RepRap 3DR Delta printer Released

Quote
Romank
Had to use some M8 nuts over the FSR. otherwise they did not work.

I'm curious as to why a flat plate of glass over the FSR is not working.
Did you apply adhesive tape to both sides of the FSR.
Are you taking advantage of the dimples on the surface of the M8 nut to concentrate forces.

Yes i glued the the FSR with adhesive tape to bottom metal plate.
Also i glued with adhesive tape the M8 nut to the FSR.

If i pressed this FSR only between to flat plates they did not respond.
I dont know why. Maybe this FSR's have been a little bit special.
Maybe they only respond if the pressure is inside the outer 1mm Ring of this FSR's.
This outside Ring looks also a little bit higher then the inside of the fsr.

However. So it works very nice for me. I can aesy remove the Print Bed plate.
Exchange it maybe to some wooden plate for nylon in future. for better adhesive.
The plate is centered only between the 6 magetic standoffs.
So also the size of the print plate has not to be accurate.

I didn t want but had to change all parts. It still looks like john s simple 3DR but it is only the look.....

No exotic linear bearing system...I will use normal and cheap stainless steel shaft in combination with Brass bushings
No extrusion based or US hip hop frame but a simple 16 mm ALu-tube for the stability
No China bearings but german 8mm Brass bushings
No fishing lines...they are good for fishing...for printing...it is ok...if you have time to fish the print

The carriage has been changed to fit 8mm Brass bushings and a belt
The base has been changed to fit the alu tube
A new part has been added to form the under-base where the electronic will be housed

Still to do:

Where is the proximity sensor that will tell to the RAMPS that the alu bed has been touched....





PS: I see more and more new accounts created in the DELTA forum. The problem is that the "vocabulary" and "exprience" of the new users has nothing to do with newbies....so I am not yet sure...Is it a cartesian anxiety as Descartes described ? :-)

---

video of my 3DR

Hi all,

I am having a problem to get my 3DR printing flat. I have searched but never found something similar.

I have perfectly calibrated steps/mm and smooth rod radius. So I get the printer to accurately go to 0.1mm over the surface near each of the three towers and in the center of the build plate.

But, I still get this distortion:


As you see, the print is correct height near the towers (top, lower right and lower left). But on the bottom it is very squeezed to the build plate, and in the top left and top right the extrusion was going so high that the plastic didn't stick to the plate. Layer height is 0.2mm. So I assume that there is being a variation of more than 0.2mm lower in the bottom and 0.2mm higher in the top left and top right.

I don't know if it could be produced by unequal diagonal rod lengths. It is said everywhere that the rods should be exactly equal length, but, how much is exactly?? I have measured my rods, most of them are equal to +-0.02mm, one is -0.05mm. I don't think this could be the cause, but I'm out of ideas.

Oh, and I have just checked the table flatness with a machinists square and a 0.05mm thickness gauge and it seems to be flat enough.

Any idea anybody?

Edited 1 time(s). Last edit at 04/05/2014 06:49PM by payala.

I don t know if you have calibrated the printer, it is not clear in your post, but if you didnt this could help you:

Delta calibration by Jay Couture

---

video of my 3DR

Hi zacbot, thanks,. well, I have done what I said in the last post:

"I have perfectly calibrated steps/mm and smooth rod radius. So I get the printer to accurately go to 0.1mm over the surface near each of the three towers and in the center of the build plate." Which is what I understand for calibrating the printer,

To make it clearer I have done this:

1. Measured diagonal rod lengths and used that value in the firmware.
2. Adjusted steps/mm for each of the three carriages with the digital caliper.
3. Adjusted endstops to get accurate height near the base of each tower with a 0.1mm thickness gauge.
4. adjusted smooth rod offset to also get accurate height in the center of the build plate
5. repeated the two previous steps until accurate height was achieved near the base of each tower and in the center also

Of all these, the only thing strange is that to get the same height in the center I had to use a smooth rod offset value around 114 instead of the recommended 111, maybe that's a clue...

This is a bit more than what is explained in the video. I understand that doing this should already give accurate height and planarity, but it isnt happening. By the way, Im using Marlin latest official version but with the changes richrap did for the 3dr.

Calibrate the firmware to print flat

At this point, it is likely that if the firmware instructs the printer to move the carriage across the build surface, the nozzle tip will not stay true to the build surface. Even though the extruder will be at the calibrated correct Z height at each of the three tower locations, it will either be above or below the build surface at the center of the build area. This is also corrected by adjusting the firmware.

Run script D / click button D. The carriage should move up to home, then straight down to the center of the build surface. You will see one of three things. The extruder nozzle will be above the build surface, at just the right height (using the paper test), or will hit the build surface.

The value in the firmware constant DELTA_RADIUS (both Repetier and Marlin) controls the “flatness” of the movement of the carriage at a given Z height. If DELTA_RADIUS is too large, the extruder nozzle will track below the desired Z height inside the calibrated points A, B, and C. If DELTA_RADIUS is too small, the extruder will track above the desired Z height inside the calibrated points A, B, and C.

However, in both Repetier and Marlin (as the code is written), you don’t adjust DELTA_RADIUS directly. DELTA_RADIUS is calculated from the initial values entered earlier:

Read here the rest of this wiki

Hope it helps this time.

---

video of my 3DR

Hi Zacbot, thanks for your answer, but I already did that. It is step number 4 in the sequence I described.

Right now I'm having correct height in points A, B, C, and also point D the center of the build area.

The problem occurs at midpoint between A and B, between B and C, between C and A.

Let's assume that on your picture:
- tower X is on bottom-left and the build plate point near its base is A
- tower Y is on bottom-right and the build plate point near its base is B
- tower Z is at the top and the build plate point near its base is C
- centre is D
- the build plate point as much opposite of A as possible is α
- the build plate point as much opposite of B as possible is β
- the build plate point as much opposite of C as possible is γ

You probably have bad tower position Z (seems like it's radial position should be more far away from centre in firmware (because point γ is too low)). You probably also have a slight error in both diagonal rod length and delta radius errors which compensate each other.

If you are good at math and know a little bit of C and have a z-probe then read some maxima tutorial and try to execute this notebook (requires single-point measurements from a z-probe): [github.com]

If you are not good at math and know a little bit of C then read the comment cells at the beginning (till the first "input start" cell) of the notebook and try to calibrate it manually.

The bit of C knowledge is there so that you can set tower positions explicitly and for each tower differently in your Configuration.h.

If you are lazy to do it manually (or you do not know a little bit of C) and have a z-probe then you can try the firmware published in this thread: [groups.google.com]

Good luck.

This sounds familiar somehow :-) ... as mentioned quite some time ago, I have the same symptoms here, though - luckily - not too serious so I can live with it.

This is something even the mose careful calibration is unable to correct, and is caused by mechanical imperfections, usually towers not being spaced or angled perfectly. Alas this is hard to fix unless the design allows mechanical adjustments for those (and even if it did, it wouldn't be easy to measure it precisely with amateur means).

Ever wondered why Johann's autolevelling hard- and software doesn't go for dialing in the three tower home locations only (which you can to with careful manual calibration already), but probes (and then applies) a transformation matrix at many points all over the bed?

hercek, appreciate your update, I admit I haven't followed recent firmware & fine tuning efforts too closely. Do those already implement a way to compensate not only for incorrectly positioned, but for slanted towers, too?

As much as I admire Johann's ingenuity, I didn't implement his autolevelling system because of the added processing load is requires during printing (applying the transformaton matrix for every point on the tool path) on a CPU which is already working at its limits, and also because of the added complexity (and, to some extent, flexibility in Z direction) of adding FSRs to the three bed mointing points.

Quote
HaDe
hercek, appreciate your update, I admit I haven't followed recent firmware & fine tuning efforts too closely. Do those already implement a way to compensate not only for incorrectly positioned, but for slanted towers, too?

I'm not aware of any firmware which would compensate for slanted towers. New cartesian-delta conversion equations would need to be derived for this. If payala is good at math and can do a bit of C then he can derive them and update calculate_delta function accordingly. Correcting incorrect tower position is easy within the current cartesian-delta conversion equations.

Quote
HaDe
As much as I admire Johann's ingenuity, I didn't implement his autolevelling system because of the added processing load is requires during printing (applying the transformaton matrix for every point on the tool path) on a CPU which is already working at its limits, and also because of the added complexity (and, to some extent, flexibility in Z direction) of adding FSRs to the three bed mointing points.

CPU overhead of Johann's autoleveling is not that bad. It is only a few comparisons, additions and multiplications. It has much bigger problem though. Incorrect calibration will lead to errors in X, Y, and Z coordinate. Johann's autoleveling will mitigate only the Z-coordinate portion of the error. That is the reason I do not use it.
Richie's autoleveling tries to find the correct calibration parameters; i.e. it is better than Johann's if it does converge. Some people did not have luck with it but many reported good improvements.

It would be helpful if you post a video while the printer is printing. A copy of your configh.h will also help.

Anyway, since it is a 3DR, using printed parts, problems related to geometric of the printer are the last I would look after.

All my "flatness" issues were related to the bed surface flatness, leveling of the bed, rods length (firmware and physical), joints and the reflector.

For example if you change the original reflector and take the mini kossel reflector....you will have a flatness problem because in this case the rods of one arm will not be parallel to each other. If you change the reflector, you must change the carriage, so that the rods stay parralel to each other. If not, you get something like the above picture. In the middle ok but the far you go from the center the worse it becomes, specially in that zone between Tower X and Tower Y.

Update: if you make the video, please do not acivate the two fans you have installed at the bottom of the printer. When I look to the picture avove, I think that the fans are blowing the PLA away from the print surface before it lands on the surface of the bed.

Edited 1 time(s). Last edit at 04/07/2014 09:00AM by zacbot.

---

video of my 3DR

>> since it is a 3DR, using printed parts, problems related to geometric of the printer are the last I would look after <<

Dont't get me wrong and please note that I'm not about to question the quality of *your* (or anyone's) parts - but in general, with printer parts that came off another hobbyist's printer with unknown calibration history and accuracy, those would actually be the *first* thing I'd look after.

For example, I found the extrusion cavity in the base parts of my own 3DR very slightly leaning to the left, so slightly that you couldn't possibly detect this by eyeballing (and just barely by using a machinist's square), but still this slight imperfection would have a noticeable effect some 400 mm upwards and perfectly explain phenomenons such as "unexplainable" Z height deviations.

Edited 1 time(s). Last edit at 04/07/2014 09:16AM by HaDe.

Wow guys! thanks for that feedback!

All very good ideas, especially, the idea about bad z tower position looks really good. I'm at work right now, so I can't look into more details. But I will do when I get home, and you've given me some good ideas to try out, and some measurements to make.

Regarding math, C and measuring capability, it should be no problem in theory... I have a home workshop, and I work as an electronic engineer doing hardware and some software also. BUT (there had to be a "but") I am also the proud father of a 2yr old and a 4 month old that keep me really busy.

About the fans... I have to run them low speed if I don't want them to blow me away ;-) Actually, they are not running in the first layers, so that did nothing in this print.

I'll keep you guys updated on the results.

Yes, I agree with this too and I had also my "FUN" with my printed parts.....Thats why I exluded mecahnical printed parts later in my build, like the spool for the fishing line....But look, what I mean is, even if the printed parts that holds the frame are a bit smaller or a bit larger than the original design, then there shouldn t be a problem with the geometry as long as they are identical to each other.

Let us wait for the video and see...I think it is the reflector. I believe there is a fan at the hotend and the way it is hold makes me think it is the mini kossel reflector that has been installed but the carriage is the original from Rich. If this is true, this will explain why changing rods length in the firmware made things better, because the reflector has a different offset than the original and this will change the result of the radius calculated by the firmware. Anyway if it is the reflector, the problem will not disapear with calibration but only by either changing the carriage or -as I did- adding nuts and longer M3 at the carriage to add at least 10 mm spacing between the parallel rods.

---

video of my 3DR

Quote
payala
so that did nothing in this print.

ah ok, I just thought it was the fan, cause I didn t find any other logical explanation for the fact that the printed filament has left the circle exactly there where the first fan (on the right side of the picture) is blowing and came back to the track after leaving the blowing zone of that fan.

When it arrives to the second fan, this does not happen, probably because that fan is not blowing towards the print surface but away from it. It is aspirating and not blowing. It means you planed to blow frish air with one fan and aspirate the hot air with the other. That s why we see the front side of one fan and the back side of the other one on the picture.

But since this is all not correct, I will wait and see. :-)

Edited 1 time(s). Last edit at 04/07/2014 01:39PM by zacbot.

---

video of my 3DR

@zacbot

what carriages did you use for your belt drive on the 3dr? did you get better print quality with the belt?

I didn t change the carriage of the my 3DR. It is still the original one, just attached a belt to it instead of the fishing line. There was no problem here.

I didn t notice a big difference in terms of quality, but with the belt the setup time is much shorter, the maintenance is simple and the motor steps are correct. The fishing line is cool, cheap, makes less noise than belts.

---

video of my 3DR

@hercek So yesterday I had some time to tinker a bit with the printer and the maxima notebook you suggested. I modified m665 to get tower position corrections and store them to eeprom. But I still dont have z probe, so iterations are going slow.

Am I right that the process is iterative? You have to measure, feed the notebook with measured data, get new params, measure and start again?

Anyway, this has got me thinking about the problem of auto calibration in general, you just need to optimize your parameter vector to minimize z error across the build plate. And I was thinking about implementing it in firmware, so the printer gets in autocalibration mode and does the iteration to optimize the parameter vector. So, in fact, I was thinking about implementing the optimization with a genetic algorithm. As bad as the printer calibration can be, you start from a not so bad situation usually, and you just want to find a better point, the problem is that it is not intuitive what is the effect of each parameter change, and especially, the amount of change. Maybe an iterative optimization algorithm could find a good solution for itself. This would be very useful for people that cant get through the math to calibrate their printers, or people that have children like me and have no time ;-) What do you guys think?

But maybe, this should start to be talked about in a different topic.

Oh, and of course... thanks a lot for the resource, it was great!

Edited 1 time(s). Last edit at 04/09/2014 01:34AM by payala.

Quote
payala
Am I right that the process is iterative? You have to measure, feed the notebook with measured data, get new params, measure and start again?

Correct, unfortunately it turned out to be iterative for me. In ideal situation, it should spit out the correct values in one iteration. That was my goal. But it did not happen to me. My guess would be that mostly because of measurement errors. Since getting the data to the notebook without a z-probe will be even less precise, then I'm not sure how much it makes sense to try to feed the data with just manually finding the touch point and copying the data from the get current position command to the notebook. Maybe you can test the precision of your manual method by measuring a set of spots more times and check whether you are getting the same results. This would also require you can move each tower carriages by one micro-step (pronterface has minimum of 0.1 mm which is about 8 micro-steps for me). But even with imprecise data it may give you better hints how to move tower positions. I did not try it so I do not know.

But it did converge quite well for me (3 iterations were needed) and it did converge for Matt Peterson (I do not know how many iterations did he use). This are two cases I know about which showed good convergence. Richie did not finish his attempt to use the notebook. He did not use either maxima or the notebook correctly. Richard wanted to use it but he has given up before actually starting. Nobody else reported any feedback. The only difference (I was able to notice) between these people is that Matt was really good at math. Probably better than me (his responses were quick and insightful). He even made a Sage port of the notebook since he likes Sage Mathematics more than Maxima. This indicates that some math intuition to guess what to make it to converge may be needed.

If it does not converge for you then it can be also because you have a printer geometry error which the notebook cannot fix (e.g. skewed tower or build plate far from perpendicular to towers).

Quote
payala
Anyway, this has got me thinking about the problem of auto calibration in general, you just need to optimize your parameter vector to minimize z error across the build plate. And I was thinking about implementing it in firmware, so the printer gets in autocalibration mode and does the iteration to optimize the parameter vector. So, in fact, I was thinking about implementing the optimization with a genetic algorithm. As bad as the printer calibration can be, you start from a not so bad situation usually, and you just want to find a better point, the problem is that it is not intuitive what is the effect of each parameter change, and especially, the amount of change. Maybe an iterative optimization algorithm could find a good solution for itself. This would be very useful for people that cant get through the math to calibrate their printers, or people that have children like me and have no time ;-) What do you guys think?

Richie tried to find some heuristic to do this after he had given up his attempt with the maxima notebook. I know his first attempt had some errors I pointed out to him. He reported fixing it but I do not know how and I did not review his code. If you add a z-probe you can just try his firmware.
If you (or anybody else) would want to implement this then there are at least two options:

• Simulate manual callibration in firmware. This is what Richie is doing. This is not computationally intensive but it will take a lot of iterations. If you like this approach it may be a good idea to just branch his firmware.
• Use the approach in the maxima notebook. This is computationally expensive. You need to do it in a PC application communicating with the firmware but it should take only a few iterations to calibrate the printer.

Of course other approaches can be used.

....Of course other approaches can be used.

What is the difference between the actual repetier firmware and Marlin firmware regarding the Delta?

Do they have different approaches to drive, calibrate and correct Delta moves?

Edited 2 time(s). Last edit at 04/09/2014 12:21PM by zacbot.

---

video of my 3DR

repetier-firmware vs marlin vs sailfish
[forums.reprap.org]

Many thanks for the link. The thread started October 16, 2012 and seems to compares the firmware for cartesian printers. It would be nice to know, if since that time, the actual repetier firmware could resolve issues like the above and if yes how.

I saw in repetier firmware option that are not in actual Marlin but don t know for which purpose they are there, like Angle of column A/B/C 210 330 90....

Is this used to get tower position corrections with Repetier firmware ?

*/

/** \brief column positions - change only to correct build imperfections! */
#define DELTA_ALPHA_A 210
#define DELTA_ALPHA_B 330
#define DELTA_ALPHA_C 90

/** Correct radius by this value for each column. Perfect builds have 0 everywhere. */
#define DELTA_RADIUS_CORRECTION_A 0
#define DELTA_RADIUS_CORRECTION_B 0
#define DELTA_RADIUS_CORRECTION_C 0

Edited 1 time(s). Last edit at 04/09/2014 05:27PM by zacbot.

---

video of my 3DR

@hercek

I've been trying today with the manual method, going through the notebook went well and things were converging up to a point where I started to oscillate. This would be fine, but the errors were still very high, so, probably what you said about the hand measurements is happening here. But it was nice to go through the notebook and see your approach on the optimization problem.

I have done some measurements on the machine, and sure enough there are some significant diferences in the distances between the towers, so definitely that is at least one of the problems.

I will build a z probe for the machine and retry the notebook and maybe Richie's firmware, it also looks very interesting. And as much as I would like to try out an implementation for it, after more detailed look at the problem, I'm starting to appreciate that it is more complex than it looks at first sight. Though it is a very interesting problem!

Any recommendations about the z-probe?

Found a similar issue....

But at the ends repetier said:

Quote
Repetier
To be fair, the radius correction is not mathematically correct. It only computes the column coordinate based on the correction and uses this. I think the avr is not able to handle the correct compensation in a timely manner. So correction should really be small or it will hurt more then help.

---

video of my 3DR

And I must say, I choosed to build the 3DR, because of this matter. The frame is build with printed parts, so the error marge is very very small, compared to any hand made part.

Most starter will not be able to produce printed parts. It means someone else did the job. The buy it as a kit or get it from a profi. It would be interesting to know, who made the parts of the above printer. For me, they look very nice, I can t imagine, they have a wrong tower position.

That is, I still think the problem is coming from a part that has been made by human, like the diagonal rods, or a different reflector than the original planed in the design of the 3DR.

Edited 1 time(s). Last edit at 04/09/2014 08:47PM by zacbot.

---

video of my 3DR

When I talk about tower positions, I mean virtual tower positions. They may be wrong because of something of this is wrong:

• actual real tower position
• carriage dimensions
• platform dimensions

Correcting for different diagonal rod length (i.e. all of them do not have the same length) requires changing the math for converting cartesian coordinates to delta coordinates.
It is interesting that Repetiers radius correction is not mathematically correct. It is easy to do it correctly. At least in Marlin. I doubt Repetier uses same different math for delta conversion. One of the difference is that Repetier uses integers internally so it can be quicker but will have sooner problems with bigger printers. People on google deltabot group also claim Repetier uses acceleration and speed limits for platform while Marlin uses them for carriages.

I used only a temporary mounted probe only. I do not think a permanent probe makes sense. My experience indicates it is needed only once when calibrating. Some people need to recalibrate after taking out the glass and putting it back or after transporting the printer. This was never needed in my case.

Very good.

And that s why, i said at the beginning of this discussion, "it is the last I will look after". I didn t say, "it is not worth to look at it".

First check reflector, rods, carriage. And because they are not present on the picture of the print above, I asked for a video to see the whole machine build and all components at work.


Update:

But I must say, that your posts about tower position, made me aware of the next problem I may have with my new build of 3DR simple. The base frame is composed of 9 plastic peaces and not 3 pieces like the original 3DR. Here there is more room for erros, spcially tower position. Thank you.

Edited 1 time(s). Last edit at 04/10/2014 07:59AM by zacbot.

---

video of my 3DR

He!
I'm want to built a 3dr, I'm from France. I wonder if there is any frenchy around who want to do a group order for all the parts with me.