Manipulating welder output with Arduino

[joshuab]

A lot of digital logic chips are tri-state designs, so good practice is to use appropriate pull-up or pull-down resistors to control the floated state.

Ugh. What a PITA to have to install resistors on all the un-used pins. I'm inclined to trash the 4066 altogether and just go with an NO relay with built-in opto-isolator. It's really the right thing to use. I'm just using the 4066 because I had it on hand.

Meh. I guess I could just install one resistor and bridge all the pins through it to ground.

[joshuab]

Here are two more update videos. I scrambled to get the project into a state where I could demo it at the welding workshop I'm going to tomorrow. It's still got more work to do, but it's looking really good!

[joshuab]

[Rambozo]

2T also can also use up and down slope, just how the button works is different. Press and hold vs. press and release. Another feature might be to add a jack for the pedal so you can switch between them without having to swap cables around.

[everlastsupport]

Ugh. What a PITA to have to install resistors on all the un-used pins. I'm inclined to trash the 4066 altogether and just go with an NO relay with built-in opto-isolator. It's really the right thing to use. I'm just using the 4066 because I had it on hand.

Meh. I guess I could just install one resistor and bridge all the pins through it to ground.

Why not use jfet or mosfet to drive them? If you take them all to ground via one R any one pin going high will pull them all. They make a DIP with resistors that bridge across, like pin 1 to 16, pin 7-8. You can use those. 4066 is a CMOS series chip (analog switch if I recall) and static, and maybe HF will be a problem down the road (maybe).

[Jason]

This thread is very interesting. Also to me, the bug fix was not boring

[Rambozo]

Something I was just reminded of was that instead of using analog inputs and multiplexing them, you could go with digital inputs and use rotary encoders instead. Cheaper than pots and much more flexible as you can emulate multi turn devices, and do things like acceleration sensitivity. Just in case you feel like writing more software. The downside would be that you would need some kind of display as knob position wouldn't work for you.

[joshuab]

Why not use jfet or mosfet to drive them?

Because I bought a MOSFET to try just that, but I accidentally bought a power MOSFET that requires 10v to activate the gate, and the Arduino only puts out 5v max. I also bought some relays to try, but I accidentally bought NC instead of NO. I still have a lot to learn about using the Mouser search engine, and reading the data sheet of ICs before I buy them.

If you take them all to ground via one R any one pin going high will pull them all.

Sure, but since these are all unused pins, I think that doesn't really matter.

[joshuab]

Something I was just reminded of was that instead of using analog inputs and multiplexing them, you could go with digital inputs and use rotary encoders instead.

True, but the mux is actually not much of a restriction in terms of use, as long as I don't exceed 8 inputs. Once I exceed 8 inputs, I have to build an array of muxes, controlled by another mux, and that's just silly. Or maybe I could find a 16-channel mux. I dunno. For now, I have enough pins and am doing fine.

Cheaper than pots and much more flexible as you can emulate multi turn devices, and do things like acceleration sensitivity. Just in case you feel like writing more software. The downside would be that you would need some kind of display as knob position wouldn't work for you.

Ugh. No thanks on the display. I have enough work to do without having to write a screen interface for this project on top of that!

[Rambozo]

Fair enough, but you really should add the up and down slope to the 2T. That is very handy and used all the time.

[joshuab]

Fair enough, but you really should add the up and down slope to the 2T. That is very handy and used all the time.

I'm so confused. I need to go back and look up what the difference between 2T and 4T really is. Or maybe just label that button something else and forget about it ;-)

[Rambozo]

I'm so confused. I need to go back and look up what the difference between 2T and 4T really is. Or maybe just label that button something else and forget about it ;-)

I started to draw a diagram, but of course there is a good one for both 2T and 4T on THIS page. Scroll down a bit for the diagrams. Jodi also explains the technique I mentioned before to go in and out of the slopes to change rods or reduce current. Something that works really great in 2T. It also covers where the gas solenoid is triggered. Something to keep in mind for a future mod project.

Two points for using 4T is being able to hold the arc at the start amp value to get positioned and ready to weld, and not having to hold the button down for a long weld. I can't say I've used the 4T function much at all, but it is pretty standard for any automatic welding systems. Maybe others can give better examples of when it works good for manual welding.

[joshuab]

Maybe I should just rename my "2T" function "spot mode" or something. I certainly think there's a lot of value in a mode that is instant-on, full-output. This is purely a consequence of my lack of experience with TIG welding, but I don't really see the advantage of 2T mode (real 2T mode, not my mistaken believe that 2T = spot mode) over 4T. With 2T, you have no control over when the upslope starts, and you have to hold the button down the whole time you're welding. It seems like you could get exactly the same experience out of 4T simply by pressing the switch and then immediately releasing it. No?

[everlastsupport]

Because I bought a MOSFET to try just that, but I accidentally bought a power MOSFET that requires 10v to activate the gate, and the Arduino only puts out 5v max. I also bought some relays to try, but I accidentally bought NC instead of NO. I still have a lot to learn about using the Mouser search engine, and reading the data sheet of ICs before I buy them.

My son just went up to Tenn. Thursday with his girl friend, he will be coming home today. Not sure where they went, but would have been funny if he could have brought you come 5vdc NO relays. I have tubes of them. Old dip style.

[Rambozo]

Maybe I should just rename my "2T" function "spot mode" or something. I certainly think there's a lot of value in a mode that is instant-on, full-output. This is purely a consequence of my lack of experience with TIG welding, but I don't really see the advantage of 2T mode (real 2T mode, not my mistaken believe that 2T = spot mode) over 4T. With 2T, you have no control over when the upslope starts, and you have to hold the button down the whole time you're welding. It seems like you could get exactly the same experience out of 4T simply by pressing the switch and then immediately releasing it. No?

Yes you can get that effect by just tapping the switch. Just like you can get your version in 2T by just turning down the slope controls to 0. The big thing for me in 2T is being able to release and press the switch with a slightly long slope to let you run a portion of the weld at a lower amp setting. I guess it would be possible in 4T but lots of switch pressing and I'm sure I would lose track of where I am in the cycle. In 2T with short slopes I find I will press the switch 1 to 3 times per second to lower the amps for a short distance. With up = press and down = release it's easy to keep track of what you are doing. It's harder to explain than to just try it and see. But basically you release the switch and wait for the amps to slope down to the level you want. Once there you start cycling the switch to hold it at that level. Then when you want full power again, just stop cycling and hold the switch down again, and you will slope up to full power again. How quickly you cycle the switch depends on the slope times you have set.
Again I'm not a big user of 4T, so maybe someone else can chime in on what it's great for. Maybe if I welded hours on end with a torch switch my finger would get tired. I use a pedal more than anything, whenever possible.

[joshuab]

[joshuab]

Well, folks, I am ready to call it an alpha build. I got the last of the major bugs worked out--at least the ones I know of--and gave it a thorough shake-down. The only thing left is to actually weld with it, which is for tomorrow.



The video is a little long because I did a full review of all features, as well as some background for those who may not have been watching every one of the videos until now.

[joshuab]

The attached video is the first time I've demonstrated actually welding with The Box. It demonstrates pulsing, since that's the most visually interesting thing that The Box does.



As you would expect, moving to a new frontier of testing (actually welding) revealed a new frontier of kinks to work out. The first kink was that, the whole time I have been developing, I have been using 2.7 volts as the reference for the welder's pedal circuit. In other words, the welder feeds 2.7 volts to the pedal, and the pedal shifts that around using a potentiometer, and the welder reads the resulting voltage on the "wiper" pin (from 0-2.7 volts) to determine what output it should use. All fine and good. Because I only recently learned how to build a proper digital version of that circuit, I have been approximating it by simply driving the wiper pin with a signal from 0-2.7 volts, generated directly from the Arduino.

2.7 volts is a little unusual, since most signaling circuits are at either 3.3 volts or 5 volts, and today I found out why. I had been thinking of it like this: welder measures voltage on the wiper pin and takes that as a percent of 2.7 volts, then maps that to a percent of max-output, where max-output is 96 amps on 110v power and 160 amps on 220v power. Nope. That is not how it works. The way it appears to actually work is that the welder takes the incoming voltage on the wiper pin as a percent of 5 volts (normal signaling voltage--ha ha!) and maps that to a percent of output, where output is 0-160 amps. When the welder is on 110v power, it reduces the signal it feeds to the pedal from 5v to 2.7v (probably nominally 2.5 volts), thereby reducing the welder's output range. I never noticed this because this is the first time I have taken the welder outside and plugged it into 220v power. So as a result, The Box only ever allows the welder to output up to 96 amps, even when the welder is on 220v power. Ha ha.

One fix for this issue would be to read the voltage that the welder is sending to the circuit, and then use that to determine whether the welder is on 110v or 220v power, then adjust The Box's output accordingly. Unfortunately, this would require another analog input. The right way to fix the issue is to build a proper digital potentiometer circuit, which a helpful soul on another forum has given me a schematic for. I have already ordered the parts and should have them by next week.

Issue number two: Because the Arduino doesn't have a proper digital-to-analog converter, it uses pulse-width-modulation to generate analog voltages. Basically, it switches a 5v signal on and off very fast, and the average on/off time determines the resulting average voltage that is seen by other devices. The base frequency of the PWM signal is adjustable, and is currently set to around 3 kHz. The welder seems to have no problem "averaging" out the PWM signal and setting the desired output level, which is how PWM is supposed to work. But when welding, there is now a "singing" tone that can only be the PWM frequency getting through. I compared the tone in the video to a 4 kHz test tone, and they are dead on. I doubt this is going to hurt the welder, but just in case, I only ran it for a minute or two, and won't be doing it again until I can properly smooth the PWM signal. Even if it's harmless, it's annoying to listen to, and it's just not good practice to allow it to happen.

[joshuab]

For reference:

[sportbike]

Good progress.

I'd throw an RC filter on the PWM output to stabilize it. Put the output frequency to pretty much as high as you can and there will be very little ripple in the DC output.