This has to do with people who have no experience building with linear actuators designing and building things with linear actuators. Like the rubber motor mounting plates, they make assumptions as to how they're meant to be used.
I spent close to a decade working in semi-automated manufacturing related to welded product mass-production... building welding/drilling/machining jigs and parts placing/moving linear actuators was a big part of that work, as was building with 8020 extrusion. Not all of it used this bronze-nut rubbish; much of it required recirculating ball leadscrews like the one above. But for applications where space did not permit, or dusty/dirty environments, or high precision in both directions of travel wasn't necessary, bronze nuts/leads did get used.
I learned quite a bit in that apprenticeship... some of it translates directly to 3DP; however not nearly as much as I'd originally taken for granted when I got into this hobby. But yeah... that is a pretty fundamental design principle for these things. The lash goes on the unloaded side.
Picture your printer stock, with no spring & lash on the nut. The weight of the X-axis carriage presses down on the nut pretty much all the time. Most of the time, that constant downward force is all that is needed to get relatively consistent results. Force from the leadscrew is applied upward. You want the force of the spring under the lash to be always pressing in the opposite direction to that of the applied force from the lead; in other words, if the lead is pressing upwards, the base of the spring needs to be pressing downwards.
All the lash & spring are supposed to do
in this case is
resist any tendency to push back in the wrong direction (unless your actuator needs to be able to apply force against a workload in both directions; then the preload of the spring under the lash needs to be greater than the maximum applied force in either direction)
and to prevent bounce. If we designed this with a proper bidirectional anti-backlash (the spring applying greater force than any anticipated loading in either direction) it wouldn't matter. But that would require a much stronger motor to overcome the drag of such an arrangement, and wear would be greater, so larger leadscrews, nuts, motor control electronics...
So, in this case, we design so it works with the prevailing preload of gravity instead of against it, and the lash & spring go on the unloaded side.People who don't know any better design printers with the nut on the Z-axis lead going either way, and
if using a standard nut it doesn't matter. They don't take into account that someone might want to put a proper anti-lash nut on there. That is when you run into these problems.
Another design consideration that gets completely ignored is proper thrust loading geometry.
Steppers are not meant to handle ANY thrust loading; they are supposed to only provide torque. A properly designed actuator has a leadscrew which is fully supported by bearings in all axes such that those bearings take all thrust and lateral load. The stepper is then pretty much an afterthought.
It will get mounted solidly to a bracket, then connected to the leadscrew with a flexible coupler that makes up for any slight misalignment between the centerlines of the two shafts.
Partly out of ignorance, and partly out of cost-cutting, most cheap 3DPrinters pretty much completely ignore all of this basic engineering. They use no thrust bearings, and in fact not only apply thrust load directly to the Z-axis stepper, they apply it through a flexible coupler with all the thrust-loading capacity of a wet noodle.
"You can't push a string." is the engineering maxim which applies here; they may not be pushing a string, but the flexible couplers which are cheapest are pretty close; they're a (more or less precision) helical-cut spring which connects a stepper to a leadscrew. These are intended to handle torsional load
only. they are not intended for even a gram of thrust loading.
Worse, the cheapest ones use grub screws against the shaft; this, combined with usually abysmal precision in the manufacturing of the bores, makes a coupler that by its nature goes off-center as soon as you start tightening things down. Their only saving grace is the fact that they are essentially a spring, and can make up for several thou out-of-round and even a degree or two off-axis. But that generally self-curing nature of the design does not cure the problem of trying to apply thrust load through a spring.
Knowing this,
you can then either completely redesign the X-Axis actuator with proper bearing plates and thrust-loading design, or you can take some minimal steps to ameliorate the impact of the poor design endemic to the product. I choose the latter course.
The fact is that the thrust loading on the stepper motor is pretty low; low enough that even if the bearings do fail as a result, it will be long after the rollers and belts and pretty much every other part of the printer wear out from just plain moving constantly on a chassis made with pretty poor tolerances. And a replacement stepper is $10-20, so really almost a consumable supply anyways.
So then we look at the coupler. There are two basic ways to resolve this issue: either spend the money on a proper semi-rigid coupler that is designed to handle thrust load, or modify the springy one to handle thrust load.
CReality, to their credit, figured this one out a long time ago and went with the first solution. These semi-rigid couplers have a slot cut approx 75% of the way through the body. This will flex enough to make up for a few thou misalignment, however not enough to give at all under thrust loading. The clamp-style attachment used on these by its nature self-centers on the centerline of the shaft it is attached to, and once tightened down, does not work loose like grub screws in this application.
The single drawback of this approach is that to work, these couplers need to be machined to pretty high precision. An order of magnitude higher than most of the cheapo China-direct stuff out there. That is not hard to get... it just costs money. These couplers may not look like much, but they cost easily 3-5x as much to make as those helical-cut cheapies.
I ran up against this issue ages ago with my Tarantula builds... and
I came up with a solution back then that still works, for the most part; but you have to have balls. Loose ones. Being a tinkerdwagon with interests in a multitude of disciplines means that one of the things I keep in my little bag of dirty tricks is an assortment of loose ball-bearings. When I first started building the Tarantula, I could literally see the coupler bunging up & down while it was running (The Tarantula was a horrible design... it loaded up the Z-Axis really badly), so I tried to find something just like the CReality coupler above. It simply was not available, except in SAE sizes from US machinery suppliers for almost as much money as the whole printer.
So I started thinking, and I realized a thrust bearing was needed here... but it just needed to apply against compression, so a single ball situated between the ends of the stepper and leadscrew shafts was all that would be needed. Even then, most 3DP used a 5mm x 8mm coupler here (due to using the cheapest leadscrew vs cheapest steppers); I found that a 5/16" ball-bearing was just small enough that it would drop inside the coupler and would self-center without binding anywhere.
I found that by adjusting the placement of the coupler so that the ball was a little off-center in the coupler vertically, then pulling the bottom of the coupler down before tightening, I could put a little preload on the ball, and all thrust-loading would be translated directly to the shaft of the stepper, with zero load taken by the
spring errr, coupler.
I've been doing them that way ever since. But I still prefer the CReality solution of simply using a better quality semi-rigid coupler. And my CR-6-SE uses two of em.
mnem