After 3 months of wrangling with RS Components (after 10 weeks, they advised that the 10 x 2 trapezoidal leadscrew I ordered was no longer available) I fitted a ballscrew shaft and nut from them, also 10mm diameter with a pitch of 2mm (Nut Ref 481-2661 & Shaft Ref 341-1778). They are made by SKF and appear to be of good quality. There are many other shafts available, but 300mm long; the k8200 requires 400mm. The SKF shaft is 1m long, so it was cut to length, leaving enough to make a second leadscrew. If anyone would like to buy this off me, make me an offer.
After installing the parts, I undertook a measurement of the accuracy, using a dial gauge as before (see viewtopic.php?f=51&t=12609#p54160). The results were good, but not what I had hoped for. With the standard 8mm threaded rod, when the command to increase Z by 0,1mm, the extruder beam moved 0,087mm to 0,12mm (average 0,975mm). With the new ballscrew, this was sometimes 0,08 mm and again 0,12mm, but the average was much better, at 0.0995mm, (maybe my old dial gauge is not so good anymore). There was still some cyclical variation, but more random. However, the first test print looks much better (how do I link photos from Dropbox?), so I will continue for a while and see how it goes. Was it worth the 230 euros? I think not, but could not find a viable alternative.
The coupling is one that is typically used on shaft encoders and also sold by robotics shops. It is pre-machined for 5mm and 8mm shafts and came in a packet marked MR56137.
The SKF shaft was machined down to 8mm - it was too hard to cut with a hacksaw, but machined ok with a new carbide tip, taking small cuts. Care was taken to make sure that the shaft ends met in the middle of the coupling, as these couplings are not stiff in the axial direction, so any deflection in the coupling would result in an error. If I were to do it again, I would use a 10mm x 5mm coupling and save on the machining, as the shaft seemed to be accurate enough, despite the catalog picture showing otherwise.
The upper end was left floating, with no bearing installed - I have been using the printer this way for a while, to good effect.
The nut was mounted to the Z axis carriage using a printed 2-piece bracket which allows lining up in 3 planes, using the stock M5 tapped holes. All need a bit of cleaning up before sharing on Thingiverse.
The coupling is one that is typically used on shaft encoders and also sold by robotics shops.
The upper end was left floating, with no bearing installed - I have been using the printer this way for a while, to good effect.[/quote]
You now have your Z axis sitting on top of a spring; the coupler.
The extruder produces an upward force depending on the extrude rate.
I have my stock screw supported by the top bearing and actually extending the coupler slightly. I notice a drop in quality if the bearing support slips down at all.
Thanks for the comment. You are right, the coupling acts as a spring, but that is why I said that I made sure that the ends of the stepper motor shaft and the ball screw shaft touch. In fact, I machined the end of the ball screw shaft with a chamfer at 30 degrees, leaving only 2 or 3mm diameter flat to minimise any error.
I will make up a bearing adaptor to go into the stock top bearing mount and try it out to see. Dial gauge measurements under the extruder beam before and after will give the answer.
That’s better!
I discovered that the new SKF ball screw nut had some balls dislodged, apparently from new. There are two rows of balls in this nut and I was working with only one!
RS Components kindly replaced it and I tested it again with the dial gauge.
The results were very good, with the average (of 89 readings) increment being 0,0991mm for a command of 0,100mm. The standard deviation was 0,0057mm. If the 3 worst readings are eliminated (I suspect measuring error), then the standard deviation is 0,0043mm.
This compares very favourably with the original equipment lead screw, which had an average increment of 0,975mm with a standard deviation of 0,0295mm, making the SKF unit six times more accurate in my opinion.
Was it worth the 230 euros? Yes, definitely! The quality of the next few prints will tell.
My K8204 Kit just arrived and i mounted it following the Instructions.
Looks not so bad, but i will seriously change the mounting, so it goes top down, with hanging z axis.
That micht even allow to get some extra Z travel than with the original motor mounting
as the leadscrew is only 300mm in length.