Delta cal torture

Hey all.

So I set out to design and build my own printer a while ago. It's been a long, time-consuming process, and I've been stuck here at the finish line for a while now. I've tried everything I could thing of but for the life of me I can't get the damn thing to print flat. It shows some peculiar behavior while calibrating. Quick list of things that seem relevant before I go into the details. It's based heavily off of the 3DR and the Kossel.

-Ramps 1.4
-Pololu drivers
-Spectra line drive
-CF rods/ Traxxas joints
-Cerberus pup carriages w/ openbuild wheels
-3DR Idlers
-Marlin
-Heated plate fastened to Kossel-esque frame
-Opto-endstops

A couple examples of weird behavior.

-When I get to M666 calibrating my Y tower, it will be wayyyy lower than it should be and I have to send the extruder to the center and back, or home and back, and it will come back to the Y tower at the right height. Like I had to shake something off or something. Really inconsistent.

-Sometimes one the M666 offset values for one of my towers will require an unusually large adjustment and then after adjusting it finer finer, I end up back close to or at the original value.

-When all 4 cal points pass the paper test(post-it note folded in half), my prints still seem to come out way off. Like adjusting doesn't have any logical effect on it. I'll attach a picture of one of my cal prints.


I've tried switching surfaces, switching carriages and wheels, switching from magnet joints to the traxxas joints, etc....Does anyone have any advice? I'm about to lose it. Thanks in advanced.

Edited 1 time(s). Last edit at 10/07/2015 09:11PM by Tennysballs.

Have you checked that the bed is totally flat?

Are the arms exactly equal length?

Do all towers have the same offset to the center?

Eliminated all play/sloppyness in joints carriages etc?

Are you heating up your bed when printing, from how you have fastened it it might get deformed as it heats up.

I am not familiar with the spectra-line but in my head it would be very hard to get it to be real tight.

How accurate is the spectra line assembly? Sounds you might have precision problems in your stepper-drive train.
Can you send several 0.1mm z-moves and actually feel every one of it?
Then send the same moves in the opposite direction. Do you see/feel the nozzle move immediately?

Try to mod the drivers with the "R4 decay hack" and or use 0.9° steppers to get finer and trustworthy microsteps.
-Olaf

Thanks for the replies.

I have not check the bed. I guess I had assumed it was flat. Is that really something I need to do?

As far as build related problems, I have tried to be as precise as I could, and the spectra lines are nice and tight.

I noticed from a previous design that using the heatbed messes with things way too much, so I haven't even thought of trying it yet.

Olaf. I think you might be on to something here. I'm not sure why I just brushed it off, but I did notice sometimes I will have to press the .1mm button on pronterface several times before it makes a definite movement. I'm currently using 1.8 degree steppers. I looked up the decay hack but I don't quite understand the need for it. What is causing the problem? Do I need to tune my steppers? Because I haven't even really touched them other than to make my motors quiter....Did mess with the precision when I did that?

What is the current rating of your stepper motors, and what is the motor current set at? You should aim to run the motors at 60 to 80% of their rated current. Much lower and the microstepping will be less accurate.

---

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].

I have my steppers set 16th steps. Which I kind of arbitrarily chose. Poor choice?

I'm having a hard time finding instruction on how to tune the A4988s. Just a lot of discussion. How do I go about that?

Quote
Tennysballs
I have my steppers set 16th steps. Which I kind of arbitrarily chose. Poor choice?

I'm having a hard time finding instruction on how to tune the A4988s. Just a lot of discussion. How do I go about that?

Check under "Tuning motor curent"

And as dc42 mentioned 60 to 80% of your motors rated current is what you should aim for but not more than 1-1.2A though that might cause overheated drivers.

Did you notice, when you sent a few 0.1mm steps, that the motor wouldn´t move much but suddenly jumps 0.3-0.4mm?
The decay hack is made to improve this microstepping torque.
Another problem is missing steps from undercurrent, then the stepper also wouldn't move, but does a 0.1mm step eventually.
But then you wouldn´t have the chance to move "somewhere" and return to the tower and be spot on. ( like you described it)

Especially for the DRV8825 the hack made a huge difference for me.
I´m currently writing a weightlifting test procedure with several different drivers and hacking/cooling methods. # see attached pic

What are your steps set to in the firmware? And how many teeth are on your pulleys?

Quote
davidf01
What are your steps set to in the firmware? And how many teeth are on your pulleys?

Not really applicable here as he has built using spectra line.

Olampe, Yeah that does sound pretty similar to what I am experiencing. Could be an under current issue. When you say decay hack, is that the same thing as the r4 hack? Should I try and tune the pots before I do that? Because from what I have been reading, I can't really find a solid value for what vref should be.

Edited 1 time(s). Last edit at 10/09/2015 03:19PM by Tennysballs.

Quote
Tennysballs
Olampe, Yeah that does sound pretty similar to what I am experiencing. Could be an under current issue. When you say decay hack, is that the same thing as the r4 hack? Should I try and tune the pots before I do that? Because from what I have been reading, I can't really find a solid value for what vref should be.

What Vref you should set is depending of what your motors are rated for, as mentioned earlier you should aim for about 60-80% of your motors rated current.
Wiki about the pololu board, check under "Tuning motor current" for the math to calculate appropriate Vref for your motors.

I don't trust the wiki-formulas regarding the DRV8825 too much. I had a DRV8825 running stable at 2.4V on a 1.7A stepper. Regarding the formula, this would have been 4.8A ?!
IIRC the formula was Vref = Imax *2....
But the A4988 formulas look right to me as long, as you identify the Rs value of your drivers.
-Olaf

Quote
o_lampe
I don't trust the wiki-formulas regarding the DRV8825 too much. I had a DRV8825 running stable at 2.4V on a 1.7A stepper. Regarding the formula, this would have been 4.8A ?!
IIRC the formula was Vref = Imax *2....
But the A4988 formulas look right to me as long, as you identify the Rs value of your drivers.
-Olaf

Take it from Pololu themselves then: [www.pololu.com]

Edited 1 time(s). Last edit at 10/10/2015 04:32AM by Koenig.

I have a Rostock mini with the same problem. I did the A4988 mod like o_lampe suggested but nothing changed (besides motor sound?)

Do you have any backlash in the traxxas joints?

If I barely touch the effector at Z0 the nozzle will move and change position, I can feel it grabbing the paper harder.

My data sheet doesn't say anything about Max current. Just current per phase. 1.5A. Does that sound right at all?

And I'm not sure about the backlash. Maybe I don't understand backlash, but I thought it was only really applicable to gears

Quote
Tennysballs
My data sheet doesn't say anything about Max current. Just current per phase. 1.5A. Does that sound right at all?

That would be it.

Quote
Tennysballs
And I'm not sure about the backlash. Maybe I don't understand backlash, but I thought it was only really applicable to gears

Translate it to "play in moving parts", all of that needs to be eliminated to be able to do an accurate calibration.

Edited 1 time(s). Last edit at 10/10/2015 02:47PM by Koenig.

That was super helpful. Thank you. I believe I got rid of all the backlash. I'll work on it tonight and report. Thanks again.

So this kind of sucks. But in a way it's good because it's something. So I tuned my motors to ~480mv Vref using the following data.

Current/phase of motors = 1.5A
70% of 1.5A = 1.2A

Vref = I_TripMax * 8 * Rs
= 1.2A * 8 * .05ohms
=480mv


Here is the interesting thing I noticed. The carriage only moved every other time I clicked the .1mm jogger, with the exception of the 11th time I pressed it. I used M114 to confirm however that each press was being registered as shown by a decreasing of my Z height by .1mm.

So if each character in this string represented one instance of me pressing the .1mm button, and a O = no movement and X = movement of .1mm, it would look like this. Consistently.


X O X O X O X O X X

Did I tune them right? Any ideas what would cause this behavior?

Edited 1 time(s). Last edit at 10/11/2015 06:14PM by Tennysballs.

Quote
Tennysballs
So this kind of sucks. But in a way it's good because it's something. So I tuned my motors to ~480mv Vref using the following data.

Current/phase of motors = 1.5A
70% of 1.5A = 1.2A

Vref = I_TripMax * 8 * Rs
= 1.2A * 8 * .05ohms
=480mv

...

Did I tune them right?

You have assumed that your drivers use 0.05 ohm sense resistors. A few do (AFAIR the Pololu Black Edition and T3P3 Ice Blue), but most others use 0.1 ohm resistors. So you may be running your motors at only 40% of their rated current. If you really are running them at 1.2A then the motors will be getting warm and the drivers will be getting very hot. If everything is running cool then try 800mV instead.

In your calculation, I presume you meant 80%, not 70%.

Edited 1 time(s). Last edit at 10/12/2015 03:39AM 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].

I assumed that because that's what was in the equation in the reprap wiki. I'll put it through a dry run tomorrow morning, adjust accordingly, and report. Thanks.

Oops. That was absent minded. Thanks for pointing that out.

Quote
dc42

Quote
Tennysballs
So this kind of sucks. But in a way it's good because it's something. So I tuned my motors to ~480mv Vref using the following data.

Current/phase of motors = 1.5A
70% of 1.5A = 1.2A

Vref = I_TripMax * 8 * Rs
= 1.2A * 8 * .05ohms
=480mv

...

Did I tune them right?

You have assumed that your drivers use 0.05 ohm sense resistors. A few do (AFAIR the Pololu Black Edition and T3P3 Ice Blue), but most others use 0.1 ohm resistors. So you may be running your motors at only 40% of their rated current. If you really are running them at 1.2A then the motors will be getting warm and the drivers will be getting very hot. If everything is running cool then try 800mV instead.

In your calculation, I presume you meant 80%, not 70%.

Check which you have, look under "Current sense resistors" on this page for the how and where to check that.

EDIT: This link will take you directly to the "Current sense resistors" part.

Edited 1 time(s). Last edit at 10/12/2015 04:03AM by Koenig.

So I do indeed have the .1ohm resistors. I re-tuned everything it's still doing the same thing as before......Just a little louder.

Since the problem was persisting, I jumped into the firmware(Marlin) to look around. I'm using 16th steps. Since I am using Rich's spool design from his 3DR(and his default steps per unit), I double checked to see if maybe he was using a different step rate and if changing it would make a difference. I couldn't really find any solid information on it, but it seemed that he might have been using 8th steps. I doubled his 55.5 steps/unit just to see what happened and my problem of inconsistent response went away-it would move every time I would press the .1mm button. The problem was that when I got to what should be Z0 in M114, I was at about half of my total Z height. So I guess I had the right step rates. I'm not sure what to make of all that....

Quote
Tennysballs
Since the problem was persisting, I jumped into the firmware(Marlin) to look around. I'm using 16th steps. Since I am using Rich's spool design from his 3DR(and his default steps per unit), I double checked to see if maybe he was using a different step rate and if changing it would make a difference. I couldn't really find any solid information on it, but it seemed that he might have been using 8th steps. I doubled his 55.5 steps/unit just to see what happened and my problem of inconsistent response went away-it would move every time I would press the .1mm button. The problem was that when I got to what should be Z0 in M114, I was at about half of my total Z height. So I guess I had the right step rates. I'm not sure what to make of all that....

That is pretty much what is expected to happen if you double the steps/mm...

Then the question is, which of the two scenarios match your physical Z-height?

However, when I come to think of it, I had a similar problem a while ago when I changed from 1.8 degree steppers to 0.9.
Turns out I had a slightly loose pulley.....

55.5 seems to be the correct one. But it is also the one where I'm having the inconsistent response from pronterface. So I'm not sure what to do from there. Could have something to do with some other firmware parameters? Acceleration or feed rate?

I know nothing is loose because the problem follows a consistent pattern that I previously outlined.

55.5steps/mm is very low. Most delta printers using 1.8deg/step motors have either 80 or 100 steps/mm. What pulleys and belts are you using?

---

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].

18mm diameter spools with spectra line. 16th steps.

Quote
Tennysballs
18mm diameter spools with spectra line. 16th steps.

I guess your 18mm diameter spools are bigger in diameter than the 16 or 20 tooth pulleys generally used with GT2 belts on Kossels and their derivatives.

---

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].

Guess so.

Either way I'm still stuck on my responsiveness issue.

You have two choices now:
switch over to GT2 belt and 16 tooth pulleys
or
Buy 0.9° steppers

Both options help you to increase accuracy.

What was Rich thinking, to build a pulley with xx.5 steps? What would the driver do with half a (micro-) step?
-Olaf