controller freezes after firing up plasma arc

[mauritsw]

I built a CNC plasma cutter that is controlled by an arduino with grbl installed. The arduino controls the motor drivers: Y-axis: DM556, X-axis: TB6600, Z-axis: DRV8825.
I'm using a 24V 15A switching PSU as power source.
The limit switches are NO and switch to GND via an optocoupler.
The plasma cutter trigger is also switched with a relay behint an optocoupler.

Everything works great as long as I don't turn on the plasma cutter. Once I turn it on, the arduino crashes immediately and corrupts it so badly that I can only reconnect if I unplug and reconnect it.
The plasma cutter is a low-frequency with a pilot arc, it even says "CNC ready". Therefor I didn't pay much attention to shielding and avoiding EMI/RFI. Unfortunately...

Now my question is, where should I start? The electronics are in a metal case that is bolted to the frame and grounded (earthed) on the poweroutlet via the PSU mains wire, but I can organize this grounding better.

What are my best options? Ferrite clips on the GPIOs? Shielding the motor cables? Shielding the plasma cable? And what should I connect the shielding to so that I don't create a ground loop? To earth or to the GND of the controller? Can I do something with HF filters in the form of capacitors with a low Farad value? And do I only put them on the GPIOs or also on the motor power cables, even though they are galvanically separated?

I'm curious about your experience/advice!

[mactec54]

I built a CNC plasma cutter that is controlled by an arduino with grbl installed. The arduino controls the motor drivers: Y-axis: DM556, X-axis: TB6600, Z-axis: DRV8825.
I'm using a 24V 15A switching PSU as power source.
The limit switches are NO and switch to GND via an optocoupler.
The plasma cutter trigger is also switched with a relay behint an optocoupler.

Everything works great as long as I don't turn on the plasma cutter. Once I turn it on, the arduino crashes immediately and corrupts it so badly that I can only reconnect if I unplug and reconnect it.
The plasma cutter is a low-frequency with a pilot arc, it even says "CNC ready". Therefor I didn't pay much attention to shielding and avoiding EMI/RFI. Unfortunately...

Now my question is, where should I start? The electronics are in a metal case that is bolted to the frame and grounded (earthed) on the poweroutlet via the PSU mains wire, but I can organize this grounding better.

What are my best options? Ferrite clips on the GPIOs? Shielding the motor cables? Shielding the plasma cable? And what should I connect the shielding to so that I don't create a ground loop? To earth or to the GND of the controller? Can I do something with HF filters in the form of capacitors with a low Farad value? And do I only put them on the GPIOs or also on the motor power cables, even though they are galvanically separated?

I'm curious about your experience/advice!

If you are using 2 power supply sources to your machine, then you will have created a Ground loop which will be where most of your problem is coming from, would have to see photos of all the wiring to get a better idea of what is going on.

[dieimison]

Troca o arduino por uma placa de porta paralela, aquelas DB25 que resolve o problema.
a interferência do plasma trava a comunicação USB
Pode utilizar um notebook também ao invés de um PC normal.

[JR107]

Now my question is, where should I start? The electronics are in a metal case that is bolted to the frame and grounded (earthed) on the poweroutlet via the PSU mains wire, but I can organize this grounding better.

What are my best options? Ferrite clips on the GPIOs? Shielding the motor cables? Shielding the plasma cable? And what should I connect the shielding to so that I don't create a ground loop? To earth or to the GND of the controller? Can I do something with HF filters in the form of capacitors with a low Farad value? And do I only put them on the GPIOs or also on the motor power cables, even though they are galvanically separated?

One option would be using a break-out-board to get as many signal wires into differential pairs as you can. You will also need a down-conversion board to get them back into single signals. Also you get what you pay for in terms of the wires themselves. Good double shielded twisted pair cables are usually found direct from a cable / wire company and are not cheap.

[Al_The_Man]

Very important, especially with Plasma M/C, is proper grounding to a star point, and carry out equi-potential bonding of all metallic parts of the machine.
There is at least a couple of commercial Companies that use a PC for control with no ill effects from the Plasma side.
The motor cables should be at least twisted pairs the whole length of the cable.

[servtech]

Very important, especially with Plasma M/C, is proper grounding to a star point, and carry out equi-potential bonding of all metallic parts of the machine.
There is at least a couple of commercial Companies that use a PC for control with no ill effects from the Plasma side.
The motor cables should be at least twisted pairs the whole length of the cable.

I have worked on CNC machines in almost every environment and without doubt the worst situations are either in, or close proximity to, welding machine shops and induction welders are the worst offenders.

Screen all signal cables, axis motor cables and ground the body of the machine separately to ground also. I'm of the school that believes all screens should be connected at the ground end only, some might disagree.

That said, I don't think the GRBL board is protected well enough for close proximity plasma welding, the line noise generated could easily corrupt or damage components. A mains filter with correct choke might help.

[joeavaerage]

Hi,
plasma is by definition 'contained, continuous lightning' and is always going to produce lots of electromagnetic noise. Many useful strategies have already been advised, and you'll probably
want to implement all of them.

If I'm not mistaken GRBL is connected by USB? USB can be subject to noise, I would advise the shortest possible USB cable, preferably with one, two or even three ferrite split cores
on it, evenly distributed.

Craig

[Al_The_Man]

I have worked on CNC machines in almost every environment and without doubt the worst situations are either in, or close proximity to, welding machine shops and induction welders are the worst offenders.

Screen all signal cables, axis motor cables and ground the body of the machine separately to ground also. I'm of the school that believes all screens should be connected at the ground end only, some might disagree.

.

I also have worked with CNC machines for quite some decades, I agree with the Siemens paper on Equi-potential bonding, this practice allows the preferred method of grounding both ends of shielded cable, which originated because of ground loops and the necessity to ground one end only.
In all the decades I have worked with both retrofitting and developing custom systems, I have never experienced ground noise problems.