Folger Tech Kossel Delta 2020 Full 3D Printer Kit w/Auto-Level

I just know, that I downloaded it from the so-called "Google Drive" on 31. July 2015, and that it was the latest version then.

When checking the manual you are mentioning, I am seeing that it shows metal parts where my own printer (Kossel) has black plastic parts. My own manual shows black plastic parts, too.

But besides, no, my kit seems to be without that Z servo. The mentioning in the instructions irritated me. Alright, so I don't need there any jumper.

Side question 1: what is the price for getting such an auto level Z servo? Any URL?

Side question 2: if extruder E1 is not existing, I don't need to mount a small green stepper driver board on top of its slot, right? Or do I have to?

Edited 2 time(s). Last edit at 02/02/2016 07:26PM by Sven.L.

Regarding the servo - it's regular 9g servo, a lot of RC hobby shops sell them. Like this one [www.hobbyking.com]

But from my point of view this servo construction is not so good...

No need to install 5th driver, keep it as spare part.

Quote
Sven.L
But besides, no, my kit seems to be without that Z servo. The mentioning in the instructions irritated me. Alright, so I don't need there any jumper.
Side question 1: what is the price for getting such an auto level Z servo? Any URL?
Side question 2: if extruder E1 is not existing, I don't need to mount a small green stepper driver board on top of its slot, right? Or do I have to?

---

My [www.thingiverse.com] stuff, my configuration of Repetier 0.92.6: [forums.reprap.org] (use Arduino 1.6.5 to compile)

Thanks. That helped me. :-)

Quote
paul_delta
Just for information, here is my measurement of standard FT effector and carriage using provided STL files.
The mount with must be 40mm for both - the effector and carriage.

[attachment 71488 FT-Kossel-effector-carriage.jpg]

It's looking more and more like my effector plate is also to blame. Mine measures fine in the dimensions you show, but I think it's the alignment of the holes / warping that is to blame.

I'll know for sure when the aluminum one arrives Thursday.

-EB

Here is the response from Repetier regarding Heat probe issue in 0.92.8:
"I see. I have now modified it to heat only when enabled and then go to max(probe height, heat height) so that it should work in any case. This all came originally for sensor inside nozzle so zprobeheight was 0 and the problems never appeared (ok and I needed to heat anyway). "

So, it's fixed now (now it pays attention to Z_PROBE_REQUIRES_HEATING) and committed to Github:

[github.com]

---

My [www.thingiverse.com] stuff, my configuration of Repetier 0.92.6: [forums.reprap.org] (use Arduino 1.6.5 to compile)

Nice! Thanks for submitting that to them

In the mean time I wrote a little app to take the G33 L0 output from one configuration, normalize the values, and generate the commands to input the matrix back in. I'm sure this is a problem that's been solved a million times already, but it was easier to just write it than to go searching for it.. Anyway, I pulled in my previous Z correction and I'm getting near perfect prints. My only issue now is to fix the stepper drivers (I never set the voltage, and up until last night when I printed for 5 hours straight I never saw any issue).

Quote
paul_delta
Here is the response from Repetier regarding Heat probe issue in 0.92.8:
"I see. I have now modified it to heat only when enabled and then go to max(probe height, heat height) so that it should work in any case. This all came originally for sensor inside nozzle so zprobeheight was 0 and the problems never appeared (ok and I needed to heat anyway). "

So, it's fixed now (now it pays attention to Z_PROBE_REQUIRES_HEATING) and committed to Github:

[github.com]

I had an idea to write something like this - to communicate with printer via COM port, measure the matrix using G30 and use dc42 calculator algorithm to adjust eeprom values.
Can you share your solution?

Quote
spiffcow
Nice! Thanks for submitting that to them

In the mean time I wrote a little app to take the G33 L0 output from one configuration, normalize the values, and generate the commands to input the matrix back in. I'm sure this is a problem that's been solved a million times already, but it was easier to just write it than to go searching for it.. Anyway, I pulled in my previous Z correction and I'm getting near perfect prints. My only issue now is to fix the stepper drivers (I never set the voltage, and up until last night when I printed for 5 hours straight I never saw any issue).

Quote
paul_delta
Here is the response from Repetier regarding Heat probe issue in 0.92.8:
"I see. I have now modified it to heat only when enabled and then go to max(probe height, heat height) so that it should work in any case. This all came originally for sensor inside nozzle so zprobeheight was 0 and the problems never appeared (ok and I needed to heat anyway). "

So, it's fixed now (now it pays attention to Z_PROBE_REQUIRES_HEATING) and committed to Github:

[github.com]

---

My [www.thingiverse.com] stuff, my configuration of Repetier 0.92.6: [forums.reprap.org] (use Arduino 1.6.5 to compile)

Heh, I thought about that too.. But then it turned out to be a lot less work to just run the script, collect the values, and use the web page to calculate it. The dc42 calculator is good, but it could really use a genetic algorithm component.. Of course, that would only be practical with such an app to automate the process

Let me know if you embark on that idea.. An auto-calibrator for delta machines *would* be a handy tool.

Quote
paul_delta
I had an idea to write something like this - to communicate with printer via COM port, measure the matrix using G30 and use dc42 calculator algorithm to adjust eeprom values.
Can you share your solution?

Quote
spiffcow
Nice! Thanks for submitting that to them

In the mean time I wrote a little app to take the G33 L0 output from one configuration, normalize the values, and generate the commands to input the matrix back in. I'm sure this is a problem that's been solved a million times already, but it was easier to just write it than to go searching for it.. Anyway, I pulled in my previous Z correction and I'm getting near perfect prints. My only issue now is to fix the stepper drivers (I never set the voltage, and up until last night when I printed for 5 hours straight I never saw any issue).

Quote
paul_delta
Here is the response from Repetier regarding Heat probe issue in 0.92.8:
"I see. I have now modified it to heat only when enabled and then go to max(probe height, heat height) so that it should work in any case. This all came originally for sensor inside nozzle so zprobeheight was 0 and the problems never appeared (ok and I needed to heat anyway). "

So, it's fixed now (now it pays attention to Z_PROBE_REQUIRES_HEATING) and committed to Github:

[github.com]

Quote
spiffcow
Heh, I thought about that too.. But then it turned out to be a lot less work to just run the script, collect the values, and use the web page to calculate it. The dc42 calculator is good, but it could really use a genetic algorithm component.. Of course, that would only be practical with such an app to automate the process

Please explain how you think a genetic algorithm could improve on the optimised least-squares fit that my calculator implements.

Genetic algorithms are generally applied to problems for which there are no classical solutions, which is not the case here. If you take the view that minimising the sums of the squares of the errors is not an appropriate goal, then please suggest an alternative goal and why you think it it better.

---

Large delta printer [miscsolutions.wordpress.com], E3D tool changer, Robotdigg SCARA printer, Crane Quad and Ormerod

Disclosure: I design Duet electronics and work on RepRapFirmware, [duet3d.com].

It works great, but seems to get stuck in local minima. I was incorrect about my terminology though. It just needs a degree of backtracking and randomization of points in order to make it suitable for a 'hands off' tuning approach. I did not mean to imply that the algorithm itself was lacking.

Quote
dc42

Quote
spiffcow
Heh, I thought about that too.. But then it turned out to be a lot less work to just run the script, collect the values, and use the web page to calculate it. The dc42 calculator is good, but it could really use a genetic algorithm component.. Of course, that would only be practical with such an app to automate the process

Please explain how you think a genetic algorithm could improve on the optimised least-squares fit that my calculator implements.

Genetic algorithms are generally applied to problems for which there are no classical solutions, which is not the case here. If you take the view that minimising the sums of the squares of the errors is not an appropriate goal, then please suggest an alternative goal and why you think it it better.

And to clarify -- it is a supurb calculator and I am very grateful for it. I meant no offense whatsoever, and was only speculating about how it could be used in an automated environment.

Quote
spiffcow
It works great, but seems to get stuck in local minima. I was incorrect about my terminology though. It just needs a degree of backtracking and randomization of points in order to make it suitable for a 'hands off' tuning approach. I did not mean to imply that the algorithm itself was lacking.

Quote
dc42

Quote
spiffcow
Heh, I thought about that too.. But then it turned out to be a lot less work to just run the script, collect the values, and use the web page to calculate it. The dc42 calculator is good, but it could really use a genetic algorithm component.. Of course, that would only be practical with such an app to automate the process

Please explain how you think a genetic algorithm could improve on the optimised least-squares fit that my calculator implements.

Genetic algorithms are generally applied to problems for which there are no classical solutions, which is not the case here. If you take the view that minimising the sums of the squares of the errors is not an appropriate goal, then please suggest an alternative goal and why you think it it better.

Thanks for the clarification. How far off were your initial parameters? Do you have any evidence that there is another minimum other than the one that was found, that would give better results?

My calculator assumes that the machine obeys the equations of delta geometry, at least approximately. If it is used on a build that is so far off that those equations break down, then I concede that it could fail to give optimal results even after several iterations. Under those conditions (i.e. unknown kinematic equations for the printer), a heuristic or genetic algorithm could give a flatter printing plane. However, the movement algorithms implemented the printer would not match the mechanics either, so quality prints would not be possible.

Edited 1 time(s). Last edit at 02/03/2016 03:06PM by dc42.

---

Large delta printer [miscsolutions.wordpress.com], E3D tool changer, Robotdigg SCARA printer, Crane Quad and Ormerod

Disclosure: I design Duet electronics and work on RepRapFirmware, [duet3d.com].

After solving my initial problems with an incorrect effector width (40mm at the carriage, 42mm at the effector) my initial parameters were very close to the final result. I did find that the angle corrections were problematic and gave me slightly twisted prints. I checked other points that were the same radial distance as the suggested parameters and got a much larger error at that point. For the moment I set the angle correction to zero and have been getting very good prints (my entire frame is metal and leveled, so the angle error should be pretty minimal). There are any number of things that could be causing poor results -- minor variations in rod length, tilting of the effector, warped bed, etc. It may be that one factor is responsible for a larger error in another (e.g. concave bed measurement causing tilting of probe).

I have to run, but I'd love to pick your brain on this topic if you have the time

Quote
dc42
Thanks for the clarification. How far off were your initial parameters? Do you have any evidence that there is another minimum other than the one that was found, that would give better results?

My calculator assumes that the machine obeys the equations of delta geometry, at least approximately. If it is used on a build that is so far off that those equations break down, then I concede that it could fail to give optimal results even after several iterations. Under those conditions (i.e. unknown kinematic equations for the printer), a heuristic or genetic algorithm could give a flatter printing plane. However, the movement algorithms implemented the printer would not match the mechanics either, so quality prints would not be possible.

I'm glad you have improved your results. To keep the angle corrections at zero, set the number of factors to calibrate to 4. You can still use a lot more than 4 measurement points if you wish, to spread the height errors more evenly over the bed.

Another common problem is towers not quite perpendicular to the bed (leaning towers). The calculator will adjust the tower angles to correct for this to a first approximation, but it is only an approximation.

Yet another cause of calibration problems is that the 3 lines through the pairs of carriage bearings do not intersect at the same angle as the 3 lines through the pairs of effector bearings. It's not too difficult to print an effector with the bearing lines at 120 degrees to each other, but various errors can cause the carriages to have bearing lines that are not at 120 degrees to each other. An extra washer under one of the wheels of one of my carriages caused this type of error until I tracked it down.

---

Large delta printer [miscsolutions.wordpress.com], E3D tool changer, Robotdigg SCARA printer, Crane Quad and Ormerod

Disclosure: I design Duet electronics and work on RepRapFirmware, [duet3d.com].

Hi.

I have reached the point of construction, where it says I have to drill a pilot hole for the "screw of the servo bag". Does anyone have, by chance, a detailed photo of that screw and the hole to be drilled? There are two small screws there inside, and one very small screw. Correct? Which one to use, the smallest one?

What servo is that blue item, by the way? I assume, it is not a Z probe servo, or is it?



Edited 1 time(s). Last edit at 02/03/2016 05:46PM by Sven.L.

Plus one more source of problem - some sliders in FT kit are very tight, so it's impossible to put them on extrusion without slight bend. In this case the carriage will not be parallel to the extrusion surface and it will sit slightly off-center.

Quote
dc42
I'm glad you have improved your results. To keep the angle corrections at zero, set the number of factors to calibrate to 4. You can still use a lot more than 4 measurement points if you wish, to spread the height errors more evenly over the bed.

Another common problem is towers not quite perpendicular to the bed (leaning towers). The calculator will adjust the tower angles to correct for this to a first approximation, but it is only an approximation.

Yet another cause of calibration problems is that the 3 lines through the pairs of carriage bearings do not intersect at the same angle as the 3 lines through the pairs of effector bearings. It's not too difficult to print an effector with the bearing lines at 120 degrees to each other, but various errors can cause the carriages to have bearing lines that are not at 120 degrees to each other. An extra washer under one of the wheels of one of my carriages caused this type of error until I tracked it down.

---

My [www.thingiverse.com] stuff, my configuration of Repetier 0.92.6: [forums.reprap.org] (use Arduino 1.6.5 to compile)

Quote
Sven.L
What servo is that blue item, by the way? I assume, it is not a Z probe servo, or is it?
[attachment 71621 Photo7461.jpg]

Read carefully my messages, I gave you the link to exactly the same servo [www.hobbyking.com]

---

My [www.thingiverse.com] stuff, my configuration of Repetier 0.92.6: [forums.reprap.org] (use Arduino 1.6.5 to compile)

PS: And yes, this is z-probe servo. HXT900 is one of the best for money (in $3-$5 price range) , I use them a lot in my RC planes.

Edited 1 time(s). Last edit at 02/03/2016 06:04PM by paul_delta.

---

My [www.thingiverse.com] stuff, my configuration of Repetier 0.92.6: [forums.reprap.org] (use Arduino 1.6.5 to compile)

Thanks again, Paul. I saw that URL, but sometimes, when at work, I am using a company computer to keep me up to date, and for visiting some URLs, I have to request it each time from a system administrator. That is, why I sometimes have to wait until "home hours" to follow up an URL or for a lunch time.

Okay, seeing this blue part now on Hobby King's website, I got a similar one with the Kossel kit. And I think, that this means, I have to set some sort of jumper on that triple connection of the RAMP board, right?

If so (please verify), then I would use three of my interconnected hollow needles. Or must I connect only pin 1 and 2, or 2 and 3? Please advise. Their PDF manual doesn't give here any visible or described detail, unfortunately.

My instruction PDF showing the Z probe...


Edited 2 time(s). Last edit at 02/03/2016 06:07PM by Sven.L.

Quote
Sven.L
Okay, seeing this blue part now on Hobby King's website, I have a similar one. And I think, that this means, I have to set some sort of jumper on that triple connection of the RAMP board, right?

If so (please verify), then I would use three of my interconnected hollow needles. Or must I connect only pin 1 and 2, or 2 and 3? Please advise. Their PDF manual doesn't give here any visible or described detail, unfortunately.
[attachment 71622 Photo7461b.jpg]

The z-probe switch should be connected between ground (-) and signal (S) pins. Regarding the servo - it's better to power +5 servo from separate source and wire only ground and signal wires to the RAMPS.

---

My [www.thingiverse.com] stuff, my configuration of Repetier 0.92.6: [forums.reprap.org] (use Arduino 1.6.5 to compile)

For +5v source you can use 12v->5v DC-DC step down converter, something like this one [www.ebay.com]
It gives 3A and it's enough for this servo.

---

My [www.thingiverse.com] stuff, my configuration of Repetier 0.92.6: [forums.reprap.org] (use Arduino 1.6.5 to compile)

PS: make sure that Z-min (z-probe) pin in firmware configured with pullup option #define ENDSTOP_PULLUP_Z_MIN true

Edited 1 time(s). Last edit at 02/03/2016 06:27PM by paul_delta.

---

My [www.thingiverse.com] stuff, my configuration of Repetier 0.92.6: [forums.reprap.org] (use Arduino 1.6.5 to compile)

Paul, that means, that the RAMPS board cannot handle the additional energy demand of that probe? Or what is the reason not to connect its power cables to the RAMP board?

I am asking, because it means I would first have to purchase an energy converter that changes 110 V to 5 V.

NOTE
Saw your two additional answers after I posted this.

Edited 1 time(s). Last edit at 02/03/2016 06:30PM by Sven.L.

Actually, it can, but if you will add LCD that draws a lot - you may experience arduino resets.

Quote
Sven.L
Paul, that means, that the RAMPS board cannot handle the additional energy demand of that probe? Or what is the reason not to connect its power cables to the RAMP board?

I am asking, because it means I would first have to purchase an energy converter that changes 110 V to 5 V.

---

My [www.thingiverse.com] stuff, my configuration of Repetier 0.92.6: [forums.reprap.org] (use Arduino 1.6.5 to compile)

Thanks for the information. Now I understand the situation.

Hmmmm, I assume these resets of the Arduino board will happen during printing, right? And thus cause loss of filament if this type of printer or the software do not start where it was interrupted before.

But I will give a try (as shipping of Ebay items to Jamaica takes usually 3 to 4 weeks). I got an LCD screen with the set, and take the risk. If it happens too often or the results cause too much loss, I will switch to that external power supply for the servo. By the way, that particular one you linked to on Ebay, it does not ship to Jamaica. But I am sure, there are many others. But I am making tests first with the regular construction before switching to this external solution. Keeping it in the back of my mind, of course. :-)

Alright, what about that triple pin for the servo on the RAMP board? How to I have to connect them with a jumper? All three of them, or only two of them (and which ones)?

Edited 2 time(s). Last edit at 02/03/2016 06:44PM by Sven.L.

That's what I did toward the end. What was interesting to me was that the rod calculation was actually incredibly helpful. I got the best results by doing several iterations of 7 factors with 10 points, then improved it by removing the angle corrections and going back to 4 factors. I can't seem to get my measured results better than 0.04 though (with or without the angle corrections). I have one rod that's harder to turn than the others, and I suspect it's ever so slightly longer than the rest of the rods. I'll take a look at the towers to make sure they're perfectly level though. I frequently use a level to check everything, but maybe something is slightly off. I'm also wondering if I'm just bumping up against the accuracy of the measurements and/or the flatness of the bed itself. I'll also look into the angles of the carriages (not quite sure how to get an accurate measurement on that one though).

Is your calculator code available? I'm wondering what would happen if I just plugged 100+ points into the algorithm..

Quote
dc42
I'm glad you have improved your results. To keep the angle corrections at zero, set the number of factors to calibrate to 4. You can still use a lot more than 4 measurement points if you wish, to spread the height errors more evenly over the bed.

Another common problem is towers not quite perpendicular to the bed (leaning towers). The calculator will adjust the tower angles to correct for this to a first approximation, but it is only an approximation.

Yet another cause of calibration problems is that the 3 lines through the pairs of carriage bearings do not intersect at the same angle as the 3 lines through the pairs of effector bearings. It's not too difficult to print an effector with the bearing lines at 120 degrees to each other, but various errors can cause the carriages to have bearing lines that are not at 120 degrees to each other. An extra washer under one of the wheels of one of my carriages caused this type of error until I tracked it down.

Edited 2 time(s). Last edit at 02/04/2016 12:26AM by spiffcow.

Quote
Sven.L
Alright, what about that triple pin for the servo on the RAMP board? How to I have to connect them with a jumper? All three of them, or only two of them (and which ones)?

RAMPS has 4 servo ports near reset button. Just connect the servo wire (with extension) paying attention to polarity. From the side of board pins are -,+,S.
HTX900 servo wire colors are like one in the middle:

---

My [www.thingiverse.com] stuff, my configuration of Repetier 0.92.6: [forums.reprap.org] (use Arduino 1.6.5 to compile)

BTW using the servo, it's easy to make an arm that will hold the switch right under the nozzle. This way you can eliminate the probe offset errors because of effector tilt.

---

My [www.thingiverse.com] stuff, my configuration of Repetier 0.92.6: [forums.reprap.org] (use Arduino 1.6.5 to compile)

Thanks, but your information is about the connector (plug). But I need to know about the triple pin, and how to apply a (self-made) jumper to any of those three pins. See the photo, top right. The Kossel instruction manual says, a jumper has to be set there for the Z probe, but doesn't give any more information...

Quote
Sven.L
Thanks, but your information is about the connector (plug). But I need to know about the triple pin, and how to apply a (self-made) jumper to any of those three pins. See the photo, top right. The Kossel instruction manual says, a jumper has to be set there for the Z probe, but doesn't give any more information...
[attachment 71625 Photo7460.jpg]

I have no idea what manual are you referring to, but the only jumpers you have to set on RAMPS - are jumpers for drivers. 3 per driver. That's it.

---

My [www.thingiverse.com] stuff, my configuration of Repetier 0.92.6: [forums.reprap.org] (use Arduino 1.6.5 to compile)