1 Topic by MolecularConcept

  • MolecularConcept
  • Member
  • Offline
  • From: Pennsylvania
  • Registered: 2013-06-03
  • Posts: 392

Topic: Set it and Forget it

Or not... Ive had my printer for a while now and I have my extrusion and flow rate locked down.  The X and Y axis are pretty much right on the money. My Z axis on the other hand is really driving my nuts!

Say I print a cube thats 5mm tall and it prints 4.9 and i do the math print it again and its 5mm, sweet! So then i print a calibration vector next and its 30mm tall.      Having JUST calibrated it it should be on point but no it prints out to be 31.49 or something. I do the math AGAIN print it out and its good. So now it should be good right ? I print the cube again and its 4.9 or something again!

Same thing happens with different layer heights. Ill have it printing 5mm cubes in ~.3mm but in ~.2 or ~ .1  it needs re calibrated.

Is this because of the 5/16ths lead screw? Ive even tried using the adjusted layer heights  like .3034 , or .1058 still get this inconsistency in my Z axis.

Or is this just how it is at this point in these machines? There is no truly calibrated machine?

2 Reply by IanJohnson

  • IanJohnson
  • Moderator
  • Offline
  • Registered: 2012-08-07
  • Posts: 2,815

Re: Set it and Forget it

Do your layer heights divide evenly into the Z dimension of the print? If not it will have to round one way or another.   Combine that with the first layer being squished shorter.

3 Reply by MolecularConcept

  • MolecularConcept
  • Member
  • Offline
  • From: Pennsylvania
  • Registered: 2013-06-03
  • Posts: 392

Re: Set it and Forget it

even if the first layer is squished that only going to give a difference of a fraction of a millimeter.

im seeing differences of whole millimeters  in some cases. id be happy with + /- .1  but when something is suposed to be 30mm and im getting 32.4 thats not cool!

4 Reply by Hazer

  • Hazer
  • Hazer
  • Member
  • Offline
  • Registered: 2013-07-19
  • Posts: 478

Re: Set it and Forget it

You just said yourself, your 5 mm cube comes out 4.9mm. What exactly do you do your first layer huh?

For instance, here is my settings:

My Z home stops just 0.2mm away from the nozzle. I have this height for any setting whether its 0.1 or 0.3 layers.

Now, in Slicer I have a 0.3mm first layer.

What actually happens is that when my start gcode sends G28 and homes everything, I end up 0.2mm gap between my glass and my nozzle. Then, just after extruder warm up but before the real printing starts, my Z axis lowers 0.3mm. My first layer height STAYS at 0.2mm because I have (through experience) a backlash of exactly 0.3mm on my stock screw. So physically that first Z move down does not move the bed.

So. Calibrating Z height using a 5mm cube is rediculous. Calibrate a 100mm column for accuracy, and expect that short prints will have some form of error to them until you can lock down your mechanical slop either through backlash compensation in Slicer (like I did) or through a threadless ball screw (which may still have backlash, I wont know until I install mine).

Chuck Bittner is a quadriplegic gamer who is petitioning the major console developers to include internal button remapping in all console games. You can help.
Sign Chuck Bittners petition

5 Reply by spapadim

  • spapadim
  • Member
  • Offline
  • From: Jersey City, NJ, USA
  • Registered: 2013-05-16
  • Posts: 185

Re: Set it and Forget it

Yes, 30 -> 32.4 is a bit much.  I never had to calibrate the Z-axis (beyond the theoretical steps/mm for 5/16-18).

.1/5 = 2% and 2.4/30 = 8% so it's not even consistent (although I wouldn't give much weight to that .1, for reasons already mentioned).  Dunno, but some things that come to mind:
- Always the same G-code?  Check the last layer's G1 Z command (and measure deviations against that, not against the STL)?
- If you do repeated trials (trimming out any error measurements in the .1-.2 range) *without* changing calibration, is error biased (e.g., always positive), or random?  If yes, is there any pattern (e.g., constant %)?  If you ask for, say, 10, 20, 30, and 40 (not just 5 and 30), what do you get?
- Or, perhaps better (or a little less tedious), get a dial gauge (or this) and measure error at, say, .1mm moves (again, measure several times, without changing calibration)?  How does plot of actual vs target look like?

6 Reply by MolecularConcept

  • MolecularConcept
  • Member
  • Offline
  • From: Pennsylvania
  • Registered: 2013-06-03
  • Posts: 392

Re: Set it and Forget it

im guessing your an engineer?  i havent even thought to look at those things before. what am i looking for exactly in the gcode? and other than downloading an STL i dont really have any of looking at the file.

im geting a metric screw hopefully this gets rid of this ... hows you guys accuracy?

7 Reply by Hazer (edited by Hazer 2014-02-07 03:08:13)

  • Hazer
  • Hazer
  • Member
  • Offline
  • Registered: 2013-07-19
  • Posts: 478

Re: Set it and Forget it

So. As mentioned, you should get to know how Slicer puts together your gcode. I did not notice that the 'first layer height' setting was actually making an initial Z move before the first extrusion until 3 months after I had my printer. I did not know because it was automatically adjusting my screws inherent backlash. Now I use it constantly.

If you want to really see what your getting for Z height accuracy, make yourself a structure that has 10mm stepped heights that you can measure. For instance, a 10mm cube joined with a 110mm column so you can measure the 100mm difference between them. This will give you a true place to adjust your Z accuracy without having to factor in the first layer.

Oh, and I never had to change my Z steps either.

Chuck Bittner is a quadriplegic gamer who is petitioning the major console developers to include internal button remapping in all console games. You can help.
Sign Chuck Bittners petition

8 Reply by MolecularConcept

  • MolecularConcept
  • Member
  • Offline
  • From: Pennsylvania
  • Registered: 2013-06-03
  • Posts: 392

Re: Set it and Forget it

is there a book i can read on this stuff? where do you guys find all this information and become this good?

9 Reply by MolecularConcept (edited by MolecularConcept 2014-02-08 02:27:58)

  • MolecularConcept
  • Member
  • Offline
  • From: Pennsylvania
  • Registered: 2013-06-03
  • Posts: 392

Re: Set it and Forget it

G1 Z9.979 F7800.000

does this mean that my print should be appx 10mm tall?

and how does my first layer impact the print is it counted in the overall height? if i pting a 10mm cube with a .3mm first layer should it be approximately 10.3mm?

10 Reply by Hazer

  • Hazer
  • Hazer
  • Member
  • Offline
  • Registered: 2013-07-19
  • Posts: 478

Re: Set it and Forget it

MolecularConcept wrote:

G1 Z9.979 F7800.000

does this mean that my print should be appx 10mm tall?

and how does my first layer impact the print is it counted in the overall height? if i pting a 10mm cube with a .3mm first layer should it be approximately 10.3mm?

No, it will not be 10.3mm. Its the first layer, so if you are printing in 0.3mm layer hieghts it is meant to be the first 0.3mm layer. But, there is a combination of two things there: Your 'Z-tab' and Slicer 'first layer' setting.

Z-Tab is the physical height the Z axis homes to. You can make this 0.3mm, 0.2mm, or even 0.0mm (nozzle touches the bed).

Then Slicer has the first layer height which it purposely moves the Z axis before moving.

So if everything was mechanically perfect, the ideal setup would be to home the nozzle to touch, and have the Slicer perform a 0.3mm move. BUT: As I said with my unit, I have a reliable backlash of 0.3mm. If I tried this, my first layer would be smashed into the bed (and filament would get grinded in the extruder instead of actually coming out onto the bed). Also, an actual first layer of 0.3mm would be bad. To get a good stick and guarantee your print does not get unstuck mid-job, you want more of a 0.2mm first layer. Thats why Solidoodle recommends a business card for homing your Z, it is 0.2mm.

Chuck Bittner is a quadriplegic gamer who is petitioning the major console developers to include internal button remapping in all console games. You can help.
Sign Chuck Bittners petition

11 Reply by tealvince

  • tealvince
  • Member
  • Offline
  • Registered: 2013-01-16
  • Posts: 319

Re: Set it and Forget it

It doesn't sound like software to me.  If the results are repeatable with the same height part then my guess is that your z rod threads vary in their spacing over the length of the rod.  If the results are more random then it sounds like backlash.  In either case, converting to a z rod with a finer thread pitch will solve both problems since the tolerances will be much tighter.  I am very happy after switching to an m3 sized zrod, with the bonus that the nyquist cutoff for the .5 mm pitch of an m3 is .25mm, so with .3mm layers z-wobble is mathematically eliminated (and at .1mm effectively so).

12 Reply by Tomek

  • Tomek
  • Member
  • Offline
  • Registered: 2012-10-01
  • Posts: 710

Re: Set it and Forget it

tealvince wrote:

It doesn't sound like software to me.  If the results are repeatable with the same height part then my guess is that your z rod threads vary in their spacing over the length of the rod.  If the results are more random then it sounds like backlash.  In either case, converting to a z rod with a finer thread pitch will solve both problems since the tolerances will be much tighter.  I am very happy after switching to an m3 sized zrod, with the bonus that the nyquist cutoff for the .5 mm pitch of an m3 is .25mm, so with .3mm layers z-wobble is mathematically eliminated (and at .1mm effectively so).


I've been debating upgrading to a M3 or M5. M3 just looks so puny, from a mental perspective it's hard to imagine all the thrust on the thin M3 rod. Can I just have a comforting word from you that M3 really doesn't buckle?  I guess it shouldn't, but it feels funny. 

Also, what did you use to couple to your stepper? I was thinking about using stiff hosing.