Manipulating welder output with Arduino

[everlastsupport]

Another interesting twist and no warranty issues A servo and pulse and it turn a pot. Servos are cheap now. Then go wi-fi on the pedal or bluetooth That would be slick.

[joshuab]

Another interesting twist and no warranty issues A servo and pulse and it turn a pot. Servos are cheap now. Then go wi-fi on the pedal or bluetooth That would be slick.

I have thought of the servo/pot approach, but I'm not sure what the realistic modulation frequency of that would be. When I talked to Mike on the phone, he said that he wouldn't recommend high-frequency pulse through the pedal input, but that pulsing in the range of up to about 10 Hz to maybe 30 Hz would probably be okay. Could a servo and pot operate around 10-30 Hz? Seems like that'd be asking a lot of that little pot, to wiggle back and forth that fast, even if the servo could pull it off (and surely there's a servo out there that could do 10-30 Hz without breaking a sweat).

If all I wanted to implement was 4T and, maybe, 1 Hz pulsing for timing filler-rod dips, a servo/pot setup would be fine.

[Rambozo]

There was a guy that rigged up a cable operated foot pedal to a HF welder front panel knob as a way to get a foot control without opening up the unit. Seemed like a lot of work, but it did work for him.
30Hz would not only take a good servo, it would take a pretty tough pot. Maybe a linear motion pot would survive better, but again more of a Rube Goldberg methodology. A digital pot would provide most of the same stuff without all the stress, but without all the fun, too.

I know it's easy to implement all this with a micro controller, but you could also do the same thing with analog electronics. My only concern is that since you are sending a PWM signal to the welder, there is a possibility that if the micro controller locks up from HF or other issues, you could end up with your pulse stuck on it's peak, and any software safeties you have will be moot. One way to prevent that is to take something like an LM317 and set it to your max safe voltage, then switch that with PWM to get the lower voltage. After working with a lot of CNC and seeing digital glitches translated into real world disasters, I've become something of a belt and braces kind of guy with regards to digital machine controls.
Are you adding adjustable up and down slope with start and end amps?
Seems like you might as well. I guess you could even add a spot timer, so you would end up with a 185EX.

[Jason]

I have considered taking some time to work out a wireless pedal. This would make it much more simpler.. Nowto justify voiding a warrenty and finding the time to actually do it..

[Rambozo]

I have considered taking some time to work out a wireless pedal. This would make it much more simpler.. Nowto justify voiding a warrenty and finding the time to actually do it..

A high frequency welding arc is a pretty effective radio jammer. Combine that with open brush motors on grinders and things and your typical shop environment is not very friendly to low power RF. Miller worked on that for a long time and took out a number of patents to bring their wireless pedal to market, and Lincoln has yet to match it. I wonder if anyone has measured the frequency that Miller used? You might look into the FCC documents for it as well.

[Jason]

A high frequency welding arc is a pretty effective radio jammer. Combine that with open brush motors on grinders and things and your typical shop environment is not very friendly to low power RF. Miller worked on that for a long time and took out a number of patents to bring their wireless pedal to market, and Lincoln has yet to match it. I wonder if anyone has measured the frequency that Miller used? You might look into the FCC documents for it as well.

I remember the talk about Miller having complications and patents and I know there are regulations around what is ok to do. It wouldnt be anything that I would be bringing to market but more something that I could play around with and personal use. Hard to justify the warrenty with something like that but when the warrenty runs out, that is one less thing to worry about.

[joshuab]

Hard to imagine how the warranty would be involved: a receiver that plugs into the pedal receptacle,and another that attaches to or is integrated into the pedal. Just don't put high voltage on the lines and you should be all right, no? Maybe you'd void the warranty on the pedal, but if this kind of project is up your alley, there's nothing in the pedal you couldn't replace for $5 and a trip to Radio Shack.

[joshuab]

If anybody out there has a Miller wireless pedal and can get me a photo of the plate with the FCC ID on it, I can look up what frequency it uses, just for fun. Honestly, the frequency alone won't tell us much, because the modulation and encoding techniques are really where the magic happens--unless they're using pure analog modulation, but I would bet not. If you knew the frequency AND you had a signal analyzer, you could pretty quickly figure out what encoding/modulation they're using. Realistically, this stuff is probably all defined in patents anyway, so if you wanted to reverse-engineer it for personal use, that would be where to start.

EDIT: Never mind. I found the FCC ID on the web, although I'm having trouble looking it up in the database.

[joshuab]

I know it's easy to implement all this with a micro controller, but you could also do the same thing with analog electronics.

It's true, but I know almost zero about analog circuits, and I can already write code. So a code-heavy solution fits my skillset.

My only concern is that since you are sending a PWM signal to the welder, there is a possibility that if the micro controller locks up from HF or other issues, you could end up with your pulse stuck on it's peak, and any software safeties you have will be moot.

Interesting kink, that I hadn't considered. Just a few minutes ago, someone on another forum suggested that you could emulate a pot by using a fixed resistor hooked to a transformer, and then feeding the transformer a PWM signal. PWM duty cycle maps to pot position. I think that is the best solution so far, because it separates the Arduino's PWM output from the welder's control circuit, and the worst-case scenario is that the transformer fails somehow and the welder's control circuit goes over to one of its rails. Basically the same as if you floored the pedal, which isn't so bad. Given how simple this solution is, I'm sorry I wasted so much time on digital pots... this is the kind of thing I would already know if I had more experience with electronic circuit design.

Are you adding adjustable up and down slope with start and end amps? Seems like you might as well. I guess you could even add a spot timer, so you would end up with a 185EX.

The box will have adjustable max-out, upslope time, downslope time, pulse frequency, pulse balance, and pulse ratio. At least, that's what the design doc says! We'll see what happens when the rubber meets the road. Basically, I looked at the front panel of one of Everlast's TIG machines and took all the relevant controls. The 4T start/end amps are going to be fixed, although it would be relatively simple to add pots to control them later if I want. For now, I think I will probably be fine with setting a fixed 20 amps or something for the 4T start/end.

A spot timer would also be easy to implement. I had the idea that one could dial in a fixed output time, maybe even longer than a spot-weld, so that people who are practicing could work on consistency in travel speed. For example, dial in a 10-second arc time and then weld ten beads. Compare the travel speed to your desired travel speed, and also look for consistency in bead length between them. That kind of thing.

[Jason]

I believe it would void the warrenty as you have modified or custom components connected to the machine and things like you mentioned could happen. Someone with Everlast would have to confirm that but it would be my initial guess as to why it would void the warrenty. I am sure there could be problems that arrise that clearly have no relation to a modification like these but it is just easier to say they can't warrenty the machine when someone is using custom components to control or send voltage/signals into the machine.

[joshuab]

I believe it would void the warrenty as you have modified or custom components connected to the machine and things like you mentioned could happen. Someone with Everlast would have to confirm that but it would be my initial guess as to why it would void the warrenty. I am sure there could be problems that arrise that clearly have no relation to a modification like these but it is just easier to say they can't warrenty the machine when someone is using custom components to control or send voltage/signals into the machine.

IMO, Everlast seems pretty copacetic with not being hardasses about warranty issues just on principle, but it certainly is possible that, once a problem arose, they would change their tune. If you wanted to be absolutely sure that your warranty was in force, the best thing to do is not to modify your welder at all.

[everlastsupport]

Well, I can tell you we could never tell what broke the unit based on this mod. We would fix it, but you would have to cover the price of shipping.

On Miller, wireless, FCC, etc. Under a certain power range, there is no FCC. So I see nothing there. And if making one just for yourself, no one will bother you.

[joshuab]

Another update for y'all.

[Rambozo]

You might want to rethink what happens when you interrupt an up slope or down slope. While this technique is usually done in 2T it can also be done in 4T. One thing that is somewhat common is to vary the welding amps on the fly by going in and out of the slopes. So you might set your machine for 100 amps, but while welding you encounter a thin section where you require less power. So depending on the mode you are in, you tap or release the button to start the down slope, then when the power is at the level that is good you tap or hold the button to go back into up slope. By repeating this you can hold you power at a lower level to weld the thin section or fill a gap from poor fitup, then once past that section, you allow the machine to go back to the full power setting. It gives you something akin to a remote amperage control that is pretty easy to work with in out of position situations.

Of course the nice thing about making your own unit, is that you can have it do whatever you want it to do.

[joshuab]

Interesting, Rambozo. I'll keep that in mind.

[Ron66]

I don't know if Arduino has the option but all Atmel AVR parts have a watchdog timer built in. Your software needs to write the watchdog register before the timer expires or the micro resets. This is the normal way to protect from code lock up.

[joshuab]

I don't know if Arduino has the option but all Atmel AVR parts have a watchdog timer built in. Your software needs to write the watchdog register before the timer expires or the micro resets. This is the normal way to protect from code lock up.

The Arduino uses Atmel AVR chips, but I don't have to write the watchdog register manually in my code, so I'm betting that it is handled by the compiler in the background.

[joshuab]

I sat down tonight and started working on the code that manipulates the welder's output when actually welding. If pulse was ignored, this would be pretty straightforward, but with pulse in the picture, it gets a lot more complicated. Certainly, one way to handle it would be to write separate routines--one for welding with pulse and one for welding without--but that gets pretty complicated pretty fast. One for pulse in 2T; one for pulse in 4T; one for pulse with pedal control. Etc... I wanted to come up with a single routine that handled welding output, regardless of other parameters. It's absolutely possible to come up with a Grand Unified Algorithm of welder output, and to then create cases like plain old 2T by setting some parameters to zero and having them drop out.

I think I made a big leap towards cracking it with this information--the Rosetta Stone of coding pulse, at least for me.

[Rambozo]

Looks really good there. There are a lot of conditions for you to deal with. However, if you really want to spin your head around a few times I have a welder for you to emulate. It adds optional pulsing settings during up and down slope, up slope and main pulsing delay timers, post heat time, main weld time, auto stitch repeat timer, and some automated frequency tweaks that require a module that I don't have.

Have fun!

[joshuab]

Thanks... my head is spinning enough already. I just finished getting the code written, compiled, uploaded, and... of course it didn't work. But I'm going to call it a night and save debugging for tomorrow. At least I got the design part done. Also, I have a bad connection to one of my pots that I have to sort out.

Once I get the main welding algorithm debugged, which will include the pulse functionality, that will actually be a huge step towards completing the unit. I still need to do some tweaking on the user interface (lighting up LEDs when buttons are pressed); I still need to implement the multiplexer chip so that I can actually read all the inputs at once; and I would like to try out the "transistor plus fixed resistor equals digital pot" trick to see if it works. Short of wiring everything up and putting it in a box, those are the last main steps until I have something that I can call an alpha product. All in all, not bad for... what is it, a week's work?

I'm really psyched to profile my pulse algorithm to see what the lower limit on pulsing frequency would be. At some point, the code itself is the limitation. So far, all of my output involves using the serial interface for debugging, which introduces a lot of delay.