[Oscar]

So a member on WW piqued my curiosity as to what the "real world" thermal limits are on my Invertig 221. By "real world" I don't mean worst-case scenario, but rather what the machine would experience in an average shop.

This was rather impromptu, so I don't have video footage (yet), but the data I'm presenting is factual.

Ambient temperature:



Picture showing machine at thermal overload:



test piece to absorb a lot of heat:




Machine showed 217A on the display. Not sure why not 220A, never really payed attention to it, perhaps just production tolerances.

Time until thermal over-load/shut-down (minutes:seconds):

1st & 2nd done back-to-back, 3rd-5th back-to-back, meaning as soon as the thermal overload light turned off I went back at it at full power.

1st: 4:32
2nd: 3:38
3rd: 3:20
4th: 3:02
5th: 2:51


Those time durations for arc on-time aren't all that surprising seeing as how the machine is rated at 20% duty cycle @ 220A @ 104°F, and the ambient temp was well below that.

BUT...........

I initially "felt" like the cool-down time (until the thermal switch turned off and I was able to strike an arc again) was in the neighborhood of 5 or 6 min, but I wasn't paying too much attention to that part; I was more focused on how long the arc was on. But as I did more and more of those over-loads, I started to pay more and more attention to just how long it actually took to restore itself back to full operation. After the last thermal overload duration of 2:51, I let the stopwatch on my phone keep going to then subtract the time and see just how long it takes to recover so one can get back to welding. Shockingly, the thermal light turned off at 6:30! Subtract the 2:51 that the machine actually had the arc on, and that means that it only took 3:39 to cool down!

So if you go back to the 4th thermal overload, I had the arc on for 3:02, machine cooled down for about ~3:39, and then the 5th arc duration was 2:51 before it overloaded again; that's a total of 9:32. At 9:32 it was overloaded (and would have stayed that way for the next few minutes), so say 10min flat, and out of those 10min, the arc was on (albeit not consecutively) for 5:53. So almost 6 minutes it was arc-on during a 10min span.

Before I jump to any conclusions, I speculated what would happen the next 10min. It would have been off for about 3min (because it was already overloaded), been able to weld for about 2:45 at the current rate of decrease, then another ~3:40 to recover, then on for maybe 30sec before the 10min span is up again. So the 2nd 10min would be about 32% duty cycle. Average that out with the 1st 10min span, and one would be looking at an average of about a 45% duty cycle (over a 20 minute span) at an ambient temperature of 75°F. Not too bad.[/QUOTE]

[Poland308]

Lots of thermal overloads are bi-metal. They get weaker every time they trip.

Edit: side question. What was the input voltage. My experience is that when machines are pushed to the max they often t4ip sooner on lower input voltages. Also discrepancies in top end out put might be attributed to actual input voltage. IE 220v instead of 240v.

[Oscar]

Input voltage was 244V AC.

[BugHunter]

I've never done a "Test" like this, but some years back I did weld a 3" diameter alum shaft to a 1" thick plate with my Dyanasty 210DX. Even with some preheat, that was ridiculous. Stand on the pedal and hope. Now, they rate that machine higher duty cycle, mine is rated 175A @ 100% and 210A @70%. All I know is that was the longest I ever had a uninterrupted arc and it never said a thing. The fans were admittedly running faster. But it cooled off quickly and the fans went back to quiet so I figured it wasn't sweating a bit. I don't know if the extra noise was the water cooler spooling up or the welder itself, never really cared I guess. I'd have to guess I was approaching the duty cycle limit, but given it's supposed to have a little more headroom with 70% at full power, maybe I was ok. I know it was on a lonngg time. forced to guess, it had to be approaching 10 minutes, but as in your case, the ambient temp wasn't anywhere near the 105* where they qualify the D-C rating.

[Poland308]

Even all the antidotal evidence in this thread leans towards the actual overload trip being above the manufacturer tested trip point.

[Oscar]

Even all the antidotal evidence in this thread leans towards the actual overload trip being above the manufacturer tested trip point.

Actually I would disagree to a certain extent. It is above the "rated point" because of the ambient temperature of my test. A lot of people think the "20% duty cycle @ 220A" is a limiting factor, but in reality where it's not always 104°F, it leads to alot better performance.

[Spartan]

Relative humidity levels are significantly more important than air temperature for fan cooled systems. Dry air at 70 degrees F will cool way less efficiently than wet air at 100 degrees F.

[BugHunter]

Relative humidity levels are significantly more important than air temperature for fan cooled systems. Dry air at 70 degrees F will cool way less efficiently than wet air at 100 degrees F.

I'm not totally convinced with this... Correct me if I'm wrong... Yes, I know, I did a bunch of google searches and all sorts of results come up concerning home heating systems and what we "Feel". That ins't the situation with a welder. There's no skin sweat to dry and cause cooling (evaporation causes cooling) and the idea of water in the air is unintuitive. You have to really search with good search terms to find someone actually discuss "Cooling" physics without all the touchy-feely stuff about how we perceive heat.

Humid air doesn't have water in it, it has water vapor in it. Ie: Water in a gaseous state, already evaporated. Humid air is LESS dense than dry air. Hence, I'd say the thermal mass per unit of volume would be greater, and the cooling capacity of dry air greater than that of humid air. I've never had a car overheat in dry daytime conditions, but I surely have had that happen when the dewpoint and temp coincide at night and keep in mind, that's after the temp has actually fallen some.

I'll be happy to be corrected, but I'm pretty sure you want dry air for cooling...

[VA-Sawyer]

I think the important thing to remember here, is that heat is detrimental to solid state electronics. The hotter you run an electronic device, the shorter its expected lifespan. Better heatsinks can extend durability, but they cost more money, so it becomes a tradeoff for manufacturers. The same is true of duty cycle ratings on welders. They could post a higher duty cycle to increase sales, but that will decrease the reliability, and increase warranty claims. Another tradeoff.
For maximum durability blow the dust out of your welders on a regular basis. Keep them cool. Buy enough capacity that you aren't constantly pushing the limits.
It is like tires.... overload them, and they won't last as long. Doing tire spinning starts won't make them go flat today, or even tomorrow, but each one gets you closer to needing new tires. Same with tripping the thermal limits on a welder. It may not die now, but that day just got a little closer.
Bottom line..... don't abuse your welder, if you want it to last.

[Oscar]

Bottom line..... don't abuse your welder, if you want it to last.

Except in the name of science!

[VA-Sawyer]

Oscar,

I understood that you was doing it as research. I also believe that you are knowledgeable enough, to understand it could hasten the welders last day. We sometimes make sacrifices to gain knowledge.

[sbaker56]

The overload "should" kick in before the components are at their temperature limit, regardless of if it's designed to trip when it reaches a temperature that would shorten the lifespan if sustained on a daily basis or if it only kicks in at the critical point, a few isolated incidences shouldn't be a big deal.

There is definitely a discrepancy between given duty cycle and when most machines will trip the thermal overload. I've noticed it myself and I've heard countless people say the same, I'm just curious if it's due to ambient temperature or if the thermal overload is actually set far beyond the given duty cycle.

I'm curious though, did it feel like it was hard to hit the cut off or could you see yourself hitting it fairly easily doing actual welding with the machine maxed out?

[Arno]

I'm just curious if it's due to ambient temperature or if the thermal overload is actually set far beyond the given duty cycle.

AFAIK the standard test performed for the duty cycle by the manufacturers is at a very high ambient temperature.

Can't find it at the moment, but it was something like 45C/113F or more.

Think of it like a 'worst case scenario' test, so in many (most?) cases the machines will perform better/longer in regular conditions, or much better when for instance used in cold climates in winter when it can shed the heat faster.

At least on 'brand' machines you can fairly safely trust that it was tested and desgined that way. On no-name/cheap machines the numbers may be more 'wishful thinking'

Bye, Arno.

[BugHunter]

I'll address 2 things below.

The overload "should" kick in before the components are at their temperature limit, regardless of if it's designed to trip when it reaches a temperature that would shorten the lifespan if sustained on a daily basis or if it only kicks in at the critical point, a few isolated incidences shouldn't be a big deal.

The datasheet for most any power handling solid state device will have a myriad of graphs dealing with temperature derating, meaning reducing safe output levels depending on the device temp. They will sometimes even have info about how many times you can over-stress it before it is considered compromised. Each time you take a device past the design limits, the junctions break down and it's no longer as 'young' as it used to be. It'll carry less and less power till finally it dies.

When a mfgr puts a thermal overload temp sensor in, it's generally not looking at individual devices, but simply taking a temp near the action somewhere. So, it's not perfect, and still not so even if it is on the device. Maybe they can put a sensor on each major component, but I'd be surprised to see that. Still, if you ever worked with a PC that had a CPU overheat, it may have 'functioned' yet, but it's easy to see its days are numbered. They do all sorts of crazy stuff afterward.

There is definitely a discrepancy between given duty cycle and when most machines will trip the thermal overload. I've noticed it myself and I've heard countless people say the same, I'm just curious if it's due to ambient temperature or if the thermal overload is actually set far beyond the given duty cycle.

Usually they simply watch the thermister and when it says it's over temp, the machine shuts off. They don't use a timer and gauge it against output amps (though they easily could), because there's more to it than just time and amps. Mostly, ambient temp.

[Poland308]

I feel this discussion is akin to one I had with some colleagues about mechanical equipment recently. This experiment is a gain in knowledge. But there comes a calculation from knowledge that requires a person to decide on reliability compared to necessity. If your machine (whatever it is) is important then you calculate the difference between the value of the operating machine compared to the $ loss of a non operating machine.

[sbaker56]

I'll address 2 things below.

The overload "should" kick in before the components are at their temperature limit, regardless of if it's designed to trip when it reaches a temperature that would shorten the lifespan if sustained on a daily basis or if it only kicks in at the critical point, a few isolated incidences shouldn't be a big deal.

The datasheet for most any power handling solid state device will have a myriad of graphs dealing with temperature derating, meaning reducing safe output levels depending on the device temp. They will sometimes even have info about how many times you can over-stress it before it is considered compromised. Each time you take a device past the design limits, the junctions break down and it's no longer as 'young' as it used to be. It'll carry less and less power till finally it dies.

When a mfgr puts a thermal overload temp sensor in, it's generally not looking at individual devices, but simply taking a temp near the action somewhere. So, it's not perfect, and still not so even if it is on the device. Maybe they can put a sensor on each major component, but I'd be surprised to see that. Still, if you ever worked with a PC that had a CPU overheat, it may have 'functioned' yet, but it's easy to see its days are numbered. They do all sorts of crazy stuff afterward.

There is definitely a discrepancy between given duty cycle and when most machines will trip the thermal overload. I've noticed it myself and I've heard countless people say the same, I'm just curious if it's due to ambient temperature or if the thermal overload is actually set far beyond the given duty cycle.

Usually they simply watch the thermister and when it says it's over temp, the machine shuts off. They don't use a timer and gauge it against output amps (though they easily could), because there's more to it than just time and amps. Mostly, ambient temp.

In my experience I've generally seen PC components come out fine after momentarily hitting throttling temperature, provided the issue is immediately identified and fixed. But in every computer I've ever built I've generally set up the cooling and when possible had software throttle back the GPU and or CPU at least 10C below the hardware limit for components. Which is basically what I was curious about. Did they set it with a safety cushion or no. But like you said it's not a perfect system and I would imagine ambient temperature may affect it's accuracy substantially.