How to correct tilted verticals
I am designing a part of a mechanism that requires two metal rods to be inserted and to end up parallel.
In the above image, a 16mm rod fits into the tube on the left - it's a tight, no wobble, friction fit. A 12mm rod fits into the right hand end of the arm - the fit is slightly looser, to allow up-down adjustment via the two pinch bolts - it's nevertheless quite an accurate fit even before being clamped.
I'm printing this in ABS - 4 perimeters with 40% cubic infill, to give the part sufficient rigidity. I print it in the orientation shown in the photo.
The problem that I'd appreciate some advice on is that the two rods aren't perfectly parallel when fitted into the printed part. In the photo, let's call the face that the part is resting on the bottom, and the rods extend in the positive Z direction, i.e. 'up'. I measure an approximately 0.25 degree 'splay' - the tops of the rods are very slightly further apart than the bottoms.
I suspect that this isn't anything to do with inaccuracy in the printer, and it's more to do with the shrinkage of the ABS as it cools. The normal problem with ABS is that it may lift from the build plate - if it did that I might expect the tops of the rods to be closer than the bottoms due to the horizontal part 'curling up'. But bed adhesion is very good and there's no sign of lifting, and my distortion is in the opposite direction. If it's due to part distortion during cooling it would be consistent with the part 'arching' slightly, rather than 'curling'. (I know those descriptions aren't strictly sufficient, but hopefully you get my drift)
I've printed the part a few times and the result is quite consistent. One simple fix would be to rotate the axis of the right hand hole by the measured 0.25 degrees to compensate for the expected distortion. But my reason for the post is to get advice on whether there's anything else I can do to fix this. Shrinkage compensation in the slicer is unlikely to correct this tilting of the verticals I think. More perimeters and more infill are likely to make it harder to print without lifting, and I think it might exacerbate any differential shrinking. Can I fix the part after printing by annealing? Or simply by heating and bending it with the rods in situ?
I do have some CF filaments - some PC-CF arriving today and some PETG-CF, so maybe I should just try those to see if the distortion is any less.
RE: How to correct tilted verticals
CF filament would definitely help to reduce shrinkage. It looks the problem here is that when the left tube upper part is printed the base part starts to cool down and shrink producing something like this \__. I guess you could try to split your design in two parts and print the left tube separately, this will require slightly widening the base so the left tube can be later inserted there (and glued if needed).
RE: How to correct tilted verticals
Thanks, yes, I'm about to try a PC-CF print to see if the problem goes away when the shrinkage is less severe. Just running through a couple of calibrations with the new filament.
RE: How to correct tilted verticals
How about just laying it on the side, if the holes for the rods prints horizontally shrinking should at least not affect them being parallel. You may need supports though.
/Anders
RE: How to correct tilted verticals
If you need the rods to be parallel to (significantly?) better than 0.25 degrees, the clamping distance in the right-hand hole seems rather short. And the unsupported, somewhat thin tube on the left-hand side might not be stiff enough, depending on the length of the rod and the forces acting upon it?
To phrase it another way: Even if you can reduce the initial deviation after printing and assembly, how sure can you be that the rods will stay parallel within that tight spec later in operation?
RE: How to correct tilted verticals
I designed and noticed something similar with a part that I initially printed in PETG-CF, although on a much smaller scale. My part has 2 x 4mm steel dowels fitted after printing. It's very hard to measure the deviation with the tools I have available, but just looking at it there it is just about noticeable that the dowel pins aren't quite parallel.
They're close to parallel but not spot on. The part couples with another part that houses 2 bushings, but inserting and removing them should be with little resistance, but I'm feeling some resistance, not enough to stop the parts coupling, but with more resistance than is ideal. Again Im putting this down to the dowel pins not being quite parallel.
I have no idea what's causing it or how to solve it, I've been over my design countless times and everything there is spot on....so I'll follow this thread with interest. I will also try a print with PCCF and see if the result is different from PETG-CF
Speaking of interest, if the rods you are fitting are 16mm, what is the diameter of the hole in your design? Is there a certain spacing % you use to allow for tolerances?
I've been designing as a hobby for a few years, but with stuff like this I've always winged with trial and error to get nice fitting part... I aim for a nice friction fit so I can avoid adhesives and salvage hardware from my earlier prototypes.
RE: How to correct tilted verticals
Thanks for all the suggestions - much appreciated.
I should explain the purpose of the part. It's for a circular sock knitting machine (CSM - no, I don't know where the 'k' went either). It's what's called a 'ribber' arm - its purpose is to suspend a circular ribber disc of ~100mm diameter just above a cylinder of a slightly bigger diameter (it's shown upside down in the image above). The ribber disk and the cylinder both hold knitting needles that have to open and close at the right time as the machine is cranked. The mounting point of the ribber arm rotates around the fixed cylinder when the machine is cranked, and it's very important that the position of the ribber disc stays still with respect to the cylinder. Any off-centring or misalignment of the verticals results in the ribber disc doing a 'hula' as the arm rotates. This creates a varying gap between the cylinder and the ribber disc, which causes all sorts of problems when the machine is knitting ribbed socks. It's for my sister's CSM, and I've now developed good models for the cylinders (with different needle counts), the ribber discs (with compatible needle counts), and the ribber 'tappet' (which pushes the ribber needles in and out). Those parts are all working well with the original machines. Getting the ribber arm to provide this perfectly static positioning is the final tricky bit.
The 16mm rod locates into a hole in the base of the machine - it has to be a good wobble-free fit, so I've had to source a rod with a tight tolerance on the diameter. I've settled on bearing rod, which is a very nice fit - just as good as the original cast iron ribber arm. The other rod is actually 14mm (not 12mm as I said previously), and its tolerance is less important - I'm using aluminium tube so that I can fix a base plate to the bottom to support the ribber disc and tappet.
How about just laying it on the side, if the holes for the rods prints horizontally shrinking should at least not affect them being parallel. You may need supports though.
I did try that previously, but the holes didn't print as accurately as they do when printed on the face shown. Fit is important in this application.
If you need the rods to be parallel to (significantly?) better than 0.25 degrees, the clamping distance in the right-hand hole seems rather short. And the unsupported, somewhat thin tube on the left-hand side might not be stiff enough, depending on the length of the rod and the forces acting upon it?
To phrase it another way: Even if you can reduce the initial deviation after printing and assembly, how sure can you be that the rods will stay parallel within that tight spec later in operation?
The clamping distance on the right is 20mm, and I've found it to be a very solid fit in practice. I tried a couple of tests - I tried squeezing the ends of the rods together before tightening the clamping screws, and I tried pushing them apart, but after clamping the results were always the same. I take this to mean that the clamping enforces the alignment rather well.
The wall thickness for the taller tube is 2mm, and prints entirely with perimeters. The hole for the 16mm rod extends into the horizontal beam to a depth of 10mm. You're right about it needing to be rigid, which is why I was favouring ABS over PETG, with plenty of perimeters and infill. The whole part, with the rods inserted, does feel very solid - I can force a little bit of flex, but it takes a lot more force than the part will experience in use. There should be little to no lateral force when in use - maybe a little bit of 'pull' from the knitted sock+weights below the cylinder, but that load is shared with the cylinder needles as well, so it's not really significant.
The 0.25 degree misalignment can cause an offset of almost 0.5mm at the end of 100mm, which means the gap between the edge of the ribber disc and the top of the cylinder varies by nearly 1mm as it 'hulas' around (+ and - 0.5mm).
Yesterday I printed the ribber arm in PETG-CF and the result is definitely different. If anything the verticals are slightly tapered together rather than apart as they were with the ABS, but it's difficult to measure. It tells me that the effect is definitely related to the material. The PET-CF is every bit as rigid as the ABS I had been using (that CF makes a BIG difference when the base material is as flexible as PETG), perhaps even more, and being a slightly lower temperature material I was able to influence the alignment by immersing it in boiling water and forcing the ends of the rods together or apart. I think this may be the way to go now in fact.
They're close to parallel but not spot on. The part couples with another part that houses 2 bushings, but inserting and removing them should be with little resistance, but I'm feeling some resistance, not enough to stop the parts coupling, but with more resistance than is ideal. Again Im putting this down to the dowel pins not being quite parallel.
I have no idea what's causing it or how to solve it, I've been over my design countless times and everything there is spot on....so I'll follow this thread with interest. I will also try a print with PCCF and see if the result is different from PETG-CF
Speaking of interest, if the rods you are fitting are 16mm, what is the diameter of the hole in your design? Is there a certain spacing % you use to allow for tolerances?
I've been designing as a hobby for a few years, but with stuff like this I've always winged with trial and error to get nice fitting part... I aim for a nice friction fit so I can avoid adhesives and salvage hardware from my earlier prototypes.
I wonder if flexing the part in boiling water may help in your case too.
I'm with you on the trial and error. The modelled size depends on the material I'm printing with. I have a small part that interlocks with the main ribber arm that I use as a test piece. With the ABS I had to model 16.15mm to get a snug 'grippy' fit on my highly accurate 16mm rod - that seemed to take account of the tendency for holes to print smaller AND the shrinkage of ABS. With the PETG-CF I had to go up to 16.2mm, despite the lower shrinkage rate, and I put this down to the slightly rougher surface.
RE: How to correct tilted verticals
Another crazy idea would be to cater for the error by design, simply tilting the 16mm rod say 0.25 degrees in the design. I am too new to 3d printing to know if the printer is accurate enough to replicate the design though.
/Anders
RE: How to correct tilted verticals
Another crazy idea would be to cater for the error by design, simply tilting the 16mm rod say 0.25 degrees in the design. I am too new to 3d printing to know if the printer is accurate enough to replicate the design though.
Yes, I considered that, and I still haven't ruled it out. It would be a model that is then specific to how I know it will print in a certain material. Provided it's always me printing it, always with the same material, that would be fine.
I was about to try that yesterday (my OpenSCAD model now includes that tilt as an option), before I decided to print it in PETG-CF, and I'm rather favouring that material now because I can manually adjust the alignment in hot water.
RE: How to correct tilted verticals
What doesn't help in my case is my steel dowels are the cheapest ones I could find on Amazon, and my bushings are...yeah I think you can guess the rest 😂
RE:
I made an alignment jig, in preparation for correcting the alignment using boiling water. I think I've found where the problem lies.
The smaller clamped tube is actually nicely perpendicular to the horizontal beam, but the 16mm rod inside the printed tube is not - it leans out, as shown by the tapering gap in the photo.
The dark blue plastic is PETG-CF, so I'm hoping I'll be able to permanently correct the alignment by heating it in boiling water. But first I'm printing one in some PETG-GF that I had delivered today, to see if this misalignment is repeatable, so that I can maybe compensate in the model.
RE: How to correct tilted verticals
I realised that the above dark blue PETG-CF was the print that I'd been using to experiment with boiling water realignment, before I printed the alignment jig, so that distortion possibly wasn't there before.
I tried the Tinmorry PETG-GF - it really is gorgeous stuff. It printed perfectly with the 'Generic PETG' profile, and with a default extrusion multiplier of 1.00 it seems dimensionally accurate. And most importantly, the two rods in my design appear to be sufficiently parallel, as far as I can tell with my alignment jig. The test will be when it's fitted to the machine, but I'm optimistic this time.
The print finish is a lovely semi-matte sheen - a product of PETG's natural shine and the slight roughness caused by the GF filler - and it comes in loads of lovely colours, unlike most CF filaments. Nice price too. This is Frosted Red.
RE: How to correct tilted verticals
I've got Tinmorry PETG-CF and as I've mentioned before, I'm a massive fan....never tried their GF. It looks absolutely great and is cheap enough that I can use it as my daily driver for prototyping etc....I think I'm on my 8th or 9th Kilo of the stuff. Tip if you buy on Amazon, about once per month they'll drop the price even further, last time I bought the CF I got 2 x 1Kg for £34, which for filament of this quality is unmatched IMO.
I'm gonna try Tinmorry PETG-GF now, you've got me intrigued...have you had any issues with splinters? Only GF I've tried is Sirayatech ABS-GF and I have to rub all printed parts down with a block of silicone before handling, glass splinters are really hard to remove from skin 😂 Problem is I love the ABS-GF. It looks great and parts are damn strong, as well as dimensionally spot on....half my E-bike is held together with the stuff !
RE: How to correct tilted verticals
I had your earlier recommendation for their PETG-CF in the back of my mind when I saw their PETG-GF. It was the greater choice in colours that caught my eye, and the price helped too. I've not noticed any splinters yet. With the generic PETG profile there was zero stringing, except for a tiny whisp inside the tall cylinder - the rest of the print is just smooth and rather pleasant to handle. The finish hides the layer lines nicely.
RE: How to correct tilted verticals
Not sure if you're interested but there are profiles available on their website (for MK4S), I presume for Orcaslicer and Bambu as they're in Json format..., I find it interesting to see what they recommend wrt retraction/EM/cooling etc.
RE: How to correct tilted verticals
Ah, I hadn't spotted that. I did notice that their suggested nozzle temperature is quite a bit higher than I used (260 versus 240), but my prints look great. They also suggest drying before first use, but I saw no signs of damp-related problems, despite the red roll being an Amazon Resale 'used' purchase that turned out to have been opened, as in the sealed bag had been cut open (at the wrong end!). Despite all that, the prints still look great. I think it's quite forgiving.
RE:
I just checked what I use for PETG-CF and its 250, I think they use 260 in their orca profile, but I find with so many factors that can play a part as long as I'm in the same ballpark and I'm happy with the results then its all good.
This stuff is ridiculously tough for PETG, normally I can work a screwdriver between layer lines and crack the part so I can recover hardware from old (or more commonly failed😅) prototypes, I just can't with this stuff....and filled filaments are meant to have bad layer adhesion !
RE: How to correct tilted verticals
I'd consider annealing the parts here - that can ease stress and help things go back to "normal", among other things.
GF can be more colors because the fibers are clear. Might be worth exploring ASA-(C/G)F here as well (or even PA). Really depends what you need out of the part.
RE: How to correct tilted verticals
From what little I've read about annealing, I understand that some distortion is likely from the process, isn't it? PETG maybe suffers less than other materials in this respect, and the GF will help too, but I want to avoid even minimal distortion.
My main requirement is for the part to have accurate geometry, and to retain it long term. The part won't be under any great stress, except at the pinch bolts and maybe from the weight of knitting and some knitting weights on the end of the thinner spindle. I may consider ABS/ASA-(C/G)F, depending on how I get on with this PETG-GF, but since both those base materials shrink more than PETG while cooling I think it will be harder to get the required accuracy of geometry in the first place. And being higher temperature materials I may not be able to adjust them simply with hot water.
I more or less discounted PLA because, although it's more rigid than PETG in its raw state, it will eventually relax under stress, e.g. at the pinch bolts and maybe under the weight of the knitting, but might some GF or CF in the PLA prevent that?
My feeling at the moment is that the PETG-GF is about right - minimal distortion of the base material while cooling, improved even further by the GF, easy to print, less natural flex thanks to the GF, and some ability to adjust the shape in hot water if necessary.