Looking good man! How's that table working out? I think that's a great idea. I do OK with the t-slots but it seems easier just to have a threaded hold ready to go. Less messing around.
-jd
The switches look GREAT!! Very good job!!
Looking good man! How's that table working out? I think that's a great idea. I do OK with the t-slots but it seems easier just to have a threaded hold ready to go. Less messing around.
-jd
John Delaney
www.rwicooking.com
The table is working out better than I had hoped. I just finished a bunch of parts made out of 1/8” aluminum sheet metal that I had to clamp against a piece of wood to cut. The larger width and all the extra threaded holes made it very easy to set up.
Vince
Any news on the backlash front? Great project, thanks for documenting it.
Also, where did you take your CNC course?
Best,
BW
No news on the backlash. No one seems to be able to pin down its source. 4 sets of bearings later I still have the same problem. For now I am making parts and learning to live with it.
I took my CNC class at DeAnza college in Cupertino, CA. They have a great CNC program.
Vince
Backlash = a the clearance/difference between motion input value asked for as opposed to the actual motion achieved from a given stop point, or when back and forth motion is asked for but not experienced.
The elimination of backlash is not a "magic bullet" type of fix. Eliminating it involves a systematic and methodical elimination of clearances/slop in the drive system achieved thru lots of perseverence and work.
For the moment, I"m going to assume that the system is ball screw drive, via servo motors by way of a 2:1 cogged belt drive as this is a typical CNC drive scheme.
First: if you are working with ground screws, first you have to check for wear, especially uneven wear. If the screws are NOT worn or NOT worn irregularly, it may be possible to reload the ball nut with oversized balls.
Don't go crazy - go oversize by 0.0001" dia or so and do it is stages. As you add larger balls, the screw gets more of a "grinding" feel to it as you rotate. You should feel more drag but you should NOT feel tightness or binding.
If you can picture a bowling ball rolling down a gutter, you can easily see why a larger ball would allow less axial play in a ball screw.
Second: gibbs. This adjustment takes a lof of finesse and patience. If you simply reef down on the adjustment, you'll tighten up the system BUT it will be hard to move - sticking becomes a real problem, especially at direction changes.
You have to come to a compromise adjustment that removes slop yet still does NOT provide appreciably drag increase.
Assuming that you do NOT have uneven gibb wear, plan on going through several iterations of adjustment. Do it with dial indicators to see that the table moves in proportion ot input at direction changes - you can't rely on 'feel'.
Plan on spending lots of time as this adjustment involves lots of compromises and measurements and a dash of 'feel'.
Ball screw bearings: once you get the slop out of the ball nut and the gibbs, it is now time to address the ball screws. Here's where some real science comes into play.
Ball screw support bearings rely on bearing "stiffness" to resist axial deflection. A preloaded 15 deg contact angle bearing is stiffer axially than a preloaded conventional deep groove bearing (the OEM BPT bearings were preloades ABEC 1 6204's in a typical mill).
It follows that a 25 deg bearings will be stiffer than a 15, a 30 stiffer than a 25 and a 60 deg ball screw bearing stiffer than any other.
ALL bearings have a load/deflection curve. That is, the more you load them axially (as a ball screw would) the LESS they continue to deflect. This should be intuitively obvious as you try to shove the ball farther and farther up the side of the raceway, the harder it becomes to do so because you get more and more solid.
You can take a OEM style 6204 psuedo ball screw bearing pair and shim or offset grind the raceways to affect an axlial preload and make the bearing stiffer (less delfection for applied axial load).
However, you start to pay a higher and higher price in friction as the ball is now being forced to ride in a flatter portion of the raceway that it wasn't intitially desinged to run in.
Now, do the same with a preloaded A/C bearing. Yes, the ball is shoved up the side of the raceway more BUT the raceway is now skewed 15, 25, 30, 60 deg. Instead if rolling in a "flat" zone, it is now rolling in a radiused portion of the raceway. Higher preload, but low rolling friction.
BUT, for the same given preload, a 15 won't be as stiff as a 25 which won't be as stiff as a 30 and none will be as stiff as a 60 deg ball screw support bearing.
Effectively, you want/need a heavily preloaded, high contact angle BALL bearing in a ball screw support position (tapered rollers have more drag do to end thrust agains the rollers under high thrust conditinos).
A/C bearings roll with the least amount of friction under the preload needed to assure good, solid,deflection free motion that you're trying to achieve so as to eliminate "backlash".
Cost: How accurate do you want to get????. The following bearing P/N's are ranked in order of stiffness (least deflection) and cost (highest first):
20TAC47BDFC10PN7A (stiffest, most expensive, lowest friction ball screw bearing)
(7204BYGDUH) mounted in DF, custom made
7204A5TYDUHPA7 mounted in DF stock item
7204CTYDUHPA7 mounted in DF stock item
(6204DBL'spreloaded ABEC 1 oem style bearings mounted DF, custom made)
Note: the bearings in the ( )'s are not a standard production item and would have to be custom ground. The 7204 BYG uses "hardware grade" electric motor style 30 deg bearings which are NOT ABEC 7 in accuracy - all the others are ABEC 7's. This BYG version is essentialy a low(er) cost ball screw substitute that would have to be custom ground by a qualified bearing rework facility - no guarantees.
Cogged belts: I didn't find that a properly tensioned, cogged belt that was not worn or was not mounted to clabbed out sprockets contributed much if any backlash. Spring couplings WILL add backlash and are to be avoided like the palgue. But there is not so much slop added with a decent cogged belt drive (IE: HTD or equivalent).
Caution regarding mounting bearings: the pocket that the BPT used in my mill required special bearing widths. If you did NOT fill up the space with appropriate spacers, the proper preload could NOT be achieved when the clamp ring plate was tightened back into postion - I chased that problem for a least a day and a half. Swap bearings until you're blue in the face and you won't fix diddly. Pay attention here, I didn't and I should have known better.
WHen you get the inner and outer rings properly preloaded and clamped into postion properly so as to affect the a reasonable and proper preload, you should end up with a mechanical 'feel' when you turn the feed handles. You may encounter some drag but you shouldn't have slop, backlash or tightness. Backlash should be a thing of the past.
The above ARE the sources of backlash. Now that they are pinned down, you should be able to find and eliminate the problem.
I have a question about this picture: http://www.cnczone.com/forums/attach...2&d=1144985099
Are the pipes and milled slots on the sliding surfaces for lubrication? Is this recomended for a cnc machine, or does it have something to do with flood cooling washing away the lubrication?
Cheers
Way lubrication. It's important, expecially for CNC, because there is typically much more movement, faster moves, and use of the axes. Its important that the ways are lubricated often(usually cncs have a timed way lube dispersion) as to decrease wear and prevent loss of accuracy of the machine.
It is hard to get oil to get to ALL corners of the rubbing surfaces on the ways, especially if they are ground and lapped.
The distribution grooves you see in the foto were put there by the OEM to facilitate oil distribution between the rubbing surfaces for the reasons already cited in post #128
I recently installed Hiwin screws in my bridgeport and had some backlash on the y axis. I took things apart and put them back together and the problem went away. At present, i am running less than .0005" backlash and I have not messed around with gibns or anything, and that is testing only with my cruddy old DRO that only reads +/-.0005"
Anyhow, there are a couple things to check, and some things I noticed taking mine apart and installing it:
The woodruff key was tight, and I mean tight, going into the screw shafts. Also, the collar/shafts for the timing pulleys fit on very tight...and again I mean tight. No movement either radially and the only way they moved axially was with a puller and the outter nuts that tightened the handle/timing shaft onto the screw.
On the x axis, you should have the pair of bearings on the right side. You need to put the shaft through them and then attach your handle/drive shaft extension and tighten it so that there is no play between the screw shaft and the drive shaft. (I.e. you pinch the pair of bearings in beween the shafts.) Both the screw shaft and the timing belt shaft have to come in contact with the inner raceways of the bearings...and the outter raceways should be held in place with other retainers to prevent them from doing any movement......
Also, the cnc systems I have worked with all have timing pulleys that have a key as well as taper lock onto the shaft. If you do not have that taper lock, you will probably have slippage.
-Ron
Does this make any sense?
r zerr : when it comes to backlash, EVERYTHING that results in lost motion reductions makes sense.
When you get down to 0.0005" backlash, small differences in gibb slop and/or bearing stiffness can make BIG differences in how quick and how quick backlash will diminish further if at all.
TOO TIGHT a gibb adjustment will result in LOTS of turning of the screw to affect a TRUE direction change - afterall, you DO have to input force to overcome friction and screw movement does eventually overcome friction as opposed to generating 1:1 motion reactions.
Some time spent working on optimizing gibb adjusment may pay off handsomely in table motion performance enhancements.
For the benefit of those trying to eliminate backlash, I have attempted to gather up as many ideas and thoughts as I could, and combined them into a semi-coherent pair of articles on the subject.
Eliminating Backlash Part 1: The Basics
http://www.thewarfields.com/MT/CCBacklash1.htm
This is a basic treatise on ballscrews and using angular contact bearings to mount them.
Eliminating Backlash Part 2: Refinements
http://www.thewarfields.com/MT/CCBacklash2.htm
This assumes you've got your ballscrews installed and you are still not happy, which is the case here.
There are some things I've run across that have not been mentioned here, but not an awful lot. Still, I hope this will be helpful to some that run across the thread.
Best,
BW
I have spent the summer hang gliding and did not have a lot of time to work on my mill. Now I have a littel more time available. I have the limit switches in place so I have increased the speed to 90 IPM. I have uploaded a video of the mill at 90IPM while running Art's road runner test program. I still have the backlash but have learned to live with it for now. I do have great repeatability though. After running the road runner programs (5 minutes and 30 seconds), my digital read out displayed 0.0000, 0.0000. Ocasionally it will read 0.0005 after running the program but never more than that even after running the program 4 times in a row.
http://www.flyingcritters.com/video/90IPM.wmv
Vince
NICE! Hey, those motors are kicking ass! I may invest in new motors after seeing that.
Been off the machine myself. Taking some .Net courses that are taking all my time.
Best,
John
John Delaney
www.rwicooking.com
Video looks good. However, I would be really careful working around that machine with the handles spinning like that. You could easily end-up with a broken hand or something worse.
Did you put a ballscrew on the Z-axis? The photos look like you are just driving the crank-handle for the knee- did you do anything to reduce the gear-lash in the bevel gears?
Im doing a Jet knee mill now and I am driving the knee.
I am just driving the crank handle shaft and turning the original acme screw and bevel gears. The weight of the assembly seems to keep out most backlash. In fact I believe the Z axis has less backlash then the other two axes that have precision ground ballscrews.
Vince
I decided to try tapping with Mach 3 today using a tapping head. Several months ago I purchased a Procunier size 3 tapping head on eBay for $90. At the time I had no idea how big a size 3 was. It turned out to be huge. I just put it in a drawer and forgot about it. Fast forward to this weekend. I decided to make a welding table out of a 35”x15”x1” piece of aluminum with holes drilled and tapped on 2” centers. I had made a similar plate for my mill while I was taking a CNC class and had use of a Haas CNC mill. This time I would use my converted Bridgeport. Due to the size of the plate I had to drill and tap it in 4 quadrants (after tig welding 20 holes closed that came in the plate and surfacing it).
I used a G73 peck drilling cycle for the holes and then just G1 and G0 moves for the tapping. The Procunier tapping head reversed at twice the RPM as the feed. I was running it at 210 RPM to it would reverse at 420. I have read that I should feed it at 95% of the calculated feed rate to account for any possible slippage or slop in the system. I used a forward feed rate of 15.2 IPM and a reverse at 30.4 IPM. The size 3 is so large that it took all but the last 1” of my knee travel to clear the table. I am always nervous any time I try some new program but everything went perfectly. The holes were drilled at a rate of 1 every 36 seconds and the tapping took 12 seconds per hole. There are 112 holes in the plate. I posted a video here:
http://www.flyingcritters.com/video/Tap.wmv
I know some people will cringe at using WD-40 for tapping fluid, but it works for me. The ½” tap I am using has tapped over 250 holes 1” deep and it still looks sharp.
Vince
I'm talking about real cookies, the type you eat. I have had a bunch
of ideas running through my head about what uses I can find for my
CNC mill. I hit on chocolate molds, but I found I needed a flexible
mold to get the chocolate out. I then hit on the idea of using the
molds I made for making cookies. I first tried a batch of sugar
cookies but found that they rise too much and I loose the image on
the cookie. Spritz cookie dough will work much better since they
done rise much.
I first imported an image of our hang gliding club (a red tail hawk
holding a hang glider pilot) into Lazycam. I had Lazycam scan the
image at .01" grids X and Y and a depth of .1". I used a 1/2" 90
degree pointed mill to cut the nylon plastic. I was able to cut at
60 ipm. It took about ½ hour to cut one mold (about 2"x2.5"). After
the finished cut I cleaned it up and used a propane torch to smooth
the tooling marks a little.
To put the image on the cookie I dusted the mold with corn starch and
also dusted the bottom of a glass that I used to press the dough into
the mold. Gently peel off the cookie and bake until brown on the
edges, enjoy!
For all of you who have spouses that say you spend too much money on
your hobby, this is a chance to win them to your side.
The finished product waiting for the oven.
Vince
Vince, I've been following your progress with keen interest. Thanks so much for keeping us in the loop.
On the tapping head, Wow! That's a big tapping head!
I had a friend recommend that Harbor Freight's head is actually decent, so I bought one new. Haven't had a chance to try it yet, but I will.
On the cookies--brilliant idea!
I would think you could make a cottage business out of custom cookie molds if you wanted to. Businesses and organizations of various kinds would love to have their logo on one. Or, you could just do a selection of interesting motifs and sell them on eBay.
Best Regards,
Bob Warfield