Vibration fixed! More improvements, but new limitations.
Oh man finally.
Yes, completely gone! Here’s what I did since the last post:
I ordered the i3 Plus version with 3 bearing block mounts, and it was a drop-in replacement for the original. It’s made of aluminum and some composite material sandwiched together. To compare the stock carriage to the new one:
|Y carriage version||Thickness||Weight||Stiffness|
|Stock||1.34 mm||376 g||Flexes fairly easily|
|Tehnologika||6.40 mm||276 g||Barely flexes|
That’s a lighter carriage with less flex, there’s basically no downside. Due to the thickness of the plate I had to buy 15 mm M4 screws for the bearing blocks and belt bracket, and I also bought 35 mm M3 screws for attaching the heated bed. The 30 mm screws that came stock with the machine still worked, but I wanted more protrusion beyond the springs so that it would be easier to drop them through the holes in the carriage plate.
In terms of reducing vibration, the bed performs beautifully. I’ve watched the free corner of the heated bed very closely, and it’s hard to see vibrations being transmitted into it. The proper placement of the 3rd levelling screw definitely helps with that.
With regards to that levelling screw, I originally expected to drill a new hole into the right side of the heated bed, but I found that the hole would have gone too close to the heating traces of the PCB. However, there is actually such a hole already drilled next to the heated bed’s wiring. I had to disconnect the wires from the Melzi board, rotate the bed, and secure the wires in their new orientation. There was barely enough wire to cross that distance, actually, but I don’t think the new arrangement is putting any more mechanical stress on them than they experienced in their stock configuration.
The last advantage of a stiffer Y carriage is that bed levelling is really really easy now. Not only is it a proper 3-point suspension, but the stiffness of the bed means that I can use a micrometer to set each corner individually and be assured that the other corners will not go up or down by a metre. This was a big problem with the flexible stock carriage because overtightening one corner would pull or push all of the others, which is why Wanhao instructs users to screw the bed down as far as it will go, and then raise all of the corners by the same amount.
The logic is that moving components should be as light as possible so that it takes less energy to get them moving and they don’t bleed lots of energy when they come to a stop, and static components should be as heavy as possible so that it takes more energy to get them moving (i.e. vibrating). I previously had the printer sitting directly on a yoga mat, which did reduce the severity of the vibration; wall thickness of the TEST object had an average deviation of 10 microns when the printer was on top of a yoga mat, versus 17 microns if it had rubber feet and was on the desk. However, I didn’t like how the entire printer sank into the mat’s surface, it just didn’t gel with me.
I eventually found jordan.c’s suggestion to put the printer on top of a paver, which tantalisingly hinted, “Prints, noticable better” but without actually showing comparison photos of the prints. I didn’t do any direct comparisons myself to see whether the paver made a difference to print quality or not — and the thing weighs 25 kg so I don’t really feel like moving it right now! — but it helps with stability and noise, and that’s good enough for me.
Finally, I noticed that the extruder was having more trouble feeding plastic than it should, so I took the thing apart and did my best to clear it. It misfeeds less often now so I must have fixed something, but I don’t think it was a big problem to begin with.
My old enclosure was a cube that I made from 12 mm plywood, lined on the inside with 10 mm thick foam from an exercise mat. The front panel was attached via magnets and completely removable, which was a great idea that I’ll be implementing again. It was sized to fit my Printrbot Simple Metal and although my larger i3 Plus fit inside it, it was getting real cramped in there with all of my abubots:
Most importantly, I couldn’t access all sides of the printer because the enclosure was a solid box, which is a real pain for a machine that I bought as a tinkering project.
I started designing a new plywood enclosure with drawers and shelves and everything, built around 20 mm square tubing joined with printed connectors, but the cost for just the structural parts of the build (tubing and plywood) was already approaching $200. By comparison, the usual IKEA Lack enclosure uses two Lack tables for a grand total of $16. I do dislike having the same enclosure as thousands of other people, but you can’t argue with that price difference — especially when that money can go into nicer features like full acrylic sides, or an endoscope, or improved lighting, or adjustable ventilation grilles.
The running cost so far is:
2 × IKEA Lack tables $ 18.00 1 x 42×42×1200 timber $ 8.00 8 x Corner bracket $ 5.60 1 x White spray paint $ 3.10 1 x Magnetic catch ×20 $ 18.20 ======= $ 52.90
The timber is for extending the legs because I don’t feel like waiting 12 hours for 15 cm extensions to print (sadly, they did not have 50×50 mm to match the Lack). The brackets were used to reinforce the legs of the Lack table at the bottom of the stack, stopping it from wobbling under the printer and helping it deal with the 25 kg paver (which I took from my backyard). The magnetic catches will be used to hold the side panels onto the enclosure so that all of them can be removed for access to the printer. The top table is also completely removable using the couplers from the Tinker-Friendly Lack Enclosure project; the absolute ledge included the Fusion 360 file so I was able to easily modify it to mate with my 42 mm risers, which I will paint and glue on.
I haven’t really decided on what the side/back/front panels will be made out of. I have some options:
Having the printer fully visible with clear acrylic sides is attractive, but one of the features I liked about the old enclosure was that it could be shuttered to make the box nearly light-tight. That’s nice for night-time printing when I still want to film time lapses. I’m also concerned about acrylic’s ability to keep heat and noise in the enclosure, and of course there’s the price.
MDF is starting to look good. It’s nice and dense too, I guess, although cutting it might be a pain.
I test-fitted the Lacks together and found that my risers are about 5 cm too tall. Getting rid of that 5 cm might make printing the risers entirely more bearable.
I’ve temporarily removed the paver to see if I can get away with not using it, which would save me another 5 cm off the risers. I’m printing TEST and comparing the vibration that’s left behind, but so far it’s looking like the paver is having a positive effect on print quality after all. I’m finding it hard to tell the difference in noise, however.
Yup, the paver improves print quality a liiiiiiittle bit. It seems like ringing is reduced a bit and the layers lie more evenly. So the paver is staying.
I bought a $4 picture frame from Kmart and started printing on it. I clean it with acetone before each print and the plastic sticks without warping as long as I have the bed perfectly levelled — which is easy with the new Y carriage.
I don’t really like the bottom finish that is left by glass (it’s too shiny unlike the rest of the print), but the biggest advantage of glass for me is that the print releases itself as the bed cools down, which means that I don’t have to wrestle it off, which means that the bed doesn’t get jogged out of level.
So I cleared the nozzle for the first time and the experience was uniquely bad after coming from the Printrbot Simple Metal’s Ubis ceramic hotend. The PTFE liner was all in the way and it put up a fight before I could pull it out.
But what’s truly bad about it is how wimpy it is. Now that my printer is dialed in and I’ve gotten a good variable acceleration and jerk profile, I’ve been trying to push faster printing speeds but the hotend just can’t keep up. I hear the extruder motor clicking and failing to extrude all the time, because really the only part of the hotend that is actually melting the plastic is the nozzle, and everything in the PTFE liner stays basically solid.
I bought a Micro Swiss all-metal hotend from The 3D Printer for AU$85. Why the Micro Swiss? Because it’s the kind everyone upgrades to, so I can just follow someone else’s instructions! An all-metal hotend will also let me experiment with tougher filaments like the IGUS stuff I was talking about in the previous post.