[MechanoMan]

I'd kinda assumed that going with a ballscrew on either side of the mill and stepping them together would be "the way to go" for performance in moving the gantry along the X-axis (the gantry's ballscrew being the Y).

The single screw underneath and in the middle looked problematic. When the Y axis is at the extremes, it's putting the cutting head far from the ballscrew axis, so there will be a torque on the guides with the ballscrew pushing the gantry assembly in the middle and the cutting force pushing back from the edge of the table. Seems like that force would result in some racking against the guides.

I was planning on a ~6ft x 5ft work area. Primarily to cut wood... although being able to cut aluminum would certainly be nice. I have a CNC Taig mill for that but it's limited to much smaller pieces.

And I started to wonder, since the ballscrews don't have a zero tolerance here. If the left and right sides are turned perfectly even at one end of travel with zero racking, and stepping them together, by the other end one ballscrew will have traveled more linear distance creating a small amount of racking- which could cause it to jam if the linear guides are tight.

But looking at builds I've seen so far, I didn't see where ANYONE went that direction. I guess it's not that necessary? What are the pros and cons here? This arrangement does cost more for ballscrews, motors, and drives.

[louieatienza]

I'd kinda assumed that going with a ballscrew on either side of the mill and stepping them together would be "the way to go" for performance in moving the gantry along the X-axis (the gantry's ballscrew being the Y).

The single screw underneath and in the middle looked problematic. When the Y axis is at the extremes, it's putting the cutting head far from the ballscrew axis, so there will be a torque on the guides with the ballscrew pushing the gantry assembly in the middle and the cutting force pushing back from the edge of the table. Seems like that force would result in some racking against the guides.

I was planning on a ~6ft x 5ft work area. Primarily to cut wood... although being able to cut aluminum would certainly be nice. I have a CNC Taig mill for that but it's limited to much smaller pieces.

And I started to wonder, since the ballscrews don't have a zero tolerance here. If the left and right sides are turned perfectly even at one end of travel with zero racking, and stepping them together, by the other end one ballscrew will have traveled more linear distance creating a small amount of racking- which could cause it to jam if the linear guides are tight.

But looking at builds I've seen so far, I didn't see where ANYONE went that direction. I guess it's not that necessary? What are the pros and cons here? This arrangement does cost more for ballscrews, motors, and drives.

Almost all of the commercial tables I've seen use only one leadscrew per axis. But they're also using expensive linear components and CNC'd parts.

There's quite a few bulds here (mine included) that use dual leadscrews on the X axis, some are belt or chain linked, and some use slaved steppers.

One advantage of dual leadscrews is that there are no mechanical components under the table. Another advantage, the leadscrews help keep the gantry from twisting which helps with home-made linear components. On the downside, it costs more. Belts can stretch. And you're paying more for components.

With the large table you mention, maybe a rack and piion or belt drive might be another consideration?

[MechanoMan]

Well, I'm good with the ballscrew. It's affordable, accurate, and simple. Assuming that linearmotionbearings2008 guy will be able to supply a longer-than-listed one, which is said he could. They say his ballscrews are great quality.

The dual axis would be the shorter one, since that would involve buying two ballscrews of that length.

Is there a guideline, some kind of rule-of-thumb, for how far apart the linear bearings riding on one rail should be in relation to the gantry width-of-span?

[johnmac]

If you are talking about true ball screws, and not just acme lead screws, price everything needed to use dual screws first.

When I built my 36"x43" router I found some real nice ball screws on ebay. That's why its 36"x43", not 36"x48" as planned. I like the results, but if I were to build a bigger machine I would use rack & pinion.

John

[MechanoMan]

Well, I've priced it already. An extra ballscrew assembly from linearmotionbearings2008 is an extra $300, if I can't get a price break on the shipping by combining. But I believe it'll allow me to reuse my G540 and only have to buy some cheaper NEMA23 motors as slave steppers, because the job is divided among the two of them.

I don't like the idea of the belt or chain connecting them. Belts and chains are somewhat prone to slack and involve a number of specialized pieces which will increase the cost and complicate the build.

I guess it depends. Does the second ballscrew actually result in a better machine, or simply cheaper? I'm not sure what to think, because it seems like "a lot" of commercial builds don't use anything but singles. But, then again, "a lot" of commercial builds are crap so that doesn't necessarily mean it's right.

And if it can resolve possible stiffness issues in keeping this wide gantry from racking, that could be worth it. Stiffening improvements to the frame could cost me more than the second ballscrew here.

Sorry, I know I'm being difficult, but I'm working hard to draw up a well-informed, cohesive construction plan before ordering parts.

[ger21]

If you can build your gantry rigid enough so it won't rack when cutting at the ends, then you need one screw. Two screws will allow a relatively "flexible" gantry to always stay in the correct position.

On my mdf/plywood machine, the gantry is only attached to the sides with a few screws in the center. With the screws removed, it'll easily "rack" by 3-4", with very little force. With 2 screws, it's rock solid.

[Tartan5]

You might want to take a look at this thread:

http://www.cnczone.com/forums/showthread.php?t=51485

[johnmac]

Don't forget how you will drive the second screw. Stepper and driver, more $$.

John

[Torchhead]

There are several advantages of the two motor/dual drive approach:

1. No under table components (as mentioned). Big advantage for larger tables and any plasma or flame cutting design.
2. Ability to fine tune each side separately so even minor variations can be accounted for
3. Double the torque
4. Ability the have separate home switches to force the gantry to square every time you do a REF.
5. Less mechanical complexity. Less mechanical loss.
6. Easier to do independent mounting and gear mesh control for R & P designs
7. Less rigid gantry construction and rail systems needed
8. Proven design. It is reliable and efficient.

We have four production tables. Three are dual drive and one is a router only single axis drive. The single drive table has a single screw down the middle drive and an Box frame gantry (Top and bottom cross structure). We are limited on speeds and depth of cut on the single drive axis and even with linear precision slides there is some racking when doing heavier cuts. The Dual drive tables have been running for years and are reliable.

TOM caudle
www.CandCNC.com

[hemsworthlad]

I use a twin screw setup with ballscrews from Chai at linearmotion.
Both screws are connected with a belt between and i dont have any trouble what so ever with the belt and i like the piece of mind it gives me that it cant rack.!!
It's absolutly rock solid with fantastic repeatabilty and super accurate, to be honest it's completely blown me away with how good and accurate this machine is considering it's diy built and not in my wildest dreams did i expect to cut to the tolarences that i do.
90% of the time it's cutting Aluminium and i even cut steel with it with out any trouble so long as the correct cutters for dry steel cutting are used.

Personaly i like the mechanical advantage and repeatabilty that screws give but with that said if i was just cutting wood 90% of the time then i'd probably go with rack to save money and up the speed.?

After building this machine NO WAY would i build a large foot print machine with a single screw and to me the only way to go is to drive both sides. . . dont matter how belt, rack, screws what ever just use 2.

[johnmac]

I agree with the design advantages of a dual drive for the gantry. But with a 6 ft. x 5 ft. table, one ball screw large enough to span 6 ft without sagging or whip is going to be heavy and expensive, let alone two.

John

[MechanoMan]

Please enlighten me as to these "sag" or "whip" issues. I've heard them mentioned; but I can't figure out if they're referring to something real or one of those things- don't take this wrong- that get misunderstood and passed around as dogma.

I was indeed concerned about the ballscrew performance in this length, and came up with two resources describing critical speed and how to calculate it:
http://medmaninc.com/rbs/ball-screws.htm
And this delightful Excel spreadsheet:
http://www.decsar.com/Desont%20Desig...-010-0510A.xls
Critical Speed is the speed at which the ballscrew will resonate, presenting a speed limit for a given type.

The two resources give different numbers. The first uses a widely quoted equation based on root diameter of the ballscrew, the Excel needs that AND weight/length. I believe the first equation makes assumptions about the weight based on a guess of how much steel is outside the root diameter, but that depends on ball diameter. So the Excel specifically asks for that and I suspect it's more accurate.

Chai was able to provide the root diameter and weight/length of his 2510 ballscrews (yay!) From that, when I went to 72" and two Fixed ends, the MedManInc equation got me a critical speed (not using Safety Factor) of 698 ipm, the Desont Excel gave 558ipm.

That is basically within the conceived 500ipm target I was designing for, although with a safety factor it won't quite hit it if I believe the Desont Excel over the MedManInc eq. That's not to say the machine MUST be able to do 500ipm, but rather anything more than 500ipm probably will not be used. This did show that the 5mm lead version is RIGHT OUT though. CS is actually an RPM, so the 5mm lead will resonate at half the linear speed.

Also I don't quite get whether one of his pillow blocks is a Fixed or Simple. It is not "free" to move per se but it's only got 2 screws holding it down. That... seems to invoke the Simple description and that factor would lower the speed by 34%, which IS problematic.

And I get a column strength of 6089lbs for the 72" shaft. That is, ">> more than I would need".

Is there anything technical about the"sag" or "whipping" factor that would suggest it's something OTHER than this well-defined ballscrew "Critical Speed"?

[MechanoMan]

I use a twin screw setup with ballscrews from Chai at linearmotion.
Both screws are connected with a belt between and i dont have any trouble what so ever with the belt and i like the piece of mind it gives me that it cant rack.!!
It's absolutly rock solid with fantastic repeatabilty and super accurate, to be honest it's completely blown me away with how good and accurate this machine is considering it's diy built and not in my wildest dreams did i expect to cut to the tolarences that i do.
90% of the time it's cutting Aluminium and i even cut steel with it with out any trouble so long as the correct cutters for dry steel cutting are used.

Personaly i like the mechanical advantage and repeatabilty that screws give but with that said if i was just cutting wood 90% of the time then i'd probably go with rack to save money and up the speed.?

After building this machine NO WAY would i build a large foot print machine with a single screw and to me the only way to go is to drive both sides. . . dont matter how belt, rack, screws what ever just use 2.

Well, what I saw here is that the machine won't rack due to cutting resistance not being inline with the underneath-middle ballscrew, but it MUST rack with ballscrew inaccuracy.

That is, we can start with absolute 0" racking with it parked on the stepper end, but no pair of ballscrews will have PERFECT matching. Chai's stuff is "precision rolled", but they're still rolled. And thermal expansion of the steel alone makes 3.2mils per degree of temp difference between the screws by the time it gets to the other end 5 ft away. That does concern me. Well, I strongly suspect that the flex of the 6 ft long gantry and tolerance of the guides will be such that a few mils of difference between Y1 and Y2 length won't create such a large force on the linear bearings that it'll want to jam from the racking.

But still, it's on my list of notes for concern.

[Tartan5]

One option is to used a fixed ballscrew with a driven nut. Then the whipping issues go away since the ballscrew is not turning......

[hemsworthlad]

And thermal expansion of the steel alone makes 3.2mils per degree of temp difference between the screws by the time it gets to the other end 5 ft away. That does concern me. .

I live in UK and the temps change wildly here, one day it can be red hot and next day it can be raining and freezing cold and i can tell you no way does thermal expansion make any differance and if so it will be in the 1000ths of mill's not full Mill's and it wont show or will be detected when cutting.

I think your looking far too deep into the theory.!!

My screws are 20mm and 1500mm x 5mm pitch i dont have any whip and my rapids are 7mt/min and maxout at 8.5 but i do have it geared 1/1.5.
Most of the time i'm running at 5mtr/min because this is a comfortable speed for the 5mm pitch.

I can run from end to end and if there is any racking due to screw differances then i cant detect it with the measuring equipment i have available.!!

Yes i've spent meny meny hours, infact far too meny hours tweaking, aligning, measuring then aligning & measureing again.
I,ve made various rig's to work as referance points that hold dial gauges etc so i can test for racking, repeatabilty etc and belive me at worst where talking 100ths of mill's and over 1500mm length for a DIY built machine using cheap screws and no special tools i dont think you could ask or expect more.? . . . . . I certainly dont.!!

I do feel this is the longest i would push 20mm screws and if starting over i would buy 25 or 32mm 10mm pitch screws and certainly would if going longer.
Or i would either fix the screws and turn the nut . . Or use R&P.

Another thing you must consider and the reason i run at 5mtr/min is that ballscrews esp the cheaper end IE: Chai's are not rated to go much over a 1000rpm and if constantly run over this speed will wear out in short order.

So even on a 10mm pitch then 500" rapids wont safely (for screws sake) be achievable.?

IMO i say . . . Dont get caught in the calculations and theory trap and just get on and start building..?? . . . . . Overs will disagree.?? . . But i bet you a dollar to the pound that 90% of them aint built a machine yet.:argue:

[johnmac]

The weight of a long lead screw can cause it to sag like power wires do between power poles. It may not be enough to be noticeable, but when the screw spins, it can try to flop like a jump rope. Thats the whip. It can result in bent screws.

A stationary screw with a spinning nut will solve the whip problem. A bit of sag will be a non-issue. But with a dual screw setup, building such a drive could get complex.

John

[gio666]

racking is 100x more to with wrong bearing , movement in the bearings , rails out of line from one end to other , not level with other one , one rail longer than other , wrong alignment at each end , than a ballscrew , if you have all designed right , one ballscrew is more than enough , and without crossmember going across the bottom , you have no box for strength , on router ,not plasma cutter

[louieatienza]

One option is to used a fixed ballscrew with a driven nut. Then the whipping issues go away since the ballscrew is not turning......

Tartan, I've heard of that concept before, but don't a lot of manufacturers recommend the ballnut be oriented a certain way?

MechanoMan, my math may be off, but if your motor is running at 500RPM and you have 5mm leadscrews then you're getting just under 100IPM.

I don't see how you would attempt to spin a 5ft ballscrew at 2500RPM and expect it not to whip. Unless you could get one with a higher lead....

[MechanoMan]

10mm lead... so the peak 500ipm would be 1270rpm.

[MechanoMan]

I live in UK and the temps change wildly here, one day it can be red hot and next day it can be raining and freezing cold and i can tell you no way does thermal expansion make any differance and if so it will be in the 1000ths of mill's not full Mill's and it wont show or will be detected when cutting.

Well, steel's thermal expansion is ~12e-6/deg C (stainless is notably more). At that rate, a 72" screw WILL grow by 0.48mil per deg F (not sure how I got 3.2mil before). Regardless of the screw diameter or quality.

Oh I don't think I'm mistaken in thinking I should understand, think, and rethink before pouring money into parts.