Tim's 3d scanner hardware thoughts / review thingy

My 3d Scanner arrived! And it’s officially the most awesome one because I had an extra Lulzbot sticker and mine has one now and the others theoretically don’t! (unless they do?)

So this thread is going to be a review / thoughts / things I would change piece, and is going to evolve over time. I am thankful I get a chance to review this unit, so please take any criticism I post here of the unit as constructive to try and improve the end user process going forward.

First off, this thing is awesome. it feels solid, it goes together quickly, and it seems to work just fine so far. I just got it all put together so I’m not really far into the process of taking my first 3d scans with it yet. I’m hoping to scan my little crystal ball there (what, no one says it has to be a translucent crystal, pyrite still counts!) .

So starting with packaging, the unit is very well packaged. The only two things I would probably change on the packaging, is the steel threaded rod clanks alarmingly when you pick up the package. wrapping them in some sort of inexpensive noise dampening packaging would probably calm fears of the unit being delivered broken. The other change I would make is the multi language assembly guides are probably not needed for scanners primarily shipped to a U.S. market. There was enough paper there between them all that you are probably looking at an extra 0.5lb per unit shipping weight. There were a couple minor issues with the manual that I would probably change, some word choices, calling out some of the assembly notes in larger font, etc. They aren’t quite as polished as the usual Lulzbot manual yet, which stands to reason since BQ made them and they were translated into English. A revised lulzbot branded manual might be nice though.

From an assembly standpoint, everything goes together well. The larger threaded rods and the nuts and washers for them do have a bit of a residue on them, oxidation in the case of the nuts, and remnants of the black coating I think on the threaded rods. Purely cosmetic and nothing that affects the function, but also something that someone putting together a precision optical scanning instrument might not expect.

The threaded rods themselves might be worth the time to design out. They are very, very heavy. I believe 20mm x 20mm mitsumi extrusions could be adapted pretty easily to all the places the threaded rod is in use and save at least 3-4 lb per unit on shipping weight. The extrusion costs should be comparable. I plan on creating that very modification here over the next couple weeks to see how it goes. The large steel bearing could also easily be replaced with a lighter unit that would still bear more weight than the crush strength of the PLA baseplate. The design will work just fine as it is, but one modification I am toying with that might be worthwhile is a folding portable unit. With T slot extrusion you would only need 1 extrusion per laser arm, and could make them smaller in width so they would fold down against the main frame, which in turn would fold down towards the rotation plate. A couple latches, maybe some Velcro and a carrying case and you have a scanner you could bring to the customer, for example taking the unit to a museum.

The power supply that comes with the unit as is really isn’t going to work. with the required adaptor for U.S. use its just too tall. Otherwise it’s a standard 12v 1.5a power supply with a standard barrel plug, so it should be pretty easy to get a more compact unit that works. It might be worth opening the packages before they go out from Lulzbot HQ, swapping the power supply out and removing the unnecessary user manuals before they go out

The electronics chamber is the other area I have some concerns about. The pins to mount the main circuit board look really really fragile. A couple of strategically placed heat set inserts and nylon retention screws would probably make me feel a whole lot better about that particular assembly. It would also be nice if the power plug was accessible without taking the device apart, and if there was a power switch. I plan on making a bottom frame piece to cover some of the wires , provide a strain relief point for the two USB cables, and integrating a separate power barrel and switch. Possibly with an inline fuse as well.

The Calibration target was the only piece of the whole thing that gave me any trouble on assembly. The holes in the acrylic plate on mine are off slightly. It’s not an issue though because the friction fit of the base holds the unit in place with quite a bit of force. I would like to see 2 additional holes in the black round baseplate that would line up with 2 posts (maybe M3 screws?) so you could place the calibration screen in the same exact spot repeatably as needed. That or some sort of edge guide

Anyone who read all that, please again, take that in the context of providing asked for feedback. Yes I am being nitpicky here. The unit is solid, is going to work great as a 3d scanner based off other samples I have seen in action, and many of the issues I bring up here are either easily solvable, or can be remedied by modifications after the fact. So don’t let any of this dissuade anyone from getting one of these. But if you put a sticker on it you totally have to put it in a different spot because I called dibs on the front spot!

Folding Anti Tip legs. Also probably a good idea.

The one I put together was missing the power plug adapter. But I had a substitute to use in its place.

Did my first scan right before I came home. now to finish installing the other software to finish getting it ready to test out.

I was having no luck running the calibration until I set the check-box to reverse the motor direction. then it ran perfectly.
NOTE: The camera requires a good light source to function reliably in the calibration wizard. (As I found out.)

Having issues with laser alignment. The point cloud just ‘kind-of’ looks like the scanned object so far. :frowning:

Improved my laser alignment (~5th time was a charm!)and now have a rough looking scan to play with. :astonished:

Attempted my first scans today, There is definitely a learning curve on this, and I don’t think I have one of my lasers quite right yet, because I am not getting really great scans yet. It’s definitely user error, not the device. I was ending up with a weird false “wall” return on them until I figured out that I could narrow the field that it was trying to scan. So far scanning simple geometric objects I keep having weird pock marks in otherwise square surfaces. I definitely think I need to adjust my scanning.

I did find several open source c# libraries to convert a PLY file directly to an STL file. If I can get accurate enough scans dialed in this weekend to where it looks like a model might have a chance to print without toying with it, I may add an "export to STL file option to the Hourus source and see if I can make it work.

I also learned that scanning a shiny object is not a good idea.

Bed position of the object seems to be very important. That could just be mine however. I am considering adding centering and targeting marks to the bed surface for alignment.

The name “Cyclop” may not be a viable scanner name in the U.S. In 2010 Makerbot had a kit called the Makerbot Cyclops (http://www.coolthings.com/makerbot-cyclops-is-a-diy-3d-scanner/) which is no longer on the market. Depending on whether they trademarked the name, it might be too close to use.

There is also a provision on the board and mention in the software of a LDR light sensitive diode add in for future models. It would be interesting to find out which model diode they are intending to use and maybe start playing with those. Camera light levels and contrast seem to be key to getting this thing to work right, and so far I have not figured it out properly, but I’ve only been using it for a few hours too. It is improving though. I can definitely see a need for a better quick start guide and some sort of default calibration scan object to be included to show people “Here is an object, here is a scan of that object with this type of scanner, if your scan looks worse than this one then something is wrong and please calibrate more”

A turn the lasers on to align them button would be useful somewhere in the software. if you skip past that option on the initial calibration run you don’t get to go back to it easily. There are workarounds.

Some sort of back light absorbing screen (matt black felt" Might be a good add on to keep laser reflection from walls to a minimum. Will have to try and see if that helps.

I found black to also have a issue on the scanned object. Sometimes it sees it and sometimes not. But some work with the camera settings may improve that.

I also agree about the manual alignment button. The time it took to line up the laser was bad since I was not really ready to align the lasers the first time I started up the program. :blush:

This post is fantastic, thanks for sharing :slight_smile: There’s a lot in here, but I’ll speak to this part quickly:

Good catch! We have already seen some units from bq arrive damaged due to the threaded rods getting loose from their initial packaging. So, beyond inducing fear, this can be problematic for the equipment.

First, if anyone receives a damaged product out of the box because of something like this, please contact bq using the technical support information on the Ciclop product page on LulzBot.com https://www.lulzbot.com/products/ciclop-3d-scanner-kit-bq

Second, we have communicated this information to bq and recommended changes to the packaging to help mitigate this. So, this feedback is already being incorporated!

P.S. Well done with the sticker placement!

Thanks! The sticker makes it go faster (because stickers make everything faster. Ever seen a race car without stickers on it? )

Another packaging thought if any of these are going to be stored as warehouse stock, a pack of silicone desiccant would be a good idea for the electronics shipping box just in case any of them end up in humid warehouses. They should only add a few cents to the packaging costs.

Heh it’s true about stickers. I’m envisioning LulzBot green octopus tentacles wrapping around the scanner, like flames along the side of a race car… :mrgreen:

That’s good feedback about desiccant. Another thing we’ve discussed with their team is the fact that these units shipped individually, versus possibly shipping the retail box in a larger box with more padding.

To your point earlier, these units are arriving safely and working well upon arrival, we’re really just nitpicking and sharing all the ways to improve 3D scanning together. Packaging / shipping is the easiest thing to start with in terms of feedback, but looking forward to seeing hardware mods and passing along some of the other things you wrote up too.

We passed a link to this thread to their team, and who knows they might even jump in!


wtf who puts parenthesis in their url?

This stuff:

results in more diffuse and less specular reflection of the laser light and better data.

Anything like it works, baby powder, foot powder, etc

note: check to make sure it’s ok to coat the object in developer before covering the object in developer.

What happens if you spray it on a software developer?

Started work on the scanner frame redesign today. I tracked down the STL files on Thingiverse http://www.thingiverse.com/thing:740357 and it looks like there are a handful of existing modifications (http://www.thingiverse.com/thing:740357/#remixes) . I really like the Idea of adding LED lighting powered off the scanner as well, so that might be something I include as well.

I’m tackling the turntabel base today. I’ve got the 20mm extrusion pockets done, but then i got to looking how they were printing that thing, and it would be really difficult to print them as is. They also are using a bunch of support. Since there isn’t a whole lot of torque there, and since I kind of want to put a different bearing arrangement there anyways, I also decided to slice the turntable base in half and mount the upper bearing and motor plate to the lower section using m5 heat set inserts. That will allow both halves of the resulting assembly to be printed without support, and will allow for easier modification / upgrades going forward. I also eliminated the third rod entirely. the motor cable will be ran via the top unused Tslot channel, and covered with tslot way cover plastic sourced from Mcmaster carr at $0.10 per foot. You can Kind of see where I am heading in this unfinished rendering.

For the scanner main body itself, printing that large of an assembly is going to take alot of time, and be prone to a higher failure rate than something composed of smaller assemblies. I still want to try and make it folding, but either way I want to make the overall unit easier to print in several assemblies. to that end, the base will become a separate hinged unit with anchor points for the power, on / off switch and 2 USB ports. The mid section will become a standalone enclosure, probably with a flat back panel for ease of printing. I like the curved one, but that has to be a pain in the ass to print.

The upper section will now bolt on to the mid tower, probably with M5 heat set inserts. This will allow replacing the upper segment to accomodate camera / laser upgrades in the future without needing to reprint the entire body. The arms will become 2omm extrusions on the same sort of hinge as the lower frame. I plan on using lever cam nuts to lock the arms and base in place, along with integral physical travel endstops, maybe a magnet or two. On the arms, the main unknown for me is laser adjustability. Would it be better to allow the lasers to be easily moved along the arm to adjust the aim point, or would it make for an easier user experiance to lock them in place at the fixed end of the arm like they are now? For the laser mount itself I plan on using a clamp style mount rather than the existing single point screw. Building in threaded adjustment points is possible, I just don’t know if they are needed.

The folding anti tip legs will bolt onto the lower frame T-slot. With the amount of mass being removed from the lower assembly, they may become more necessary.

I’ll probably make a separate build thread in here once I get parts further along to separate it from general observations. I need to figure out next which angle and intensity of lighting work best for best scan resolution.

It is nice to see your excitement over your new toy however I think your initial assertions and expectations are far to high. Please don’t take this too critical after all you opened this discussion in the same manner to point out those aspects in which you would change. Might I say it’s an open source project meant for you to build and discover your own process while offering an example of its basic function. Also…, you knew the design you were purchasing before you ordered it so what you are suggesting is more of an “after the fact”, in receiving it. Not so much an unboxing review. Kind of like buying a tv without evaluating it until after you bring it home, or going to the scrap yard picking up some alloy and then telling everyone afterwards it has a bend in it? It isn’t useful to the community. This to me is more so a kind hearted complaint by you that it couldn’t be better for the money you paid. Again, this is what it sounds like in your statements. It could further be argued this should have been a preemptive discussion while researching prior to your purchase and asking others opinion about practical changes beforehand. When looking online at the design I had pointed out to myself (privately) many issues from my own personal assessment including it’s obvious inability to scan effectively…, so I would not purchase one. That scanner was developed for educational and developmental purposes so that an individual could make changes on their own and perhaps later share those ideas with the community if they so choose. The suggestions in using extrusion etc… you could have purchased those parts yourself and built one from scratch using less than 3 meters of material, perhaps 2. Or…, you could’ve approached the concept by using the STL print files to print off the parts to assemble one and evaluate in private before purchasing the hard components? Then you could have interpreted your changes and worked your modifications in a new structure or printable design; in the end having spent less money and a product that suited your needs. You’re also not taking the cyclops to a museum for scanning; regardless of it being an example in conversation. First of all they have scanners that cost tens of thousands of dollars. I’m certain that they use multiple scanning processes including structured light. We are building a scanner ourselves however as stated not this model. Not having a full box enclosure makes this type of scanning inadequate. We will be implementing a combination of scanning technologies ourselves including structured light and laser for developmental purposes. Using various techniques and trialing multiple high quality line lasers, industrial grade cameras, and a carefully lit extruded frame enclosure. We will look at the zum board among pi, and some of the micro pc based boards in various aspects to see where the most reliable and effective platform fits within the realm of 3D scanning. Look into the David structured light scanner that was recently acquired by Hewlett-Packard and also various similar technologies while you google and then imagine scaling one of those technologies into a 900 x 700 x 850 printer enclosure with a 750 x 600 x 600 print bed and scan platform mounted inside using two operating systems interfaced into front of the enclosure. The ability to scan, print, and rescan to check product tolerances, quality, etc…, and also replicate your existing products that may be machined in other applications for ease of prototyping changes in design in a cost effective manner. When it comes to scanning it is important your hardware isn’t low quality. Keeping in mind much of the process beyond hardware is done in the software which also requires many aspects that Arduino will struggle to achieve. Perhaps the Due, Samd21, or similar platform may do an ok job but it is more likely a Linux or python based 64 bit system with at least 2gb of Ram is used and a minimum 32gb hard drive if one plans to compile those scans into one finished product within the scope of a efficient 3D scanner. Or a 3rd party program will be used to compile the scans together and stitch them into a high resolution file that can be replicated exactly or better than originally scanned. To emphasize this takes some processing power to properly achieve positive results. If you haven’t yet seen it google the Fabscan pi. Currently the product is in development of version 3 which is actually promising for hobby level 3D scanning. I beleive if they were to move to an orange pi or other 64 bit Multi core board that they will start seeing huge improvements in scan quality. It would also help a developer significantly when a smaller commercial unit can be purchased and present for comparative analysis. This could save them considerable time and money. Trialing hundreds of cameras, dozens of lasers, and so on adds up before they realiz they have $15,000 in hardware laying around to turn around and open source their hard work so you can do it free or on a friendly budget. The absolute dedication from these men and women out there is outstanding and because of them and other developers we have all of this open sources technology currently available and being consistently updated to the community. Hats off to them for their dedication and hard work while the vast majority of the community takes advantage of discount products and Free software/firmware. I urge you to consider delving into your new found toy and finding new and innovative ways to expand upon it. I’m sure if the community agrees and responds in kind you may find your efforts being passed along.