[braidmd]

I'm having a bit of trouble with backlash on the X axis of my DIY-CNC router (http://www.cnczone.com/forums/showthread.php?t=43748). While I noted this recently in the thread, I don't expect everyone to follow it. So I wanted to post my request for input separately...

Here's the situation: My gantry is heavy - at least 100 LBS, perhaps 130 LBS? I'm running dual lead screws on the X axis with 425 in/oz steppers (Xylotex), 48V/12A power supply, with a G540 driver. While the setup isn't a perfect match, it seems to work well. The screws are ACME 1/2"-10 TPI single start from Enco (not precision, by an means). And the lead nuts are the AB type from Dumpster CNC. That said, I can rapid at 100 IPM, though the screws do whip a bit.

I've noticed that the backlash is getting worse with machine use - I can now shift either side of the gantry 1-2 mm, and have isolated the movement to the lead nuts, at the threads. I have a buildup of black, semi-greasy, soft plastic in the slots of the AB nut. I'm not seeing this on the Y or Z lead nuts. So I'm wondering if the problem is one of the following, or a combination of two or more:

- The mass of the gantry is too much for the delrin AB nuts to handle.
- The speed of the screw during rapids over the length of the axis is generating heat from friction, thus softening the delrin and increasing the rate of wear.
- The screws are cheap and are a bit rough (not polished).
- The screws are not well-aligned (need to verify this - it might be true of my machine).

That noted, I am planning to upgrade to faster screws. But I don't want to by parts that can't handle the mass of my gantry (no complaints about Dumpster's products - I like his stuff). So I'm requesting input on what multi-start screws & lead nuts would function "best" on my machine (for both the X & Y) - keeping cost within reason (I can't afford ball screws... $500 US, give or take, is my limit). I know this type of question comes up time & time again. I've been searching through the threads, but haven't found anything focused on screws/nuts for a heavy gantry... on the DIY router, at least. So I'm hoping this round of query will be a good reference for others. Thanks in advance!

[stk2008]

Hi there

Im no expert but im very confused why you went for plastic nut?
why not bronze ones?.

I to went for acme thread from marchent dice and got the bronze nuts and have tbh 0.001" last time I checked back lash maybe I got lucky.

Do a search to see the size of my machine its all Aluminum so aint light.

http://www.worldofcnc.com/products.asp?recnumber=47

keep us posted

[ger21]

I can now shift either side of the gantry 1-2 mm, and have isolated the movement to the lead nuts, at the threads.

This means the nuts aren't working. The spring should be pushing the ring, which forces the nut tightly around the screw, preventing movement. Is the ring stuck? Perhaps from the grease?

What are you using for lubrication? Try cleaning the nuts thouroughly (remove and soak them if you have to) and use a light oil for lubrication. I use air tool oil and it works very good.

I guess that it's possible that the nuts are badly worn if the screws are rough. But with multi start screws, you shouldn't see much wear, as they are extremely smooth. A thorough inspection should tell you if they are worn.

I'd clean them, inspect them, and try them without mounting them to see if you can feel any play in them.

I use 1/2-8 2 start screws with 250oz motors and a Xylotex running on 24V. My gantry is about 80-90lbs. I can get 155ipm reliably, and am limited by my 60" screws whipping. I can get 190ipm on my Y axis with a 40" screw.

With those same screws, and the same rpm you 're getting now, you'll get 2.5 times more speed.

[jsheerin]

I have about the same setup, a bit heavier gantry, but with 5 start screws and don't see problems after several miles moved (I have about .002"-.003" of backlash). You do need to keep the nut pressure * surface velocity (PV) lower for plastic nuts than metal. A quick calculation shows that to keep PV below 10000 (a limit I found on a manufacturer's website in units of psi*fpm), 1/2-10 5 start would give you 135ipm while single start would give you 27ipm, so it is very possible you are melting your nuts. ;> Seriously, check for heat. If the screws are rough, that's also a bad thing. I'd go for the precision multi-start screws.

[lunchtrayrider]

i recently melted a z-axis nut. exactly as you described, cause: too much speed.

[braidmd]

Thanks everyone for the input! I just ordered 1/2-8 4-start screws from Roton, and the compatible AB lead nuts from DumpsterCNC + collars & couplers... all for just under $400, shipped. If I'm lucky, I will have the parts in time to do the upgrade this weekend. I am still a bit concerned about the plastic nuts. But my gut feeling is that a combination of speed/heat & low-grade screws killed my first set of lead nuts. Knowing so many other builds are running on plastic nuts with great success, I think my worries are for not. I'll post the outcome once the job is done...

[braidmd]

Here's an update on the issue with the backlash in my X axis. I still have yet to install the new screws and hardware. I have been holding off until I know the root-cause of the problem. I still don't have it sorted out fully. But I have done a bit of experimenting and wanted to share with everyone.

First, to test the potential for backlash at the lead nuts, I removed, cleaned & inspected them. The spring and tensioner seem to be fine on each nut, and the nuts had no noticeable play when loaded on a screw. So I put them back on & gave it a go. I still had backlash. But it seemed the degree of movement changed sides, now being worse on the right instead of the left (though both sides had noticeable play).

My next test involved the "shade tree" fabrication of some anti-backlash nuts that could be installed on the already-assembled machine. I had come across a DIY instructable while looking for a suitable design that involved sandwiching two blocks of plastic together, drilling a hole for threading between the layers, and then placing a scrap of threaded rod in the channel & heating it to soften the plastic - thus allowing threads to be compression-formed in the heated plastic (see pics). It isn't the most elegant product. But it works and was actually very easy to fabricate (I used UHMW). Unfortunately, it didn't solve the problem on my machine.

The next "fix" involved the "moving knot" method that I had read about here on the forum (http://www.cnczone.com/forums/showthread.php?t=51485). I didn't have enough room under the table to comfortably run the cables & pulleys. So I ran them around the outside and rear of the machine (see pics). This didn't solve the problem either. However, it did help me get the gantry very square & added noticeable stability too. So I won't say it was a waste of time.

After all of that work, I still had 0.022" of play on the left side of the machine and 0.029" on the right. So I proceeded to test the movement for hours to measure the backlash & the repeatability of the unwanted movement. Once I had it down and realized that backlash values were consistent, I tried compensating for it through the anti-backlash function in Mach3. And that did the trick! Using the magic value of 0.023", I was able to get the accuracy down to 0.002", even over long runs on the rails (i.e. feet). I'm still perplexed over the source of the problem. But at least I can start cutting now, with less worry of backlash being an issue. I ran a quick test-cut to verify the improved performance - the circles were round, the lines straight and the squareness was really square. Sweet!!!

[ger21]

Did you check the bearings for backlash?

[braidmd]

Did you check the bearings for backlash?

Hi Ger - Yes, I did check them with a dial indicator. I could move the gantry, but only read a thousandth on the dial at the end of the screw. I'm also wondering if my cross-brace under the table might be part of the problem. It is square aluminum tubing with 1/8" thick walls. Perhaps there could be some give or spring to it that could be the culprit???