Hacker News new | past | comments | ask | show | jobs | submit login
Continuous 3D Printing Possible With Transparent Polygon Scanning? (fabbaloo.com)
90 points by hexa_storm 11 months ago | hide | past | favorite | 68 comments

The sad part of this whole article is this method of scanning a laser beam had to be devised to side step a patent that uses common galvanometers instead. So there isn't any new 3d printing technology here. Only a clever method to circumvent a silly patent.

I work in the area of new intellectual property generation, and you (as was I) would be surprised to learn this is not actually a sad feature of the system.

There have been countless advancements in the sciences when people have been forced to work around existing IP and have uncovered hidden benefits when working through the actual implementation.

It is a hidden benefit of the western system of IP that China et al. have not yet caught on to.

The problem is that, in 2019, most obvious methods of scanning a laser with moving mirrors should be covered by expired patents from 40-30 years ago.

You have to specify application in a patent. Riken desribed it for 3D printing in 1997 see https://sffsymposium.engr.utexas.edu/Manuscripts/1997/1997-2... but did not specify the use of laser diode as it was quite new at the time. It got only invented in 1996 by https://en.wikipedia.org/wiki/Shuji_Nakamura who got a patent for the laserdiode but little rewards and fought with his employer over it.

how is "obvious thing ... with a laser diode" any different than "obvious thing ... with a computer"?

Obvious things are not patentable in the first place, so it doesn't matter what the "with a X" refers to. That the patent office continues to patent things that are obvious to a normal practitioner is a serious problem. I think part of the problem is that the field of technology is so wide right now that most patent examiners aren't good at judging whether something is obvious or not -- so they don't generally attempt it. If it's novel, then it's not obvious, is what I think they are doing.

Having said all that "with a laser diode" is categorically different than "with a computer". In the latter case, you can do any general purpose computation you like. So if you have a machine driven by a computer, you should not be able to patent it if you would not ordinarily be able to patent the machine. In other words, the driving of the machine by a computer should not be patentable, IMHO. Of course it gets complicated as you add more and more machinery to the machine -- should "computer driving a car be patentable"? Probably it should as it is not just software, even though it requires software. I think it's tricky.

If a laser diode allows you to do things you wouldn't otherwise be able to do and that use of the diode is not obvious, then it probably should be patentable. However, generally if "machine with a laser" is not patentable, then "machine with a laser diode" should also not be patentable. There's got to be something special and non-obvious about the use of the laser diode.

Patent offices are patently not doing the job of rejecting obvious modifications. Of course lawyer mangling of the documents does not help with discerning whether something is new or not.

Rotating prisms for scanning lasers are old technology, they were used extensively in photo imagers used to produce film artwork for the printing industry, the most famous of which was the Lasercomp. I was working with these devices in the early 80's. I suspect there is a lot of precident patents and prior art.


It's true... if I wasn't forced to avoid the IP.. It would have been unlikely I would have made such an effort or this thing build.

Indeed, the ability to "invent around" a patent is essential because it means that patent-holders can't completely monopolize an idea. Instead, they only get to monopolize their specific inventions disclosed in their patent, and only for a limited time.

So you're saying that duplicating effort everywhere is good because sometimes the duplication leads to some benefit?

Personally, I'd take the "have dirt-cheap home SLA printers ten years earlier" option, thank you.

>It is a hidden benefit of the western system of IP that China et al. have not yet caught on to.

Oh China has caught on. When they see some new IP they want, they acquire it or steal it. Why bother with all those silly rules the West thinks are so important.

That is indeed.. how i decided to brand it and this seems working. In reality both systems have different optical properties; this does matter if you push them to the limits. Galvanometers are slow, require more calibration. Rotating devices don't stand still at the end of a sweep and are much faster (10 times).

Galvanometer based systems often suffer from acceleration issues at the start and end of the swing which limits the usable spread.

the concept is really cool, congrats! Can the prototype actually print? Do you have some photos of the printed stuff?

This is how it prints on UV paper, https://cdn.hackaday.io/images/1013221548797737200.jpg . This is the calibration data recorded by imaging with the camera directly on a CMOS chip. You can count pixels.. this is what an expert needs to determine the quality. https://cdn.hackaday.io/files/21933938381472/calib_data.zip

More data is here https://cdn.hackaday.io/files/21933938381472/21measurements.... pixel 4.65 microns. You can use opencv to do the trick. I added a little script in this one.

The patent (https://patents.google.com/patent/US7892474B2/en) doesn’t seem to mention galvanometers at all.

It uses the phrase ”digital light processing/digital micromirror device”.

I don’t know how that slash should be interpreted. If it’s “aka”, the patent doesn’t cover using prisms.

I wouldn’t bet on that, though, noting that the patent is careful to avoid the term “light”, using the broader “electromagnetic radiation” instead.

the rotating reflective polygon patent is this one US9079355 I think it is even more relevant but somehow it did not get posted. https://patents.google.com/patent/US9079355B2/en

claim 1 in US7892474B2... by means of mask exposure ... so prisms brings you around it.

Only proves that patents are a hindrance to the progress, not a solution to speed it up.

Why people are so greedy? Let's say X make an innovation that necessitate the use an older patent from inventor Y.

Three things: 1) X has not to pay anything to conceptualize his innovation. 2) X can make a nested patent. 3) a user of the innovation of x must pay a royaltie (why is that an issue?) to Y. If X made a nested patent, the user must pay X and Y. Is this right?

BTW who fixe the amount of royalties?

Presumably each chunk of royalty makes it that much harder to sell a successful product. Often these fess are significantly higher than nearly any other component on the device (think Qualcomm for cellphones and Fraunhofer for mp3 decoding)

Nobody's going to buy the $100 item if it provides $15 amount of value, and nobody's going to make it for $15 if $9 of it is going to licensing.

In which case licensing fees should be explicitly capped. E.g. to a percentage of revenue you make off the device.

That'd stop patent trolls and silly patent holdings dead. You don't use it, you can't charge others.

If someone starts using it, you would charge a percentage royalty.

Exactly, there should be a political push for this!

Yes, but I observe on the net what seems to be an aversion to pay all patents even if it only cost a few cents. The price per unit of the concerned patent is never talked about and this irrational aversion is an issue. Websites displaying patent content should display price too. I don't know if patents are generally overpriced but patent owner have an incentive to maximize their benefits which means they probably want to maximize the number of users of the patents by making it cheap.

laser printers use rotating mirror instead of galvanometers

I was to a couple additive manufacturing exhibitions and conferences, and generally the people from the metal 3d printing industry seem to call all laser pointing devices "galvanoscanners", regardless of actual construction.

rotating mirrors are too fast for laser industry, it requires a very fast switchable laser, like the one developed by european union https://www.razipol.eu/

Rotating mirrors here means continuously rotating, not precise rotation to some particular point? I misunderstood then, yeah, that kind of mirrors aren't used there (I think).

they use glass lenses and infrared, cross scan error must be quite significant due to long separation distance between polygon and substrate.

If this is about photolitographic 3D printing with resin, then I say not much is being disrupted. Those printers aren't that expensive now; the resin itself is quite expensive though.

The price of resin isn't directly comparable to the price of plastic used for FDM printing because one can print most models nearly hollow with FDM in an automated way (e.g. with honeybomb-like infill patterns), but I'm not aware of doing the same thing with resin with ease now, making the resin prints 10X the price of PLA ones.

Better software would disrupt that tech though.

expensive is a matter of taste. The envisiontec 3SP costs well over 5K euros and can go up to 100K. You can only get a a price if you ask for a quote. Carbon 3D does not sell their printers you have to lease them. Formlabs does sell you a printer at 3.5K but you can't use third party resins, so you decide. I actually know some open-hardware printers which sell for 5K.

you can get an Anycubic Photon under $500, its a decent resin LCD printer.

The patent mentioned in the article is specifically for continuous SLA printing (no pauses for lifting/peeling between layers), which the photon and other similarly-priced models don't do. I have one of the cheap SLA printers, and the lifting can occupy over half the printing time depending on layer height/resin type. The continuous printing machines seem to start in the several-thousand-dollar range.

the rotating reflective polygon patent is this one https://patentimages.storage.googleapis.com/29/68/46/da50635... I think it is even more relevant but somehow it did not get posted.

You'd think laser printers would be prior art - a rotating mirror (usually a hexagon) was exactly how older laser printers would scan across the drum to form out the print.

You have to specify application in a patent. Riken desribed it for 3D printing in 1997 see https://sffsymposium.engr.utexas.edu/Manuscripts/1997/1997-2.... but did not specify the use of laser diode as it was quite new at the time. It got only invented in 1996 by https://en.wikipedia.org/wiki/Shuji_Nakamura who got a patent for the laserdiode but little rewards and fought with his employer over it.

true it is a nice machine.. but laser bundles have advantages. You can direct other wavelengths, sinter powders and get to much smaller features. All these LCD printer, do show a pixel size but not a real spec or accuracy. If you go really high end even for mask projection with UV led you need expensive lights see https://www.idonus.com/activities/products/mems-products/uv-... ... costs around 20K i believe.

I'm not sure which resin printing process you're referring to. Most of the time for consumer-grade resin printers like the one I have, you design your model with a drain hole and print it hollow so the resin can flow out (or add a hole before slicing). Printing fully solid is generally only for when you want it specifically to be solid.

> you design your model with a drain hole

That's indeed what I was talking about. It's a manual process to my knowledge.

And then you end up with a shell, which is not as strong.

To my knowledge, you can't really have this with resin:


Please correct me if I'm wrong :)

I use FDM and resin printers regularly. You can definitely do any infills you want, such as the ones shown in your link. It's not a manual process any more than it is in FDM printers.

Yup, just to add onto this, software like Meshmixer will automate most of this for you. It's still possible to get voids where resin will pool depending on your model, but there's not a whole lot of manual effort involved these days.

Thanks! I am behind the times on resin printers. What hardware/software combo would you recommend? (With FDM, I've happy with Printrbot/Cura).

Netherlands-based Rik Starmans has devised something he calls “Transparent Polygon Scanning”, or “TPS”.

Does this mean the linked article is a TPS report?

should have named it differently; I use a laser scanner with a prism instead of a mirror. A detailed description is here; https://reprap.org/wiki/Transparent_Polygon_Scanning

should have named it differently;

Don't sweat it. I would guess that the name will amplify the amount of media attention it gets.

I don't think it's that disruptive. One thing that the 2009 roadmap for additive manufacturing has emphasized is the need to move from point to line and area deposition[0]. This still scans a point like stereolithography has done since the beginning. We now have stereolithography machines capable of area deposition by using a DMD projector[1], often called DLP, that are approaching manufacturing products at high volume[2]. Some DLP stereolithography machines now have resolution on the order of 10 micrometers over a 160 mm by 50 mm area[2]. We're having to develop new ways to describe 3d geometry just so that we can take advantage of this resolution. So there is potential here for both fast and incredibly high resolution using DLP stereolithography. We've also seen the development that the 2009 additive manufacturing roadmap couldn't anticipate: volumetric deposition[3].

In addition, there are some potential disadvantages to prism scanning. Prism scanning can only do raster scans, we can't define the outline of a part in vector mode. Although I will admit that this may not be as much of a problem if we can control the brightness of our laser. In addition, in stereolithography it is advantageous to carry out raster scans perpendicular to each other, this is because the photopolymer contracts as it cures and scanning in both directions helps make this contraction uniform.

[0]https://pdfs.semanticscholar.org/4716/c69f0b90a158589e54248a... [1]https://www.youtube.com/watch?v=UpH1zhUQY0c [2]https://www.carbon3d.com/case-studies/adidas/ [3]https://www.youtube.com/watch?v=31xHZvUP1Ls

#3 is pretty neat. Paper: https://www.epfl.ch/labs/lapd/wp-content/uploads/2018/11/ms3...

* * *

Inre #1–2, Is anyone selling custom parts produced on that Carbon3D printer? That is, someone that ingests a shapefile (or whatever) and mails back a part printed on one of these machines?

DLP projection has a lot of advantages, the area projection is one of those. For large area beamers are moved; see moving light technology by prodways https://youtu.be/1AQJaqdrzzE . There is no longer a real difference then. Lasers can go below 200 nm using dual photon (nanoscribe) or sted lithography.

Hardware is the new frontier of open source, I'd like to see schematics/PCB versioning on GitHub, with diff viewer and stuff.

These guys have been trying to do something like that: https://upverter.com/

I'd settle for better PCB design tooling (or maybe I'm just unaware of good options). I feel like the tools are pretty archaic/inconvenient compared to stuff we have for software. Something like Eagle vs. some IDE feels very different.

That sounds like an amazing idea and is pretty close to what I want as well.

There is a blog here https://hackaday.io/project/21933-open-hardware-transparent-... also provides all the link to github files. Design are made in Freecad, Kicad, C++ and Python. My main focus is on developing prior art. This needs to be done to prevent IP and "liberate" the technology from patent trolls.

Of course, what’s worse than a weak start is simply not starting at all. Moreover, if you’ve publicly disclosed your invention, then you have only a one-year grace period to apply for a patent. You forfeit your patent rights if you fail to file within the 1-year grace period. http://www.patenttrademarkblog.com/first-to-file-patent-rule...

In the EU.. you destroy immediately.. there is no grace period.

You might want to consider publishing it in a prior art database like https://ip.com/ (no affiliation, and I don’t have any reason to endorse them over others.)

You’ll likely want a lawyer, etc to make sure that it is properly protected, especially now that the US is “first to file” not “first to invent”. (I’m not a lawyer, you might be fine already...)

Interesting, I've used something quite similar to the scanning-through-prism technology a few years ago, though with an 8-sided polygon for the application of exposing PCB masks. We were looking at 3D printing too IIRC.

To get to the desired ~80x80cm PCB size and the accuracy needed, we had to achieve super high throughput and super low timing jitter on the UV-LEDs. Very interesting technology, unfortunately we took too long to get a smaller scale proto working and funding was cut right after we delivered that proto :-(. Calibrating the whole machine from scratch was clever too.

Learned a lot there.

do you have a link?

The domain name has lapsed, but some articles remain: - https://www.alexandergroenewege.nl/wp-content/uploads/2016/0... - https://www.linkmagazine.nl/maskerloze-pcb-machine-ldi-syste...

In Dutch unfortunately. Looking at OP's article, it could very well be that we had licensed some of the tech of the Rik Starmans mentioned int he article.

Are there things beyond modeling where 3d printing is useful? The parts I've seen using dlp or sla never seem structurally useful to me. Sintered metal printing seems like although it may be less precise might be more useful structurally?

Small scale stuff you want on premise is a decent use case. The prime example being medical/dental. Imagine a patient with a crushed jarbone. You can model and print a custom prosthesis on premise and print it in maybe 16h-38h. Compare that to sending the job to some lab, custom molds being created and waiting for the item to be delivered (while the patient is waiting).

I know it's used quite a bit for custom tooling as well. Think special tools needed on an assembly line in automotive where pausing the assembly is very expensive or lead times for some custom tool to grab a wheel while working on it and whatnot are expensive.

I think one of the biggest success stories is cutsom made hearing aids. Return rate a lot lower, easier to iterate and produce the perfect fit etc.

Another example (maybe a bit more high end) is turning some item that is made out of many parts into a single item both for cost saving and other reasons. Iirc NASA uses ultrasonic welding (which is 3d printing of sorts) for this.

I'm guessing the resin used for medical applications isn't the cheapo toxic stuff.

Neat use cases though, and certainly makes some sense. I guess I would've expected precision cnc to be faster/better for such cases though, unless you absolutely need the impossible to create voids

I've used nylon SLS for lab and commercial optical work. It's strong and you can tap it. It's much cleaner than dealing with resin prints. The quality is fine if you're making simple polygonal shapes and unless you need really fine detail it's enough. You can also do things easily with SLS which are really annoying to do with printer technologies that need removable supports, like cavities and overhangs.

you can also combine it with optical coherence tomography and use it for detecting eye diseases or laser surgery...

Hmmm... Seems interesting (and perhaps questionable?) from an infringement perspective.

What possible recourse would the original "continuous printing" patent holder have now that this prismatic design has been open sourced, rather than used in a commercial application?

patents are expensive and you don't know the results.. Patent in my name is here https://patentimages.storage.googleapis.com/43/72/4e/1ee7792... .. but is for a plurality of sources. So I avoid my patent, I am author of :P

Came here for the 3D printing of unicorn animals using lasers. Disappointing.

I prefer highly significant Bladerunneresque origami unicorns.


Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | Legal | Apply to YC | Contact