PDA

View Full Version : Yet another gantry CNC router.



cambesol
13-06-2015, 01:07 PM
Hello Forum,

Just thought I'd post a picture of my almost complete router. Works well on aluminium although it generates far more filth than I expected which is causing problems with limit switches.
It uses a HSD liquid cooled spindle and overall I am very happy with it.

CAM workflow is: Draftsight -> LibreCAD -> DXF2GCODE -> OpenSCAM (sanity check) -> LinuxCNC.

All open source and avoids any Microsoft products.

Thanks,

Neil.

routercnc
13-06-2015, 07:33 PM
Neil,

Very nice little machine well done!

Are you getting swarf in the limits switches or are you having false e-stops? Noticed you have a single energy chain so are running stepper power and limit signal lines together.

Out of interest, what is the electrical set up - is that the box just out of shot at the top of the picture?

cambesol
13-06-2015, 07:52 PM
Yes the enclosure contains stepper drives, VFD, contactors etc.

The problem is due to swarf and I conclusively proved this last night. Should be easy to fix though.

Thanks,

Neil.

Boyan Silyavski
14-06-2015, 12:44 PM
Couldn't you just rise 2 plastic or other panels both sides, height as just the gantry to be able to pass, so all swarf stays in?

cambesol
14-06-2015, 05:20 PM
Couldn't you just rise 2 plastic or other panels both sides, height as just the gantry to be able to pass, so all swarf stays in?

The problematic limit switch is right at the bottom of the z-axis so adding panels to the side would not prevent the problem. In reality this switch is almost useless because if it ever gets activated by the plate that supports the spindle, bad things will have already happened to the table as it is very low. I only kept it there for reasons of symmetry (I have a similar, more useful limit switches on the x- and y-axes). I have two choices, remove it or mount it higher and on an insulating base as well as shielding it from all the chips that get blown around. I will make that call tomorrow after deciding if it is worth having it at all. Moving it would involve a severe strip-down and I am not too enthusiastic about that at the moment.

One other big mistake I made was to wire the limit switches into the stepper driver enable line to remove drive on whichever axis has tripped it. This leaves drive to the other two and as it is the z-axis which stops, x and y are free to bury the cutting tool into my work piece which is reasonably catastrophic. You live and learn.

routercnc
14-06-2015, 05:57 PM
I'm pretty sure Dean runs without a lower Z limit switch for similar reasons you have mentioned above. I also ran without limit and Z home switch for a while as I kept getting false e-stops (longest cables, direct into BoB), but my 24V upgrade should sort that out. You'll have to decide what you are comfortable with.

You could locate the 'lower' limit switch right at the top near the stepper motor housing, then connect a long upside-down 'L' shaped bracket to the Z axis so that when it gets near the bottom the bit of the 'L' which sticks out comes down and presses the top of the switch. This might look a bit untidy but its an option and is well out of the way of the swarf.

If you've been following the recent build logs you'll see some useful circuit diagrams for wiring limits etc. which avoids the problem you mention above. Once you get the bugs out (which is normal by the way!) I think you have a nice little machine there.

Been meaning to ask (as I'm about to do the same) - did you finish the top of the bedplate yourself on another machine or did you have it polished? I can't see machining marks or witness marks at the extremes of travel so I'm guessing you haven't leveled it using the cnc machine?

Clive S
14-06-2015, 06:38 PM
There is not a lot of point in my opinion of having any limit switch on the bottom of the Z axis as you always have different length tools in the collet etc (so how can you set it) . Also if any limit is tripped the machine should be set up to stop all action on all axis as you will have lost position. ..Clive

cambesol
14-06-2015, 07:08 PM
Absolutely agree. Now that I have seen the affect of two axis still running. Problem with using the enable line is that it fails very quietly. I only figured the problem out because the z-axis status LED on the stepper drive was solid and the other two blinking (status enabled and ok). I blew all the cr@p off the limit switch and all three LEDs started flashing. I had lost z position and tried to save the job but forgot to home the z again (after clearing the fault). Suffice to say it all ended badly.

cambesol
14-06-2015, 07:12 PM
The bedplate is 12mm ecocast. It is mounted in a way that I can adjust in situ then clamp it tight. I use a digital indicator mounted in the spindle and have achieved a parallelism of +/- 0.05mm across the table.
I didn't like the finish on the ecocast as it showed every little touch (just wiping swarf left marks) so I used a scouring pad + WD-40 to buff if to a uniform bright finish. This can be achieved every time I want to give it a quick polish.

I am undecided about z-limit. Will have a long look tomorrow.

Boyan Silyavski
14-06-2015, 08:00 PM
Isn't it designed to hard stop? I design Z axis so that it looks it would fell from bearing rails, but in reality at a certain moment the ball screw would stop it up or down so the drive will trip if we come to that.
So i agree with all, no switches are needed. Plus you have software limits. Plus you are with Lunix so no excuse on part of software.

mitchejc
15-06-2015, 09:19 PM
Congrats, its a VERY nice build, well done! I can imagine lots of hours going into those great looking alu parts! I just love the neat looks of an all aluminium build. Very good suggestion by Silyavski, maybe put a little block of wood or plastic to hard stop it just to prevent it running off the rails if required and ditch the lower Z switch.

Sorry, I don't want to stray too far from the topic but since you have first hand experience, whats your opinion of LinuxCNC so far?

cambesol
15-06-2015, 11:08 PM
Congrats, its a VERY nice build, well done! I can imagine lots of hours going into those great looking alu parts! I just love the neat looks of an all aluminium build. Very good suggestion by Silyavski, maybe put a little block of wood or plastic to hard stop it just to prevent it running off the rails if required and ditch the lower Z switch.

Thanks! From start to finish has been around eighteen months which included the control system and liquid cooling. The machining took around 9 months but this is only five hours per week due to work/family commitments. I prefer face-milled aluminium rather than lots of extrusion because it looks nice. Not the fastest way to go, but worth it in the end. It does have a hard-stop, the spindle will not drop onto the table.


Sorry, I don't want to stray too far from the topic but since you have first hand experience, whats your opinion of LinuxCNC so far?

I use Linux a lot (I develop embedded software, drivers etc running on Linux of all flavours) so for me it was a natural choice. I really like LinuxCNC, the work flow I described in the first post has yielded good results although I haven't produced anything that I want to post on this forum yet due to the f*****g limit switch causing random paths to be followed (design flaw, please don't ask - too embarrassing). Today I have drawn up some replacement parts made in acrylic which should fix the problem. I will post a picture when I have my final test piece finished this week.

Linux is not for everyone as it can be frustrating trying to figure out what you need to know to fix a problem. LinuxCNC, once configured, works very well indeed.

In a few months I want to dig deeper into BlenderCAM as it can produce some incredible results. The UI on Blender is a bit of a nightmare though and the learning curve is significant so for the moment I am sticking to 2.5D designs.

mitchejc
16-06-2015, 10:43 AM
Good to know. Are you running via a parallel port BOB or are you using one of the smarter usb/ethernet controllers?

I have not really looked into it but first prize for me would be if one can run LinuxCNC on the new Raspberry Pi2 to do the trajectory planning and maybe offload the pulse generation to something else. With your experience maybe something you can do for us :friendly_wink: