@NeverDie, I just uploaded my model here: Another Shrinkage Compensation Model by Alex_vG - MakerWorld
1 Like
I stumbled across this, and it’s interesting because it takes less than 11g of filament to print it at 150x150mm:
https://www.printables.com/model/362362-xy-calibration-cross-150x150mm
Edit: Makes me wonder whether maybe some kind of non-obvious generative design could produce the optimal print for this purpose? i.e. strong and rigid enough to take accurate measurements, but using the least amount of filament to accomplish that.
Ah, looks interesting too. I’m wondering if it is stiff enough. Did you try that too?
You see, this is the kind of frivolous model that I believe should be flagged on these model sites, allowing us to filter out such garbage.
So, what did the uploader prove? That they can create a pretty picture in CAD? There should be a setting within each of the upload sites where one could filter models based on whether they are “Theoretical,” as is the case here. What’s the difference between theoretical and real? It’s simple: post a photo of the output model on the upload proving that the model can indeed be printed.
If you go back to the original post that this Printables model was uploaded there isn’t any more information there either.
This uploader, FUSIONPAUSE has a link to an empty YouTube page. Again, all a waste of time and I wish that these upload sites had a way to block bad uploaders.
So if I were to give this uploader the benefit of the doubt, I can only guess that perhaps this person has a descriptive post somewhere, such as Reddit, and was simply using Printables and/or Thingiverse as either a file sharing repository or, in the case of Printables, to earn points for freebies. Either way, in my view, this model is a waste of pixels and bandwidth if it doesn’t provide instructions or any statement for its intended purpose.
Now in contrast, let’s look at this calibration model which popped up underneath the Thingiverse listing for 2434692. This model exemplifies a quality post intended to serve the 3D printing community.
What makes this model great? First, the uploader gives credit everywhere. He states that the original Gcode is from Ender. He also prints out the actual model and shows the photo on the build plate which further enhances understanding of how the model is to be uses. It also proves that the model has actually been printed, a sort of “proof of life” exercise. He also includes a statement indicating the models calibration goals and then last but not least, includes links to the YouTube discussion demonstrating how to use the model and what one might derive from its use. And finally, he includes a GitHub link with the Gcode. To me, that’s a quality post. 
Those are the gems I bookmark.
I guess this topic really strikes a nerve because of all the bullshit garbage that one has to wade through in order to find genuine content. I hate it when I have to wade through it on Google and Reddit, and I don’t like it any better when I see it on sites like Thingiverse. Makerworld, with its allowance of spamming, has predetermined its trajectory as a low-quality repository for precisely this reason. That’s not to say that one can’t find any gems on that site, but one has to wade through a sewer of garbage to find them not to mention the force-feeding of content as Bambulab feels you should consume it.
2 Likes
I can see I’m out-voted on this, which is fine: I’ll go with the majority opinion. I just thought it was an interesting idea that doesn’t use a lot of filament. I wasn’t looking at it as necessarily anything more than that.
In practical terms, we know that @Alex_vG 's model is meant to average inner and outer measurements to compute the difference between that and 150mm, and he provides a convicing argument for why that is the better way. The model you didn’t like probably does outside dimensions only, though that could be corrected if someone wanted to re-model it from scratch. Either way, I don’t claim it to be optimal, but the idea of using the long, skinny, parallel walls of filament might hold some merit as a way to avoid wasting filament. I didn’t do time modeling things with popsicle sticks, so maybe it has no merit and it would be obvious if I had. But since I didn’t, I just can’t tell for sure purely by looking at it. For instance, don’t some beam calipers take that approach? At root level, it’s kinda the same idea, isn’t it?
All in all, I’m surprised there isn’t more general interest in measuring filament shrinkage as part of calibrating each filament that you use. For most people, it seems it’s not even on their radar.
1 Like
I can’t quite figure out if what you said here indicate you are aware of this or not, but Orca actually has exactly this setting in the filament profiles. If you measure 99.92 mm for a 100.00 mm test, you’d set it to 99.92 - simple as that.
On a random note I’ve had the idea to print two test lines/walls, high enough for a good caliper reading, 150 mm apart (actually if they are 0.42 mm wide they’d be 149.16 mm apart) or rather four of them, one set for X and one for Y, and then measure after cooling but while they’re still adhered to the build plate (using the Cool plate would probably be a good idea). This should mostly show the printer’s own error rather than shrinkage. It might be nice to correct that with M290.2, and only then measure filament shrinkage and compensate for it with the mentioned Orca setting.
[Edit: Corrected M292.2 to M290.2, typo in original thread]
1 Like
Yes, that is exactly what I plan to do. I was trying to explain why I was searching for a model to take the measurements from (and consequently why others should want to do the same), but apparently I did a poor job of explaining.
1 Like
I followed this thread very closely. And I’m wondering, why the filament vendors do not publish this data. They should know best. That’s one of the reasons why I like to use the filament from Extrudr and it’s my number one filament for technical prints. They have the shrinkage in the technical data documents of their filaments. The other filaments I’m using mostly for fun and deco prints, where this doesn’t matter. 
Maye not a good strategy. The build plate will shrink as it cools (the coefficient of thermal expansion of the build plate material would tell you by how much), and so even just that alone would throw off the measurement even if the print stays attached to the build plate throughout the cooling. You could estimate by how much and then add that back in, but ideally you want hard numbers and so you don’t want to be estimating anything.
I agree. I think the filament vendors who sell to commercial buyers probably have to do exactly that if they want the business. The guy on youtube who speaks for slant3D, which claims to be the biggest print farm in the USA even says that unless you have the detailed datasheet for the filament you’re printing, right down to its exact chemical composition, then you really have no idea what the hell you are printing with. I’m paraphrasing, but that was the gist of it. Of course, someone like him may have the buying power to demand that info, and get it, while you or I do not, so maybe that partly explains it. Also, if you’re running a big print farm, you legitimately really need to know, because any variance could completely disrupt your production, not to mention repeatability and uniformity of parts produced.
The same guy, by the way, thinks filamints are way overpriced and that a reasonable retail price is about $10 per roll for something like PETG, because the plastic itself is used in so many things in industry that it is literally a commodity.
True. If trying that strategy at all, I should probably print on PEI and measure while still hot. And I’m not sure the whole idea has any merit anyway - but if nothing else, it would be interesting to see how much of the error is from the printer’s mechanics as opposed to filament properties.
1 Like
I did the measurements today using @Alex_vG 's model and method for Overture CoPa, and the shrinkage compensation factor came out to 99.91%. Very little shrinkage. 
IMHO, CoPa is an under-rated material. I almost never hear of anyone using it, yet it’s easy to print, and now I know it also doesn’t shrink by much. 
2 Likes
Cool 
I have seen Polymakers CoPa and didn’t pay much attention. It’s not nearly as cheap as the Overture one. I will keep it in mind when I need the properties. Thanks for pointing it out!
1 Like
Short follow up: I have uploaded a new version of my model to make taking measurements a lot easier. I always would have needed 3 hands for my old one or CaliFlower. This new version has guides so that you can just throw it on your calipers and you have both hands for handling the calipers. I also reduced the weight to less than 20g.
Have a look: ShrinkingTest V4 by Alex_vG by Alex_vG - MakerWorld
3 Likes
Worthy of note: Having tried it, it turns out that Overture CoPa is rather “bendy”. i.e. not as rigid as, say, PLA, or other non-TPU filaments. I hadn’t seen it mentioned in any of the reviews, so it came as a bit of a surprise. Probably everyone but me already knew this, but, if not, then worthy of note.
Thanks for updating your shrinkage model! I view it as a necessary tool when printing parts that need to fit somebody’s else’s accurately dimensioned parts, such that holes align correctly, etc.
P.S. Thanks for including the measurement photos. I think maybe I had been taking the measurements upside down in the prior model!
1 Like
@Alex_vG Thanks for the great tool. I’ve used it quite a lot, and I’ve come across one small issue with it that may only affect some users, but I thought I’d pass it along to the originator.
The issue has to do with the beveling. Both top and bottom are beveled. In principle, this is a good idea, as you don’t want, say, elephant foot or other potential first layer or top layer artifacts throwing off the measurement. I get it. I would keep the top beveling as is, but I’d suggest maybe (?) the bottom beveling to not start until maybe a few layers up. The reason: I’ve been experimenting with using a 0.3mm slicer layer height for the first layer rather than 0.2mm, as it seems to give a more perfect first layer. I keep subsequent layers at 0.2mm Unfortunately, it throws a bit of wrench into this model. A 0.2mm first layer plays nice with the model, but a 0.3mm first layer height seems not to. The inside wall of the holes doesn’t attach to the build plate like it should. This is in the beveled area.
Like I say, it’s a small thing. What do you think? Now that I’ve identified the issue, the short-term workaround is just to set first layer to 0.2mm.
I think I haven’t fully understood the problem yet, but I will try to remedy that. To replicate the problem, I just take my print profile and set first layer height to 0.3 instead of 0.2mm?
1 Like
AFAIK, yes. I’ve tried it that way now on a couple different printer models with the same result.
BTW, I’m using Orca Slicer for the slicing. Sunlu HS-PETG.
OK, I found another way to overcome it, which is to slow down the first layer print speed by a lot. So, maybe this is one of those gray zones where one could argue it’s not a model problem but a slicer setting problem. I’ll post some pictures when time permits.
The other issue with using a 0.3mm first layer height is that it gives only one layer of webbing at the bottom instead of the two that one gets with a layer setting of 0.2mm for both first and subsequent layers. but, again, I suspect I could overcome that by maybe forcing another layer of webbing in the slicer settings–I just haven’t tried it yet. Anyway, maybe one layer of webbing is enough, so just an observation at the moment.
Again, at 0.2mm first and subsequent layer height, the model is sound and works great, so I hesitate to even bring this up. I’m not really sure what the “right” perspective is in which to frame the problem. Perhaps there is no one right answer at all.