Disassembling and Reassembling of the Form 2 / re-calibrating Tank-Tray

Experimental
1

I have put together a video on how to disassemble and reassemble the Form 2. This is not the only way but hopefully a useful guide to anyone who needs to open up the machine for whatever reason.
I did not take the optical compartment apart as those components are readily accessible (like stepper motors, gear rods, laser unit) once getting the major components disassembled.
Here is the link:

The tank-tray does not have to come off for most tasks. If for whatever reason it has to come off and one does not have any special equipment (like height gauges) and needs to re-calibrate the tray, using the most basic materials, and some patience I have a short guide here:

3 Likes
2

Great job with the videos! The first could serve as a useful aide for users looking for video explanations of certain suggestions by out support team. For users having printing issues, I’d recommend getting in touch with our support team first before attempting these techniques. Our support team will be best suited to diagnose and advise next steps.

I’m a bit worried by the techniques used in the second video. While having simple techniques for tray carrier calibration would certainly be ideal and make repairs easier both for us and for users, it doesn’t tend to be possible. A number of different components are calibrated relative to the tray carrier and some of those calibrations rely on factors like the absolute distance between the carrier and other components. While this technique should work for simply leveling the carrier, it could throw off the calibration of other components and lead to other issues which would be much more challenging to correct. I’m glad to hear that this worked in your case but would advise against it for other users.

3

Frew,
Let me say this up front, Formlabs has the best support team I know - period. So yes I would think that the majority of users would contact support first anyway, should there be a problem.
Now having this said - I was in contact with the support team, because of the resin sensor error, this often happens right in the middle of an important build - but until the resin sensor error was fixed I tested the machine to see if there were internal components like the sensor board (mine is already the revised version) or the resin release at fault - of course they weren’t if they would have been I could have let the support team know what the problem was. So in my mind it would have made things easier and faster, to get my machine back on track - in the end it was a firmware problem - or let’s say it was fixed with the latest firmware version. By that time I had my machine back running in open mode again. I do understand that Formlabs technicians don’t want to see users opening up the machine and try to troubleshoot it, for obvious reasons we all know.
There are most likely few people like myself who would take the machine apart and search for problems and rather have spare parts available before going through that long, costly process of shipping the machine out to Formlabs, waiting for it to be fixed and then most likely getting a different machine back.

When it comes to dialing in the tray, there are some basics that one should know about how the machine reacts when building a part at certain distances, what needs to be done when this or that problem occurs. That’s why I said with some knowledge of the machine… Now you are referring to “different components” that are also calibrated in relation to the tray.
I can’t quite follow there - the distance between the tray and the building platform is what relates to everything else. If that distance is off everything else will be off. Please do correct me here if I am mistaken.
The laser is set, the mirrors are set so whatever hits the large mirror will hit the building platform at that spot. Wherever that building-platform is that’s when things become important. If the tray is too high we’ll have pressure on the building platform and the connecting components that put extra stress on the mechanics - if the tray is too low we simply have no adhesion on the building-platform.
My method is really not that simple, the tools are. I also believe that most people would not “mess” with the tray. The key is to have support platforms build properly as well - if they are hanging off - which can happen the machine has a built in Z-Axis correction feature which - of course is controlled by the software, can also cause complete building failures if overdone. Even if the support base hangs off on one side only (let’s say left) the Z-axis correction feature will not only move the building-platform down on the left but entirely, causing an imbalance right there. Once the machine leaves Formlabs one should not have to mess with this feature in first place - the crazy thing here is that this was the case for one of my Form 1+ machines - nothing would adhere to the building-platform properly right out of the box, so I had to use that Z-axis correction tool. Instead of that one theoretically and practically could also “just” move the tray up or down as needed.
Which shows that this is what controls the calibration. Of course the laser itself needs to be calibrated also - I have not taken a closer look at the laser unit and whether you guys changed it in terms of assembly from the form 1+ where the laser had some play and didn’t “lock” into place when inserted.
So I am not sure which other components would be jeopardized, except for those being directly “manipulated”. Again, please let me know if I am missing something here.
The machine is quite simple in the way it’s put together and laid out - which is not to downgrade the machine and those who worked on it - as a matter of fact the best machines are simple yet effective, like the Form 2. My machine is not dialed in 100% as can be seen on my tests but it is within that window where it performs flawlessly. I could waste more resin on tests and dial it in even closer - Considering the machine can build 0.025mm layers I am off by 0.07 on the test in the video and got closer on another test to 0.06 (not on the video) I am sure I can get even closer by wasting more resin but the machine operates within that window (flawlessly from 100-25microns). I also tried to get some info from the support team on the actual length of these square columns, once built (these are from you guys - optics test file), but I never got an answer on that, so I relied on parts I had made before and how they should look coming out of the machine to get the initial setting.
I don’t think you guys need to worry too much as I doubt many users will actually need to have the tray re-calibrated.
Those who do and use the built in Z-axis calibration feature are probably more experienced anyway or were advised by you guys to use it. And most of them probably wouldn’t “mess” with the tray itself. I also keep saying in both videos that it is not recommended to take the tray off so those who do know what they are in for - I assume.
Some people are even reluctant to take the glass off even if advice by you guys…now that’s a crazy one right there.

2 Likes
4

I absolutely agree that there’s a great deal of utility in having resources available that show users how to disassemble the machine. I largely want to make sure that new users don’t attempt this as a first step and instead troubleshoot with our team. For less experienced users, disassembling the machine could potentially cause more errors than it corrects.

One of the main reasons we’re unable to provide tray calibration in the field is because vertical towers at each corner provide an imperfect measure of tray level. They have a great deal of utility for other troubleshooting steps but don’t dial in tray level to the level of precision we aim for. For your applications, it sounds like the process has been adequate and I’m glad that’s the case, but it’s not quite accurate enough to achieve our desired tolerances.

As an example of relative measures, the tilt of the tray is calibrated relative to the tilt of the accelerometer. Slight deviations in tray tilt likely won’t cause issue here but it’s something to watch out for. Different positions of the tank relative the capacitive level sensor is a bit more worrisome as slight deviations will have a more significant impact. In any case, I’m excited to hear that this worked for you, and this technique could provide reasonable levels of accuracy. It just might be something to err away from for newer users and factory calibration will provide greater levels of precision.

5

Is the tray calibrated relative to the accelerometer of is the accelerometer calibrated relative to the tray? :slight_smile:

From a mechanical point of view, there is a fixed position of the build platform that can not be calibrated in any way, as the holder, arm and linear slider are all fixed and assumed reasonably perpendicular. So this would be the reference point/datum.

From there you have the resin tank where the PDMS needs perfect alignment and be parallel to the build platform or you will cure an uneven layer of resin. This should be the biggest influence factor on accuracy, taking into account that you didn’t touch the calibration of the Z axis and optics(not influencing laser focal point).

From there you would have to align/calibrate the accelerometer to the resin tank, to make sure the resin is perfectly level, as the accelerometer is on an undefined space on the PCB which is mounted to the bottom of the machine.

Then, when everything is level, you need to lower the build platform relative to the capacitive sensor so it senses a full tray… this is in my opinion less critical.

If you have print layers of 25 micrometer, you might desire and accuracy in parallelity of 1/10th of that, which I fully agree is not possible to do with a standard calliper(check the datasheet on accuracy!) and you need to make an accurate measurement block to do this. I could imagine you guys have a block with dial gauges(or some other electronically measuring magnetic sensors) on each end. For us end users we could do something fairly simple: calibrated, spring loaded pins that make electrical contact once touching the build plate. This technique is used to calibrate industrial machines as well.

Correct me if my order is wrong, but this is how I would do it. :slight_smile:

Anyway, @Jens_Schnabel, thanks for sharing the video’s! We had a very bad resin spill, all electronics, steppers, gears and sensors were filled with white V3 resin after the parts broke off mid-print. We managed to completely disassemble the machine, rinse it and put it back together without much trouble. -> Not something I would recommend for the casual/regular normal user!!

1 Like
6

Very Good !!!

7

Accelerometer - The Machine level sensor has nothing to do with the absolute accuracy of a built part. Which is what’s at stake when “messing” with the tank-tray as I call it - there is the carrier, which is mounted to the optical housing then the tray that holds the tank, which is the only part that can be adjusted. Just to clarify my use of words.

The machine level sensor or better it’s display is not particularly accurate as the system accepts the machine to be level even if the small circle is not dead center. Even if it looks to be dead center visually, it still can be off slightly, since it’s a visual calibration in the end… so we have an inaccuracy here already.

Further the accelerometer has to be calibrated to the building platform not to the tank. The building platform is fix and therefore the only reference to the accelerometer, and tray. The accelerometer is dialed in by the factory so I assume it is right on out of the box, if that is so this won’t cause any problems when re-calibrating the tray as everything else stayed the same (Machine level sensor, laser unit), unless it’s been “messed” with.
Possible that when reattaching the base plate it won’t be in the exact same position as before therefore throwing the accelerometer off. Unlikely and if it’ll be so slight that my machine didn’t detect it.
The Z-Axis is fix, the carrier is fix, the PCB is fix - as you say @fantasy2 we assume the Z-Axis is perpendicular to the PCB and all other horizontal planes (to absolute precision).
I also assume here that we have an accurate PDMS layer - if we don’t that is not too critical as, again the tray can be adjusted to a slight imbalance, unless we have “waves” within the PDMS, which can happen but is unlikely within such a small surface area. This however is another point of possible inaccuracy. The play of the Z-axis and the tank unit within itself is rather small but there is some play, not so much the tank unit, but Z-axis. More potential inaccuracies.
The Form 2 still kicks out some layer shifting, right out of the box, rarely but it still happens.

Having this said - it is important to have the machine level or else we might encounter a resin level imbalance, which could cause spilling, therefore the accelerometer. If the machine is not level it won’t work - or better if it’s not level enough…
For the machine to spill resin (without parts or particles floating around) the tray would have to be off by quite a bit, we’re talking millimeters.

Yes @fantasy2 essentially the final calibration is the distance PDMS surface to build(ing) platform. Even with my crude calibration techniques. Let me add here that my brother bashed me for trying what I did and wanted to drag in his calibration equipment - he also was the one checking the PCB and resin sensor board in the video - I wouldn’t know where to start looking if I had to test the PCB myself or any other circuit board for that matter, so I do have some good help if needed in that field.
The machine nevertheless was dialed in with my crude tools as this is what I wanted to test. Most don’t have any calibration equipment and for those this is a possibility to get the machine back on track. The final question, which I don’t seem to get an answer for is how close is this calibration. Far off or not quite accurate enough to achieve the desired tolerances (that doesn’t help) what are these tolerances @Frew? :slight_smile: I understand that you most likely are not allowed to tell me (us).

Of course I agree 100% that with my crude ways I won’t be able to get it dialed in to absolute perfection.
The machine - as all machines, operates within a certain window of accuracy - that window starts at a certain point and stops even if the machine can be dialed in better. The question here is how far am I away from that window with my crude way of calibration - it looks like I am at least close to it, judging by the parts it kicks out.
Thanks @fantasy2 for the numbers - 1/10th of 0.025mm I am not so sure if the machine can be dialed in to that then again maybe it can, maybe even more accurately …@Frew?
In any case thanks for chiming in here @fantasy2 - crazy the spilling on your machine - I heard that from a few people - I did have one part breaking off but luckily no spilling even when resin was stuck on the PDMS. The new tank is apparently spill proof…and lasts 20 times longer - i want to see that - 20 liters instead of 1 (which is typically what I get out of a tank 1-1.5 liters), that would be something!
I did have the resin sensor error and I also get - off and on - the no resin cartridge error, even though the cartridge is inserted…and we are talking about accuracy here…:slight_smile:

Here is an Optical test, using a file my good buddy from the support team sent me. There are 2 tiny parts on there which I need for my latest project - no problem for my machine as “off” as it is, to get this optical file done “perfectly”.

Optical test 1
Optical test 11618Ă—1191 318 KB
Optics Test 1b
Optics Test 1b1920Ă—1280 389 KB
Optics Test 1c
Optics Test 1c1920Ă—1280 413 KB
Optics Test 1a
Optics Test 1a1920Ă—1280 475 KB

3 Likes
8

can you post the calibration spheres you used in the corners?

9

The ones I used in the beginning of the video? Those are elements I took from a model I was working on. They are basically half spheres with an outside ring attached to the flat side and rather useless compared to Formlabs’ optics test file.
I suggest just using Formlabs’ test file instead.

I tried to attach the optics file but it’s too big in terms of file size. Ask Formlabs’ support for the optics test file, they’ll be happy to give it to you.

10

i ended up printing 10mm cubes and got them calibrated within +/- 0.25mm.

This inaccuracy should only affect the first few mm printed right (i.e. only the first few layers should be distorted)?

thanks for posting your video.

11

You’re welcome (for the video)
When you say +/- does that mean you have a combined difference of 0.5mm? If so try to get to a maximum of 0.25 all together. Or simply test your machine with a good part and see how well it comes out. Every machine is slightly different so what might be the right number for me might not be right for you. Once your machine builds a good base and connects all supports properly you’re in the right ballpark with a good working machine.

12

Do you have these print test files?

13

Antonio,
I have the file but it’s too large to upload here - but you can go to Formalbs’ Customer service and ask them to give you the optics test file (which is exactly what I ended up using), they’ll be happy to do so.
You can also simply make 4 small cubes in your modeling software and place each in one corner of the building platform. That’s besiclly all it is. Let the machine go and you’ll be able to measure each one and adjust the machine accordingly.

1 Like
14

OPTICS_Test.form (274.2 KB)

Optics test file attached.

Chris

2 Likes
15

Chris,
thanks for uploading the file.
Interesting that your version is only 274KB mine somehow is 5 times larger (1.1MB)but show the same objects as I checked briefly (4 columns and the center object). Byt that’s the one. :slight_smile:

16

I was a little confused when you said the file was too big…LOL, Just for your reference the file I uploaded was supplied to me on 2/1/18. When did you get your version?

Chris

17

August 16th, 2017 is the date on mine.
Jens

18

Many thanks for submitting the file!

19

I found a much faster way to align the corners of the tray carrier with the build platform:
A. Remove the resin tank.
B. Set the printer to open mode.
C. Start a print, the build plate will come all the way down.
D. Unplug the printer.

Now that the plate is stopped at the bottom, get out your trusty spark plug gapping tool,
set the tool for a 7.5 mm gap. Adjust the gap at each corner by inserting the tool between the tray carrier and the bottom of the build plate, then adjust the screw at the same corner. You will need a #10 Torx Security driver, not a plain Torx and you should go around more than once because each adjustment can affect the other corners.

At this point I printed a test file which is shown below.

The result I’m getting is 4 good pyramids on the corners but the center pyramid doesn’t print, just leaves one layer stuck to the bottom of the resin tray. That’s the problem I will tackle tomorrow. Any suggestions?

pyramids
pyramids1209Ă—1312 42.9 KB

2 Likes
20

Increased the gap between the build plate and the tank carrier to 8.4 mm in several steps and it had no effect, corners still good, center still bad. I checked the optical window, the main mirror and the acrylic window on the tank, can’t see anything wrong. I’m beginning to think the PDMS layer on this tank has been damaged in the center.