MechMate CNC Router Forum

MechMate CNC Router Forum (http://www.mechmate.com/forums/index.php)
-   Driving Mechanisms: Rack/pinion, gears, screws, belts & chains (http://www.mechmate.com/forums/forumdisplay.php?f=24)
-   -   Accuracy in racks/Mechmate (http://www.mechmate.com/forums/showthread.php?t=3854)

davtuner Sun 22 July 2012 10:23

Accuracy in racks/Mechmate
 
Hey everyone,

I finished building a Joes Hybrid 4x4 about a year ago, and decided to go with Moore Racks. I noticed the precision (repeatability) of the machine cuts were around 0.001", but the accuracy was terrible, around 0.001" error per 0.1" of travel. Granted, this error was non cumulative so larger parts turned out alright, but I'm unable to get interference fits for parts (need 0.001" accuracy). Using a dial indicator, accurate limit switches, and accounting for backlash (very small amount), I traced the errors to the racks. I was able to differentiate this error from the gears because I was able to find series of repeatable errors independent of starting gear position, and the errors in position were repeatable on the rack. Yes this means I could fix it using screw mapping, but its not time/cost effective.

To defend the electronics, rest of the drivetrain, etc. the Z axis, which uses ACME lead screw, is highly accurate, and has an accuracy of around 0.001". This is including using microstepping on the Z axis, which can obviously produce positional errors over small movements.

The guys over at the Joe's forums suggested I might be a leadscrew guy (probably am), but I figured I'd try buying a new rack for a single axis first before retrofitting the whole machine for leadscrews. I noticed some manufacturers like Atlanta gears and http://www.schlenkent.com actually quote accuracies, similar to McMaster does with leadscrews. Moore gears does not, for obvious reasons. FYI, i'm using the cncrouterparts rack and pinion transmission, which are 20 degree pressure angle, 20 pitch, 0.5" width gears.

My question is, are any of you able to get 0.001" accuracy (not precision) with a rack? If you've used a dial indicator to verify, thats a big plus. I'd really appreciate any leads on sources/advice.

Thanks,
David

Gerald D Sun 22 July 2012 11:36

I don't think that anyone here is using a micrometer for checking their cuts.

davtuner Sun 22 July 2012 11:43

Gerald,

I guess the root problem is I'm not able to get interference fits between parts, and when doing double sided cuts like PCBs the through holes aren't lining up. Also, since I'm usually cutting in wood products, I'm not micrometering the parts, but I used a dial indicator on the machine to flush out the root of the problem: inaccuracies on the rack. Any suggestions on how to get interference fits, or where to go where people ARE talking about levels of accuracy like this would be appreciated; I know it's possible on the leadscrew version of the 4x4 Joes Hybrid, at the cost of speed.

Also, any advice on whether or not I can even get this level of precision with a rack and pinion set up would be appreciated.

Thanks,
David

Gerald D Sun 22 July 2012 12:04

While a good leadscrew may give you the single thou's, the flex (plus other factors) in the rest of the machine will mess with those thou's. I don't think it is worth considering leadscrews in machines with a relatively flimsy structure, like the MechMate and the Joe's (relative to the "big-iron" machines).

davtuner Sun 22 July 2012 12:13

I figured I could compensate this by using shallower passes, and I actually get consistent results. They're just consistently wrong, the exact same dimensions of wrong. Do you have any suggestions on how to get interference fits or accurate throughhole passes on PCBs?

I can't imagine doing a surface etching or doing light engravings is going to put substantial flex on the machine. Especially when my precision is consistently spot on, but my accuracy is just poor. I know other things like temperature control of the room, vibration, etc. could be effecting it, but I don't get chatter and the room is under AC.

What other factors would be affecting it? I work as a QC/QA manager at a specialty gas manufacturer, details tend to get under my skin. :D

KenC Sun 22 July 2012 20:47

There really isn't much except to make a good jig to line up your job accurately. Any error here is human error in my book. nothing to do with the machine.
Also, if it is a machine issue, it is a glaring one, i.e. your X & Y axis isn't square to each other.

Did you factor in fits & limits in your actual physical cut dimension when you want interference fit?
Just in case you don't understand what I'm trying to say. Engineering Tolerance

*PS, armed with the right limits & fits knowledge, I get whatever fit when it is called for, any time of the day.
*PPS, Dump your Joe's hybrid & build a Mechmate.

Gerald D Mon 23 July 2012 00:34

If you are sure that the racks are the error, get better racks, or select racks by checking how well a pair of them mate together at different positions. (My experience of sticking Italian racks into each other is that they are very good).

Have you checked if your pinion gears have the hole on center? An eccentric gear will give you the type of thing you are describing.

JasonC Thu 26 July 2012 08:16

What would you need .001 of accuarcy for?


Jason

davtuner Thu 26 July 2012 18:05

Also, Diane Spackman, the nice lady at Moore Gears, recently said the racks have a total composite error of 0.002". I'm not sure what this means in respect to a rack (but do in respect to a gear). The rack is a 20 DP 20 degree pressure angle rack, so 6.36 teeth per inch. Based on what I found online, does this mean she just said there's a 0.002" variance between teeth?? I feel like a fool for all the work I've done...

JasonC:

I wanted to cut an energy chain (similar to an igus chain for carrying cables) for the machine, and to have the pivoting joints ride smoothly and the top and bottom plates to interference fit the two sides together, for easy assembly/reassembly.


Ken:

I remember reading that accuracy of 0.001" was required to get an interference fit. I suppose this means around +-0.0005". I'm familiar with tolerancing in product control, but not specifically fits and limits. Is there another way to get consistent interference fits? I've been using aspen plywood, and a downspiral carbide bit so the faces that mesh have smooth cut surfaces. Others seem to have been able to do it, so its discouraging after all this work. After about 10 different cuts, making an energy cable chain, I was able to get most of the links to fit. Some of this was also trying looser pieces with tighter pieces, which corrected (and proved) locational variance of the axes.

I don't think its an issue with the table being square. I assumed the piece of MDF I got from Lowes for the table base was cut square with the huge bandsaw that they have. I validated this assumption by looking at two cuts from two different stores, and compared their edges when sitting flat against the floor. I then flipped one of them 180 degrees and compared their edges. There was no parallellogram so to speak, so I assumed across the 54" they were square. I then mounted a dial indicator on the spindle, put the bed on the table, and ran the dial indicator along the x axis touching the parallel face of the board (w.r.t. Z axis slide), and made it so that the dial indicator (0.001" accuracy) showed no movement (other than small random surface roughness) across the 4' of travel. I then traveled along the y axis (two motor axis) and measured the change of the dial indicator. I made small steps with a single y axis joint in EMC2 to get the gantry square to the x axis, at both ends of y axis travel (to make sure the two y axis rails weren't bowed w.r.t. each other).


I performed a similar operation to verify the table was not racked, by travelling to each corner of the spoilboard and having the dial indicator ride on the surface. It did not vary more than the surface roughness error (<0.01" across 4' x and y directions).


I squared the z axis then to the table using a similar method.


Gerald:

I didn't check the eccentricity of the actual pinion gears. I assumed measuring each axis's movement independently with the dial indicator over a 0.1" distance (to avoid microstepping error), in the same direction (to correct for any backlash error) and across lengths longer than the circumference of the gear would have shown a repitition in the errors, which would have proven the error was with the gear. Instead, it showed a repeatable position error w.r.t. the individual axes, and I assumed this was what was causing the seemingly random variance in the cuts. The spindle has a static runout of <0.0005" (measured with a +-0.00005" dial indicator).

Now that you mention it the racks don't seem to fully mesh together, I have a small scrap section, there seems to be small spaces left randomly to the left or right of each tooth. I'll measure the eccentricity of the gear relative to its mounting plate when I get some time tonight or tomorrow, thanks for the suggestion, this has all been very frustrating given the amount of effort I've put into this.

JasonC Fri 27 July 2012 08:13

I really think, myself, its over kill. You will fight forever trying to achieve that amount of accuracy. Buy the chain...lol.

KenC Fri 27 July 2012 08:37

devtuner,
Your problem is not your machine accuracy. Its your limits & fits of your components.
You are almost there with limits ^ fits, but 0.001" isn't an universal cure, it can be as much as 0.1" if you are dealling with different size & machining accuracy. Also, it is not always +/- a number, in limits & fits, it ias alway 2 numbers with +/+, or -/- or 0/+, or -/0.... in front of it.
Why don't you post your cad drawing here & let see if anyone here can help you with out.
Machining tight fitting stuff is fun, & educational. I get full satisfaction everytime when things went spot-on.

ss82 Thu 29 November 2012 16:17

Hi, first time poster, so excuse me if I breach the forum norms by bringing life to an old thread:)

I appreciate the design of the Mechmate and I have almost committed myself to building one. However, I have some questions regarding the above mentioned topic of accuracy and precision. I do not have any specific target for accuracy because I simply do not know what it is demanded. Let me elaborate:

http://www.renovobikes.com/ These bikes frames are first machined in two halves and then glued together. Then the additional seat and chain stays are glued to the main frame. For bikes like the Renovo the tolerances will of course be paramount to structural integrity. So the basic thought is; higher accuracy is equal to better fit and the stronger the bike will be. But at one point one might start to chase diminishing returns with regards to the accuracy. After all wood expands and contracts with the surrounding air moisture content, which might skew or destroy a perfect fit anyway.

So my question is will a proper Mechmate be sufficient to make structural designs, exemplified by these bikes? Revono has an industrial CNC machine http://www.renovobikes.com/shop-gall...allery/1307237 and claims tolerances of .005”/.13mm http://www.renovobikes.com/making-a-frame/ Is this really necessary?

Is Revono heavy over-equipped or could I forget machining these kind of designs on a Mechmate?

Thanks

Gerald D Thu 29 November 2012 23:24

Not entirely fair to compare a 9 ton Italian to a half ton South African! :D

But, I don't think you will have a problem with small-scale production of this type. Even the Italian needs sanding afterwards....you will probably have to spend double the sanding time, I don't think much worse than that.

However, you will have to pay special attention to rail straightness and gantry squareness when you want two mirrored halfs to fit exactly together. Even without super-special attention, you should get to the order of 0.5mm on that size of product and you can look at that on a ruler and decide if you can sand that way.

Regnar Fri 30 November 2012 01:57

Quote:
But it takes meticulous and painstaking hand craftsmanship to turn those precision parts into the beautiful bicycle you’ll love like no other.
And now we decipher the code.

No matter how much programming, setup, machine calibrating we cannot for the life of us get the 2 parts to line up after machining.


All times are GMT -6. The time now is 17:20.

Powered by vBulletin® Version 3.8.3
Copyright ©2000 - 2024, Jelsoft Enterprises Ltd.