Idea for stopping the flexing in a mendel90

hi folks, the mendel90 was my first 3d printer and i can honestly say that i make it the "benchmark" as far as engineering goes. i ran mine at 200mm/sec, with no modifications, and it worked perfectly for over 18 months.

now, the one thing that i really couldn't work out how to stop it doing is: wobbling about. if you lift one corner of a mendel90 the other three corners actually stay on the ground even if you lift by over 10mm! yesterday i have a brain-wave on how to fix that. it's actually very simple:

• a second upside-down U-shape mounted right at the back, fixed to the base and all the way up the side panels
• a top panel spanning right across the two U-shapes, fixed to both and fixed to the tops of the side panels as well

the end-result would be a completely rigid box at the back half of the mendel90, where the base, protruding from the front, would be supported by its aluminium box-section on the base. the entire printer would then no longer flex about *at all*.

also if instead of a U-shape it was made a full panel at the back, it would be a partially-enclosed box that would retain heat slightly better than the existing open frame. i no longer have my mendel90 so i cannot recall if the printbed, at full extent on the Y-Axis, protrudes over the back edge of the base, but if so that obviously would need either the U-shape at the back or a slot for it to protrude through.

i would be fascinated to hear if anyone considered this a worthwhile enough improvement to consider trying it out: i no longer have my mendel90 (i sold it to the person who initially evaluated it and recommended it to me) so cannot try it out myself.

lastly i have to say, the mendel90 *really is* an iconic design of superb engineering, the only "downside" that a number of people think it has is the direct-drive (it's not), and that can be fixed (actually *massively* improved on) with this drop-in upgrade - a direct-drive flex3drive NEMA17 adapter: [flex3drive.com]

All you need to do to prevent it wobbling is to place it on a firm flat surface. I have never seen the corners lift while it is printing.

---

[www.hydraraptor.blogspot.com]

Quote
nophead
All you need to do to prevent it wobbling is to place it on a firm flat surface. I have never seen the corners lift while it is printing.

allo nophead, nice to hear from you on here. i sold my mendel90 to the person who reviewed it in the first place. you may recall from our conversations mmm... at least... 2 years ago now? i was the one running my mendel90 - unmodified - at 200mm/sec, without significant degradation in the really quite complex parts. 250mm/sec i found was a bit too much, so i backed it down to 200mm/sec and it did really surprisingly well.

when we talked, i know you recommend only up to a maximum of 150mm/sec, and expressed surprise that the prints regularly came out ok at 200mm/sec, but i am primarily thinking in terms of how to improve on the mendel90 design, such as by using the NEMA17 drop-in replacement "Direct Drive" Flex3Drive adapter (instead of the known problematic bowden concept). mutley3d has a youtube demo of running at 400mm/sec just for a laugh on the MendelFlex, the sound is hilarious. he also says that 300mm/sec is perfectly fine, but when printing parts for customers he prefers to ramp it down to "only" 150mm/sec so as to get much much better quality as opposed to "higher speed for the sake of it".

so, anyway, i'm just thinking of how to push the boundaries, especially on this iconic well-engineered design which i still absolutely love. and you happened to have made

Are you saying the corners of the base lift if you print at 250mm/sec?

I am confused by what you mean by "instead of the known problematic bowden concept" because Mendel90 doesn't use Bowden drive to feed the filament. Yes you can use a Flex3drive to make the X carriage lighter but the Y carriage is the heaviest axis.

---

[www.hydraraptor.blogspot.com]

Quote
nophead
Are you saying the corners of the base lift if you print at 250mm/sec?

ah. apologies but it's been over 18 months now, and i knew less then than i do now so wasn't specifically looking for that as a symptom. what i *did* notice was that the cardboard on which the printer sat, on a high (polished) table, would actually move about! the table's legs were clearly undergoing vibration.

also i ended up with both "rounding" *and* "bulging" on corners that should have been square, indicating that the linear bearings were rattling considerably as well as there potentially being other "flexing", but the main reason i stopped running at greater than 200mm/sec was "zagging" on extra-long parts. during the high-speed travel the extruder simply couldn't push out enough filament and it would snap, leaving a gap on one layer that then interfered with the printing on the next, leaving these quite pretty "Z" patterns on the outer layer.

Quote

I am confused by what you mean by "instead of the known problematic bowden concept" because Mendel90 doesn't use Bowden drive to feed the filament.

i recall - some time ago - our discussions about bowden tubes. i found your insights absolutely invaluable. i mention bowden simply because the flex3drive has all the advantages of the bowden concept... with *none of the disadvantages*.

Quote

Yes you can use a Flex3drive to make the X carriage lighter but the Y carriage is the heaviest axis.

ahh interesting, and thank you for drawing my attention to that, i'd forgotten to consider the Y carriage weight. let me think... ok i'm not so hugely concerned about the Y axis because mechanically it just travels back-and-forth. it's the X-axis which has the extruder (and 3D plastic), and a much smaller distance between the bearings, so the additional inertia of a higher speed would result in amplified effects of mechanical strain and tolerances.

on the Y axis the distance between bearings is.. what... 120mm? so any "rotation" caused by play in the actual bearings would be minimised. conceivably you could replace the Y belt with 8mm or 9mm and higher-torque motors...

There shouldn't actually be any play in the bearings unless they are worn out. Linear bearings have pre-load, which means the balls are under tension between the outer shell and the rod. That is why they feel a bit lumpy because when the balls recirculate they have to squeeze into a gap smaller than their diameter. On the other hand, bushings are much smoother but they needs some clearance, so always have some play.

---

[www.hydraraptor.blogspot.com]

yehh i've found that as a general rule LM8UU bearings always have some sort of play (guestimated 0.2mm), and LM8LUU even more so - you can feel it by wobbling whatever the assembly is. the preloading makes a lot of sense. i _did_ consider trying out those plastic LM8UU bearings at one point (made in japan) but honestly i'm so fed up with rods i'm going to be using Hiwin MGN rails in future.

If you use LM8UU bearings with proper h6 tolerance precision ground rods there should no radial play because of the preload. With a single linear bearing you can get angular play but when used in a pair that disappears. I get about 0.2mm play with Igus RJMP bushings but not with LM8UU.

---

[www.hydraraptor.blogspot.com]

thanks nop, that's good to know: i was just about to try using RJMP igus bearings. also i've dropped LM8LUU in favour of 2 LM8UU and yes i can confirm it makes a staggering difference.... although where the hell i got such rubbish quality LM8LUU bearings which are *literally* rattling on the rods (not even angular rotation, just... moving laterally) i really can't recall.