Manipulating welder output with Arduino

[joshuab]

I had a four-hour flight today, so I spent some time playing around with DipTrace trying to figure out how to integrate the Boarduino PCB into my PCB. Finally, I figured it out. First, I opened the Boarduino schematic and pasted my schematic into it, then hooked up my components to the Boarduino schematic's pins in the appropriate ways. Next, I opened the Boarduino PCB file. The key step was to select and "lock" the Boarduino PCB before doing a back-annotation to the schematic. Before, when I back-annotated to the schematic, it would erase all of the Boarduino PCB's nice, professionally-designed routing. With the Boarduino locked, all of its routing remained, and my parts were added to the board with rat-lines, so I could finish routing them. This is a good step if I ever get to a point where I want to produce the boards on a larger scale. For one thing, it wastes none of my valuable PCB space. For another thing, I can order the components of a Boarduino from Mouser (qty. 1) for about $7, whereas they cost $18 from Adafruit. $5 of that $18 is the Boarduino PCB itself, which I'll no longer need.



Keeping the Boarduino layout is very desirable, as it includes a nice, regulated power supply, ICSP header for programming, and and all the other support miscellaneous that are needed.

[joshuab]

Back from vacation. Got the PCBs in the mail. Spent a little quality soldering-iron time tonight.



Pretty satisfying to see the board looking all professional-like. I haven't tested it yet, as I want to get the metal enclosure set up instead of the plastic one I have been using. More to come!

[joshuab]

I went to the hardware store this morning for a 5/8" hole saw. I needed a 5/8" hole to mount my buttons in the new, metal enclosure. On the plastic enclosure, I used a spade bit, but I didn't think those could be used on metal. Well, the hardware store didn't have a 5/8" hole saw, but the helpful clerk said that a spade bit could be used on aluminum no problem, so I gave it a go, and it worked okay.

Basically, now I am back where I was 3 weeks ago, but this time with a PCB, a nice metal enclosure, properly-mounted board, spring terminals instead of solder joints that are going to break, and so on.







I have done some rudimentary checks of the PCB, and it is working fine so far. Knobs, buttons, and lights are working fine. I found that I forgot a smoothing capacitor on the PCB, but I soldered it between two of the leads so no problem there. Hopefully, that turns out to be my one mistake for this PCB.

Once I have confirmed that the PCB is correct, I'm going to continue working to refine the interface circuit with the welder. I implemented a small change in the PCB that is supposed to improve the performance of the optocouplers a smidge over what I had with my prototype board. I also pulled the input line to the interface circuit out to one of the un-used spring terminals so that I could come up with a wholly new circuit on a daughter-board if I wanted to. I have some high-speed optocouplers that I may give a try to. I have passed them over so far because they require 4.5v minimum on the welder's side of the circuit. The welder puts out 5v to the remote when it's on 220v power, but only 2.7v when it's on 110v power, so it's possible the high-speed optocouplers wouldn't work on 110v power.

[joshuab]

The theme of the night is "debounce". While I was working on the bench, for whatever reason, I never had to debounce my buttons. As soon as I hooked up the actual torch switch, I got all kinds of crazy bounce issues. For those who don't know, "bounce" is when a switch, in the fraction of a second when it makes or breaks contact, causes the input to rapidly cycle between high and low. Without appropriate countermeasures, this registers as several switch presses occurring very quickly, and it screws up the response to whatever you were trying to do by pushing the button. So I have been working to implement software debounce, and have pretty much finished that.

My RC filter circuit is working well. I have got the PWM frequency on the Arduino maxxed out at around 32 kHz, and you can baaaarely hear the whine coming through when you're welding. The changes I made to the welder interface circuit have improved the optoisolators' response to the point where pulsing seems to be right on the verge of working properly. Because I still have a few software issues to work out with pulsing, I can't be sure whether any remaining issues are related to the optoisolators or the code.

One thing I discovered was that, if I had debugging turned on in the code, the 9600 baud serial output took so long to write that it was causing the code to miss pulses if the pulse frequency was too high. This was one cause of the messed-up pulsing in my previous demonstration video. Also, I seem to be having an issue with pulse balance, where it isn't going past maybe 50%. This is almost certainly some code-related thing.

2T functionality and pedal functionality are basically at 100%. 4T functionality is also at 100% as of tonight. I'm still using only lift-arc, because I haven't even started to broach the issue of shielding from interference. I get much better results with lift-arc in 4T mode, where I can come in at something like 15 amps and then let the arc ramp up to full output. There is MUCH less sticking and contamination of the electrode at 15 amps than at higher amperages.

I have not implemented the "goes back and forth" feature that has been suggested to me for 4T operation--where if you press and release the switch during upslope or downslope, you can ramp back and forth. Currently, any torch press or release during upslope or downslope ends the arc. I personally like this better, because it means that I can kill the arc definitively any time I need to, without having to wait for a downslope. I might change my mind if I had more field experience with "real" 4T operation. I thought that adding the "goes back and forth" feature would be a trivial re-mapping of my state diagram, but I tried that, and 4T got all fubar, so I just dropped it, because it's not a feature that I'm dying to use anyway.

[joshuab]

Here's my workspace setup when debugging requires the use of the welder. Fortunately, it's TIG, and I'm really just striking a low-amperage arc on a flat plate, so I'm comfortable doing it inside.



I've been tempted to just set the laptop on the welding table, but the potential for disaster is just too high.

[joshuab]

WAHOO! PULSING! THIS IS IT, FOLKS! THE HOLY GRAIL!

[Depdog]

Very cool read through and watching the videos. Made any more headway on the box? I have the new version of the 185 and would love to add pulse to it.

Glenn

[RichardH]

Oooh, hardware hacking - very cool project, and right up my alley! Thanks Depdog for giving it a bump. I have some reading to do here and I'll look forward to its progress.

I wonder if Everlast has considered building an open source version of a welder... As example, Canon offers this ability in their cameras, and it can be pretty cool to see what folks come up with.

Cheers,
Richard

[joshuab]

Very cool read through and watching the videos. Made any more headway on the box? I have the new version of the 185 and would love to add pulse to it.

Work has gotten pretty crazy since I posted that last video, but I feel like I reached a pretty good plateau, developmentally speaking. The next steps would be to work on hardening against interference so it could be used with HF start, but as long as you're doing lift-arc, the thing is pretty solid.

I'm happy to share parts lists, schematics, and source code with anybody who's interested. I also have two extra PCBs printed up that I'd be glad to sell you (for cost, plus shipping). If you don't already have a pretty solid foundation in electronics, I wouldn't recommend it, but if you know your way around a multimeter and a soldering iron, you could probably give it a go.