The Ender 5 PlusPlus on dual linear rails (#E21F7)
A little update on my big Ender.
While I’m indeed satisfied with how the thing works with the somewhat recent upgrade to linear rails on the Y axis (moving the X carriage) in #E21F6 last August, it’s still an old Ender. It sounds like an old Ender and it moves like an old Ender, because despite all the axis changes, mainboard swap and Klipper upgrade, it’s still an Ender chassis with Ender motors underneath.
The other day I had one of these brainfarts at work: We also run machine parts on linear rails, very high acceleration machine parts to be precise. And on quad carriages, meaning two parallel rails with two carriages each. The slowest axis does like 2-3g, so 20-30k acc in 3D printing terms, total acceleration of the 3D movable toolhead is an order of magnitude higher than that. We use Rexroth or some Japanese ones, not sure if NSK or THK. Good stuff.
Sooo…the Ender already has two parallel rails installed…why not upgrade to two carriages per rail?
Of course the seller of my rails on Amazon doesn’t (currently) offer single carriage blocks, so I’m out of luck right away. These original blocks did fit nicely, I have to say, especially for the price. Instead, I bought single blocks off AliExpress, hoping they would be compatible. It’s MGN12H after all, isn’t it?
I paid 21.88€ including a lot of shipping for a pack of four, reason being that I could swich the dual Y axis to those blocks entirely if necessary, plus single blocks in case of catastrophic failure are like a tenner, so I just ordered one more than needed for a full swap.
Well – they do fit. But seating on the rail isn’t snug, there is audible play. If these were the original blocks, I would have returned the entire rail, but as an aftermarket part, they would rightfully point out that maybe my cheapo rail isn’t up to spec and their carriages are. Well, I’m already twenty bucks in, so why not try them.
A little bit of CAD work and the realization that a slightly asymmetric mount would be best later, I had a working prototype for the left hand side, just like I did last time when installing the rails. Version 2 solved minor fitting issues with the LED panels, but V1 already worked nicely. V3 is basically a mirrored V2 with the cutouts for the motor mount added, suitable for the right hand side. Although these mounts are in fact very wide, I did not lose any printing area due to exact positioning – very tight job with like 2mm to spare on each side, though. Mounting system is the same as before – two M8 heat inserts and regular M3 for the blocks (2×4). The center hole is for accessing the 20×20 aluminium profile screw that sits inside the metal mount where this part bolts onto.
Here are the thingiverse renderings:
Now, I don’t think anyone would want to use these exact parts – but maybe this inspires someone to do the same, or similar things with shorter MGN12C blocks. I can supply STEP files of these if needed, but for now the STLs on thingiverse (thing 7009462) or direct download V3_Motor / V2_Pulley have to do. I once again printed this in PETG (and transparent PETG is all that I currently use, 90% of my prints are done in PLA nowadays)
Photos from the actual machine while printing: Left side with the pulley system:
And right side with the motor:
Results in terms of printing? Well, I wouldn’t say printing quality has taken a hit nor has gotten significantly better. It was already decent and still is. Maybe it’ll show on large (time and scale) prints.
Printing speed however is affected: Previously, the system maxed out at 145mm/s speed and 6500mm/s² acceleration (Klipper speed test, limit = max values before losing full steps). With just one side upgraded, things went up to 155mm/s at 10500mm/s². That tells me that there was some wobble in the system that did get minimized due to better guiding, even if the second block has a couple dozen µm of rotational play over the rail profile. Once both sides were upgraded, the speed test maxed out at 160mm/s and 12500mm/s – so speed didn’t gain much (that’s likely a motor thing), but possible acceleration almost doubled. In terms of actual values used during print, it’s very similar – I kept speed at 140mm/s, but upped acceleration from 4500 to 8000 mm/s² and that is working very well. I absolutely could not print with 8k acc before, but now I do…
…and that’s not all! With some accelerometer tuning and, more importantly, removal of all the small bits the carriage mount was chafing on at the extremal values – the printer is now at over 20k acceleration without losing steps, so more than triple the original value. That doesn’t mean I should use that value for prints, but dang, that’s an upgrade.
The Klipper spectral response graphs also provide insights on what’s the biggest mechanical issue on that printer. Given the green solid line is tiny compared to the blue one – we’re good on Y, but Z needs some work next 
