There seem to be two schools of thought on z-backlash. One is that it is not an issue because gravity takes care of everything. The other, that the z-bearings (even Misumi's for those that upgraded!) require a certain amount of pressure before they slip into motion and that sometimes they might have moved when we rather hoped that they would. All of that I can imagine being sensitive to the parallelism of the z-rods, the lubricant and the temperature. The other day a bloke told me an interesting thing. You can make a leadnut with almost zero backlash by taking a piece of Teflon (eg rods available on aliexpress), drilling it and tapping it for the leadscrew dimensions. Because teflon is spongy (by comparison with metal) it stretches a bit as you tap, then springs back as the tap comes out of the hole. Which means you end up with a threaded hole that will fit the leadscrew really snugly. Moreover, Teflon is self-lubricating. Felt I had to share this in case someone thought it was worth a try.
This is similar to using the POM (Delrin) leadscrew nuts RobotDigg has. I have a set on order we're going to try in a from-scratch BigBox. Delrin isn't as slick as PTFE but it's much tougher.
Sounds intriguing, not the easiest stuff to tap but should work fine. There's a few different ways I've seen backlash stopped. Pointless in our case but an large machined threads you can cut at a slightly larger pitch so that the top of the nut pushes against the underside of the rod thread. The easiest way is to use two i.e. a nut and a locknut . Position one and fix in position. Add a second nut and turn until finger tight. The second should then be pressing in the other direction. Maybe a spot of glue to the nut face to stop it turning or a washer in between them? Difficult to get the right tightness on the nuts without stopping the rod from turning though. I've pondered adding a weight to the bed to keep the nuts under load too. A few options anyway.
haha - I tried the weight thing (and thought I was mad). Thing is my wobbly leadscrew problems are probably greater than any backlash issues there might be. Another thing one would maybe wonder about with a weight would be the inertia of the z moves?
This guy makes a backlash free delrin nut from scratch --> . The technique should be the same for Teflon, but in my opinion it isn't worth it since most of the lead screws in the production big boxes, as @R Design says, are wobbly, so even without backlash it still wouldn't be perfect. My testing of the ballbearing screws from robotdigg is showing to be very promising and fixes all of the issues related to z-axis motion. I will be posting here: https://forum.e3d-online.com/index.php?threads/z-axis-ball-screws.2224/ when I finish testing.
Most of the wobble comes from the couplers and other owners have said that exchanging these for plumb couplers apparently improves the situation. I spent time adjusting the clamps on my original couplers and have reasonable performance.
This was done in the assembly stage by rotating the lead-screw and tightening the clamp then turning the coupler and observing the movement at the top of the screw. After many iterations I managed to get the eccentricity down to an acceptable level and my lead-screws are straight. I will, one day(!), consider making a solid coupler carefully matched to the motor shaft and the screw such that the axial centre of both are perfectly in line and concentric. The real problem is that the length of the lead-screw covered by the coupler is quite short, so I will try to make it as long as possible. The spring coupling is not needed as the frame construction is good enough such that any misalignment is negligible.
That's really interesting @mike01hu as I swear I didn't have this problem at one point. At some point, I took things apart for a full clean and it's possible that my screws were straighter before the rebuild. Thanks for sharing your insight into this. I looked over the assembly last night and wondered to myself why the lead screw is not held at the top, and the same goes for the top of the corner rods. I guess that's in order to permit some sort of movement to occur, and that's what we see in our z banding, and potentially x and y banding could be explained by the same thing; slight movements caused by the lead screw which are regular due to precession of a fault/misalignment. In theory then, if the screw can be made truly straight, could not the tops of the screws and rods be more rigid than they are?
Yes! If the damned leadscrews and guide rods were straight, I'd say constrain the top of the leadscrew in a bearing and get rid of the coupler. Sadly, that's not practical with most part sets. If you did constrain the top of the leadscrew, all bends in the leadscrews would have to be taken out by the couplers moving sideways and the platform would still wiggle, with that getting worse and worse as the platform approached the couplers..
Then sadly, for us, Z=0 is closest to the point where leadscrew wobble will have the maximum impact, thus affecting every print. At least if it were the other way round, only really tall prints would take the hit. Will be interesting to see how this is all impacted by @Greg Holloway's promise of screw and motor integration. If that creates something straight, then perhaps more constraint will be possible for the top of the rods.
I've not read the entire thread, but I've had experience with Delrin nuts so I'll drop my two pence in. I originally started with brass nuts on my printer(cartesian prusa style machine made almost entirely of 2080 v-slot) which gave a lot of layer misalignment, although the prints were still of a decent quality. At this point I assumed that gravity was removing any backlash present. However, I started to notice that one side had more backlash then the other. I switched to a single motor for my Z-axis, assuming that it was a motor misalignment issue. This probably had some effect and I swear by using a single motor as I haven't had to recalibrate since the upgrade but the bad layer alignment persisted. I switched to Delrin anti-backlash nuts and the improvement is tremendous. What I actually discovered in the end was that The X axis motor which is placed to one side actually caused a 'see-saw' like effect, where the heaviest side would fall but the lighter side would slightly rise and float a bit like the needle on a record player. Another fix on this would be to distribute the weight more evenly but the simpler fix was anti-backlash Delrin nuts for me. Here is an up close image of the layer alignment of one of my prints. The layers are 300 microns and I can recreate this alignment at 200 or even 100 microns. I daresay I could go to 50, but I rarely print at that resolution.
