Upgrading Chinese Plasma?

Hi all,

My boss recently started a new company, and I was his first employee, I'm in charge of design, and we have a couple more employees manufacturing. We got a gantry plasma from China, but the more we use it, the more problems we are having. Here's the main ones:

There is no active height control, it only gets a height before piercing, and stays at the same height through the cut, we cut some 6m sheets and if it's warped it wont keep a constant height. There is also no way to set the height in software, there is a little knob for height "delay" on the controller that determines how high it will go before each pierce, so we have to literally play it by ear to set it. We often had problems on more warped sheets where the middle is bowed out, and the torch hits it, halting the cut. On a few occasions I had to stand on the sheet as it was cutting to keep going.

As far as I can tell there are no encoders on the motor, I'm not sure how it knows where the torch is, but as far as I can tell the controller doesn't get feedback from the motor to tell it if it stopped moving. On several occasions the torch or the drill head got caught on steel, tried to move but didn't, after raising it back up again and starting the program it would cut somewhere else entirely, because the torch isn't where the controller thinks it is.

There's no way to control air pressure, although I'm not sure if you have to. Current and Voltage are also controlled by manual knobs, voltage knob is on the control panel, and the current is set on the plasma controller, so there's no way to adjust these settings through the controller/program.

There was no real table for cutting settings supplied, they supplied one we were using but the manufacturer pretty much just told us that they aren't necessarily the ideal settings, just a starting point and to experiment with them. What we have is a list of steel thicknesses and corresponding nozzle diameters, speed ranges, voltages, and currents recommended. But I don't trust them that these settings are optimal in any way.

There isn't even a way to turn off the torch as it's moving, I found the M code responsible for turning off the torch, but if fired it stops movement too, rather than turning off on leadout as I wanted.

There's a drill too but it's a bag of shite too, it's a spindle mounted beside the torch, the plunge speed is controlled by very inaccurate knob on this pneumatic/hydraulic ram, and the RPM is controlled on an inverter inside the main electronics panel of the plasma. I literally measured the plunge speeds to create this graph that I can reference to get approx. mm/sec for plunge speed, and I have to convert frequencies into RPM for speeds.
Even then it's not designed well enough to know to speed up until it reaches the steel then slow the plunge, and only go as deep as it needs to. So usually you wait 30 sec for it to plunge down to even start drilling, then once it goes through the steel you either have to wait another 30 seconds before it bottoms out, or you have to hit pause on the controller so it retracts and goes to the next hole.

So yeah, we make do with it, anything that isn't very critical can still be cut and worked with, but parts are bevelled, every hole needs to be reamed out, and so on.

Anyone ever tackled a job like this? In my head it sounds like it should be possible by scrapping the original controller and adding a linux cnc one, and then any other electronics/sensors required to get the functionality we need. Mechanically it seems to have all it needs.

If anyone's interested the controller on the machine looks identical to this:
www.collectcnc.com/starfire-sf-2300-cnc-...roller-p1333316.html

with a THC very similar to the one in the photograph mounted below it in their custom housing I suppose.

The plasma itself is almost the same as this one, except it doesn't have all these gas nozzles, just plasma, single oxy torch, drill spindle and marker. But the plasma supply, gantry and everything else is the same:
www.oltcnc.com/nry.asp?bt=cp&dbtid=120&xbtid=81&id=681


We aren't at the stage where we can invest in a top of the range plasma, but I was hoping that by investing a bit of extra time and money we could get way more value and quality of the machine than we do currently.

I've read your big screem topic.


And I must say that

I'm not sure how it knows where the torch is

then change controller from Chines to linuxcnc by your own forces is a VERY BAD IDEA.

I found the M code responsible for turning off the torch, but if fired it stops movement too, rather than turning off on leadout as I wanted

I want to asure you that you found bad code, keep going in your search to find more suitable command (for example if M3 - turn on, than M5 - turm off). Look G-code quick reference for more details.

For other questions you must ask supplier for help and consultate you with G-code for controlling their plasma.
For example automatic tourch height control should be activated/deactivated with special G-code commands.

Retrofitting to PlasmaC I feel would be the best option as far as performance will go for low cost.

First thing, Can you take good pictures of the motors on the unit and the drives so we can determine what signals they use.

Once we know that we can make recommendations for a Linux-CNC based controller that will run PlasmaC.

I found a couple of manuals for the Starfire SF-2200. when I searched for the SF-2300. Its a 2 axis controller
It seems the Axis motor controllers are Step and direction.

It seems there is a 24V DC motor on the Z axis Lifter. I'm pretty sure Alibaba is awash for this type of lifter. at one stage I did manage to track the motor part number down and it was a pretty readily available 24 volt geared motor. So the X & Y axes are going to be pretty easy to control from Linuxcnc but you'd need to explore the Z axis further.

I don't know enough about servo controllers but you might be able to add one that controls the DC motor, otherwise I would be exploring how to fit a Small servo in place of the DC motor. It probably needs a reduction gear set. Failing that, I would explore replacing the whole Z axis lifter mechanism.

Plasma usually has a bevel on it and there is a diagram in the Plasma primer that explains how this works. You may get improvement by changing the cut height.
linuxcnc.org/docs/devel/html/plasma/plasma-cnc-primer.html

But that does not address the plasma unit itself. It will take some time and effort to dial in cut speeds.
The lack of air pressure control should not be a concern as that is linked to the nozzles you are using.
I'd look on line for the manual for the Thermal Dynamics A120 for guidance on air pressure. It has manual pressure settings up to 120 amps.

So in summary, I would say if considering a retrofit, you need to come up with a strategy for the Z axis.
LinuxCNC is multi spindle so properly controlling the drill should not be too hard. There are some features in Plasmac to do this.

Another thing worth adding,

If you are making actual money with this machine it might be worth doing the numbers on what kind of return you will get from a well known plasma unit such as a Hypertherm or Thermal Dynamics machine.

I can understand the uses of a china machine for the hobbyist but when you are destroying good steel you paid for because of the bad cut quality. it does not make sense to use a china machine for business.

If you buy a brand name it will come with a cut settings chart and will get you great cuts out of the box.

Thanks for the replies guys. I didn't mean to come off ranty or like I'm complaining too much. While I do find issues with the plasma, 80% of the time the results are satisfactory, so I'm not going to go in guns blazing and tear it all up and start changing things. First I wanted to know if it's possible to do something about it, and it seems like it is, but perhaps I can get an improvement with some minor changes first, like dialling in the right settings alone.

Anyway next time I'm in the factory I will get some info on motors, drives and the Z arrangement. It would be great to at least figure out a way to add automatic height control, as currently it only probes it, and it seems the height is controlled by a timer, there is no way to actually input a height, I'm sure this is the main issue causing bevelling problems. I understand some bevel is unavoidable, but often times it's ridiculous, for example we had 20mm holes in 10mm plate come out 15mm on the bottom side.

I can understand your position,

What might be the limiting factor is your controller. I am just making assumptions but I would guess it is a proprietary design that most likely wont allow for the addition of a third party THC. Maybe the manufacture offers such a feature as a upgrade? Might be worth looking into.

The best thing about LinuxCNC is its ability to be modified to suit different hardware. In the age of proprietary design it is really great to have that flexibility.

thefabricator03 wrote:

I can understand your position,

What might be the limiting factor is your controller. I am just making assumptions but I would guess it is a proprietary design that most likely wont allow for the addition of a third party THC.

I think this is on the money but the limiting factor stems from the lack of motion control of the Z axis. Which in turn means they only used a 2 axis controller. If you had step and direction signals for the Z axis, there are external THC controls designed to be retrofitted to machines in this class that act independently of the motion control. Robot3T may have a solution that works with your Z axis motor.

I think if you replaced your lifter with a purpose built plasma lifter stage (thefabricator03 might have some insight here) and adopted LinuxCNC's Plasmac interface you would be rocketed into a different world.

But then again, I wonder if linuxCNC could not control your lifter by a PID component that uses torch voltage as a feedback signal. THen the limiting factor is how to set torch height