Anyone familiar with LinuxCNC?

[jimski...@gmail.com]

Wondering if anyone is familiar with LinuxCNC and willing to give me some pointers on getting it going. I have it installed but need to configure it for my motor controller (Gecko G540) and minimill.

[Andrew Ricke]

I'm not familiar with LinuxCNC but with configuring CNC settings for my controller. What's the questions?  Maybe I can translate the particulars over. 

On Thursday, May 5, 2016, <jimski...@gmail.com> wrote:

Wondering if anyone is familiar with LinuxCNC and willing to give me some pointers on getting it going.  I have it installed but need to configure it for my motor controller (Gecko G540) and minimill.

--
You received this message because you are subscribed to the Google Groups "Arch Reactor" group.
To unsubscribe from this group and stop receiving emails from it, send an email to arch-reactor...@googlegroups.com.
To post to this group, send email to arch-r...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/arch-reactor/91de0adf-4fc2-4d9f-817b-509d58351745%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[jimski...@gmail.com]

I just got the axis motors running. What is the standard convention for X, Y and Z. Z is up and down - which way is positive? Y I think is left to right (looking at the mill), again which is positive direction? Same for X. I am guessing that this is somewhat arbitrary but also that there is also a somewhat standard way that most machines are set up. I ran across this that seems reasonable based on what other things I have seen...

- Walk to your machine and stand in front of it (from where you will operate it)
- X- is left, X+ right
- Y- is towards you, Y+ away from you
- Z- is down, Z+ is up

-------------------
Also need to figure out home/origin position. I don't have home or limit switches (yet - I plan to add them but not immediately). Will be setting origin on each job relative to the work piece. Any advice along these lines will be helpful.

[Ray Scheufler]

On the two machines at area515 controlled by Linux CNC we have X as left to right with X+ being right, Y as towards and away with Y+ being away, and Z as up and down with Z+ being up.

Our laser cutter dies not have any limit switches. We just home to to work piece every time. Out mogul CNC (CNC wood router) has X and Y limits and a touch pad for Z. The process on the mogul is to home the machine and then set a local coordinate system for the work piece. If you want more details I can probably get them, otherwise join our Google group (dsmhackerspace) and ask away.

Ray Scheufler

--
You received this message because you are subscribed to the Google Groups "Arch Reactor" group.
To unsubscribe from this group and stop receiving emails from it, send an email to arch-reactor...@googlegroups.com.
To post to this group, send email to arch-r...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/arch-reactor/81409c31-3ef8-4dfa-99a9-6d75f1f17ab4%40googlegroups.com.

[Andrew Ricke]

Yup, I concur with those orientations for X, Y, and Z.  It'll align with most CAM software expectations, though that can also be configured otherwise.

I also don't use limit or homing switches on my machine.  It's mostly due to laziness, but the steppers I have are not strong enough to harm the steel plates of my shapeoko, and it's rare that I crash to cause undo strain on the belting.  Setting machine axis where you want it on your table saves you the hassle of trying to understand the relationship of machine 0 (G53) vs work 0 (G54), which just keeps it easier in my mind to avoid the 2 coordinate layers in the gcoding.

Have you tested your distance / step calculations?  Sometimes with belts and screws, there's a variation beyond the (not so) basic math.  For example, based on your machine, you think it's 40mm/rev * motor_steps (or step_angle) * microstep_adjustment which with...but when you measure it out to calibrate, you actually need 37.85 steps / rev.

To view this discussion on the web visit https://groups.google.com/d/msgid/arch-reactor/CACVcErWf_FmvsTzszObtX8aPi0h9Zn%2BDfgYx%2B9ME%2BEAp7sphMw%40mail.gmail.com.

[Jim S]

I tested the distances and they match what I expected(moment of joy/relief).  If the expected numbers don't match the actual then something is wrong.  Maybe that's your point as it is a check of calculations and I would say a required check.  But I wouldn't call a difference "variation" - I'd call it an error.  Maybe I am missing something.  In my case:

5 TPI ball screws = 0.200"/rev
4:1 belt reduction
200 steps/rev
10 microsteps

Movement per step = 0.200 / 4 / 200 / 10 = 0.000025" 

Steps per inch = 1 / movement per step = 40,000

The machine originally had servos but was converted to steppers when I got it.  The accuracy is nowhere near that so the microstepping and/or possibly the belt reduction is overkill but that's the hardware I have.  It's a bit slow but it's a small machine so I don't think it will be as bad as I first expected.  Once I get it running I may look at ways to speed it up. 

[Andrew Ricke]

That's a good quality machine you have there if the .200/rev is true on the ball screws. You could get variance in manufacturing from .196 to .205, etc.  It's worse with belt or pinion setups than linear screws.  Hard to measure until you run your machine and measure actual results with a caliper. 

With a 40,000 steps per inch that's a lot! I'm only in the ~250 steps per inch range and maybe with a jump to ~500 steps per inch I could handle aluminum better than just "sorta". Maybe I should ask what your are CNCing, I'm scaled more for wood/plastic sheets on a 1x3ft bed for comparison. 

On first guess, I'd drop the microsteps. That'll give you better torque (though you don't likely need that with the belting gear-down), smoother steps, and less heat on your steppers and controller - without having to change anything mechanical. ~4000 steps per inch is still plenty to spare. 

FYI, for your context, zero of this conversation has been about LinuxCNC so far. All this is universal to any CNC and even obscured into 3D printers. 

--
You received this message because you are subscribed to the Google Groups "Arch Reactor" group.
To unsubscribe from this group and stop receiving emails from it, send an email to arch-reactor...@googlegroups.com.
To post to this group, send email to arch-r...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/arch-reactor/c11d5bfa-0236-4468-a0d3-5cf42a7e49c2%40googlegroups.com.

[jimski...@gmail.com]

On Monday, May 9, 2016 at 6:05:43 PM UTC-5, Andy Ricke wrote:
> That's a good quality machine you have there if the .200/rev is true on the ball screws. You could get variance in manufacturing from .196 to .205, etc.  It's worse with belt or pinion setups than linear screws.  Hard to measure until you run your machine and measure actual results with a caliper. 
>
>
>
> With a 40,000 steps per inch that's a lot! I'm only in the ~250 steps per inch range and maybe with a jump to ~500 steps per inch I could handle aluminum better than just "sorta". Maybe I should ask what your are CNCing, I'm scaled more for wood/plastic sheets on a 1x3ft bed for comparison. 
>

I plan to cnc mild steel. Some aluminum too. Probably a little wood and plastic. Envelope is 10 x 5.25 x 3.5 inches. Positioning when new was speced at +/- 0.0002. I am guessing I won't be able to measure backlash. First project is an embossing die set for some lettering in aluminum sheet. Well, actually the first one will be something very simple to check things out but the first REAL project is the die set.

>
> On first guess, I'd drop the microsteps. That'll give you better torque (though you don't likely need that with the belting gear-down), smoother steps, and less heat on your steppers and controller - without having to change anything mechanical. ~4000 steps per inch is still plenty to spare. 
>

I have a Gecko G540 with 10 microsteps. The G540 can't be adjusted in this regard. So I will live with it until it becomes a real issue.

>
> FYI, for your context, zero of this conversation has been about LinuxCNC so far. All this is universal to any CNC and even obscured into 3D printers. 
>

Yes, I realize that. But it didn't seem that important to start a new thread for general questions and keep this thread to pure linuxcnc questions. :) And since the systems of others seem to be other than linuxcnc that seemed appropriate. Part of my reason for posting was to simply see what others were doing for milling. I got my linuxcnc questions answered so far by google and such like.

[Andrew Ricke]

You can make a test square. Or the circle inside a square (or the diamond inside a circle inside a square). That will give you points to measure.   30mm sides or so will do. 

Make the origin start at the mid point of a side. If both halfs of that midpoint don't measure == 30mm then that's your backlash.   Rotate 90 for the other axis. 

 

I don't trust my calipers to measure .002" correctly, so make sure you get a nice one for .0002"!  Being a perfectionist may drive you insane at that precision levels or make you a better machinist. Little of A, little of B....

--
You received this message because you are subscribed to the Google Groups "Arch Reactor" group.
To unsubscribe from this group and stop receiving emails from it, send an email to arch-reactor...@googlegroups.com.
To post to this group, send email to arch-r...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/arch-reactor/e217a3b5-0f50-4bd8-b1a8-f7f1297990f7%40googlegroups.com.

[Hroller McKnutt]

Speaking of tolerances and perfectionism...

http://metalcutting.com/your-tight-tolerance-could-be-seriously-affecting-part-cost/

Don't call out a part for  0.0002 when 0.002 will do.

Mike

To view this discussion on the web visit https://groups.google.com/d/msgid/arch-reactor/CABOBYTwr64Cx575dfXYvTeuO3hB%3DsDZJhkNz-OkzNAFOqDXa9A%40mail.gmail.com.

For more options, visit https://groups.google.com/d/optout.

--

Nullo edactore ligni, consilium cedet. - Countess Magda

[Jim S]

You both make good points.  If I can't measure the tolerance with my current tools I think it is good enough - no need to get better measuring tools for the projects I have in mind.

[Andrew Ricke]

Yeah, so many things contribute to the insanity. Sure my steppers reliably move to 0.1mm (I mostly do metric for sanity, it's ~ .004"), but I'm 100% sure my cutting table is not flat to 0.1mm tolerance or any material I use within that. Idealistically I'm working at 0.5mm (~0.02") if I setup well.  Realistically...probably 1-2mm...

Sigh, I'll never have this: http://m.imgur.com/Pn5pIjy?r

--
You received this message because you are subscribed to the Google Groups "Arch Reactor" group.
To unsubscribe from this group and stop receiving emails from it, send an email to arch-reactor...@googlegroups.com.
To post to this group, send email to arch-r...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/arch-reactor/5c41469f-f457-489a-857f-b2df8b74036a%40googlegroups.com.

[EschewObfuscation]

Traditional poor man's substitute for a proper granite layout block is a thick piece of ordinary window glass. It's a pretty good substitute, as long as you don't drop it.