We are going to fix my toilet’s clean water flush mechanism – with technology. There’s a little plastic part in here that connects the push plate to the flush mechanism, and that’s broken off.
These German flush boxes are all neatly built into the wall, but that means for replacing broken parts, you are dependent on the manufacturer to still make that one exact part. So I thought, why not use this as an experiment to see which would be faster – simply using CAD to draw up the part and printing a new one, or using 3D scanners to just copy a functional one?
We have the clip at the bottom, which serves as the pivot point. The pushrod, when activated, pushes down on the flush tube at the back. This is a complex component, and recreating it in CAD will require some effort, I believe. But let’s get started with the 3D scans first, using the EinScan-SE.
EinScan-SE
I was pleasantly surprised by how quickly the hardware setup went, but things took a turn when I tried to get the software up and running. The EinScan scanners require activation and licensing, which can be a bit of a hassle. I’ve found the activation process to be finicky in the past, and this time was no exception. To download the software, I had to sign up for newsletters, and then I discovered that my existing activation file wasn’t compatible with the current version. Instead, I needed to link my device to an online account – which I didn’t have yet. The problem was, their online activation system wouldn’t recognize my scanner’s serial number, leaving me locked out.
The hardware itself is fine, but the software issues were holding me back. Troubleshooting this would have taken up a significant chunk of my day, rather than the few minutes I was aiming for. Unfortunately, this is a common downside of using proprietary hardware like 3D scanners – they can be pretty restrictive.
You know, calipers? These? Nobody can lock me out of these!
3DMakerpro Lynx
Let’s hope this second one goes better, this is the 3DMakerPro Lynx. It’s a totally different scanner approach. It’s a handheld scanner, and I’m not sure if it’s going to have the detail, but we will try anyway.
And here, the longest part of the setup was just screwing in the locking USB cable, but with that, we end up at the same 1-minute and 20-second mark.
I then spent about a minute setting up the part so that I’d have access to it from all sides, it needs to be raised off the ground so we can scan the underside as well. Plugging it in, the software booted right up. However, there is a little problem: It is not picking up the red part. It’s kind of overexposed, underexposed, but no matter what I do, it does not scan. It might be because this is an infrared scanner, that it doesn’t work with the red. I am going to grab the scanning spray, but unfortunately, that does mean masking up and wearing some gloves because this stuff does contain some nasty solvents.
I’ve also tried to add as much random stuff to the area as possible to give the scanner an easier time to track.
This spray is like a very light foam that is dusted onto the parts. The scanning process itself is rather simple, you just orbit around the part a couple of times until you’ve covered all the possible angles. In my case, I did 3 separate scans from 3 different positions, which took me about 10 minutes. Then you take those 3 scans, align them, and combine them into one big dataset, which takes another 5 minutes of processing time, and then you export that point cloud into a surface model, which is another 5 minutes.
This is actually better than what I was expecting. It’s picking up all the details, and detail on the knob is pretty okay. There is a good amount of noise in the scan, and it looks like there’s some stuff going on at the edges, but overall I think the shape is ok.
Meshmixer
Next, I imported the scanned file into Meshmixer to get it ready for printing. While I could have done some post-processing in the scanning software itself, I found it easier to work with Meshmixer’s more powerful tools – even if it meant dealing with larger files. Using Meshmixer’s sculpting tools, I was able to tweak the shape of the clip area, which is a spot where 3D scanners often struggle to create accurate geometry. By making some adjustments, I was able to get the shape just right. From pulling the scanner off the shelf to having a printable part, the whole process took me just 33 minutes – a great start for a quick test print.
CAD
Now I’m going to try drawing this part in CAD. Only three features on this part matter, that is the hook, the square peg, and the fin.
For the other parts in between, I could potentially estimate and not measure, as long as the three key features end up in the right spots and are shaped correctly. However, I’ll opt for precision and measure everything to be fair to the scan.
Most of the features were pretty straightforward square features that you can pick up with just a pair of calipers, but there were some angles on the part that were a bit harder to measure, so I traced the part onto a piece of paper, extended those lines tangentially and then measured the angle of those. The total time spent here was 33 minutes and I checked the footage: It is down to the minute the exact same amount of time I had spent on the 3D scan.
Slicer
Both of these parts are now ready to drop into the slicer. I did a quick check to make sure they are roughly the same size and they matched up really well.
You can see that the original and the CAD-designed parts are nearly identical, I can put them on top of each other and the dimensions are pretty much spot on. The scanned part is more like an “homage” to the original. All the features are there, and when considering the proportions, it appears slightly wider, but all the dimensions should be quite accurate.
Fitting
First of all, I made a 1:1 replica of my flushbox’s geometry that the printed parts would need to mate to. It is basically just a 6mm rod, 12mm wide with a bit of supporting geometry around it. I added it in as close to the original as I could, but the only thing that matters is the 6mm rod and how wide it is.
When I try to attach the original part, it fits perfectly and swivels smoothly with no resistance. However, when I use the CAD-designed part, it clips on but binds up. The issue lies in an error I made while taking the dimensions from the original part, specifically at the small nub at the end of the clip. Initially, I simply extended the circle of the clip and centered the nub’s circle on top of it.
But if you look closely at how this interacts with the pivot point it’s supposed to clip onto, you can see that there’s interference between the two parts. As a result, the clip will spring open and bind. To fix this, I should have placed the nub on the outside of the circle, so that it’s tangential to the inside of the circle, allowing for a smooth fit.
I changed that, and this CAD part is ready for its second revision.
Now our scanned part: It does fit surprisingly well, it does clip on, and it swivels. Overall though, it is still too wide and that clip inside diameter is too tight, so I’m going to fix those two things.
And that brings us to the last thing that I noticed, and that’s that the part that we were taking measurements off of is also… broken. Somebody actually snipped off a side rod that’s supposed to be there.
In CAD, this was just a couple of seconds to add that back in, because all the geometry is already there that I can work off of. For fixing the features on the 3D scan part, I tried Meshmixer and Blender, but ultimately I ended up using PrusaSlicer.
It doesn’t have the most sophisticated tools to modify and work with these meshes, but it is very fast and easy to use. I couldn’t figure out how to do a boolean intersection, so I used the cut Z tool. For adjusting the diameter of the clip and for adding that rod, I added two cylinders and had one as a negative and one as a positive volume.
And with an extra round of printing, both parts work!
Which one was faster?
Actually, the 3D scanned version was faster, even though it was quite a different process than doing the CAD. In CAD, it’s one long workflow and pretty flowy once you get into it. You take your dimensions, create a sketch, create your feature, and you just keep adding on features until your part looks exactly the way you want it to. With 3D scanning, it’s more like a stages process, so there are half a dozen different tasks that you have to do in the right order and you have to be fairly competent at every individual task. Between part prep, the actual scanning, aligning parts, and post-processing the meshes, it’s quite a wide skill set that you need to have to get parts that look good. Overall, I enjoyed the CAD drawing process more than the scanning process, because I feel like I have more control with CAD. With scanning, it’s more like the hardware and the circumstances decide a lot of what your result is going to be like. But also, I must note that the part that we’re designing is actually pretty well suited to being drawn up in CAD, with all straight features and linear extrudes. Once you have something that is a bit more involved it would be a bit tougher to draw in CAD. There, the scanner would have an advantage when it comes to creating parts that are accurate.
But then the question popped up for me, is this actually how scanners are supposed to be used? Because the scan results are okay, they’re usable, but they’re not up to the standard of having a smooth surface, and crisp features. This is okay, but not super high-detail. I think, scanners at a budget level are better suited for creating references, either with a basic scan for importing into your CAD tool to take measurements or as a backdrop and then trace around it.
Scanners that can handle larger scanning areas would work well for how, for example, SuperfastMatt uses his scanners, and that is to create a reference environment that he can then freely draw his designs in. For one of his project cars, he took off the wheels, scanned the entire wheel well and all the mounting points, and then in CAD, he just drew up his own features, his own suspension parts, right in the context of the scanned reference. I think that is a much better use case for these, and for that, the quality that you get is way more than enough. This is like millimeter accuracy, and for that sort of stuff, totally suitable.
So while I did prefer the CAD and calipers approach, having a scanner at your disposal is very useful even if it’s not for directly copying a part.
Topology Optimization
Of course I couldn’t leave well enough alone, so I made a topology optimized flush lever. This thing’s never going to break. 😉
Check out @SuperfastMatt ‘s video on YouTube
Download the files on Printables
Scanned with the 3DMakerpro Lynx
💙 Enjoying the videos? Support my work on Patreon!
Product links are affiliate links – I may earn a commission on qualifying purchases (at no extra cost to you)