Page 1 of 1

Z Height Question

PostPosted: Thu May 07, 2015 6:10 am
by icecube45
Hello!

So I just have a question about some calibration... My x and y calibration are fairly accurate, (a 20x20 box prints to be around 19.95x19.95). Yet for some reason, all my prints are taller than they should be, stretched out on the z axis..

I'm just wondering where in the calibration process I should look to help narrow down/solve this issue

Re: Z Height Question

PostPosted: Thu May 07, 2015 10:17 am
by Nicholas Seward
currently there is no provision for this. I should have it do some touch points off a 10mm slab and add that to the auto calibration. Without that, I suspect the steps/mm is off for the arms.

Re: Z Height Question

PostPosted: Thu May 07, 2015 10:55 am
by timrwilliams
The XY accuracy is great. I never understood why I was printing 'tall' and only resolved it, as I have posted, by tightening and rechecking everything.
If your Z values are consistently and reproducibly high, you could prescale the original .stl file Z height in something like 'netfabb' - unsatisfactory but practical !

Regards

Re: Z Height Question

PostPosted: Thu May 07, 2015 4:51 pm
by Nicholas Seward
I should note that the autocalibrate "as is" is only concerned with tweaking all the parameters in order to get a flat print. It really has no respect for the accuracy in any direction. Due to the nature of the machine it ends up with a fairly accurate first layer but the z is more sensitive to miscalibration.

Options that I am exploring: (way down the road)
*Touch probe. (You could mount some block just outside the build area for elevated z touch offs.
*Camera Calibration. (I really want to mount a camera to the effector and have it go through a set of motions with dwells while looking at a printed calibration sheet. I could then generate corrections functions and/or use it for autocalibration.)

The cam calibration seems like it will pretty cheap. I have done some math on how accurately I can calculate the position of the camera and it is better than we need in one direction (z) and worse than we need in the other two. However, assuming the system is well behaved then multiple measurements will increase the accuracy to an acceptable point.

I know that none of that helps you right now unless you want to implement this yourself but I thought it was worth mentioning.

Side side note: Eventually, I want to get into realtime position detection. I would love to be able to drive robots with unknown or crazy or unsteady kinematics. (Squishy robots, Flexible robots, etc) You could also combine this with print quality detection.

Re: Z Height Question

PostPosted: Sat May 09, 2015 1:29 am
by icecube45
Nicholas Seward wrote:I should note that the autocalibrate "as is" is only concerned with tweaking all the parameters in order to get a flat print. It really has no respect for the accuracy in any direction. Due to the nature of the machine it ends up with a fairly accurate first layer but the z is more sensitive to miscalibration.

Options that I am exploring: (way down the road)
*Touch probe. (You could mount some block just outside the build area for elevated z touch offs.
*Camera Calibration. (I really want to mount a camera to the effector and have it go through a set of motions with dwells while looking at a printed calibration sheet. I could then generate corrections functions and/or use it for autocalibration.)

The cam calibration seems like it will pretty cheap. I have done some math on how accurately I can calculate the position of the camera and it is better than we need in one direction (z) and worse than we need in the other two. However, assuming the system is well behaved then multiple measurements will increase the accuracy to an acceptable point.

I know that none of that helps you right now unless you want to implement this yourself but I thought it was worth mentioning.

Side side note: Eventually, I want to get into realtime position detection. I would love to be able to drive robots with unknown or crazy or unsteady kinematics. (Squishy robots, Flexible robots, etc) You could also combine this with print quality detection.


All of those seem like great ideas!..

As for my steps/mm... I suppose I can check them again..

But in order to make sure I am doing it the right way, what I am currently doing, is moving the arms 100mm, then repeating the formula of oldsteps*(100/actual value) Until I measure 100mm on the arms..

Is this the correct way to do it?

Also, what is the best way of measuring the arms? I tried the caliper mount posted in another thread, but kept having issues with the caliper slipping out.


Thanks,
icecube45

Re: Z Height Question

PostPosted: Sat May 09, 2015 1:45 am
by Nicholas Seward
Sounds like your steps/mm would be correct.

Re: Z Height Question

PostPosted: Sat May 09, 2015 1:56 am
by icecube45
Nicholas Seward wrote:Sounds like your steps/mm would be correct.

Alright!
I'll go check my steps/mm again