This is a generic skill. How I acquired it is recounted following.
I have a calibrated "clamp-meter" / "tongues-tester" which serves most on-site weld diagnosis.
In most other writings, there was a specific objective within the wider world, and the story is of overcoming obstacles - technical and human - on the path to achieving that objective.
This was essentially a "just for fun" activity (!).
The endeavour was in 2011, while doing my Welding Engineering qualification at a University.
I haven't since needed to use "more" than a calibrated meter.
I helped a team use a
"weld monitor"
to measure and record the average Amps and Volts of a welding process.
(web-search if you want more information on "weld monitor" - also
sometimes called "arc logging equipment")
I realised the equipment could be the starting stage of a much more extensive analysis of welding processes, and tried whether that could be so. That did prove to be so!
The "weld-monitor" for its own purposes internally stored a numeric
text data-table of measurements on its internal storage device.
I could get at that and transfer a copy of it onto a general-purpose
computer, to post-process as I liked...
The "fields" (columns) of the data-table were the Amps, Volts and wire
feed speed at each sampled instant in time.
The "rows" (lines) of the data-table were those conditions for each
sampled instant of time - which was at the rate you chose - thousands
or tens of thousands of equally-spaced intervals per second.
Consequently, in monitoring a welding activity, the resultant
data-table had millions of lines of data.
Which I could freely post-process on a "unix" computer, using the "unix user toolkit". Expecially "awk" (as mentioned previously - see section "Post-processing the data-table").
So, I had "free use" of a capability to analyse welding processes in about as much detail as one could want to :-)
I'd extensively used MIG / GMAW as a Tradesperson welder-fabricator working in steel fabrications Companies ("FabCo's"), welding architectural steelwork. Now I could inspect the welding conditions I'd used for years in this harsh workshop world!
Using MIG / GMAW the vital choice is whether you are applying "dip
transfer" (low-energy) or "spray transfer" (high-energy).
You do this by qualitative judgement, "navigating" your way to the
welding condition you are looking for in matching what you are
experiencing in test-welds on offcuts to a "model" in your mind of how
the welding process works. A "model" of the dynamics of the arc and
the response of the welding power-source to the conditions, plus
heat-flow, surface-tension of the weld-pool, etc.
The "model" for dip-transfer is totally different to the "model" for
spray-transfer.
The quite emotional moment - given the human "backstory" to my career progression - was that it looked like I would imminently be inspecting whether those "models" as I visualised them were correct...
Broadly, according to those visualised "models", one expected that in:
The big moment - what do we actually see???
See dip-transfer datalogged and analysed .
There it is, "laid bare" to be inspected!
The "model" in my mind of dip-transfer appears to be exactly what is
actually happening.
There is the cyclical event, with the Amps and Volts each going
through their physically-linked but different cycles.
See spray-transfer datalogged and analysed .
That's it - "flat as a billiard-table" constant conditions of Amps and Volts, as visualised.
I was a little surprised that there was no roughness on the V and I
graphs. Given you "squeeze" the arc-cone to get a small concentrated
directional heat-source by very slowly trimming down the Voltage (or
increasing the wire feed speed - which is continuously adjustable even
on copper-and-iron purely electrical pre-electronics MIG / GMAW
welding machines), until there is just the onset of a ripping sound
suggesting some contact between wire and weld-pool.
The effect of that is not seen on the V - I graphs. Seems the
welding power-source is "big enough" to take that without detectable
change, if it's happening at all.
My "spray-transfer" "model" most definitely endures given the
evidence.
Sense of fun - let'd "pick-apart" a proprietary commercial pulse-mode which is supposed to be a "trade secret" :-)
Piling-up a stack of offcuts, donning my welding gear and attaching a manual welding torch to this top-of-the-range machine intended for production-line robotic welding, I got to it...
On post-processing, this is what I found: Lincoln "RapidArc" datalogged and analysed .
I was somewhat mystified. What was happening was very clear - but why? What purpose did it serve???
I got onto Lincoln Electric by phone, and emailed my graphs produced
from the datalogged sampling as attached image files.
Seeing the graphs, the person I was speaking with paused for a moment,
then made some policy decision and became discursive and forthcoming.
He explained that the purpose served by the low-voltage event in the
cycle is sensing accurately where the weld-pool surface is, so a very
short arc can be maintained - applying "tricks" in the other parts of
the cycle.
The outcome being that the welding process is stable ("clean") at very
high travel-speeds (run-rates).
I was pointed to Lincoln's brochures for "RapidArc" (external links);
one brochure
;
another brochure
(links correct at time writing, 2017).
Dashing back into the welding workshop, I got hold of the manual torch and tested that claim. Exactly so - you could sweep the torch along the joint at "absurdly" high travel-speeds and a smooth (both longitudinally (lengthwise) and transversely (cross-section)) properly-formed weld-bead resulted.
Advantage of testing conditions with a manual torch ;-)
I was "slashing" along the joint at rapidly-increasing speeds in each
run, "pushing" to find operable limits.
Full credit to Lincoln Electric - the advantages they claimed seemed "real".
This proved to be a good-fun exercise, what with the interactions and the "playing-around"!
(R. Smith, 27Jun2017)