I appreciate your sharing, thanks Bryan.
As I love to metal work, this is a big deal to me given the affordability.
I’ll study your recommendations in order to gain better perspective.
DIY is awesome, but the electronics get overwhelming for me quickly, which become really limits the rate of progress…
As a result, I’m willing to spend a bit of money for the commercial solution.
I’m 90% done with an Acorn implementation on a small mill, it’s gotten cold here, so I’ve diverted attention to a garage heating solution
This machine is definitely capable. The bottleneck really comes down to the quality of your plasma cutter. I rapid jog at 300 ipm, but can only really cut at 100 ipm. I’m working to fine tune the cut settings to get it cutting faster, but I may have reached my limit with the Lotos LTP5000D. Of course, at higher speeds you fight a losing battle with inertia and rigidity; trying to turn sharp corners and maintain cut quality.
From my short time in this domain, I’ve seen that DIY solutions for plasma cncs are somewhat primitive. Having a tightly coupled THC and motion controller like the $$$$ professional machines can make all of the difference in cut quality. Plasma cutting steel on a cnc is much closer to milling metal vs wood. The settings are more sensitive when you’re trying to maintain 0.060" cut height at 100+ ipm to prevent overheating and warping the material. An external THC solves half of the problem, but it doesn’t know what to do with an upcoming corner, or when the torch has passed a kerf that was created by something it already cut and the voltage spikes.
I’m glad to see that Acorn is working on coupling the THC into the motion controller. It’s not trivial.
As far as the hardware pairing, yes this is what we have done with Marlin/RAMPS 1.4 since the beginning; first wired in parallel and now series. You could run the LowRider with 3 axis inputs, hardware pairing the extra Y and Z motors. You would lose individual control such as auto-squaring both the Y and the Z axes. The Z axis is probably not as critical to square since you will be maintaining the cut height with the THC, but at some point you may like to have automatic Y squaring. You could do this with the 4 axis setup (XYYZ) where Y is software paired and Z is hardware paired.
First real project: 14ga steel and burned wood flag, roughly 22"x14"
Learned a lot about piercing (59 pierces on the main piece) and poor quality consumables that don’t last long at all. Bending the steel took quite a while. My FIL helped me bend the wide portion with a hydraulic press and bending die, but I bent the lower strips by myself (my own press is on the way!)
I also am working on a shinier version where I stripped the mill scale in muriatic acid (HCl) and grinded it with a flap disc. Should be finished with that one this week.
I started this build over the summer with intentions of making a few things in support of a craft fair which is this Saturday, so I’d say my planning is quite superb.
Looks nice. I bet with a little weather outside the colors will be completely different. Grey wood and red rusty steel. Awesome project, I am sure the craft fair will go well.
LOL, I couldn’t agree more, EXCELLENT project management !!!
Thanks again for share your learning and insight. It’s very helpful to me in finding direction for my effort. It’s surprising to me how much effort chasing this topic on the LinuxCNC forum, 11 pages in less than a month. I’ll cut wood for a while, that will be good experience. I have some ideas around making mold for fiberglass parts that need some vetting. Hopefully that will keep me busy long enough for Centroid to complete their plasma revision.
I don’t quite follow what you’re describing with axis motor pairing, but I need to read more. This is actually where I am in my MPCNC build. I traced out the kit wiring harness yesterday, but still don’t have my head wrapped around why the circuits work yet … but it’s cool !!
I haven’t been in tune with DIY plasma cncs for a long time, but I’ve seen a lot of development within the last month even. It’s a good time to be getting into it.
Depending on the controller board you select, it may not have enough stepper driver inputs for each motor to have its own independent control. Independent control is nice when you want to square up an axis with two motors. Ex: Move Y1 and Y2 to home switches, when Y1 switch is triggered then keep moving Y2 independently until it triggers it’s own home switch. You couldn’t do that with a hardware pairing mode because both motors see the same step/dir signals.
You can combine several motors together, such as both Y motors, by wiring them in series (Ryan has instructions on this). That Acorn board only has 4 axis inputs, so you could do independent control over XYZ and for the 4th do a software pair of the other Y motor. You’ll be forced to do a hardware pair (series wiring) of the two Z motors since you don’t have the 5th input available. Does that make any more sense?
Conceptually, yes, I can see where the logic and drivers would search and adjust to trigger both home switches simultaneously, which would then result in a square machine.
I guess where I’m confused is the MPCNC set up, I’m working on this item currently, haven’t tested the auto-square, doing the series wiring of the first axis motors…
The motors are wired in series and there are 2 home switches, 1 for each motor of the axis, each is wired back to the Rambo.
Is the firmware still able to square the axis, even though there is only 1 driver for both motors, but 2 switches?
Admittedly, I may be blind to something fundamental / foundational to understanding the solution…
That’s where I’m struggling, you’re being clear.
It’s not having seen this work yet, first hand, is where my brain is getting in a knot …
And thanks for continuing to entertain my remedial questions.
No, dual endstops requires dual motor drivers. For the mpcnc, you need 5 drivers. I always point out, there were thousands of machines that worked well before dual endstops. It’s an improvement, but it is a bit more expensive and complicated.
I actually went to the trouble of figuring out the series wiring and made a custom cable, from CAT5, for my long axis. Even though it’s a waste, it was kind of fun to drive both motors and house both endstop switch wiring in a single cable …
Back to crimping those tiny wires… At least I found my correct crimping tool, no more needle nose pliers !!!
I was wondering if you could elaborate on how you put the water pan together? Use a brake to bend it into shape or weld for miles? Use any kind of sealents other than relying on the weld?
The water pan is made out of 14 ga steel. I had a shop make two bends for a 3" tall wall in a U-shape. They did not have a box brake, so making 4 bends was out of the question unfortunately. They gave me two strips of steel, 3"x48" for me to weld on to make the other two sides.
A few things I learned from doing this: if possible, find someone with a box brake that can bend all four sides from one sheet. If you want a 4’x4’ pan, then this means you need to start with a 5’x10’ sheet in order to have the wall height included. By welding on the two side pieces, I probably should not have welded the entire length. I did it in 2-3" sections at a time but I believe it warped the pan too much. After welding I laid on a lot of 100% silicone sealant and filled it with water. It did leak in a few places so I added some more. Again, if I had all four sides bent it would have saved all the trouble.
After a second round of silicone it has stopped leaking and held water for a few months no problem. The weld beads got in the way of a proper seal due to my lack of experience welding. Perhaps brazing would have been a better way? Or fewer welds and more sealant.
For the first two months I left the water untreated and it was rusting a good bit. Nothing more than surface rust, but it will definitely not last. There are ways to go about preventing it from rusting. I decided to add sodium nitrite and it’s kept the water clear.
Oh… And instead of a 48"x48" pan I should have done about a 50" square. The sheets of steel get caught on the edges because it’s not perfect and I have to bend back the walls to get the sheet to sit flat.
Edit: I also got some flex seal liquid that I was going to try out but haven’t needed to yet. Was wondering if it could prevent the pan from rusting
I paid roughly $100 for the steel and to have it bent around 5 months ago. I probably could have gotten it cheaper but I didn’t take the time to shop around. Cutting on open slats produced too much smoke and metal powder that I wasn’t enjoying using it. Now it stays clean. I would probably pay double that if it was bent on all four sides and I wouldn’t have to worry about leakage…
If you had to guess about how many linear feet of shielded wire did it take to wire up your table?
What CAM software are you using? In an email with the estlcam developer they told me it’s not suitable for plasma, they didn’t elaborate as to why. I know it doesn’t have curved lead-in/lead-out or peirce. If that’s the only reason I can make a post processor to fix that.
Finally, do you pay special attention on keeping your electronics distanced from your plasma? If so how far?
SheetCAM has been the easiest thing to use, by far. It’s based on DXF/SVG files so it saves a few steps from fusion 360. Since I use LinuxCNC, I don’t even use a complicated post processor. There’s a script that runs inside Axis (my LinuxCNC control GUI) that has a control panel on the side of the screen where I can real-time adjust cut voltages and speeds. No need to bake that into the post processor since it can be changed so easily.
Post processing for plasma is pretty straight forward. You need to modify your torch on commands to include any piercing routine (go to Z -100 and wait for limit switch trigger, go to pierce height, dwell at pierce height, go to cut height, resume path). Lead in and outs are helpful to keep cut paths clean and prevent overblown holes from when you pierce.
What control board/firmware are you using?
I think I’ve been extremely lucky on EMI. I haven’t noticed any ill effects since the beginning (save for my wifi going out), but when I switched to the Hypertherm even that works fine. I did not drive a copper ground rod into the ground and tie all the grounds to that, but I would have if I had problems. At a minimum I would run shielded cable and ground the shielding at one end (control box) as a precaution. There’s lot of horror stories on this subject, and a lot come from HF start machines (cheap Chinese plasmas). When in doubt, ground everything.
I haven’t settled on that yet. I was hoping RAMBo with Ryan’s firmware + Proma THC SD would suffice just for simplicity. If I really do need to just go the Mesa + linuxCNC route I can do that. I’m a long time linux tinkerer so that won’t be an issue. I don’t understand at the moment exactly how the THCAD-300 THC is used. And to muddy the waters a little more I get the impression that grbl would at least be better for this than marlin?
Answered before I could even ask!
BTW the link to the wire you gave says solid core wire but the picture shows stranded. The shielding is a plastic sleeve + bare wire instead of a foil?
In my opinion, an integrated THC like THCAD-300 or (-10) is better than the “external” THCs like the Proma. The THCAD is integrated into the motion planner, so it can be conscious of current speeds. The external THC reacts to the instantaneous voltage reading and nothing more. It doesn’t realize that you are slowing down to go into a corner. With the Mesa/THCAD setup you can account for %feedrate and adjust the logic corresponding to your THC. Proma is more of an open loop control and the THCAD is closed loop.
The THCAD converts the arc voltage (or divided arc voltage if using the -10) to a frequency. This frequency is then sent to the Mesa hardware which reads it and is converted back into the voltage level. That way you aren’t sending raw arc voltage from your plasma to your control board, just a low voltage, variable frequency signal. Another benefit is that it is much cheaper than the external THCs. You could get Mesa boards + THCAD for less than a Proma.
LinuxCNC (on any hardware) > Mach3/4 > GRBL > Marlin. I would switch the first two if you didn’t have any Linux experience because it can be daunting to set everything up with the correct real time kernel and such whereas Mach3 is more turnkey. The customizability is endless with LinuxCNC if you are all about fine tuning and min/maxing your setup. As far as the last two, I never really liked Marlin for CNC router applications, I always thought GRBL was a much better solution because that’s what it was made for with multiple coordinate systems and special gcodes and Marlin is a 3D printing firmware first and foremost. I know that Marlin has expanded a little recently to include CNC support, but it has nothing for plasma machines.
I’m not sure if that was the exact product listing I bought, but it had the same packaging. Mine is definitely stranded wire, not solid. It should have 4 insulated conductors + bare conductor + foil + plastic outer sheathing. The foil is what is providing the shielding.
Sorry if I linked the incorrect one, I’ll edit once I find the exact one I bought.
Well sir, you’ve about got me talked into it! Thanks again for all of your help, you’re thread has been extremely informative.
Another question about your machine: I followed your link above to your pictures and videos of your machine. I notice your 611 plate change colors. What material and thickness is that? Did the previous plate fail?
Sorry, I mis-remembered the price of the Proma (I thought it was closer to $300???)
Anyway, Mesa 5i25 + 7i76 is ~$210, THCAD is $70. Total : $280 +tax (extra for a PC to run Linux if you don’t have one already. I use a $50 Dell Optiplex 960 off ebay)
Proma + Rambo ~= $350
The linuxcnc forums were super helpful for me when I was getting everything started, I would check them out before you buy your boards
No problem, that’s what I was hoping for. If it helps someone else get up and running faster than I did, it is all worth it.
