Few problems with custom made delta printer

The printer can't print circles properly. We've looked up solutions and the closest answer is probably the belt being too tight/too loose but we think that our carriages/bearings might be bad. I personally think that there are some errors in printing the carriages (such as the holes are crooked or inaccurate) causing the printer to print inaccurately.

If it is the carriages, what design would be best for replacements? Thanks in advance.

Secondly, we tried printing this and it looks fine but the problem is, is that the pieces don't line up at all. The curvature is different even though when placed on the print bed it is completely parallel to the bed. It has to be the printer but we have no idea where to start on fixing this.

Edited 1 time(s). Last edit at 10/30/2017 08:14PM by Lexar.

Without knowing what printer design you have, and what firmware you're running, it's hard to give useful advice.

Coincidentally, I just went through calibration of my Delta here -- maybe that's why I think your problem is calibration rather than hardware.

Here's the steps briefly. Googling for "Delta 3d printer calibration" will get you lots of hits with a lot more detail.
1. The bed must be at right angles to all the towers, and the towers must be parallel to each other. If this is wrong, you get distorted shapes (circles will print as egg-shapes, etc)
2. The number of steps from end-stop to bed must be same on each tower... i.e. the distance from end-stop to bed must be the same on all towers. If this is wrong, you'll have problems with attaching the print to the bed versus running the nozzle into the bed.
3. Correct for bowl/dome error. If your printer radius is different from the firmware configuration, you'll be printing OK in the middle, but further out you'll have problems with attaching the print to the bed versus running the nozzle into the bed.
4. Correct for tower position error using this. If your towers aren't exactly 60 degrees apart, or compensated for in your firmware, you'll potentially have all of the above problems.
5. Correct for dimensional error. If your diagonal rod length is different from the firmware configuration, you'll be printing larger or smaller than what's specified.
6. Once you've done a basic calibration, go back and do all the steps again; changing one of the calibration settings will affect the others.

We are using Marlin firmware, and this is what it looks like.

I forgot to mention that someone who has been helping us has successfully constructed a identical printer to ours, right down to the nut. Everything from hardware to firmware is almost exactly the same, which adds more mystery to this. We've been in R&D Purgatory for the past 2 months as I like to say.

We've calibrated our printer with those exact steps and still no luck.

Edited 1 time(s). Last edit at 11/02/2017 09:54PM by Lexar.

I don't think it matters too much if your carriage wheels are badly aligned or anything like that. So long as they're round and they roll up and down the tower, they're good.

I don't see any kind of tensioning on your drive belts. Any play in the belt will cause distortion. You can print some simple belt tensioners.

What is really important is that each pair of diagonal rods forms a parallelogram. If you draw a line through the bearings at the top of the rods, and a line through the bearings attaching the rods to the effector, those two lines *must* be parallel. The two rods must also be parallel, and the same length. If you get that right, the effector will remain horizontal, no matter where it is moved to on the bed. Any tilt of the effector is going to result in distortions.

Also check the plastic parts of your printer for cracks. The overall structure should be quite rigid. With power on to the motors, you should not be able to move the effector. If there's any movement or clicks (i.e. play in the bearings), you need to fix that.

When you say is *almost* exactly the same, what are the differences? Can you perhaps exchange parts between the two printers to identify which part(s) are faulty?

I think the frame corners are not so great, not very stiff or strong with all the material cut away to make them 4-5mm walls with cutout vertical holes where there should be dense infill helping provide support.

The carriages I do not think good, not fully realized design. nice idea but the amount of material and its placement, the ratio of length between contact points versus the cantilever from the contacts to the belt pulling points, it's all to compact and the belts when loading the carriage will pull it out of square to the rail. the contacts need to be farther apart, like a wide stable stance versus standing with one's feet together. There are engineering guides for this kind of thing, need to be followed. Then the design needs to be tested quite allot. I would suggest trying a better design of frame components. I am doubting the face of those carriages is square to the rails.

It looks like a school environment. I recommend getting metal corners and metal carriages and rails, or V-wheels. I experiment with a completely 3d printed delta part set, it works well but it is a smaller printer and it is intended for manufacturing in limited resource environments, uses no lubrication for 3rd world use, it has compromises, its not 1st world classroom education machine unless it's a class in engineering, industrial design, or computer science with a hardware focus, in my opinion. I get close to the intended geometry with my machines but it is not perfectly accurate, on small features in the top of the build

it takes a certain amount of expertise in the printer to make a printed corner frame work well. I went through PLA, ABS, and settled on PETG corners, went through about 10 different prototypes of each corner. tested 2 printers for a year making revisions as needed. many of the delta parts one finds posted online seem to have about a week of fanciful dreaming behind them, but they have not gone through a design and analysis process.

If you are going to use 3d printed parts in a delta, you really need a 32-bit control board with a good bed level sensor and the ability to auto calibrate the printer. It is not going to be perfectly aligned and to be honest there is a certain level of deviation one will have even on a very good compact printed corner delta. A better control board will help overcome these problems. The Duet control board will do this with one click in a web based interface, ridiculously easy and I highly recommend it. This makes operating a delta much easier, and allows one to get on with printing the things one wanted to in the first place. It's not a bad investment to upgrade an 8 bit controller to a 32 bit unit, pays off in less time wasted following calibration guides versus the computer doing it for you.

unfortunately my well tested design is for a different size extrusion, 1515. I am just starting a 2020 design this week and will post it when done enough for testing, but it could be a month or more before prototypes are good enough for any kind of public testing. Not sure that helps but it's the best I can do.

I am sure there are some better 2020 printed parts online, something with more vertical support instead of everything cut away, that would help to start. Print with 40% or more infill, a nice dense grid of internal ribbing that keeps the printed corner much better aligned. Make sure the printer that produces them is perfectly calibrated and aligned, let them print slowly and accurately with no ringing from moving too fast even on travel moves, overnight prints for each corner, one at a time, one after another, from the same Gcode file on one printer. calibrate the production printer as well as possible, print 3 corners and hope it stayed calibrated exactly through them all. Do this for each set of 3 matching parts for each tower.

It just makes them more matching to make them all in sequence with the machine as close to the same calibration and geometry. As an example on a mill working in metal, if you want accurate matching parts you set up once and make all your parts with the tooling and everything all in place, if you move it then move it back you're not going to get as exact accuracy repeating a cut again, things will be slightly different.

Edited 1 time(s). Last edit at 11/03/2017 12:59PM by Milton.

Quote
frankvdh
I don't think it matters too much if your carriage wheels are badly aligned or anything like that. So long as they're round and they roll up and down the tower, they're good....With power on to the motors, you should not be able to move the effector. If there's any movement or clicks (i.e. play in the bearings), you need to fix that.

See that's the thing. I think it is the carriage/bearings because everytime when it moves up or down, the carriages themselves wobble. Now the bearings must have a role in that, but I have no idea if the size of the carriages depends on that too (i.e it being too loose or too tight). We ordered more bearings, and going to reprint another carriage that wobbles way too much. We should know by Mon/Tues.

Quote
frankvdh
I don't see any kind of tensioning on your drive belts. Any play in the belt will cause distortion. You can print some simple belt tensioners.

This was taken during construction. This is the printer now

Quote
frankvdh
What is really important is that each pair of diagonal rods forms a parallelogram. If you draw a line through the bearings at the top of the rods, and a line through the bearings attaching the rods to the effector, those two lines *must* be parallel. The two rods must also be parallel, and the same length. If you get that right, the effector will remain horizontal, no matter where it is moved to on the bed. Any tilt of the effector is going to result in distortions.

When you say is *almost* exactly the same, what are the differences? Can you perhaps exchange parts between the two printers to identify which part(s) are faulty?

I will have to take another look at that Monday.

We can, but the differences is that he is using a LED and we aren't, we are using a computer, while he's using a Pi, we are using different type of screw for the bottom aluminum rods and that we are using a glass bed and he's using tile. None of which should have this type of ungodly quality.

Edited 1 time(s). Last edit at 11/03/2017 11:31PM by Lexar.