[travis.farmer]

This is an older picture of my machine. i have since bought an ATC spindle for it.

the ATC spindle mounted in place. it is a 4.5KW 3PH spindle.

This is the control cabinet wiring, so far. still have several things to tie in.

the CNC Router is run by LinuxCNC, using a Mesa 7i80HD-16 controller board (seen up top of the cabinet, hidden by the copper-clad ribbon cables.)

I look forward to providing more details as i go

 

[travis.farmer]

Here is a video i did (by no means a pro video maker) of a test run (no material).

 

[travis.farmer]

Is the computer directly controlling the router? Most 3d printers, and laser cutters just load the file, and run from a micro SD card (well, at least mine do, maybe not most).

How fine of a detail, or resolution can that do?

the Mesa 7i80HD-16 board is where the real magic happens. it is a FPGA that is programmed (with selectable Mesa firmware) to handle the step generation that drives the stepper drivers and stepper motors. but the computer does play a huge part in controlling the FPGA, and therefore the machine. without both working together, the magic would not happen.
the computer is running Debian Linux 11 (Bullseye), with a real-time kernel (very important), and LinuxCNC 2.9. the reason for the real-time kernel, is microprocessors have a hard time in time based processor management. after the first step, they complete the second step when they get around to it. the real-time kernel allows the microprocessor to act like a microcontroller, similar to a 3d printer controller. in other words, after the first step, the computer does what it needs to do, but when the time comes for the second step, it is near spot on timing.
the real-time kernel is most important if LinuxCNC is doing software stepping. that is when a parallel printer port is used for the machine input and output. so for this setup, one must select the computer motherboard and processor carefully and experiment with it to get the timing settings matched with it's ability. my machine used the Mesa board to handle the step generation, and connects to the computer via a Ethernet connection, directly (no network switch or router in between). because of this, there is much less load on the computer itself for timing, except that it only needs to send the FPGA instructions in time to generate the steps.

hopefully i have explained that part well. i don't know all there is to know, but i can share what i "think" i do know.

as for precision, if i did the math correctly (not my strong point), the steppers have 200 steps per revolution, and 1/8 micro-stepping, so 1600 microsteps per revolution. the ballscrews have a 5mm per revolution pitch, so every microstep, the axis will move about 0.003125 mm. though for routing wood, i am unsure if that level of precision is even necessary. ;-) i mainly run 1/8 microstepping just to make the machine run smoother. there is a lot of mass in the gantry and spindle, so if the Y axis is running a little jumpy, it would shake the whole shop (before i setup the acceleration correctly).

 

[travis.farmer]

the previously mentioned ATC (Automatic Tool Changer) spindle that i have on my machine is actually a fairly recent addition. i bought it with my IRS tax return money. i intend to make a living, using my CNC router, so i wanted an ATC spindle to allow faster tool changes, and eventually, fully automatic tool changes. it uses pneumatic power to operate a piston inside the spindle. this piston actuates a little grabber that grabs onto a little knob at the top of the tool-holder, and draws it up tight, so the cone part of the tool-holder is pulled into a matching cone in the nose of the spindle.
because this cone is the entire drive contact of the tool-holder, it must be clean and free of oil, water, and dust. so the air coming in must be clean and filtered, mainly because a charge of air is used to blow dust from the cone before locking it in. so that air charge must be water and oil free. despite having a pneumatic piston, my first thought was that it should have a few drops of air tool oil, but the directions advise strongly against this.

i am relatively new at CNC routers, but i am learning as i build mine up. i originally bought it off of Craigslist, for i think it was $600. just the machine and the original spindle (a Perske 5HP 3PH). i added the stepper motors (NEMA23), stepper drivers (Leadshine DM542E), Mesa board (7i80HD-16), and well... the list goes on. but i have been very slowly building this machine up, and while it still has quite a few temporary installations, like a homing sensor held on with tape, as i fine tune things i can mount things more solid.

i currently have it sitting in the back of a gutted out camper trailer, as it was what i had. i am building a new shed that will serve as my shop. it is a 13' X 20' platform, that i will move a tent/garage onto, and build some better end-walls. i have the platform built, but time got away from me and winter hit, so the platform is covered by a tarp, waiting for summer.

today, i was adding the air-lines that wind around my machine to get to the spindle, and once all was connected, i turned the regulators down, and added air from an air compressor. then i turned the pressure up for the lock/unlock piston to around 95psi, according to the gauge, and noted it would not quite actuate the piston enough to allow a tool-holder to be inserted. i carefully turned the regulator up to 100psi, and it just operates the piston. i suspect the gauge may be off, as 95psi is the max for the spindle. either that or the air passages may have a blockage from some packing Styrofoam. i hope not, as i am unsure how to clean them out...
my first order of troubleshooting is to replace the gauge on the regulator. there is also the potential that the solenoid air valve is not opening all the way, and therefore not letting the full charge of air through. a few things to check tomorrow.

it is my intention to be able to make some basic signs for people to pay for the upgrades, like the full ATC setup so i don't have to manually change tools. my biggest hold-back is the space the machine is in only allows access to one side (yeah, it is that tight a fit in the old camper). so i cannot even attach the spoilboard yet. i just can't reach across the table, and i don't want to lean on the table because the slides for the gantry are mounted under the table. i can't risk bending the table.
so i just putter along on the machine, and get as much done as i can before i can move it to the new shop.

 

[travis.farmer]

Yesterday, i was able to add a rack tool changer code to my LinuxCNC configuration. i didn't write that code, entirely, as it was grabbed from another LinuxCNC member's configs. i did modify it to suit my machine, though. i had to temporarily disable some safety checks to run it though, as the rack and sensors is not there, yet. once the rack and sensors are there, i can re-enable them.
Still, it was nice to see the machine going through the motions. i hope to do a quick YouTube video at some point, showing the machine going through the motions. pretty fascinating to watch. i may even turn on the Air, so the spindle tool lock/unlock can be heard to function.

i see the schematic in David's CNC thread.. i am digging the colored wire traces. i use the electronics design part of Fusion360 (free hobbyist/non-commercial license), and to my knowledge, coloring the traces is not an option. i have attached my current, work-in-progress, schematic of my machine. it is not as detailed as David's, but i understand what it means, and can find the connections as needed. color traces would be a real improvement though.

 

[travis.farmer]

other than tying up some loose ends, like proper homeing sensor mounting (the tape method keeps coming undone), and cleaning up the on machine wires, the machine is basically ready to mount a spoilboard, and start using for basic signs and other carvings.
though, there is the issue of finding T-nuts and T-bolts for the aluminum table, and other mounting locations. the table T-tracks i have measured to be 3/8" inner tracks. all the T-nuts and T-bolts i have are 1/2" wide. so to mount the spoilboard, i either have to grind some T-bolts down to 3/8" wide, or find some other way to mount the spoilboard.
I can't drill into the aluminum table top, as the Y-axis slides mount under the table, and i am hesitant to weaken the table support ability.

if i can find enough T-nuts and T-bolts that fit without grinding, i could do away with one large spoilboard, and just place scraps of wood under anything i need to cut through the material, and otherwise just mount the material directly to the table.

 

[travis.farmer]

I'm not certain that is a good thing. Having a single spoilboard that can be surfaced for a known level and coplanar surface is a good thing. You can mount your workpiece anywhere and know that it is on a reference surface.

i was under the impression that where the gantry slides were attached to the underside of the table, that the gantry rode parallel with the table. i had assumed that the table was a reference surface, but perhaps you are correct, that machining a reference surface is best. as soon as i find something for a spoilboard, i will mount it in place.

i have been doing some thinking on my rack tool changer (for ATC), and due to my machines setup, rather than the usual back mounted rack, i will do a front mounted one. i will have to reach over it, but i will build it on slides, so it slides out of the way, forward, when not in use. i could even use a spare stepper and screw and some crap slides i have on a mini CNC that i no longer use. all it needs to do is move forward and back to allow the CNC to work down to Y-zero, when not changing tools. all i would have to do is tie it into LinuxCNC as a forth axis, and set it up with the limits and homing, like a regular axis. it will need slide and screw protection from wood chips, as will the tool rack. that can simply be a cover that flips open as the rack slides forward.
each tool pocket on the rack will need a inductive proximity sensor, so before it grabs a tool, it knows if the pocket has one in it. and likewise, if returning a tool, so it knows if the pocket is full already (shouldn't be, but for human error).
it is fortunate that the LinuxCNC + Mesa board system has ways to expand the number of digital I/O, with fair ease.

It is all so simple in my head... ;-)

i remember the old days of parallel printer port driven logic, or self-contained GRBL systems. not nearly as easy to add more I/O ;-) (trying hard not to sound like i am selling LinuxCNC and Mesa... but i do like the system a lot )

 

[travis.farmer]

it just depends on how much the humidity swings in your CNC area.

it swings a LOT. Maine is very humid, and neither the current nor the future (in process of building) shed is very close to climate controlled. the future shed will just be a large garage tent that i will build end-walls for, and insulate when i can afford to.

kinda why i was hesitant at first on using a spoilboard when not cutting through. kinda has me questioning how to keep it flat with wild humidity swings. seems like i should have quite a bit of spoilboard material around, ready to go when the previous one gets too thin. ;-) MDF at my local lumber yard has to be ordered, as it is rarely used in these parts. i think Home-Depot or Lowes may carry it (located 20 - 30 miles away).

 

[travis.farmer]

minor setback... bought some hardware to upgrade my machine... and didn't realize it would require such an extensive re-wire.
i bought some Mesa boards to expand the number of inputs (for my rack tool changer), and a board for an on-machine control panel. but to use the boards, i have to change the firmware on the main Mesa board, and that will move the wiring around to different pins. like, all the wiring.
it is a hassle, but needs to be done to protect my investment in the new boards.

 

[travis.farmer]

Inside every small problem is a larger problem struggling to get out ...

probably likely, though i am unsure exactly what you mean. i have many larger problems, so it is difficult to know to what one you refer.

 

[travis.farmer]

the hardware i bought to upgrade my machine is a circuit board that handles most of the inputs from the machine. or at least, all the PNP sensors (PNP sensors pull to the logic-high when active, where NPN pulls to logic-low when active). my homing sensors and apparently my tool-length probe are NPN, so those are still handled by the main board.
the reason for the add-on board, is it works with machine voltage (24V) rather than the low voltage (5v) the main board needs, so no need for voltage level converters. this cleans up my control cabinet just a bit, with the exception of the spaghetti-wiring everywhere.

another board i bought is a remote pendant/control panel board. this will allow me to have on-machine controls for various task. the machine is large, and when i get in the new shop, the computer will not be located directly at the machine, as it would be in the way of getting around the machine. this control panel board allows me to have up to 4 MPG hand-wheels (encoder wheels), so i can jog the axis around during the setup process.
it also will have a 20 X 4 character display that can be used as a DRO, or for other status info.
there is inputs for a 8 X 8 key/button matrix, that i can just wire up to 64 buttons to.
there is also some general purpose inputs i can attach an E-Stop button on, and 2 outputs that i can use for status indication, like when the machine is turned on, and when it is actively running a program (maybe). really they could be for anything...

i have not yet bought anything to make the tool changer rack yet, simply because i can not get to that side of the machine to attach and adjust the rack. that will have to wait until i get the machine in the new shed. i may buy the basic hardware for it at some point, but i do not need it just laying around right now.

 

[travis.farmer]

What is your spoil board made of?

at present, i have not attached a spoilboard, as i can not get around the machine in the current location to attach it properly. it will be MDF, i believe.

Are you thinking of cutting 6061 aluminum?

i would like to be able to do aluminum, but i believe this machine will be strictly for woodworking. i have been told that aluminum can be milled dry with carbide bits, providing that chip clearing is very good (either air blowing, or good vacuum suction). i may try it some day, with a carbide bit i don't care about experimenting with.
the HDPE spoilboard may be an idea too. though i can envision a mess on the floor if i can't contain the spray.

 

[travis.farmer]

that is very interesting that so little coolant volume is used. i had visions of coolant dripping off the table everywhere, but it doesn't sound like that is the case at all. i may yet again look into setting up for coolant, and try aluminum at some point.

an idea i had previously was a plywood table painted with epoxy floor paint, but the HDPE sounds far better. i was thinking of mounting aluminum T-track in the spoilboard, below the surface, and using the T-track as coolant drains. but with so little coolant being used, i can envision just a curb at the sides, and back, and let it drain off the front into a channel, and catch container.

see, my new shop floor is wood (Advantec 3/4" T&G), and i would rather not get much coolant on it.

Oh! Speaking of noise, WEAR EAR PROTECTION! We had an earplug dispenser for the operator and a pair of monkey ears for guests.

already planned on it. there isn't much noise while i am making sure everything is connected and working right during the build, though there are some times when testing something that i forget, and only realize afterward that something may have been too loud.

We bought three of those 5' tall free standing office partitions, the ones with fabric.

that's not a bad idea, as those panels are usually sound-deadening as well. they are expensive, even used, in my area as they are in such short supply, but they are easy to make.

at present, my biggest noise maker is my air-compressor. it is an oil-less pump, and really grates on the nerves. particularly when i forget it is on, and it starts to re-charge. good thing the ceiling in the current shed is very low. that way i have less distance to fall when i peal off the ceiling.

i kinda wish i had a nearby CNC guru, as i have an odd issue... well two actually.
firstly the sensor that detects if the ATC spindle is unlocked is not triggering. traced the wire out, and tried a different input, but nothing. i even verified, after the pin move, that the signal is not present directly from the ATC spindle. my guess on the solution is it has to do with the second issue.
the second issue, is that my ATC spindle needs up to 95psi to unlock the tool-holder from it, on paper. yet i need to feed it 115psi for it to be able to actually unlock, and it is slow doing so. not sure if my gauge is wrong on the regulator, or what is going on. could have something to do with the spindle internals being so cold, too. no full-time heat in this shed, so the machine only warms up when i go out and turn the heat on.

it would make sense if the pneumatic piston in the spindle needed some air-tool oil, but the instructions say specifically dry air, free of oil and water. though i have thought of a drop or two of oil, and just work the piston a few times to work in the oil, and expel the rest.

...oh dear... i seem to have written a novel again.

 

[travis.farmer]

You need a pressure switch so the cnc can know when the pressure is sufficient.

strangely, i have one, though it seems not to function. i have ordered a replacement.

but the spindle sensor is quite necessary, as i don't want the machine to move during a tool change until it knows if the spindle is locked or not. otherwise, things can break.

it actually has two sensors in the spindle. one to detect if the spindle is unlocked, and one to detect if when locked, if there is a tool in it. both are important. i will have to play around some more and see what i can figure out on the non-working one. i suspect the air pressure is too low to actuate 100%. it is enough to release the tool, but not enough to trigger the sensor.
it worked when i was on shop-air, as the pressure was higher. i don't think the compressor out in my shed can put enough past the on-compressor regulator (the weak link) to operate the ATC spindle. this afternoon i will try it on shop-air again, and see if there is a difference.

 

[travis.farmer]

i have attached the diagram of the pneumatic system, from the MFG. the diagram says 6 - 6.5bar (87 - 94psi), i took that to mean that max pressure is 6.5bar. if it is just the recommend range, i wonder what the max limit really is. i would love snappy operation, but i am working with Chinese translated directions that are spotty at best. i would rather not blow a seal, as i do not know where i could get another one from (the major flaw in my purchase choice).

i could perhaps bump it up to 120psi, but that is all the regulator on my compressor will do.

as for the switch, i set it from min to max and with 120psi from the compressor (it is before the machine side regulator), it would not activate. i think it is just simply a dud. i will just have to pay close attention to the gauge until the new switch comes.

 

[travis.farmer]

well, it may have been an impulse buy... but i bought the sensors and tool-holder forks to build a rack tool changer... i really don't know why, as i don't yet have room to work around the machine to adjust the rack changer, but the items went on sale for a very good price (i think), and i didn't know when they would go on sale again. if nothing else, i can store the parts in a box for safe keeping.
at least on the plus side, having sourced out the parts i need, i won't have to re-source the parts. perhaps i am just telling myself that to justify the impulse buy...

 

[travis.farmer]

well, the compressed air switches i bought, after i read a little more closely, are analog sensors. but my setup includes up to 6 analog inputs channels. they report in volts, so i wrote a a little conversion driver to convert from volts to PSI, with a digital "safe range" output that i can tie into the configs, so that before it runs the tool-change macro, it can check the value (boolean, true or false), if it is in the safe range (currently >= 105psi).

but that is why i love LinuxCNC. i can write my own little add-on driver programs to be able to process inputs and such as i choose.

the air pressure sensors are supposedly due to arrive today in the mail, so i plan on adding those.

the worst part is, i am running out things i can do, while the CNC machine is in it's current location. but spring is comming, and i am very much looking forward to working on my machine in it's new home.

 

[travis.farmer]

just got the air pressure sensors wired up and installed in the air lines. one pre-regulator that provides my safe value output, and one at the regulator as a matter of curiosity (the gauge was in fact wrong... or the sensor is...). but i wired them all in, made the necessary program connections, and gave it a test drive... everything worked! first try, even!

feeling very pleased with myself!

 

[travis.farmer]

interesting deal at Amazon... the impulse buy of the tool-changer forks i got for my machine... i ordered 10, amazon says i ordered 10... but i have a box of 15... i know they were on sale a bit, but i didn't realize it was a buy 2, get one free sale.

 

[travis.farmer]

one of the side projects i am working on for my machine is an on-machine control panel. basically some jog encoders, and various control buttons and switches. the console will be made from whatever i can scrounge up for wood, with a plywood face. the plywood will be holed out in places to mount panel modules made of 1/8inch PVC sheet that i have.
the major thing, from my perspective, will be milling the holes in the PVC sheet to mount the dials, buttons, and such, without the PVC exploding. i suspect light DOC, and feed speeds may prevent it breaking, as well as milling holding tabs for the disc cutouts, so they don't fly out.
hopefully i can remember to take some pics as i build it. keep in mind, though, the wood console will be for function, not art, so i doubt i will spend much time on making it look great. but anything is possible. likely it will just get painted, anyway.

i have a control pendant (for Mach3, but works with LinuxCNC), but i was never fond of the latency. i push the stop button, and so it stops, eventually. not a good setup for E-Stop. but this control panel is connected to LinuxCNC in real-time, so it should respond better. to be tested, of course.

the board for the control panel includes a driver for a 20 X 4 character LCD display, normally used for a DRO, but it can be paged, so if i can figure out how to change pages, i can display a variety of information, like air PSI, and such. i got that connected and working today, and it seems to function quite well. tomorrow, i hope to connect some dial encoders, and see if i can jog the machine with them. i will also connect an E-Stop switch, so i can test the response of that as well.

still not sure how i will attach the control console to the machine, but it may just be a roll-around stand, with a weighted bottom for stability. but anyway, just thought i would share my thought process.