

The questions I'd like to ask you are:

* Will my frame assembly be rigid enough using the flanged and bolted together system shown?

* Is having the cutting table and the long rails on different steel frames going to cause me problems with sag in the middle? What is the optimal frame arrangement? I see a lot of machines such as the shopbots using the rails on the same structure which supports the cutting table and wonder if this is perhaps better, or is this just a better arrangement for rank and pinion systems?

* Can the accuracy required for such a build be achieve using a very agricultural pillar drill, a cut-off band saw and a file? I've struggled thus far to drill holes to within 0.5mm of my intended mark and would really welcome some tips on "best practices" when it comes to drilling and marking.

* Do you think my long axis ballscrew is substantial enough at 25mm D?

* What is the best way of attaching the profile rail bearings to the gantry? Does some interface piece as shown have to be used to get around the fact they bolt in from the top, but are going on the underside of the gantry..

* On the Z axis, I'm using unsupported round rails mostly for cost reasons, but also because I read somewhere that supported round rails shouldn't be used with forces as they are on a Z axis. Will 25mm rails be meaty enough? I'm also concerned that the whole weight of the spindle and Zaxis will be supported by 8 M6 bolts going into tapped aluminium, if I use the SK style mountings as shown.

* In the MadVac machine, Vaclav goes to extraordinary lengths to ensure the flatness of his steel before mounting his profile rails to it. As I'll be using rolled steel, I was simply going to "fill the valley" with epoxy. What approach would you recommend?

I have a lot of other questions, but I think that is enough to be going on with for now! Please bear in mind that I don't have any great mechanical experience being quite young and an electronics graduate in a software job, but I am very eager to learn.

Thanks a lot
Joe

[RGeo]

My first thoughts are: looks like a fairly nice machine, why not use servos instead of steppers? Also, I would opt to hide the linear guides underneath the framework, this helps eliminate buildup. I am assuming this machine is a router? One other thing, you can use floating rails to support your ball-screws intermittently, along the way. It’s really nothing more than a bushing attached to a rail that your axis drags along. Actually, there may be a better solution, although I have no personal experience with them, I have seen articles about roller pinions, I’ve included a link. I’m sure there are some articles on this site that have discussed them as well. Just looks like something to look into. Long ball screws doing work over long open spans tend to deflect a bit.

http://news.thomasnet.com/fullstory/...rnative-460795

[ger21]

First, I think your in way over your head.

I think you'd be better off building a Mechmate, but I'll give some advice here.

For the Z axis, I'd recommend 15mmprofile rails. Much less weight, more compact, and far more rigid.

To answer your questions?

Rigid enough? Probably depends on your fabrication skills.

Will it sag? It'll definitely have some flex to it. I wouldn't want an unsupported table that long.

Accuracy of the build? Again, depends on your skills. Some really skilled can do much better work with bad tools than someone with no skill and good tools.

Ballscrew? Depends how fast you want to spin it, and how it's mounted.

Attaching blocks. You can get flanged blocks that will bolt from the bottom.

Fill the valley with epoxy? Your assuming they are straight. They probably will have some twist, and filling with epoxy is probably not as easy as you think.

[Benonymous]

It would be useful to know the type of work you intend to do with this machine. When we know that, we can frame our advice to fit.

Honestly, I'd suggest a smaller machine for your first time but that said, here's my advice.

I'd probably go along with the other responses. That length screw will whip after you get bored with running the machine at very low feed rates. The gantry looks to have a nice wide base so that might prevent racking but with a big whippy screw, that's still a possibility. As far as the flange/bolt construction goes, I'd say if you can jig and weld the flanges accurately it would be OK. I'm planning to use bolts and brackets for all my frame and gantry joints, no welding at all.

The main improvement I'd make would be a rack and pinion system for the X drive.

[elitechicken]

Thanks for your replies. I agree that I'm probably biting off more than I can chew, but I like a challenge.

I agree with the suggestions to use rack and pinion on the X axis. I've designed it so far to use a ballscrew because its simple, but theres no point going to this length to build a machine like this to then cripple it with a low max speed. However, it would appear than rack and pinion systems such as the ones from Marchant Dice will cost me a lot more than the ballscrew. Do you need to have a motor/rack/pinion on both sides of the gantry?

A supported ballscrew seems attractive as I can stay in the ballscrew "comfort zone". I may try to come up with some homebrew sprung system which retracts when the gantry comes past.

I'm using steppers because I have a couple of suitable ones already, and have some geckos knocking about, but should steppers be unsuitable, hopefully the NEMA34 mounting will let me use servos without too much work.

The machine is going to be used to cut stations for building boats, a bit of furniture out of MDF, and also for prototyping for ideas I get. Ideally I'd like to be able to machine aluminium too. My budget is about £4000-5000.

This evening I shall refactor the design to use the rack and pinion as listed here http://www.worldofcnc.com/main.asp?c...ack+%26+Pinion. I'll also try to support the table a bit better and I'll post the results later.

Do you think for the aluminium pieces in the Z axis it would be worthwhile outsourcing this to be made on CNC for me? Would it work out cheaper in the long run to invest in a manual mill and learn how to use it?

[rowbare]

You might want to look at this thread for inspiration: http://www.cnczone.com/forums/showthread.php?t=30751

bob

[RGeo]

When cutting aluminum you will have more tool pressure then if you are cutting say MDF, so given the size of your machine you will deflect too much to do machined aluminum parts any real justice.
/
As for dual rack and pinions, you can usually use one motor driving a shaft which is common to both rack and pinions. One pinion will need to be rotationally, adjustable for alignment purposes.
/
Concerning the location of your rails, I would bottom mount them. It simply helps reduce contamination.
/
Also, the firm and proper mounting of linear guides is paramount to proper operation and accurate finished parts. The structural members used for your frame will not be straight enough to use as a reference. Without the proper alignment equipment, you are at the mercy of the materials, unless you know some tricks. I used to build / reassemble large precision machines in the field and build new machines in the factory and I could likely come up with a modified procedure to help you come up with a straight machine. I would begin by mounting the rails to cold roll steel and bolting the cold roll to the structural member, add some threaded holes to put some set screws and you have a surface that you can straighten mechanically.
/
Additionally, I know that the high end controllers, i.e. Fanuc, Siemens, Mitsubishi, etc have dual axis compensation, however, I am not familiar enough with the lower end controls to say which ones do. Dual axis compensation allows the machine to actively offset the position of one axis, in response to the position of another axis. In other words, if my X,Y is skewed it can be straightened by parameter rather than in the physical world. It very helpful, but it’s not a magic bullet. There is no real substitute for a geometrically aligned machine. Anyway, consider this feature when purchasing a control or software for your PC control.

[elitechicken]

rowbare - thank you for that link. I've been trawling this forum for build threads such as that and didn't read that one but it is exceptionally informative and has forced me to rethink a little bit.

RGeo - I hadn't considered a method like your idea for mechanically straightening the long rolled steel box section but if i continue down this route with this machine, I will employ a system like that. I'd be very grateful for any further ideas you have! Also, what job did you have where you were building precision equipment in the field? That sounds quite an interesting career.

I think before going off full speed down the full size 8x4 machine, I'm going to make an MDF machine as tight as I can so that I can mill out some alu parts for this one. I've been considering how I can get a precise Z axis assembly and have come to the conclusion I would either have to outsource it or acquire some significant tooling to do the job.

[Al_The_Man]

Additionally, I know that the high end controllers, i.e. Fanuc, Siemens, Mitsubishi, etc have dual axis compensation, however, I am not familiar enough with the lower end controls to say which ones do.

With Mach/Galil control and servo's, the electronic gearing command can be used to synchronize two X axis servo motors, one as a master the other slaved off it.
Al.

[RGeo]

Al, what I'm talking about is more on the control software side; the dual axis compensation function / procedure creates a matrix map of the plane you are working with and offsets the axes based on this data. For example, even when you are simply making a move in X, all by itself, and you were watching you position page, Y would obviously stay stationary; however, with this function, if you looked at the servo page, Y would be making minute corrections, depending where X is, thereby correcting for deviations in geometry relative to XY.

[ger21]

Mach3 can do that, but the correction would be linear. For instance, Y can increase .01mm for each 100mm that X travels.

It can also square a gantry by homing each side independently to separate switches.

[RGeo]

ger21, that's pretty much what I'm talking about, although Mazak, Mitsubishi, Funuc, and I believe Siemens all have something that allows you to correct in + and - directions. There is basically a password hidden area that has a spread sheet that you fill in, if you want to do just a linear adjustment, like you are talking about, you need only three rows of data, consisting of an X position and a Y compensation which is either + or -. For correcting a not so linear, linear axis, you would compensate every so often, most people use the ball screw thread lead, so maybe every 10mm. You can do all of the axes against each other and have the machine read as perfect as you want. However, I am a strong believer in getting the actual geometry correct first, then using this function just to tweak a few tenths, but you can much further than that if you want to cheat. The problem I have with relying on this function is issues with tool angle. So I see no serious negatives on machines with really short Z-axes, such as plasma, laser, punch press, and even some routers, but bad geometries on something with substantial travels in all axes results in a volumetrically bad machine, even with this type of advanced compensation. I guess my point being, I would want my control to be capable of this, but I’m still going to build and adjust for proper alignment for days, if necessary, before I resort to using the function.

[RGeo]

elitechicken,
I was thinking about the rack and pinion you chose and it makes me wonder about the accuracy needed in an application like routering wood. I would like you to consider this point: Although many controllers have backlash compensation, AKA lost motion compensation, there is a limit to the effectiveness of such a program level function.
/
First, I didn’t look up the specs on your rack and pinion, but it’s likely to have 0.7mm (.028”) in backlash at the engineered mesh. It’s conceivable that the backlash parameter could compensate for 0.7mm of back lash. So, if you made a program and your machine employed backlash compensation and you where drilling some holes; you move XY, then Z, move XY, then Z, etc. I’m sure that everything would work well, maybe even flawlessly.
/
Now consider something like cutting a planar circle. As far as your machine will be concerned, a circle has four quadrants, let’s call them (+X,-Y),(-X,-Y),(-X,+Y),(+X,+Y). So, a circle has two changes in X direction, and two changes in Y direction per full rotation, and the direction changes will occur in the quadrants of the circle. Imagine this being done with no compensation, if you conclude correctly, you will see that there will be outward spikes in the quadrants of the circle, and they will be about 0.7mm in length. This would likely be a clearly visible anomaly.
/
Now let’s consider using backlash compensation to reduce these spikes. Obviously the quadrants are where the backlash compensation will do its work, this is true because this is where the motors change direction. Also, realize that there is no such thing as zero time, so it will take a finite period of time for the each axes to correct for the assumed 0.7mm in backlash. So, it is difficult for the spike to be removed completely, although, the slower the programmed travel speed, the more relative time the compensation has to complete its 0.7mm move.
/
Now let’s take this out of perfect world conditions and add tool pressure and clime vs. conventional milling directions. In the real world the tool will apply force to your machine, and this force, will at some point, be enough to push or pull your axes in a way that backlash compensation cannot anticipate. This will create some unforeseen spikes or divots your machined profile. This too can sometimes be reduced, although not fully eliminated by slowing down the feed rate, especially during finish cuts. This is a long way of saying; with physical backlash present, undersigned features will appear in your finished parts, especially at higher travel speeds. All that being said, I don’t know if a 0.7mm spike or divot will matter in a cabinet door or the neck of a guitar, i.e. wood router type work. However, I wanted you to be aware of how backlash affects machine performance and profile geometry.
/
Here are some alternatives to accepting default backlash from a standard rack and pinion:
1) Spring load the two together. Pro; reduces backlash, con; it increases wear.
2) Use precision herringbone rack and pinion: Pro; reduces backlash, con; cost.
3) Laminate pinion with steel and slightly oversized polyurethane: pro; reduces backlash, con; results vary under high load conditions.
4) Use a roller pinion rack: Pro; reduces backlash, con; expensive (I don’t know enough about them to say if there are more cons).
5) Use ballscrew: Pro; low backlash, con; expensive, deflects over long distances.
/
I would try #4: Laminate pinion with steel and slightly oversized polyurethane layers. I have seen this done in a few applications and it seems to work well, it certainly dampens and therefore retards the appearance of spikes and divots in the quadrants of circles arcs and reduces geometric anomalies in sharp corners. I’m not sure who sells such a system, but I know someone does because I’ve seen them in use. I believe the equipment was German.
/
Anyway, excuse my long dissertation on the subject, but it sounds like you are willing to invest no small amount of time and money to build this machine, and I wanted to give you some specific design challenges to consider.

[RGeo]

Typo: undersigned = undesired

[ger21]

Here are some alternatives to accepting default backlash from a standard rack and pinion:
1) Spring load the two together. Pro; reduces backlash, con; it increases wear.

Virtually all homebuilt machines you see here, as well as ShopBot and other lower cost commercial machines use this method. Backlash can be nearly eliminated. Wear really shouldn't be an issue unless it'll be seeing heavy use daily. If that's the case, replacing the rack and pinion is a fairly inexpensive proposition when wear occurs.

[juba]

I also am in a similar position to you, with similar machine capability aspirations.

i have little experience with actual builds but am an engineer.

I think your design is sound, but obviously building accurately is the big challenge. IE linear rails mounting. I am thinking of aluminium members with capped ends doweled into an lasered/machined end plate. Aluminium should be more consistant and

I like the single ball screw despite others comments here proposing rack and pinion. speed traversing to me is not so important and a small sacrifice as cutting time is most of it! I operate a laser cutter and traversing speed is not so important.

Your machine resembles a techno LC machine I wonder why more do not talk about doing a similar design on here as seems they have a simple but robust machine (I have some experience with them)

anyway not a very helpful reply but a vote of confidence.

[elitechicken]

Thank you all for your replies. I must confess that when designing so far, I have only considered "Can I make this design accurately enough" and not "will it be slow, flexible, unfit for purpose". Having seen read your replies and reading other build posts, I can clearly see that with my limited (but growing!) metal working abillity and access to engineering tools, there is no way I can achieve the design I have there with the amount of time i get to apply to it (I only have about 5 hours per weekend access to a workshop).

I'm paticularly grateful to RGeo for his comprehensive detailing of backlash - I think I definitely have a lot more things to worry about that I thought I had!

I had been thinking about backlash compensation methods for the R+P and I couldn't visualise a homebrew sprung system but I shall do some research, as overwhelmingly seem to favour R+P for this size of machine (though thank you jeba for your vote of confidence!)..

I'm going to make an "intermediate" machine so that I can machine the aluminium parts I need for the big one, and also help make some jigs accurately. I'm itching to get making something as CNC has been all I've thought about for months. I'll knock out some more CAD and see what people think of that one!

[ahren]

I might humbly suggest the CNC Router Parts LLC Rack and Pinion drive system -- it has basically been adopted by all Joe's builders who are building 4' x 8' machines (and some building 4' x 4's just for the added speed), as well as several OEM plasma cutter vendors. We've sold over 300 of them since they were released in October of 2009. There is both a Nema 23 and a Nema 34 version available.

We should have some metric versions available soon that run on module 1.5 rack, which is more readily available in the EU than the 20 pitch commonly used in the US.

http://www.cncrouterparts.com/produc...products_id=50

Ahren
www.cncrouterparts.com