Welding -- what to do about HUGE 2" arcs

HF at all.

So, maybe, there is nothing even unusual in what I have.

i

Yes, Iggy's welder tries to maintain current at setpoint all the way up to 90 Volts. I think I have convinced him that this is not ideal in a separate discussion on SED.

Droop :-).

Yes, I think that you mentioned droop a couple of times. :) Thanks Glen. I will work on it. I am making a control board now, with some relays etc, will add also some droop stuff as well. Fortunately, many things can be added as I go.

i

And the flux covering on stick electrodes have a lot of sodium/potassium compounds for binders and arc stabilization...

Once an arc has been initiated with a small gap and perhaps HF or lift arc assistance, the voltage required to maintain the arc, even as long as 2" length is fairly low. If the voltage were 30 KV, the welding conductor insulation would be breaking down, causing great discomfort to Ignoramus8797 and the topic would be different now.

I think the challenge here might be to add software management that will sequence a 90 Volt OCV during start but shut the output down after the arc has been running and OCV rises above 60 volts. Once the machine has software controlling everything, it can be tinkered and improved forever at low cost. That would assume you enjoy writing code though. ;o)

I would also examine the design for EMI/RFI remediation and protection of the controller and output devices. Faults here can lead to expensive failures.

BTW, you have a fascinating project, ignoramus8797. I've much enjoyed reading your project blog.

Larry

Mike Berger wrote:

Thanks. Some intelligent person just made an interesting suggestion, to find some "embedded linux" system. I think that it is wise to go for computer control (maybe basic stamp or some such) rather than for a bunch of relays.

If I can find something suitable, that would be awesome, linux is an extension of my body basically.

i

How about embedding a PC compatible desktop as your controller? That completes the control panel and user interface for the cost of an obsolete PC setup. Those are sometimes found abandoned in the trash once they have aquired a fully variety of virus infestations. It wouldn't take long to clean one up with a fresh install of Linux. That would save the time needed to fabricate control switches and displays for a stamp module.

The expensive part would be an I/O card to plug into the bus. I'm thinking eBay is the place to find that item since they are pretty expensive as new items. In a pinch, a video game input card and printer port output might suffice. I would look for the I/O card(s) myself. Much of the EMI proofing of the controller system would have been handled in the PC packaging for you. Only the lines from the cyber tig to the I/O cards would require further attention.

Whatever controller you choose, the next step would be defining what inputs and outputs you require for future tinkering. I'm thinking output voltage, current sense and envelope detectors are needed. For outputs the gas solonoid and power control signals come to mind at the moment. All the indicators can be graphical items on the CRT or LCD monitor.

Larry

Ignoramus8797 wrote:

These desktops are all too big to fit into my welder. Your idea is very interesting, in fact, but I think that practical reality makes embedding a real PC into the welder impossible.

Here's what I ended up buying, rightly or wrongly:

the seler of this thing is Comfile itself. It can be interfaced to serial port or Ethernet (with an extra card), i think.

I thought that envelope detection can be done by software, no?

Yep,

- (*) on/off,

- (*) contactor control (low voltage for welding vs. high voltage for plasma),

- (*) gas solenoid+water cooler 120V outlet

- (*) High Frequency arc stabilization output

- desired current and voltage outputs

Outputs marked (*) are relay style outputs, YES or NO (or 1 vs. 0).

current and voltage are continuous.

i

That board has a nice collection of I/O's. I had been thinking in terms of graphical user interface with the PC on top of and not inside the welder. I figured the welder isn't portable anyway. On further thought, a CRT display would never work nearby the welder's magnetic fields anyway and an LCD monitor is still pretty expensive.

Ignoramus8797 wrote:

I would first worry about EMI from the welding arc and HF power getting into the computer and seriously confusing it. Build the computer into a diecast metal box with lowpass feedthroughs carrying power and digital I/O signals in and out, and shielded cables for connections to the display, et al. Start with the EMI control part of the design first, and life will be a great deal easier.

Joe Gwinn

I agree. The I/O signals needs R-L-C decoupling and the stamp board needs to be enclosed. Feedthru capacitors may be of limited use without some series R and/or L outside the metal enclosure. Series resistance is desirable so the series L and shunt C of the feedthru don't become high-Q resonant at some frequency. Series R, and a shunt feedthru paralleled with a larger value cap would work nicely on analog inputs. Feedthru caps larger than 1000 pf tend to be expensive and 1000 pf is inadequate for this application. The cap paralleled across the feedthru could benefit from a ferrite bead on it's lead so lead inductance doesn't resonant with the feedthru at UHF. Relay outputs won't need as much decoupling treatment but should not be ingored.

Because of low frequency magnetic fields from the weld> I would first worry about EMI from the welding arc and HF power getting

Guys, I read what you wrote, with great interest, but feel like I need to clarify something.

1. The welder I have has three compartments: the transformer compartment, the HF/inductor/commutator compartment, and the control compartment.

The control compartment is fully isolated from the rest by steel sheet, except for a couple of 3/4" holes where wires go through.

1. All wires that are used for voltage or current sensing are twisted (with a drill, about 2 twists per inch).
2. When I install the AC inverter (using IGBT), it will be located in the transformer compartment.
3. The voltage and current sensing is done on the DC bus right off the rectifier, before the inductor and HF section.

So.. In light of this... Is interference from the xfmr an issue?

would simply installing a resistor and capacitor be enough to filter input?

The SCR controller thingy that I have, seems to survive welding and HF, just fine. Does not mind it a bit. It was advertised for "noisy and dirty industrial conditions". The CuSB22R controller may not be such, though, so I want to do the right job.

Addition of the AC inverter could materially change things, also. Very large dI/dt.

i