Well since Eustathios does not use a coupler for the leadscrews (driven via belt, riding on a bearing), and its clearly not z ribbing as you described, or classical z-wobble then I think this turned out to be exactly as Eric described. The smooth rod was forcing the nut with a side load that was causing the z axis to ride mostly on one side of the screw. So the bed wasn’t falling down as far when the driving edge of the screw was at the other side. Fixing the alignment of the z-rod has made this completely disappear, only artifacts I’m now seeing are really minor and probably due to small filament diameter variation.
Side-load can do it too, if it has a cyclical component. Driving it through a belt with an eccentric pulley could do it too. Those were just a couple of examples of how vertical Z wobble can occur.
Can you add a section on vertical z wobble to your evernote page? Really helpful document and could save a bunch of people some headscratching calling some of these issues out a little. I had no idea it would present this way and have been chasing z-wobble for ages 
Yeah, might find some downtime for that on my trip to MRRF.
I might want to use one of your pictures (with credit) if that’s alright.
Go for it, use anything you like.
I’m jealous of all you guys headed out to MRRF, have fun!
I’ve got a similar problem before it was directly related to the z axis. Now that I’ve swapped it out for a new bit it’s become more…erratic and less patterned than it was. With a herculien
part on the right is an old part. and the ribs or pattern is exactly 2mm OC
part on left is with new screw and its more erratic. i dont have my large benchy near me, but it shows it more.
@Jim_Stone the one on the left looks like it could just be spool tension, but that’s a Bowden machine, isn’t it? Bowden extruders shouldn’t have that problem unless the tension is causing the entire frame to flex
@Whosa_whatsis yes. It’s a herculien. With a bond tech extruder and an e3d. Using atomic filament abs
Yeah, so unless your frame is flexing, that’s not it.
yeah i doubt the whole flexing thing. its a pretty solid frame.
question. would my table have anything to do with it?
The verticals on HercuLien are only 20x20 and I have seen that they can flex if the 4 feet are level. I’ve actually put levelers on my feet now to make sure that I’ve got it set on the table in a way that isn’t putting the frame and tension.
Added a section on vertical Z wobble (mostly reworked from what I wrote here, plus a couple more potential causes I noticed while looking at the variety of machines a MRRF): https://www.evernote.com/l/ANNwHDbE3dVGaaSClT2JJMcdHvmSmIKVSHw
@Whosa_whatsis Random note in the article… I don’t know if I’d recommend HDPE feed tubes, they can have quite a lot of friction if there’s much curvature in the tube. I did some drag math (capstan equation) on tube materials a while back and basically everything but PTFE and nylon will add a lot of drag.
@Ryan_Carlyle They’re certainly not as good as PTFE, but empirically, the larger-diameter tubes I recommend don’t add noticeably more friction than the tighter PTFE tubes used for bowden tubes. If they fit the filament nearly as tightly as PTFE tubes generally do, I would agree completely. Because they cost virtually nothing and are far easier to find locally, I think they’re still worth mentioning.
The stiffness of these tubes is actually a bigger concern, depending on how long they are and how they are routed. I’ve actually seen one of these tubes break off the filament at the extruder when someone tried to use one that was too short for the job.
@Whosa_whatsis Has anybody ever collected any hard data to your knowledge on drag force vs tube ID? Seems to me like poor straightness in the filament (ie mild bends where it crosses lower wraps) would be the only effect causing increased drag until you got a REALLY tight fit.
Man, I need to make a filament force sensing jig.
Not that I know of, but friction is proportional to the normal force, and it seems like you could calculate that by modeling the tube and filament as two springs pushing against one another…
Cylinder-in-cylinder physics is pretty much bread and butter stuff at my day job, but it took me a LOT of thinking to figure out any reason why tube ID would affect drag friction in the absence of filament kinking. If you assume line contact between the filament and tube, there should be ZERO impact of either filament OD or tube ID. Ideal line contact is diameter-neutral. Likewise if you assume a flat (planar) contact cross-section, as would be used in the typical Capstan Equation workup. Sliding friction of a cylinder on a plate will be approximately diameter-neutral, since Coulomb friction is a function of normal force rather than contact area or pressure. The only time the diameters should come into play is when the contact cross-section becomes large enough to cover a meaningful amount of arc curvature. At that point, there is an additional “wedging” action due to the horizontal force components from the contact on either side of the filament. That’s the only thing I can think of (in addition to filament bending/kinking effects).
That fits my empirical impressions. There’s also the fact that the difference in diameters will be subtracted from a term in the spring force equation (after all, there would be no force contributing to friction except from gravity if the tube’s ID was large enough to be effectively infinite so that the differences in their bend radii no longer matters).
Oh, and you also need to consider the second-order characteristics of the shapes. If one of your cylinders is being pulled and the other is being pushed (they reverse roles in a Bowden tube vs. A feed tube), the one being pushed will try to compress axially and it will want to buckle, adding additional side forces. This is why you need a reasonably stiff tube.
What do you mean when you say there would be no force contributing to friction if the tube ID were infinite? No additional force? I’m not sure if this is what you meant, but the primary friction force is coming from the normal force applied by filament tension (or compression) as it wraps around the large-scale tube curvature. An “infinite ID tube” is just filament wrapped around a cylinder. That’s classic capstan friction.