If someone would post complete teardown photos of the current printer like Bunnies I would be grateful. I expect given my comments on their forum that formlabs will not make getting my problem rectified when they figure out how. I do not want to make my life harder by voiding my warranty but I do want to see if I can asses what other issues I have in store.
@RocusHalbasch, @KevinHolmes, @Ante_Vukorepa, and @JoshK
I printed a collection of double helix in tubes and duplicated the result that the closer you get to the hinge the worse your prints get. I now have a hypothesis. The sloshing that goes on during the peel is more violent at the peel side and less so at the hinge side. The result is that the partially cured resin debris is washed off the part before a successive layers lasing can cure it onto the rest of the part.
Nice @EvanFoss could you post some pictures. Iâm really interested to see how significant the difference is. Also out of curiosity what resolution did you print them. I just printed one at 100um which I figured should not show any problems, but it does. The interesting part is the side where the flare is, is actually more consistently bad than at 25um but all the other sides are much better. Also coming in another post I plan on updating everyone on the status of my communication with Formlabs on this topic. It includes pictures of their attempt to print my test printâŠ
Hah, the galvo drivers are actually pretty standard fare laser scanner drivers (i.e. drivers used for DIY laser show scanners), including the 5-trimpot setup. Unfortunately, itâs unavoidable (more insight here: mg4355æŁæ”è·Żçșżapp-Apple App Store).
There will be some drift, but itâs not very relevant, as galvos themselves are liable to positional drift with changes in temperature. Thatâs why such systems usually have some sort of a feedback loop built in anyways.
(Besides, even if you were to be somehow able to magically replace the trimpots with precise, fixed resistances, tuned to every individual Form1, those resistances would drift with temperature as well.)
@Ante_Vukorepa I know they are standard fair but there is a much better way to do it and at the same time compensate for the temperature drift. I do this all the time. The laboratory instruments I design use the same kind of current sources and feedback loops (PID). The only difference is normally I have a persons brain at the other end instead of a galvo.
While I bow to your knowledge of optics sir, this is an area where I am absolutely sure I know what I am talking about.
@Ante_Vukorepa This guy posted a photo of the hidden part of the galvo More from him here describing the galvo
The one thing he has wrong in his description is that the galvoâs distance of travel is set by current not voltage. Voltage will make it move but using current drive means you get linear proportion. To product a current you use a howland current pump. Those are written up in modest detail here by the legendary Bob Pease. To reduce drift on them I use precision resistor netoworks so that the resistors all drift in almost exactly (0.001%) direction at once balancing against each other. I also use integrator circuits made from more opamps to reduce other forms of drift. As for the motion feedback and drift there the problem is even simpler since you can do things like using chopper amplifiers at power up to work out the curve for photocell output to position.
If you want more detail let me know but we should probably move that discussion to a different forum.
@Ante_Vukorepa I do have a question for you though. Is it possible that the angle of incident with the laser beam on the tank is having some effect. I mean the here are 3 boundaries there. 1) air to tank 2) tank to dpms 3) dpms to resin. Could it be there is light bouncing between those layers when the beam is off the dead center?
Ok folks I am having an impossible time uploading images to this thread. I am wondering if we hit some kind of storage limit on the thread. Could someone please try to post an image, any image just to check. I would ask fl but they appear to have tuned out of this conversation.
Youâre absolutely right, but iâm not sure that approach is feasible for a mass-produced product. Calibrating the trimpots is already a somewhat time-consuming process and cannot be automated without significant further investments (into developing the equipment and the process).
Now imagine doing the same but adding the additional overhead of soldering a resistor network per each unit produced after youâve found the correct values (for a single unit).
I believe commercial galvo systems also employ a warm-up / settling time before power-up calibration. Which brings me to the important question - do Form1(+) galvos have a position feedback loop? Has anyone here experienced positional drift over time?
Entirely possible thereâs an added effect due to media boundaries. The light bouncing between the layers (total internal reflection) should not be an issue, because i doubt the angle is ever that shallow. However, the boundaries will probably cause the flare to compress more with the change of the angle (than it would have in the simplified pic iâve doodled, just due to the angle itself).
Yup, not working here either.
Maybe you can upload them in the Public folder in Dropbox and hotlink them here.
@Ante_Vukorepa Yes the precision networks I use are a lot more expensive than the form 1âs trim pots. However, they cost a lot less to calibrate and you never need to worry about them again ever. I know for a fact precision networks are used in commercial products. Every decent multimeter uses them. By decent I mean, Fluke, Keithley, HP/Agilent/Keysight. I think they are in almost every meter costing > $100.
I did the math for my last device, no calibration should be needed for another 15 years. If I wanted to I could probably have done better but that is 5 years past itâs new owners retirement. <- ok yea I am showing off but I am really proud of that gadget.
@Ante_Vukorepa & @RocusHalbasch off to flickr I go. I really hate the forums file size limit any way.
Thanks for the heads up, Evan. There does appear to be an issue w/ uploading photos at the moment. Weâve also bumped the attachment size â but that doesnât seem to appear to be the problem in this instance.
@Sam_Jacoby I owe you an apology. Clearly you are still reading this thread.
Thank you for raising the file size limit.
So here are pics of all of the sides of the sample printed at 100um. If only it all came out as nicely as the peel side. Like I said earlier the sides not corresponding to the flare came out much cleaner than when I printed at 25um, but the flare side is more consistently rough than at 25um.
Front
Hinge
Back
Peel
And now a pic showing specifically the side corresponding to the flare. On my particular printer the flare points roughly to the back left corner, so the rough side of the print is midway between the hinge side and the back side of the print. As you can see the surface finish is awful.
@EvanFoss So from those pictures your flare points almost straight toward the back of your printer. I get flaking as well I just havenât mentioned it. The difference in quality is quite noticeable. Although you donât have nearly as many artifacts on the one from the peel side I can still see on the close up of the top that the back wall is still noticeably thicker. Did you print all three at once? What resolution did you print them at? Was the one on the hinge side rotated slightly, because all its artifacts are located on a slightly different side? I have taken to printing it with the drain hole to the front, so I can figure out the orientation. If it wasnât slightly rotated I wonder why the artifacts drifted.
Following is the state of the conversation I have been having with Formlabs support on this topic. I have previously expressed my frustration in post #185, however the progression from that point to now is sad and almost comical.
This is the same as the message in post #185 but I have included it here for context. See my previous rants following post #185 for more details. But in a nutshell I couldnât believe he was basically writing the problem off to user error and essentially denying or avoiding the topic of whether my printer was functioning correctly. He was also only offering a workaround for this situation not a fix for the problem. I decided to give him the benefit of the doubt and assume he didnât really understand the point of the test print. So I replied attempting to clarify.
I thought surely that should make it clear that I didnât want the part rotated, and didnât think it was relevant. I also thought it would make my intent here clearer.
On reading the first sentence of the second paragraph I got worried. I wondered at this point if he actually understood that I didnât care if I had a nice print of my test model, what I wanted was for my printer to print it in the configuration I supplied. I wondered if he understood the value in this was not having a pretty print of my helix, but that if my printer could print it correctly in the supplied configuration I could be confident my printer wouldnât mess up other prints due to the problem the helix was designed to exacerbate. I wondered if he really still didnât get it, or if he was trying to divert me from a real conversation about my printer and make it about my model. So I tried to make this really really straight forward.
I figured this was simple enough so it wouldnât go off course.
Nope. Not gonna veer off course.
Simple enough.
I had 4 simple questions.
When your engineering team prints the .form file included with the model without changing it, at a resolution 25um does it come out correctly?
He said it did, then explained how it came out like mine but not as bad. That is not âcoming out correctlyâ. Then he included pictures where the part wasnât awful but I wouldnât say it âcame out correctlyâ by any stretch either. From his wording I think he might be saying that the rotated versions came out correctly it is hard to tell. As for the fact that he included rotated versions despite my efforts to explain their irrelevance. He also provided no information on the contents of the pictures. For the 3 copies made straight from my .from file I donât know if those are different copies in the two pictures or the same one at a different angle. I donât know if the big scratch down one of them is supposed to represent something. I also have no idea what the other sides look like or which side corresponds to the flare. I have no idea how many different printers they tried or where they all tried on one. But all of that while useful where more than I asked for. All in all he failed at answering this question I asked clearly and correctly. He did however give me enough so I could work out the answer which is: No, but it wasnât nearly as bad as yours.
Should it come out correctly?
He didnât address this at all. Or maybe he thought saying it came out correctly meant it should. I donât know but he didnât help me with this question either way.
If it should why doesnât mine?
He maybe tried to address this, however in the process he seems to yet again strongly suggest user error, by adding in the comment âIf you feel that the mirrors and tank cannot be any cleanerâ. The condescending insinuations really arenât helping this situation. As for it being the tank, or the mirrors, I have already explained no matter where in the build are I print the artefacts are largely on one side of the print. It doesnât take a rocket scientist to understand that means it is not dust or smudges on the tank, or mirrors. Dust and smudges donât move when you reposition your print. Also the tank is brand new, remember my last one broke and leaked all over my printer, thanks to another design flaw / QA problem :). As for dust on the galvos itâs basically the same as the mirrors and tank it just canât explain the problem always being on one side no matter where I print. None of the suggestions except laser flare are consistent with the problem so why am I being treated like the fool?
If not why not?
Once again he didnât answer but maybe that is because he thinks it should.
So in summary Iâm really worried if I send them my printer they will print the same samples they printed when it got past QA the first time, and maybe throw in a reoriented version of my test for good measure and then send it back. Or they might replace the laser, re calibrate it and send me back a printer with excessive compression, or bad alignment of the tray and plate. This is the first machine I have had that didnât have one or both of the latter problems, which they donât cover under warranty as they donât consider them an issue either.