Setting up a plasma cutter and looking for some help.

I am in the middle of retrofitting an old plasma cutter to use linux cnc and needing some help. My linux skills are slim and none.

Using an old Parallel port breakout board we had laying around. The older version of this with the socketable ic's instead.

Using clearpath SC AC motors in step/direction mode. (These are servo motors but they close the loop on the motor itself so treat them as steppers than won't loose steps for the purpose of this exercise.)


I have X and Y moving and ~ accurate as to steps per inch. Currently I have the motors set to 8000 steps per rev. This equates to about 8500 steps/inch on the Y axis and 6500 on the X. Not sure if this needs to be lowered. (The motors can be changed as to how many pulses to a revolution. Currently set to the resolution that allows for full speed on the motor's side of things.)

I don't have a z axis. What I do have is an external bluescreen Torchmate AVHC . It has the following pins.

Outputs from linux CNC. (I think I gotta find relay's as I don't think I can use a direct output from the breakout board for this. Lets see what I have laying around.)

Start +/-. These are to be wired to the N.O. contacts on the cnc's start relay.

Corner +/- Disables the auto height for when making corners.

Ok to move +/- When enabled via a contact closure the AVHC assumes the arc has been established and begins the pierce delay. Says this comes from the plasma cutter if equipped however it was going to the old cnc setup setup so im not sure what if anything I should do with it.

Inputs to linuxcnc.

Pierce. The is a NO contact on the AVHC that is used to signal the CNC to start moving. As I understand it goes Closed after the AVHC finds the height of the material by pushing down on it.

Fault. NO contact that closes to let the cnc know that something went wrong.

imgur.com/a/8ToLD4E pictures of the AVHC manual and machine. Pardon the mess. It has sat for like 6 years and was old before then.

I don't know that I have enough inputs for everything. I can parallel the 4 limit switches. Can I also parallel the 2 home switches and then have it home each axis individually? This would cut home/limits from 6 to 2 inputs.

i don't think that plasmac is really the right interface to use becasue your torch/THC does what Plasmac does.

I would start with a simple XY machine with axis/gmoccappy or whatever you like. Additional connections will go something like
start = spindle.0.on
pierce = spindle.0.at-speed
fault = halui.program-pause
corner = use this component forum.linuxcnc.org/plasma-laser/32658-thc-cornerlock-component
connect to cornerlock.0.is-active
OK to move = comes from plasma cutter ArcOK (sometimes called transfer) from the plasma cutter CPC port

Re cornerlock, I would just hardcode in cornerlock.0.threshold to be say 0.9 with a setp command

So this component compares the desired cut velocity (set with Gcode F command) with the current velocity (motion.current-vel) and if it falls below a threshold, enables an output pin which can be used to tell your THC to halt adjusting the height.

We have a greater understanding of how cornerlock should be used now since that was written and the problem is that the original feed rate is thrown away after the F command is parsed by Linuxcnc. The solution is to use a remap (there is a sim example of remap to guide you). So what you want to do is remap the F command and set one of the motion.analog inputs to = the F command in the remap with M67/M68 and use the analog pin to set cornerlock.0.cut-vel in HAL.

Finally just to clarify the start and pierce signals. When an M3 is received, spindle.0.on goes true but motion does not commence until the spindle gets to speed at which time motion.0.at-speed is expected to go true. When this signal is received, motion commences. You should be able to get this going with minimal effort subject to you having enough I/O pins.

Personally with some high quality servos in use, I'd consider using a Mesa 7i76 or ethernet 7i76e which would allow you to set this all up properly without being constrained by lack of I/O. The cheaper Mesa ethernet 7i96 might have enough inputs and has the advantage of having relays on board.

And re inputs for home and limits, you could do this with shared home/limit switches. Assuming separate motors on each side of the gantry, you need separate home switches, so you would need 3 inputs as min limit and home. You are going to need to use V2.8 or V2.9 to square the gantry when homing. You could skip the max limit switches but you may be able to common them with the home/limit for the single motor axis.

rodw wrote:

And re inputs for home and limits, you could do this with shared home/limit switches. Assuming separate motors on each side of the gantry, you need separate home switches, so you would need 3 inputs as min limit and home. You are going to need to use V2.8 or V2.9 to square the gantry when homing. You could skip the max limit switches but you may be able to common them with the home/limit for the single motor axis.

Single motor gantry. The sides are tied together with a large shaft. The motor is on the left in these pictures. Each end of the shaft has a rack and pinion gear setup and the gantry rides on round linear rails on top.

I am on 2.9 atm as I was under the impression I needed it for plasmac.

rodw wrote:

i don't think that plasmac is really the right interface to use becasue your torch/THC does what Plasmac does.

I would start with a simple XY machine with axis/gmoccappy or whatever you like. Additional connections will go something like
start = spindle.0.on
pierce = spindle.0.at-speed
fault = halui.program-pause
corner = use this component forum.linuxcnc.org/plasma-laser/32658-thc-cornerlock-component
connect to cornerlock.0.is-active
OK to move = comes from plasma cutter ArcOK (sometimes called transfer) from the plasma cutter CPC port

Re cornerlock, I would just hardcode in cornerlock.0.threshold to be say 0.9 with a setp command

So this component compares the desired cut velocity (set with Gcode F command) with the current velocity (motion.current-vel) and if it falls below a threshold, enables an output pin which can be used to tell your THC to halt adjusting the height.

We have a greater understanding of how cornerlock should be used now since that was written and the problem is that the original feed rate is thrown away after the F command is parsed by Linuxcnc. The solution is to use a remap (there is a sim example of remap to guide you). So what you want to do is remap the F command and set one of the motion.analog inputs to = the F command in the remap with M67/M68 and use the analog pin to set cornerlock.0.cut-vel in HAL.

Finally just to clarify the start and pierce signals. When an M3 is received, spindle.0.on goes true but motion does not commence until the spindle gets to speed at which time motion.0.at-speed is expected to go true. When this signal is received, motion commences. You should be able to get this going with minimal effort subject to you having enough I/O pins.

Personally with some high quality servos in use, I'd consider using a Mesa 7i76 or ethernet 7i76e which would allow you to set this all up properly without being constrained by lack of I/O. The cheaper Mesa ethernet 7i96 might have enough inputs and has the advantage of having relays on board.

I may look into a mesa board. I am half making do with what we have laying around for now. (Less an issue of money as if I need to buy it to make it work my boss would not have a problem and more I have been doing this mostly seat of my pants and don't know what I need till I get to it.)

I believe the rest of what you are saying makes sense. Hopefully it still does monday as I was actually supposed to leave work 15 minutes ago. I am sure I will be back then with more questions. Or not. Might be Tuesday as I still have plenty of hardware side things to do. Ive

Cool, so 2 inputs, one for each axis, common the min and max limits of each axis and use shared home/limit input for each axis. Make sure during homing the home offset moves the axis off the limit switch so the limit error is not triggered when you come out of homing.

I would just build a basic X/Y table now using the axis gui. Get X/Y motion working. Maybe just setp spindle.0.at-speed to 1 to start with so you can move it. Then add your plasma pins as I mentioned.

I have never used a parallel port but hopefully you gain a couple of outputs from dropping off the stepgen for the third axis. You only need 2 more inputs to get this going.

You should end up with a high quality machine, having the cornerlock will be a great addition.

I used the Torchmate AVHC for Zaxis control for years and as Rod says it will not work correctly with the PlasmaC branch without serious hacking. Further more depending on the actual firmware version you have the cornerlock feature may or may not work even though the input is there. If you still really want to use the TM lifter assembly (Z axis mechanics and drive) the very expensive Smartmotor on it can be programmed to work as a simple step and direction drive like your Clearpath servos. The TM lifter is a POS really so build or buy a better one.

Just spend the money and buy a Mesa 7i76E and THCAD boards, you've already splurged on good servo motors. In the end you will be much happier with the results, actually no comparison really. Did I say the TM AVHC system is a true POS compared to the PlasmaC standard?

John

So I think where the steppers will restrict you is on the maximum speed. Looks like your servos can get up around 2800 rpm (vs 750 rpm I work on for my steppers). You want to be able to get up around 800 inches per min but if your motors can go faster, try and use the max velocity and acceleration you can without the machine shaking apart.

I thought I saw someone say parports were going to max out at step pulses at about 50 kHz and the Mesas are good for 10 mHz (subject to other constraints re pulse timings). So I suspect using a parport is going to significantly limit both the top speed and the number of pulses per rev.

Do some tests and see how fast you can get it going and compare with what the Clearpaths are capable of. Then make the case to your boss about the hardware becasue with plasma velocity and acceleration are king! The faster you can accelerate into and out of corners the better your cuts will be.

You say that you have two axis moving now. What type and thickness of material do you want to cut? What speed and acceleration can you get now? If you don't have a working LCNC configuration how are you testing you axis servos? Did you have the Teknic's sales engineer size your servos for you? If not you most likely didn't get the correct size for your table, it is much harder to correctly size servo motors for good performance than steppers. They may work now but the overall performance will never reach a servo driven systems potential.

John