A laser flare solution and initial results


#1

I’ve been doing a lot of work on a diy flare solution following on from these theads : Form1+ laser flare issues illustrated - pics and video Laser Flare & the effects on your print and the original thread when I became aware of the flare issue: Is the Y galvo mirror supposed to be transparent?

First up - I’ve finally had some success, and I think now I have a good understanding of what’s going on - this post is going to be long, and even then I won’t be covering everything I’ve learned over the last month or two, but to start with here’s two comparison pics of the tall cross print test - showing the flare side (rear) with solution in place, and without…

flare solution in place - rear “flare” side shown - notice lack of flaking

bare laser, flare solution removed - same vat, same resin, only change is to slightly move position of parts to avoid clouded pdms from previous print, and filtered resin.

As I covered here Form1+ laser flare issues illustrated - pics and video and here Form1+ laser flare issues illustrated - pics and video the laser beam is not collimated - instead it has a focal length of approximately 30cm so that it converges to the small spot as it hits the build surface. See this picture from the above link showing the laser beam profile as it changes over distance:

Now - initially I thought this meant we had to discard the simple choke/aperture/iris approach - since the flare “grows” out from the spot over distance, so it wouldn’t be as simple as cutting it off at the laser barrel opening.

However - after a LOT of experimentation - I’ve come to the conclusion that extending the beam path with a mirror and refocusing lenses is not a viable solution. I’ll go over this much more complex approach and why it failed in a follow up post - it’s a long story, so I won’t bore those only interested in the solution in this post.

I’m now quite confident that the flare issue is caused the final cylindrical lens being off-centre in the laser - I hope this diagram illustrates the problem effectively. It is crude and much simplified, but hopefully catches the main point.

In fact - having looked inside my laser again after all this experimentation I believe it’s actually clear that the lens is off-center in pics I shared earlier:

and now annotated to pick out the edges of the cylindrical lens

So - a solution? well I think in honour of @JoshK’s pioneering effort I’ll call it a choke, because it’s not an aperture or an iris, and it operates like an old car choke (albeit in reverse, push, instead of pull).

and here it is - un-installed as I’m running another test print at the moment. I’ll post again about porosity later once I can take the laser out and take some pictures of the magnified spot profile
.

The silicone ring slides into the laser barrel opening, and the choke slides between it and the barrel threads. It’s adjusted while running the laser spot test once the laser is installed back in the F1, by pushing the choke inwards with a fingernail until the flare is visibly reduced. I’ll post a video of this installation process once I’m done with the current tests without the choke, so it’ll probably be tomorrow. Here’s an STL file of the mold for the silicone ring: https://drive.google.com/file/d/0Bxi9QnsGaF7EWWRFNFdnNVZleGM/view?usp=sharing

edit - here’s the youtube install link from post 24 below https://www.youtube.com/watch?v=z7AiD0EpBzY

and here’s a before after comparison of the laser spots

before:

after:


Retro Form 1+ Guide
#2

Very interesting @KevinHolmes. I shall be waiting for the more complex/detailed description.
I’m still in contact with the support team (and as of today the engineering team) concerning this issue. Though we haven’t ensured that it’s a laser problem yet as the results don’t add up completely (in my case). I’ll keep you informed if we find an issue which is similar/related to this and check if there is a fit solution as well. If all else fails I’ll definitely wait for the detailed explanation to see if your solution is a viable one :).

Great work!

P.S. I don’t fully understand the model you’ve uploaded. It’s closed on one side, whereas that doesn’t fully make sense to me. But i’m sure the video will make it more clear.


#3

It’s a mold for the silicone ring - you pour RTV into it until it overflows, then once set, pull it out and trim off the overflow


#4

Of course it is. I must have initially misread! Thanks for clearing that up. But why a mold and not a directly printable object? Even with laser issues, the size it has shouldn’t be an issue


#5

Great work @KevinHolmes!

Your pictures speak volumes and I’m in awe at how much work you’ve put into this. Like @Alex_Vermeer I’m still working with FL support on a solution. They have a new software update due today / tomorrow which they are hoping will alleviate things, but I’m not hopeful. To me it looks like a hardware issue and your hard work and tests just support that theory.

I find it strange that they have chosen to changing the material setting (I take it that means the curing schedules), to alleviate the problem on a few printers.

But as I’ve said before, I’m a total amateur and will be very happy to be proven wrong.

Looking forward to your next post.


#6

@Alex_Vermeer [quote=“Alex_Vermeer, post:4, topic:3644”]
But why a mold and not a directly printable object?
[/quote]
Because it needs to be flexible to allow the “choke” (a simple cut out piece of soft black plastic from a CD sleeve) to slide in and out, and also because inserting FL printed parts into the laser barrel makes me extremely nervous of resin dust from scraping on the fine abrasive barrel thread (it is threaded, there’s just nothing mounted there) getting into the open internals of the laser. RTV silicone seemed the natural choice.


#7

I agree that it seems completely the wrong way to handle it, for a variety of reasons. However if they can adjust the cure times so the fill is more solid without causing more surface quality issues they can avoid replacing some poorly built lasers. It’s way cheaper. It may not fix the issues for everyone but if it fixes some or at least improves some to the point they can write the problem off as “being within acceptable margins of error” they can say the laser on those printers is just fine, and not fix the root of the problem. I do wonder if they can succeed at increasing the amount of curing on the infill without exacerbating flaking and surface problems for those suffering from them, or if this software change will result in a group of people who had been printing fine in clear suddenly having surface quality problems with the same clear resin.


#8

@RocusHalbasch

Exactly - I do more than wonder in fact :slight_smile: I predict that’s what will happen. Although hopefully they add the extra in-fill curing as an option instead of standard - In combination with the choke, extra in-fill curing should solve both flare and porosity.

Of course the only real solution is a replacement laser tested to prove it’s correctly constructed.


#9

Amazing work @KevinHolmes and @JoshK also who did figure this out a while ago.

I’ve been waiting for a replacement form1+ that has consistent good quality prints all over the build platform for a few months already, so If I could open up my form1+ that only prints successfully (semi-successfully) at the very far end of the build platform, I would try this right now. Unfortunately the refurbishing factory completely stripped one of the side screws, so I can’t even open the machine.


#10

@Monger_Designs yes the middle screws in the side are really easy to strip out because of the black paint (guess it’s powder coated) in the threads. When I stripped the screw on my first machine I dremelled a slot in it with a cut-off wheel and then used a flat head screwdriver. It’s not actually that hard to find replacements once you’ve got it out - eg : http://www.ebay.co.uk/itm/M3-4-5-6-8mm-HIGH-TENSILE-10-9-BUTTON-HEAD-ALLEN-BOLTS-SELF-COLOUR-SOCKET-SCREWS-/181321197034?pt=UK_DIY_Material_Nails_Fixing_MJ&var=&hash=item2a3795e5ea they’re M4s btw


#11

@KevinHolmes In it’s current incarnation this still seems like a, one off, temporary patch, that works as a proof of concept. Please correct me if I’m wrong as much of the details are still not known to me so my opinion is mostly conjecture. But here are my reasons for thinking so. It seems likely if the theory you propose as the reason for the flare problems is correct, and your choke is correcting it in the way you propose, very slight changes in the angle of the bent part, or positioning of the choke after installation will result in it not functioning correctly. Being as it is made from a strip of soft black plastic off a CD sleeve, which was then bent, it seems likely with time it will unbend some thus throwing it out of whack, also since it is only held in place by your silicone inset with time and vibration it may drift. In addition since different machines show different levels of severity some even seem to show two bad sides not just one it may well be that some configurations may require modifications to get the solution to work or it may be that it won’t fully work on some machines, and in any case replicating your method and getting it correct seems as though it might be a bit on the tricks side especially positioning it correctly. I know the key portion of the flare can’t be distinguished with the naked eye, as it is in the middle of a much larger blob of light that is all brighter than your eye can make out. I also know it is rather small, and close to the main beam to find something close to the edge of the center of the beam without going to far and weakening your beam or falling too short and still having some artifacts when you can’t even see with your naked eye the part that matters most sounds like a combination of art and chance.

Anyway I was curious what your thoughts are and if you intend to take this to the next step or keep it as a proof of concept.


#12

@RocusHalbasch - no I think the bent plastic will work fine - it’s the “russian” engineering solution (russian pencil vs american million dollar space pen, story is apocryphal but based in truth)

The bend is permanent - yes it’s true it’s not a quantifiable attribute, but it doesn’t need to be - you just bend it until the bend takes (stressed plastic) the bend only has to be approximate, when installing the choke you just adjust its depth until it’s right. So one piece of plastic that wasn’t bent quite as much as another will just end up in very slightly different final position.

As for the adjustment - it’s actually quite straightforward and repeatable since it’s all in one dimension - the choke is a “wall” that is coming in from the side - the line diagrams above are a top down view. All that needs to be done is push the wall up until enough the flare is cut off - note it cuts off everything on one side - and this is why it needs to operate in the laser barrel as close to the cylindrical lens as possible - to ensure that it’s only picking off light that is going to be bent too far.

Having finished the latest test after removing the choke, I was just about to install it again and record the process - that video might help answer your questions if I haven’t been clear - or if I missed a point.

It takes forever to process on you-tube though so it’ll be tomorrow before it’s up - and after I’ve run another test print to prove it of course.


#13

Note - I am not by any means recommending this as a procedure for everyone with flare issues - absolutely not, this is only for the “foolhardy” among us. For those of us too impatient to go the FL support route and willing to take a risk.

@RocusHalbasch So yes in one sense - it is a proof of concept, in that it’s mainly for me and to prove a point to FL and hopefully stimulate them to do more for others with flare issues. On the other hand - I do think for those with flare issues and who are adventurous - it’s good enough.


#14

In a past life I used Lasers a fair amount (interferometers for position control). There are any number of issues that could occur in the beam path and cause distortion of the beam. The “choke” works by blocking the parts of the beam that are more “off center”. Only the light exiting the lens at a low enough angle makes it past the choke. That proves the problem is with the laser itself, since if it was the mirrors the flare would still be there, only maybe a little smaller.

This all makes perfect sense.

The problem is that if the beam is being scattered at the lens, and the choke is blocking the “larger” scattered light, less light than intended is making it to the resin. The flare goes away and provided there’s sufficient “margin” in the power level that reaches the resin tank, things still work and even work better. But this is a compromise, the resin isn’t getting as cured as it would be if the beam wasn’t being scattered by the lens. If you block half the light, you’re curing with half the power.


#15

@Randy_Cohen this is why installing the choke is an adjustable process - you only apply it enough to cut off the part of the laser profile that is being bent too far and is causing flare - as illustrated above in the light path diagrams. It’s true that it could cut off part of the valid laser profile - but this is why it’s installed as close to the cylindrical lens as possible - inside the laser barrel - where the light that is cut off is all (or very nearly all) only light that will be bent too far and cause flare. I’ve done it twice now, and the test print for the second and different choke install is also looking good 4 hours in.

My installation process video has finished uploading to youtube, but it needs some editing - so given it’s midnight here in the UK and that it’ll take several hours to process the edits on youtube - install procedure video will be up tomorrow afternoon.

@RocusHalbasch to answer your question about repeatability I decided to make a new choke for the re-install - and proving it was a good question I did have a little hiccup with the new choke bend, initially it was too much - as you’ll see in the vid tomorrow (it’s long though!) - but it was a simple matter to adjust - and as above, the test print looks good at 4 hours in.


#16

@KevinHolmes Great work! This is some nice science you are doing. We just need to turn the lessons from this into an engineered solution.


#17

@EvanFoss the engineered solution is dead simple, no further work needed, FL replaces the lasers with ones that have been properly tested for folks experiencing flare issues. (and also porosity issues which is connected to, but not solved by this alteration)


#18

Indeed. :slight_smile:


#19

Nice work!
Kinda surprised you’re not getting any diffraction from that edge (i guess it’s early enough in the optical path it doesn’t matter).


#20

[quote=“KevinHolmes, post:15, topic:3644, full:true”]
@Randy_Cohen this is why installing the choke is an adjustable process - you only apply it enough to cut off the part of the laser profile that is being bent too far and is causing flare - as illustrated above in the light path diagrams. It’s true that it could cut off part of the valid laser profile - but this is why it’s installed as close to the cylindrical lens as possible - inside the laser barrel - where the light that is cut off is all (or very nearly all) only light that will be bent too far and cause flare. I’ve done it twice now, and the test print for the second and different choke install is also looking good 4 hours in.[/quote]

I can’t argue with results. But curing is a function of the amount/power of light reaching the resin. If the beam is partially distorted and you position a screen so that the scattered portion of the beam is blocked, that’s power that’s being lost. After all, you get flakiness to begin with because that scattered portion of the beam has enough power to at least partially cure your resin. So it’s a “nontrivial” amount of power. How much beam do you block with the choke? 5%? 10%? 20%?

I like your photo of the business end of the laser and what appears to be a misplaced lens, and your conclusion that this displacement is the reason your beam is distorted. It would be interesting to see how the laser from a printer that doesn’t exhibit the distortion compares. Assuming the interpretation of your observations about the lens is correct and it’s not where it’s supposed to be, this ought to be very easy conclusion for FormLabs to verify, diagnose and correct.

The “right” solution of course is to get the lens back where it’s supposed to be. The Choke is at best a compromise because you’re attenuating the beam. I agree it’s better to lose the power completely than have it cure resin you don’t want cured. But the curing you get via the choke is occurring at a lower power level than if the choke wasn’t there. Since it’s working for you, it’s not too little power in your case. But unless everyone else’s beam distortion is the same as yours, not everyone can count on the same degree of success. The Choke will be as effective in blocking the scatter irrespective of how much light it’s blocking. What gets through may not be enough in every case.

I have a long career in failure analysis of high reliability electronics and electro-mechanical devices. Everything has a failure rate. Various environmental and operational “stresses” can accelerate those failure rates. From my experience, the way in which printers have been described as failing would be fully consistent with a weakness in the attachment of the lens to the body of the laser, consistent with your observation. Out of box failures are often due to mechanical weaknesses exposed by shocks and vibration during shipping and handling. Longer term failures can often be caused by thermal stresses. Every time the printer is used, that laser heats up (pretty good, too, I bet), and then cools down when the print is done. That’s a “thermal cycle” and because different materials (like a glass lens and a brass laser body and the glue that holds them together) have different coefficients of thermal expansion, as they heat and cool they’re going to change size to different degrees and at different rates. They’re stuck together, though, so they can’t move with respect to each other, so the thermal changes induce mechanical stresses. Those stresses are really good at exploiting material defects or design weaknesses. Given enough thermal cycles, almost anything can be made to fail due to fatigue. This kind of failure mechanism could explain a printer that goes from working to not working from one print to the next, as well as one that degrades quickly over the course of a few prints.