Welding -- what to do about HUGE 2" arcs

Sure, a microcontroller could do this job fairly easily. You need something with 2 A/D inputs, 2 D/A outputs (12 bits would be adequate), a bit of digital I/O for the front panel interface. I'm sure you could find something suitable from one of the many single board computer manufacturers. Or use an old laptop computer with an appropriate analog I/O module such as made by Measurement Computing Inc, National Instruments, etc - most of these have a few digital I/O thrown in, and a laptop computer would make a pretty good front panel. The job is made easier since you do not need to make any "front panel" adjustments while you are welding. You would probably want the control routine to be driven by a timed non-maskable interrupt; check voltage and current from A/D, compute new controller voltage and current limit setpoints, output new values to D/A, *toggle the watch-dog timer output*, return. Where an external hardware watchdog timer will clamp both set-points to zero if the control loop stops. You probably could get away with 100 updates/sec or so.

But this does not let you off the hook for some analog circuit design. You need to get those voltage and current signals properly scaled, level shifted and protected before you can connect then to an A/D converter. This job is made a tad more difficult by the fact that the welder output is completely isolated from ground except for the arbitrary connection of either lead to a grounded workpiece, while your controller must be grounded. By the time you have filtered, protected and scaled these signals to a ground-referenced 0 to 4.096 volt (for example) signal, you are about 90% of the way to a simple analog computer, which would be easier to make extremely reliable in an electrically noisy environment. (A laptop or anything with a HDD would be the least reliable IMO.)

What is the required signal range for your controller voltage and current limit setpoints, and what is the reference for these signals? If the reference for these signals is not connected to ground you have additional isolation concerns.

Glen, thanks. I have decided to settle down on one of the Comfile controllers. It is programmable both in BASIC as well as ladder logic. It can also have serial or ethernet interface. What this hopefully means is that I could set my laptop on top of the welder and use it for debugging of the BASIC programs, or send commands (though, surely, the laptop will not be part of the final product).

Regarding isolation: right now, perhaps wrongly, I have negative side grounded. That makes some things a little simplier for me. PCTI power supply is designed to work this way, I think. To change that (unground negative side), I would need to switch to using current and voltage transducers. I already have a voltage transducer, but no current transducer yet.

i

The grounded negative side is a safety issue. You will never see this done in any commercial welder. Standard safety practices include connecting the welder frame to safety ground, and connecting all metal workbenches used for welding to safety ground. If you want to use straight polarity (electrode negative) or AC (electrode sometimes negative) then you have an immediate problem when you connect the work lead to a grounded workpiece, as full welder output current will travel through the safety ground system back to the grounded negative bus, likely starting a fire and certainly melting the safety ground wiring unless it is sized way above code requirements. Even if you only ever weld DCEP you would immediatly have a problem if the work connection falls off while you are welding, with welding current again returning through safety ground. You should definitely fix this, and in the meantime make sure nothing you weld on is grounded, and treat everything you weld on as electrically hot while you are welding.

Thank you. To resolve this, I likely need to install current and voltage transducers, to "untie" myself from the ground. Right?

i

Correct, you need isolation for both measured variables. If the PCTI controller current limit and voltage limit set points signals are also referenced to the negative bus, then you need to "untie" or isolate these control signals from ground also. You could either buy isolated transducers, or build your own.

they are.

Thank you. I will definitely do it. Tying negative bus to the ground was a temporary measure, just to get things going. Instead of using transducers, can I isolate the signals that I have, using a optocoupler or some such?

i

An optocoupler by itself will not provode an accurate transfer of your signals across the isolation barrier, although it can be done with voltage to frequency -> optocoupler -> frequency to voltage for example. You would be better off buying an isolation amplifier such as the TI ISO124:

There are lots of other iso amps but this is the cheapest suitable part I know of. An eBay deal on a relatively expensive 3-port iso amp might be even better since it transfers power across the isolation gap too; a plain iso amp like the 124 needs separate isolated power on both sides.

Got it Glen. So, the item I should look for is called "isolation amplifier", right? I will also call LEM USA tomorrow. I am wrapping up voltage switch contactor work today.

i

Glen, that item is not available anywhere, earliest it ships is August. Do you think that AD629 would be suitable? I have one right at home:

the PDF for it even shows a sample connection to a current shunt.

i

...

Yes, that looks like it should work fine. Analog Devices also makes isolation amps which are actually available now, but they are relatively pricey, especially compared to parts you already have.

That's great Glen. I will look for it. Should be easy to do, I think.

i