# Thread: Fixed Gantry - Trying again

1. Louie - yes you are right.

Chris - wow - that is very impressive software you have. I will get the files from the bosch rexnoth site and up load them. (edit - thanks, but no longer needed )

Edit - well it turns out to be easier than I thought. I assumed that the information would be in the data sheet, and it wasn't. Instead, Bosch Rexnorth has a very convenient deflection calculator that runs as a spreadsheet macro, and includes all of the info needed. Now I just need to figure out how to turn back on the macros in libreoffice and play.

For the curious, here is the link to the deflection calculator.

Deflection Calculator - Deflection Calculator - actuators, ball, ballscrews, bearings, bosch, Bosch, guides, leadscrews, linear, motion, rexroth, screws, shafting, star

Thanks

Harry

2. Since profiled extrusion is quite expensive, I decided to do some calculations to see if I could use smaller extrusion sizes and obtain the same deflection as rectangular tube.

Since the primary impact on deflection for a given material is the second moment of inertia, I am posting a few values here for easy reference. Feel free to double check my math.

For a hollow rectangular tube, the second moment of inertia is :

= BH3 / 12 (of the outside dimensions) - BH3 / 12 of the inside dimensions

From wikipedia.org http://en.wikipedia.org/wiki/Second_moment_of_area

So for a tube 4 x 4 x 1/4 inch

= [ (4 x 4 x 4 x 4) / 12 ] - [ (3.5 x 3.5 x 3.5 x 3.5 ) / 12 ]

= 21 - 12.5

= 8.5 in to the 4th

For a tube 3.5 x 3.5 x 1/4 inch thick

= [ (3.5 x 3.5 x 3.5 x 3.5 ) / 12 ] - [ (3 x 3 x 3 x 3) / 12 ]

= 12.5 - 6.75

= 5.75 in 4th

By comparison, using the Bosch Rexroth deflection calculator for a

90 x 90 H section ( roughly a 3.5 x 3.5 inch beam)

90 x 90 H = 7.2 in 4th

90 x 90 = 5.1

60 x 60 = .78

60 x 60 H = 1.25

80 x 80 = 3.2

45 x 90 = 0.56 and 2 (depending on direction)

45 x 90 H = 0.8 and 3 (depending on direction)

90 x 180 = 9.6 and 33 (dending on direction)

90 x 180 H = 13 and 51.3 (depending on direction)

This means it is possible to replace a larger tube section with a smaller extrusions and still obtain similar deflections. It also shows the value of the H type extrusions vs the standard.

edit - even better, here is the data sheet with the info on a pdf.

http://www13.boschrexroth-us.com/cat...s/section2.pdf

3. I did some more research on rack and pinion setups, and the related methods of gearing down the motors.

I am convinced more and more that pretty much any of the methods of belt, rack, and screw can work, and for a given precision, the price is surprisingly close. Nonetheless, it is interesting just how pricey things can get in the new market vs. the used.

I have some interest, although not mandatory, to be able to make a programmed step of roughly 0.001 inch, and for it to actually happen on the machine. This turns out to be much harder than I expected. In pursuit of this, I looked at both servo drives and stepper based drive systems, and the torque curves, at least to the level of my capability.

This lead me toward worm gear reduction units which can easily provide 10:1 to 40:1 ratios. At least the AtlantaDrives (economy) versions, which are really quite nice, are around \$ 1200 / each, and I need 5 of them - so much for that idea. In addition, even these remarkably well made setups will give about 0.001 inch of backlash at spec. ( I suspect that they are conservative, being a German outfit, but that is their spec )

They offer helical rack as well and when using spring split pinions (spring loaded on to themselves, not to the rack) the backlash to the rack can be less than 0.001 inch using module 2 rack.

The other method they recommend is to use 20 degree PA with hardened and ground rack and pinions. These precision ground setups will also give less than 0.001 in backlash, and similar precision per foot. Definitely in the ball screw range.

Prices start to get in the ball screw range as well, pushing toward \$ 1,000 for a 2 meter setup. They also offer (strongly) a felt gear setup which constantly cleans and lubricates the pinion to extend lifetime.

The good news, is that they are one of the few companies I have seen that actually is willing to spec the performance of the rack and pinion setup, vs the more commonly used rack I see which has - no specs at all.

They don't have a lot of good things to say about the use of spring loading the pinion into the rack, pretty much pointing out that this more or less eliminates any ability to specify what is going on at the interface.

For industrial use, I would certainly consider their products. For a DIY project like mine, \$ 1 - 2 K / drive (plus motors and drivers ) x 5 driven ends = budget breaker.

I guess I need to do some scavenging.

4. There are many things in a drive system that can (or will) have more than +/-0.001" accuracy over the intended operating range. They all tend to add up algebraically. Trying to arrive at +/-0.001" is very difficult to do, and to maintain it at that level during it's useful lifetime would drive anyone mad. Software error compensation helps, and the steps per inch setting number can be tweaked to center the remaining error around the target reference distance.

Most DIY machines realistically don't do that without help from the software on a reasonable budget, even when built by a professional machinist. The possibility of having a high end CNC milling center in a home shop or basement would certainly be nice to have. (I still haven't won the state lottery.)

You need to decide first on what materials you will cut and whether there is a real world need for 0.001" accuracy. Cutting wood based materials will never achieve high accuracy anyway. Don't let this prevent you from building a machine of some sort and learning to use it. I guarantee that your first machine is just a stepping stone to something better, or bigger, or just different in some way. It never ends until you lose interest in it.

5. Thanks C1 - certainly you are right. It is easy to get caught up in an attempt at over precision.

There are a couple of situations of importance to me, perhaps as much home political as actually important, if you know what I mean.

It is not so critical about the perfection of +/- 0.001 inch accuracy of a part measurement, but rather that a step movement is actually made when a step is commanded. I thought this would be easy to achieve, but based on what I am reading about stepper and servo motors, and what people are doing to achieve tight results, it is far from straightforward.

Example 1

I don't expect to be really milling Al, but I do want to be able to very accurately drill hole locations in it. For this, the plan is to mount a drill chuck / drill motor in replacement for the router so it goes at the right speed.

In many of the things I would like to do, the location of the first hole can be slightly off, but the pitch of the hole drilling needs to be pretty spot on.

Example 2

Photo carving and pcb carving. I expect that these will actually be the highest types of work that this hobby unit will do. For a larger photo carvings, viewed up close, very small cut imperfections will drive some people crazy. Wiggly lines and chatter will really show up, and that would mean the project dies vs. get an upgrade.

This is probably why some machines are so overbuilt, or never get off the ground. I need to physically visit some people that have working machines.

1) Fine picture carving, similar to your Mayan calendar work.

6. To add to CarveOne's comment, 0.001" accuracy over a distance of 2 meters is really pointless due to thermal expansion. If the entire machine was made of steel, only 1 degree C or 2 degrees F will change the dimensions by 0.001" (approximately). For an aluminum machine this is even twice that much and once you take large machined parts out of the shop it will change similarly due to temperature and humidity (for wood).

If you have a smaller machine in a temperature controlled environment, machine smaller parts and use ground ballscrews you can actually achieve the 0.001" at a reasonable price. I paid about \$450 for a set of 4 precision new (i.e. surplus) 15x15 mm Kuroda screws in 1.2 m length.

7. Hi Jerry, thank you for the input and advice, it is really appreciated. As you know, sometimes it helps to get re grounded on project goals to avoid too much spec creep from coming in.

I am actually having trouble staying within 0.100 inch over a 6 ft distance at this point if I use the published specs of parts and add them up. This goes away if I spend \$ 2000 / axis for all needed motion parts, but at the \$ 300 - 400 / axis, it hasn't so far. This does not include the linear rail pricing.

Perhaps worse, I had assumed that I could program in a 1/10th microstep into a 200 step stepper motor and it would move the router bit with at least 20% of the motor force rating, even if the move is imperfect. Now I am not even sure that this is always true.

You mentioned the price for the screws, but that is really just the start in many cases, as you then need bearings, mounts, rotating nuts, brackets, pulleys, etc. I am looking at NOS screws as well, but the more I look at them, the more I realize that mounting them correctly requires some real knowledge and planning as well. I understand the basic concepts of how they work, but the bearing mount details are non-trivial and add up fast.

8. Originally Posted by harryn
Perhaps worse, I had assumed that I could program in a 1/10th microstep into a 200 step stepper motor and it would move the router bit with at least 20% of the motor force rating, even if the move is imperfect. Now I am not even sure that this is always true.
That is actually not so bad if the stepper runs slowly (i.e. before going into into square-wave full-step-mode at higher speeds]. I can jog single microsteps back and forth (0.0003" on my machine) and can see the motion results are reasonably linear on a precision dial indicator. However, it takes only a pound or two of force on the bit to achieve the same movement although the machine is quite rigid.

Originally Posted by harryn
You mentioned the price for the screws, but that is really just the start in many cases, as you then need bearings, mounts, rotating nuts, brackets, pulleys, etc. I am looking at NOS screws as well, but the more I look at them, the more I realize that mounting them correctly requires some real knowledge and planning as well. I understand the basic concepts of how they work, but the bearing mount details are non-trivial and add up fast.
I had the same problem and ended up with a redneck engineering solution that actually works very well and needed little material (maybe \$20 per screw). The description is somewhere buried in my build log. It is easy to do if you have access to a lathe. But once again, if you want 2 meter length the screws are probably not going to work if not very thick.

You may want to consider if the larger parts are the focus of your interest or the PCB and like. You are really talking about 2 different machines.

9. I don't expect to be really milling Al, but I do want to be able to very accurately drill hole locations in it. For this, the plan is to mount a drill chuck / drill motor in replacement for the router so it goes at the right speed.

In many of the things I would like to do, the location of the first hole can be slightly off, but the pitch of the hole drilling needs to be pretty spot on.
I'm guessing the runout on a drill chuck is at least .002-.003", so throw your .001" out the window right there.

I am actually having trouble staying within 0.100 inch over a 6 ft distance at this point if I use the published specs of parts and add them up.
This to me is also a non issue. Move your machine 6ft. Measure. Adjust steps/inch as needed until it moves exactly the amount it's supposed to. Provided you can measure the 6ft distance to within the .001" you're looking for.

For a larger photo carvings, viewed up close, very small cut imperfections will drive some people crazy. Wiggly lines and chatter will really show up,
It's not that difficult to build a machine that can do fine carving without any visible lines. All that really matters is that the machine is rigid. I routinely do carvings with a stepover of .002", and can see the motors turn the same amount with each step. Mu machine has a full step resolution of .00125", but is used with 1/8 stepping.

10. Originally Posted by harryn
I need to physically visit some people that have working machines.
I think you're suffering big time for not having some hands on experience with a cnc machine. You'll be amazed at what's being done in practice without the agony of over thinking everything. Don't forget that 0.001" is approximately one-quarter the thickness of a human hair. "Get thee to a machine!"

11. Originally Posted by ger21
It's not that difficult to build a machine that can do fine carving without any visible lines. All that really matters is that the machine is rigid. I routinely do carvings with a stepover of .002", and can see the motors turn the same amount with each step. Mu machine has a full step resolution of .00125", but is used with 1/8 stepping.
Gerry, thanks for the reply. You have to admit that it is kind of funny for you to tell me not to be so worried about resolution, and then yours is even 10x better.

OCNC, you are right, I need to find someone locally with a cncrouter and buy them some pizza and accessories.

I found an interesting, nearly similar concept on cnczone to what I am thinking about for the motion. Also interestingly, it has the exact problem I was wondering about. Please consider to go to the nearly last post in this thread with a picture of his results:

The system uses a nema 34 motor with module 1.5 rack and pinion. The pinion is directly mounted onto the stepper motor shaft, no reduction. It started out that it did not have enough torque, but this was fixed by increasing the power supply voltage.

Now that this was fixed, he made some diagonal cuts and posted a picture. You can see that rather than a smooth diagonal line, it is made up from small line segments. This is exactly what I was worried about.

http://www.cnczone.com/forums/linear..._no_works.html

The gear reduction for nema 34 motors sold by Ahren is only 2x reduction, so I don't think that this is going to solve this problem, nor would I expect 3x to do so either.

What is not clear, is if the OP was using micro stepping or not, but I think so since he talks about setting his controller to 3200.

This is the kind of thing I am spending time calculating and wondering.

Thanks

Harry

12. Gerry, thanks for the reply. You have to admit that it is kind of funny for you to tell me not to be so worried about resolution, and then yours is even 10x better
Maybe I misunderstood what you were saying?
I was trying to tell you that you should have much better than .01" resolution. Unless that is what your design calls for.
I mentioned before that direct driving the pinion was a bad idea.
I believe you can buy 3:1 nema 34 reduction units from someone at the Mechmate forum.

Page 5 of 6 First ... 23456 Last

05-01-2012, 06:03 PM
2. ###### Thread portable and lightweight fixed-gantry router - water filled tubes? | CNCzone.com-The Ultimate Machinist Community | BoardReader
04-20-2012, 11:44 PM