Mark, my order is still awaiting fulfillment since the 18th of August while all parts were marked as in stock. When is it going to be shipped? The money is already taken of my card so where are we waiting for?

 

Anybody here that can say something about using acrylic as side and gantry plates? Thinking about 10mm thick acrylic for use.

 

I wonder why the micro-stepping has an influence on the backlash. When the steps are increased, I suppose that the accuracy is decreased (I mean that 10 mm of movement is less exactly 10mm), but I wonder if it has an impact on the backlash. If it has, I wonder if this does not mean that the motor is unable to produce the force to take the small steps.

it seems that also GRBL has some plans for the future to incorporate backlash compensation.
https://github.com/grbl/grbl/wiki/Development-Path-and-Future-Needs
In the meantime some standalone scripts might help.

Frederic

 

Derwin where do you get to block to hold the acme screw nut?

Thanks,

Alberto

 

Anyone have a simple way to get a broken tap out of some 20x60? Of course it was the last hole I needed to get done

 

I never tried it, but you may try to cool it enough (with a compressed air can upside down) for the tap to become loose in the hole and remove it by hand or with a punch, or just destroy it with the punch since it should be brittle when it's cold (I read that part somewhere...) Good luck!

 

I'm really interested in this backlash discussion. I wonder if it has anything to do with how many teeth are engaged on the drive pulley? What does your setup look like?

 

SU has extensions that will handle the true circle and arch issues. I use them all the time for 3D printing parts. Let me look……http://www.smustard.com/script/Weld is a good one to get really smooth lines in circles, arches and other forms. Smustard has also a lot of other extensions that are very handy for 3D design. Good thing is that they are free, too. There are further loads of SU extensions for practically all versions of it, that are great tools for any kind of 3D work in Sketchup.

Another one is the STL export extension I'll post a link later.

 

Weld works perfect with my 3D printer and gets circles as they are supposed to. Not using it for STL will result in seeing all the lines in the circle print. I have no experience yet with SketchUcam so I cannot say anything about the combination of the two. I will pay attention though to what you say and keep it in mind when my build is ready to cut. Thanks for the info!

 

I was wondering what everybody is using to control the OX. What is a recommended USB CNC Control board that works with Mach3?

 

It is a fully standard OX build.

Frederic

 

Alberto, you gotta make them.. I needed alot of clearence since the lead screw is twice as big.. I did alot of modifying on the zed.. I am very fortunate to have a machinist friend beside me helping with the cad & cam work. Big shout out to PMT..

 

Geckodrive g540 is (IMO) easier. Pricey but reliable...

 

I compiled a few G-code scripts that help measuring the backlash more easily. With each line, lines are shifted by an additional 50 micrometer.When this is drawn lines are first shifted to one side and later to the other side.

 

Do these plots represent backlash in its true sense or does it demonstrate the micro step error mentioned a few posts above?

I'm not quite at this stage yet so can not provide any practical feedback. But to my thinking if we are considering only the freeplay between the pulley and belt, the error should be consistent across every pair of lines. It appears on your plots that the error is increasing.

 

Just an update.

Here is my wiring so far...

Air direction-of-travel is in thru the bottom fan (mounted in the box, not the one at the bottom of the picture) across the controller boards and out thru the center of the lid. There should be more than plenty of air-flow across the boards.
The power cords to the 24v PS, computer, monitor, and router, plug into a heavy duty surge protector mounted to the underside of the table top. I am waiting on my 24v to 12v converter which will mount in the lower right corner of the grey box and will power the two fans. USB cable is not in picture because I won't connect it until I am ready to test.

Here is the Router side of the wiring with my 3D-printed chain guides and wire looms...

Everything rolls nice and smooth with no binding.

 

I'd read somewhere, not sure if it was on this forum, that you could run two 12v fans in series off a 24v power supply.

Maybe someone with some electrical knowledge could confirm this?

 

The G-code instructs to miss align all the top-lines with the bottom lines. The misalignment is supposed to increase with 0.05 mm with each line. The first picture I posted, the red lines on black background, depict the G-code; this is what the machine should draw. In reality this is not what appears on paper. Lines are first shifted to one side and later to the other side, this is because each time 0.05 mm is added. Where the lines align, this is the amount that needs to be added to compensate for the backlash.

The play on the pencil mount and minimal bending of the pencil probably adds to the error, but I suppose that the play in puller/belt is the main factor.

You mean the fact that Jeferson SImoes mentions that the backlash seems to change with the selected microstepping?
In any case, I do not loose steps. What I did not mention is, are the extra lines at the left. First bottom part is drawn before the test and the upper part after the test. Both parts are drawn from left to right so that backlash should not have any influence. As the lines align, it demonstrates that the router/pencil is still where the controller thinks it is (and no steps where lost)
Perhaps these pictures help to understand what happens:

 

This will probably work, provided the fans are identical (brand/type). 24V fans are also very common.

Frederic

 

It all looks very nice

 

Thanks for clarifying that for me. So apart from the backlash the accuracy looks quite good, in reference to where the first pair of lines started and ended.

I will try your test code when I get my machine operating and report results for comparison.

Cheers
Anthony

 

There are 2 direction changes:

1. from left to right
2. from right to left, backlash is visible
3. again from right to left, perfect alignment with 1.


Thanks,
Frederic

 

This is making me think he is loosing steps on one Y axis motor

 

Thanks for helping me looking into this.
I believe there is something you are overlooking. The G code is such that the distance between the vertical lines at the bottom are 10.00 mm apart, while the vertical lines at the top are 10.05 mm apart. So the top lines shift more and more compared to the bottom lines.
As long as the cumulative shift is lower than the backlash, the top line is at the right of the bottom line. When the cumulative shift equals the backlash, the lines align. Later the cumulative shift is higher than the backlash, the top line is at the left of the bottom line.

Frederic

 

Bench tested my gShield last night, the Windows mini tablet I am using works well with GRBL controller and will probably be much more reliable in a dusty environment than a laptop/desktop.

Has anyone been able to get TinyG2 running well on a gShield/Due combo? Using TGfx it seems to work, ie motors turn, but the software doesn't plot the gcode trace in the preview window and doesn't report the machine position.

 

i started to go on the GRBL train but kinda knew that all this is new to me and would have to rely on the limited knowledge base. i went with MACH3 and paid for the license and another $20 for the 2010 screenset. it has macros for toolchange, auto z, and alot more people using it around the world. i hope you get your machine up and working soon.

 

Really? I guess if you don't draw accurately, it delivers an inaccurate DXF file, but I use SketchUp and Autocad and the files are interchangeable and the accuracy is the same in both. I am confused as to how something that stores an element digitally can be 'out'.

I can understand that the person designing the part in SketchUp maybe had the accuracy settings turned down and so drew something that had a few tenths of a mm variance, but I have mine set to three decimals of accuracy and that's what I get.

Circles/Arcs, similarly. If someone that doesn't use SU a lot doesn't tell it how many sides to draw for a circle or an arc, it will default to 24 - which is a bit low. And if you want dimensions and centre points, put them in.

But, I am using it for controlling an Ox and a 3D printer, not creating code for an Onsrud. If my Ox gives me three decimals of accuracy, I will be damned happy.

 

Hi,
Is it possible to use a Nema 23 instead of the Nema 17 for the Z-axis?

Johan