Kossel - all axes calibrated, but height is wrong

Hi,
I'm using a mini kossel with first marlin and now reprap-host. I've got all axes calibrated very good, so any print will come out in the correct length and width - but not height! For example, the 10x10x50mm calibration object is printed 10,0x10,0x62,5.
Or even without printing, just moving: every move horizontally is made correct, every move vertically is wrong. If i let it move from z=0 to z=50, it's 62.5 when i measure it.
If all dimensions would be wrong, i'd say ok, it's the steps/mm. But this way, I've got no idea. Since all movements are interpolated between the axes, i have no idea how this can happen, please give me a hint!

Here some values from my configuration.h, maybe someone with a mini kossel with v-rollers can check if his values are in the same range (first from repetier-host, second marlin):

#define DELTA_DIAGONAL_ROD 228 // mm
#define DELTA_ALPHA_A 210
#define DELTA_ALPHA_B 330
#define DELTA_ALPHA_C 90
#define DELTA_MAX_RADIUS 150
#define DELTA_RADIUS 94.6
#define DELTA_HOME_ON_POWER 0

#define DELTA_DIAGONAL_ROD 228.0
#define DELTA_SMOOTH_ROD_OFFSET 134 
#define DELTA_EFFECTOR_OFFSET 19.9 
#define DELTA_CARRIAGE_OFFSET 19.5

The marlin config is a bit older, the repetier-part might have changed a bit, but it gave similar results.

Edited 1 time(s). Last edit at 03/10/2014 08:13AM by stahlsau.

You have probably wrong steps per mm, delta radius and diagonal rod length. Wrong delta radius and diagonal rod length together can partially compensate for wrong steps per mm in XY direction.

Level the bed first, it must be leveled at least at these points:
* as near to each tower base as possible,
* as far away (across the bed) from each tower as possible,
* in the bed center.

Your calibration object is too small to show XY error. But you probably cannot even print an object which is e.g. 0.5 mm tall but covers whole bed (because your bed is not leveled everywhere).

well I guess you are right. I've calculated my steps per mm to 100, but in fact, it should be 80. And that's exactly the factor from 50 to 62.5.
I somehow managed to get the bed leveled correctly, and the axes too, depite the wrong steps/mm. And that was it what kept me from thinking the steps could be wrong...the horizontal dimensions were right. I've printed a fan cover for a 120mm fan which came out pretty nice and fitting (it was only 10mm in height, so the error didn't make much difference), and i've printed lots of vases and stuff, where the height-dimensions aren't critical ;-).


So, back to calibrating, now with the right steps per mm, and many thanks for the idea.

I have the same problem, when I print a object X40mm Y40mm Z50mm height the XY dimensions or correct but the height is bad.
I measure Z53mm when the print is ready
I change the steps\mm of the z axis but this give not the good result.

Print a cone with bottom radius of about 1 cm and top radius about 0.5 cm. Make it about 5 cm high. Measure the height with callipers. If the height is not correct then your steps per mm are wrong. Fix steps per mm first. It does not make sense to calibrate anything else if your steps per mm are wrong. When your steps per mm are correct then go on calibrating diagonal rod length and delta radius (or even the tower positions if your geometry is not symmetric) so that bed level z-height is flat. If your bed is flat (and perpendicular or almost perpendicular to towers) then you do not need to care about X/Y dimensions. Just calibrate the bed level z-height and when done then the X/Y dimensions must be correct too. That is a feature of delta printers X, Y, and Z errors are all tied together. When one of them is correct across whole print area then the other 2 will be correct too.

I am having some similar issues.

My X and Y are calibrated almost perfectly. However my Z hight is about 6% off. I can make a 25mm cube. The hight is 26.35mm. The X and Y are 25.00+/-.03. If my steps/mm where 80 insted of 100. I should see 31.25. I tightened belts and verified my that my steps/mm to be 100.

Anybody have an Idea where I should go next?. I am thinking my current setting may be wrong.

I am using RAMPs 1.4 with 16x micro microstepping.

Quote
AdrianW
I am having some similar issues.

My X and Y are calibrated almost perfectly. However my Z hight is about 6% off. I can make a 25mm cube. The hight is 26.35mm. The X and Y are 25.00+/-.03. If my steps/mm where 80 insted of 100. I should see 31.25. I tightened belts and verified my that my steps/mm to be 100.

Anybody have an Idea where I should go next?. I am thinking my current setting may be wrong.

I am using RAMPs 1.4 with 16x micro microstepping.

It's DELTA_DIAGONAL_ROD that sets the X and Y dimensions. It's your steps/mm that adjusts your Z height and that is a function of your stepper motor's degrees per step, the type of belt you use, the number of teeth on you pulleys and the micro stepping you set on your stepper drivers

For normal 1.8 degree motors, 1/16 micro stepping, 20 tooth pulleys and GT2 belt the correct value is 80

Motor steps per revolution - (360 / 1.8) x 16 = 3200

GT2 belt has a 2mm tooth pitch x 20 teeth (pulley) = 40mm per revolution

3200 / 40 = 80

Assuming your slicer is operating correctly then you could alter your steps per mm to suit your result. If 100 steps results in a 25mm model being 26.35mm then :-

(Current steps/mm / target height) x actual height = correct steps/mm

(100 / 25) x 26.35 = 105.4

But that is a weird number if you are using belts and pulleys

Quote
AndyCart
(Current steps/mm / target height) x actual height = correct steps/mm

(100 / 25) x 26.35 = 105.4

Surely it's the other way round? He wants fewer steps, not more, so he needs to reduce his steps/mm to 25/26.35 x 100. But it still gives an odd number.

Edited 1 time(s). Last edit at 03/14/2015 05:20AM 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].

Deleted

Edited 1 time(s). Last edit at 03/14/2015 05:19AM 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].

Quote
dc42

Quote
AndyCart
(Current steps/mm / target height) x actual height = correct steps/mm

(100 / 25) x 26.35 = 105.4

Surely it's the other way round? He wants fewer steps, not more, so he needs to reduce his steps/mm to 25/26.35 x 100. But it still gives an odd number.

Oops. Of course you are right.

(Current steps/mm / actual height) x target height = correct steps/mm

(100 / 26.35) x 25 = 94.88

AndyCart,

I am using Belts an pulleys. I have GT2 Belts, 16 Tooth pulley, 200 steps per rev, and 16 uSteps per step. 6% off is just odd. I thought I was skipping steps. But after setting the current, tightening belts, and slowing the travel speed nothing changed.

I adjusted my steps/mm to 94.1 based on actual measurments. So I am 5.9% off. Now I am at 25.04mm when I set a Z travel of 25mm.

My thoughts now are this. I got this mini-kossel kit from MakerGeeks. I think that this kit is full of the low cost Chinese parts. The steppers are specked to be +/-5% per step. I believe my problem is a tolerance stack up issue.

I am totally and completely unimpressed with MakerGeeks them. Not because of my current problems but for many others I have had with the company and not just the kit. But this is not the right place for that comment.

Thanks for your help. I am re-calabrating the rest of the printer withe the new travel.

Have you measured the actual tooth pitch of the belts? The number of steps per rev on a stepper motor is not subject to tolerance.

Perhaps you have the motor jerk speed or acceleration set too high, so that you should really be using 80 steps/mm but you are missing a few steps on each layer change. OTOH then I would expect you to see problems in the X and Y directions as well.

---

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].

dc42,

I have measured the belts, and the pulley teeth. all as it should be, 100 steps/mm. I have read that MakerGeek may have given me the wrong motors. The current for these is around 400mA in steed of the suggested 1500mA.

The jerk and acceleration is a good Idea. However, I run the printer very slow because the extruder can't keep up.

Have you physically counted the number of teeth on the pulleys.

I know it sounds daft but I had a similar issue on my Prusa with the sizes until I did count the teeth and found that the Kit I had was supplied with 17 tooth pulley's

I know what an odd size but once I re-calculated it for 17 tooth pulley's all came out ok

Rep Host belt calculator gives the correct value as 94.1176.

HTH

Doug

Quote
AdrianW
dc42,

I have measured the belts, and the pulley teeth. all as it should be, 100 steps/mm. I have read that MakerGeek may have given me the wrong motors. The current for these is around 400mA in steed of the suggested 1500mA.

The jerk and acceleration is a good Idea. However, I run the printer very slow because the extruder can't keep up.

All I have to say is OMG.

I put a mark on the pulley and told the printer to move (5 rev worth of distance). The mark better end up in the proper place. I will count the teeth.

My fudge factor make a steps/mm number is so close to your calculated number that this must be the problem. It is worth taking the printer apart to count them.

Thanks for your insight.

Adrian

Any news on this?

Doug

Adrian / all,

I've got a makergeeks kossel and have exact same issue, used for months but only started printing precision parts now. 25mm object comes out 26.6mm tall.

Same deal with the motor too, 0.4A when general design recommendation is 1.5A (bought some, haven't put them in, prints well enough if i keep acceleration <1500).

....and son of a...17 teeth on these pulleys!

Another thing in a long line of 'wtf' moments on this build kit...

Going to re-compile and try now, might have just solved quite a few things with that.

Man, the same thing happened to me too. I had purchased GT2 16 teeth pulleys, or so I thought. After calibrating the Kossel for days I just could not get the proper height. After reading this post I took out the pulley and counted the teeth. It turned out the pulley has 20 teeth instead of 16. Now the print dimension is spot on.

The moral of this story is to not take it for granted to assume what you get is what you purchased. Verify everything!

Man, the same thing happened to me too. I had purchased GT2 16 teeth pulleys, or so I thought. After calibrating the Kossel for days I just could not get the proper height. After reading this post I took out the pulley and counted the teeth. It turned out the pulley has 20 teeth instead of 16. Now the print dimension is spot on.

The moral of this story is to not take it for granted to assume what you get is what you purchased. Verify everything!

This was very helpful! I have a makergeeks delta machine as well. It has been difficult and a learning experience.

The 17 teeth hint was helpful. I think that has solved my vertical height issue. I may have servos that aren't per spec either. On bigger prints they tappear to overheat and skip steps. I had put in the 1500 limit on acceleration, and have also started gradually turning the potentiometers on the motor drivers counter-clockwise to reduce the current to the servos.

The optical switch was broken, so I was unable to level the bed.

I also replaced the hotend with an E3D version. It is so much better. Next I am going to replace the carriage and the hot end holder, and I have ordered a microswitch so I can back to the original bed leveling design.

I´m confused about the 17/16 teeth counts...
Do I count the teeth of the pulley or the dents? Because a 17 tooth pulley would have 16 dents, right?

Is it worth using smaller pulleys, like 12 or 10 teeth?
I think the carrier wouldn´t like a smaller gap and the belt would scratch there.
But the carrier rarely gets close to the stepper, so it´d work?
-Olaf

Quote
o_lampe
I´m confused about the 17/16 teeth counts...
Do I count the teeth of the pulley or the dents? Because a 17 tooth pulley would have 16 dents, right?

No, the numbers of teeth and dents are the same.

Quote
o_lampe
Is it worth using smaller pulleys, like 12 or 10 teeth?

I've not heard of GT2 pulleys for stepper motor shafts with fewer than 16 teeth. To get better resolution, you can go to 0.9 deg/step motors. But then you will need a 24V supply and 32-bit electronics to generate the stepper motor pulses fast enough and counter the increased back emf from the motors, or accept a reduction in maximum speed. I have just done this upgrade to my Kossel (see my blog for details).

Edited 1 time(s). Last edit at 08/10/2015 09:05AM 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].

The tooth count was my issue.

Thank You all (very belated) for your help.

I have noticed the MakerGeeks has stopped selling this kit.

Based on my experience with them and the kit I refuse to do any business with them.

I currently have replaced most of the parts on this kit.

Adrian