Page 1 of 1

Digitizing a welding waveform

Posted: Sat Feb 24, 2018 6:58 pm
by Tigger17
Hi guys! I need some help. I'm doing some research on quantifying welding waveforms (really--the current/amperage/frequency output of the machine). I'm trying to figure out the best way to "digitize" that waveform. I'm thinking shunts and oscilloscopes. I'm using TIG machines--millers, specifically (280/Mark 8s), and maybe a Thermal Arc Fabricator 8i. Does anyone have any ideas on how to best translate the wave from the machine to something I can capture? Thanks so much! :)

Tay (Tigger)

Re: Digitizing a welding waveform

Posted: Sun Feb 25, 2018 3:15 pm
by Louie1961
go check out the "this old Tony" youtube page. has done this a number of times. Maybe Ave as well

https://www.youtube.com/user/featony

https://www.youtube.com/user/arduinoversusevil

Re: Digitizing a welding waveform

Posted: Sun Feb 25, 2018 5:02 pm
by Poland308
Not that I have any idea how to go about what your doing. Curious though, are you studying it as a fixed power source, under load, or while it maintains an arc?

Re: Digitizing a welding waveform

Posted: Mon Feb 26, 2018 3:45 am
by Arno
Tigger17 wrote:I'm trying to figure out the best way to "digitize" that waveform. I'm thinking shunts and oscilloscopes.
Digital scope with a bunch of inputs or data aquisition interface board with a PC hooked up (usually USB these days) and scope software for easier data gathering. The USB 'scope' modules can be quite cheap on Ebay and then like and if you don't need extreme accuracy they can be a good method to get a big amount of measurement inputs (if your PC is also fast enough of course :) )

Then use some 'amp clamp' style scope probes for current measurement for the least (aka. pretyty much 0) influence on the current measurements. Shunts and the like to correlate volateg drop to amps do of course introduce some extra resistance in the system. It's low, but still there. Could be another interesting test for further study to see if measured values differ when testing using an amp-clamp probe or when using a voltage drop shunt.

You do need to pay attention to the masurement range these amp-clamps can handle (often a number of specific amp ranges with differing accuracies/granularity) and frequencies it can support so determine the kind of range(s) you're interested in and research some suitable probes.

May as well get most of the 'additional' data logged at the same time, so some extra scope inputs that can also directly measure the voltage (may give interesting correlations with amps..) and perhaps also some data aquisition on environmentals with some thermocouples on air intake and exhaust of welding machines and perhaps some stuck onto heatsinks.

Can never have too much raw data.. :lol:

Depending on what your are testing for or researching it may be best to perform various tests in a single testing session/day under at least somewhat controlled/same conditions and gather up as much data as you can in one go.

Usually then the biggest amount of work needs to be done to turn the raw measurements and data into something useful/meaningful and depending on your objectives and the initial results, adjust your parameters/goals and re-test.

Bye, Arno.

Re: Digitizing a welding waveform

Posted: Tue Feb 27, 2018 4:50 pm
by Tigger17
Poland308 wrote:Not that I have any idea how to go about what your doing. Curious though, are you studying it as a fixed power source, under load, or while it maintains an arc?

I'm studying it to determine what the "baseline" looks like, so while it maintains an arc. Also, we're attempting to measure what a "good" welding waveform looks like, so that we can optimize the waveforms to give a new outlook on how math can predict what is going wrong with a weldment. The idea is to be able to quantify the waveforms mathematically into good baselines versus ugly defects (such as measuring the defects of dipping the tungsten, long-arcing, etcetera). This way we can figure out how our machine vs. operator is working out at a more optimal level! :)

Re: Digitizing a welding waveform

Posted: Tue Feb 27, 2018 4:52 pm
by Tigger17
Also, thank you, Louie1961! These are helpful; we already know how to set up oscilloscopes, though, which is great; we're just wanting to find a way to optimize how much raw data we can get! haha We're wanting as many data points as we can generate, which requires (as I'm told) the rate of output from the machine to equal the rate gathered by the oscilloscope/shunt pairing.

Re: Digitizing a welding waveform

Posted: Tue Feb 27, 2018 4:56 pm
by Tigger17
Arno wrote: Can never have too much raw data.. :lol:

Depending on what your are testing for or researching it may be best to perform various tests in a single testing session/day under at least somewhat controlled/same conditions and gather up as much data as you can in one go.

Usually then the biggest amount of work needs to be done to turn the raw measurements and data into something useful/meaningful and depending on your objectives and the initial results, adjust your parameters/goals and re-test.

Bye, Arno.
Hi, Arno! Thanks so much; I definitely need prime real estate when it comes to accuracy though. That's true--resistance adds bit of an error percentage that we'll need to calculate, because when anything translates, there's something left behind. I think what you said is a cool twist--a further study on measuring instrumentation...

Yes, you're exactly right for keeping in mind the amp clamp range--we're using Mark 8s and CST 280 Millers at the moment for Tig applications. I think sticking some probes on heatsinks and getting thermal data would be neat--any ideas how?

Truth on the raw data front! lol And, yes, I'm definitely going to be doing a lot of "meaningful transferring"!

Re: Digitizing a welding waveform

Posted: Tue Feb 27, 2018 5:28 pm
by MinnesotaDave
Tigger17 wrote: I'm studying it to determine what the "baseline" looks like, so while it maintains an arc. Also, we're attempting to measure what a "good" welding waveform looks like, so that we can optimize the waveforms to give a new outlook on how math can predict what is going wrong with a weldment. The idea is to be able to quantify the waveforms mathematically into good baselines versus ugly defects (such as measuring the defects of dipping the tungsten, long-arcing, etcetera). This way we can figure out how our machine vs. operator is working out at a more optimal level! :)
The long-arc is easy to spot on a transformer machine. The voltage increases.

When I teach a person to TIG with a machine that has digital readout. I record the voltage I weld at with proper arc length for the weld. Then I coach them on arc length based on their voltage readout while they weld.

It's faster than trying to see their arc length all the time. Seems you are heading in this direction with your testing?

Re: Digitizing a welding waveform

Posted: Tue Feb 27, 2018 6:07 pm
by Tigger17
MinnesotaDave wrote:
Tigger17 wrote: The long-arc is easy to spot on a transformer machine. The voltage increases.

When I teach a person to TIG with a machine that has digital readout. I record the voltage I weld at with proper arc length for the weld. Then I coach them on arc length based on their voltage readout while they weld.

It's faster than trying to see their arc length all the time. Seems you are heading in this direction with your testing?
Hi, Dave! Thanks for the tips; yes, I've definitely seen how long-arcing and defects are easy to spot with voltage jumps! The research I'm getting into is moreso the whole waveform, which incorporates voltage readout as well as amperage/frequency/current/etcetera. These variables are definitely linked to arc length, etcetera, as we know as welders, but for this project specifically, I want to get into optimizing those variables because as I've spoken with Miller/Lincoln/Welding Engineers, they don't optimize their waveforms using the math I'm getting into yet. They use sensors to figure out what suits a certain material/process best...just like we do (trial and error with the knobs, etc.)! I'm trying to figure out a way that we can mathematically say, "This is the best waveform, and here's what happens when it doesn't look this way," in order to help our future welders and manufacturers of those welders (and maybe even robotics welding, because Lord know robots can't intuitively "feel" a good weld like we do).

Plus, I know a bunch of my peers are always asking "how do I get these settings right?" This project is just another way to answer that, albeit I know it's a little complicated... :? But, who knows? It might amount to something good because I know Miller is really interested, which is good, right? :D

Re: Digitizing a welding waveform

Posted: Wed Feb 28, 2018 2:24 am
by Arno
Tigger17 wrote:I think sticking some probes on heatsinks and getting thermal data would be neat--any ideas how?
Thermocouples and some simple datalogging/sampling interfaces are your friend :)

A few quick examples:

https://www.omega.com/pptst/UTC-USB.html

https://www.picotech.com/data-logger/tc ... ata-logger

Should probably be more on Ebay. Some have their own memory and can datalog standalone which you download at a later date while others are intended more as a direct interface to a PC (or things like Raspberry-PI's and the like) via USB for direct sampling.

The thermocouples themselves are usually not that expensive, especially if you just need/want the common types without special protected probe tips. You can then clamp/stick probes on heatsinks and the like and datalog the info.

If you're not familiar with thermocouples.. One thing to be careful of is to not just extend cables with soldered copper wires. The seemingly 'clunky' plugs on the end are quite specific to this application as well as the material of the various conductors used as the interaction between metals across temperature gradiants is what the whole system is based on. A quick google should throw up a lot of info on that.

Bye, Arno.

Re: Digitizing a welding waveform

Posted: Wed Feb 28, 2018 7:32 am
by tungstendipper
MinnesotaDave wrote:
Tigger17 wrote: I'm studying it to determine what the "baseline" looks like, so while it maintains an arc. Also, we're attempting to measure what a "good" welding waveform looks like, so that we can optimize the waveforms to give a new outlook on how math can predict what is going wrong with a weldment. The idea is to be able to quantify the waveforms mathematically into good baselines versus ugly defects (such as measuring the defects of dipping the tungsten, long-arcing, etcetera). This way we can figure out how our machine vs. operator is working out at a more optimal level! :)
The long-arc is easy to spot on a transformer machine. The voltage increases.

When I teach a person to TIG with a machine that has digital readout. I record the voltage I weld at with proper arc length for the weld. Then I coach them on arc length based on their voltage readout while they weld.

It's faster than trying to see their arc length all the time. Seems you are heading in this direction with your testing?
I wish the digital read-out was coupled to a tone so you could "hear" the right voltage.

Re: Digitizing a welding waveform

Posted: Wed Feb 28, 2018 3:06 pm
by Tigger17
Arno wrote:
Thermocouples and some simple datalogging/sampling interfaces are your friend :)
Lol, right? That doesn't look too bad; I think that might be round 4 of our experiment thus far! Thanks, Arno!

Re: Digitizing a welding waveform

Posted: Wed Feb 28, 2018 3:08 pm
by Tigger17
tungstendipper wrote:
I wish the digital read-out was coupled to a tone so you could "hear" the right voltage.
Huh, that's a thought. We have a deaf student right now who's working on getting his certs in, and we're interested in the variety of ways people feel/hear/etcetera. Since he can't hear, he can feel the vibration of the current, and he says that he's getting used to feeling when it feels right. Really cool.

Re: Digitizing a welding waveform

Posted: Wed Feb 28, 2018 3:21 pm
by MinnesotaDave
Tigger17 wrote:
tungstendipper wrote:
I wish the digital read-out was coupled to a tone so you could "hear" the right voltage.
Huh, that's a thought. We have a deaf student right now who's working on getting his certs in, and we're interested in the variety of ways people feel/hear/etcetera. Since he can't hear, he can feel the vibration of the current, and he says that he's getting used to feeling when it feels right. Really cool.
It was really strange the first time I had to stick weld in a noisy environment with ear plugs in.

Couldn't hear the arc at all - felt like I didn't know what was going on with the weld. :shock: :)