kWeld - "Next level" DIY battery spot welder

spinningmagnets said:
One of the benefits of using supercapacitors in the kCap is that the battery (or power supply) does not necessarily need to be very high amps and very low resistance, like the nanotech LiPo that was the initial power battery for this design.

Many builders wanted to use a common 12V car battery, but if they did that, it needed to be fairly new so it had low resistance. This is also why the V4 has a power test function, so the builder can tell ahead of time if the battery they plan to use is adequate.

If using only one kCap, I believe the max voltage is 8.2V, but I would have to check that.

personally i dont liek using big power supplies to charge caps fast. that requires hard hitting CC/CV power supplies that can reliably take dead shorts on their outputs. not many will survive those beatings in the long run.
for this reason i am using a small 5A 14V power supply to keep the current lifepo4 battery topped up, with constant welding it cant keep up ofcourse but there is way more enough energy in the battery that i will never drain it to notice a power drop.

if i can really crank up to a 8S 40Ah lifepo4 battery then the power is WAY higher then i would be able to get with a kCap setup and it would not require a big power supply to feed those caps to weld at speed, a big battery bank does not have a recovery time like a kCap setup has.
it would be nice to know if something like this has been tested.

its not that i am willing to cheap out but if i already got about 30 big lifepo4 cells on the shelf i can make a battery that will probably destroy any welder i hook it up to. so i am searching for things that would be in favour of having a kCap board versus a huge 30V battery bank.
the only advantage i see is that it is smaller, but it does require a big power supply and might pop the extremely sensitive type A RCD's in my workshop. (yes, its rented to i cant replace it.)
 
flippy said:
personally i dont liek using big power supplies to charge caps fast. that requires hard hitting CC/CV power supplies that can reliably take dead shorts on their outputs. not many will survive those beatings in the long run.
for this reason i am using a small 5A 14V power supply to keep the current lifepo4 battery topped up, with constant welding it cant keep up ofcourse but there is way more enough energy in the battery that i will never drain it to notice a power drop.

if i can really crank up to a 8S 40Ah lifepo4 battery then the power is WAY higher then i would be able to get with a kCap setup and it would not require a big power supply to feed those caps to weld at speed, a big battery bank does not have a recovery time like a kCap setup has.
it would be nice to know if something like this has been tested.

its not that i am willing to cheap out but if i already got about 30 big lifepo4 cells on the shelf i can make a battery that will probably destroy any welder i hook it up to. so i am searching for things that would be in favour of having a kCap board versus a huge 30V battery bank.
the only advantage i see is that it is smaller, but it does require a big power supply and might pop the extremely sensitive type A RCD's in my workshop. (yes, its rented to i cant replace it.)
It may be hidden further up in this thread, but I have been working on just this CC/CV regulator you mention above. A first production batch is ordered and will arrive in a few weeks, and I'll put that in the shop as soon as I have time. That solution is based on a 800W HP server power supply. The system charges kCap at 70A (of course limited at 8.1V) and needs less than a second to fully top it up even at high energy settings. On the other hand, server PSU's aren't exactly known for their low AC leakage current.

I think that your idea of using a large battery and not needing a powerful charger might be unrealistic. The amount of average power that the charger has to deliver doesn't differ between the kCap and battery solutions. This is the equation that I use to estimate that power requirement:
P_average = Q * f / n,
where
Q = pulse energy
f = pulse frequency
n = system efficiency

For example, when doing 50J welds with one second interval, and taking my measured efficiency for the Hobbyking Lipo solution of 15% as a basis (that number varies with the used power supply, kCap is actually a bit better), you get
P_average = 50 * 1 / 0.15 = 333W.

This is the power that your charger has to deliver when you want to keep your battery solution topped up during long welding sessions.
 
right now i am using a CC/CV 5A regulator to keep a 14V 4S lifepo4 bank charged and it manages that just fine at 70W for .2 nickel and thinner. i do have to take more breaks because the tips wear out fast and get too hot, i do want to weld faster or at least reduce the breaks i have to take.
the charger is behind a separate DC converter for isolation and filtering. this setup helps to filter out the huge pulses that get back into the grid. i have noticed that the solar setup on the roof of the building does not like welding pulses on the grid inverters, they tend to go offline and reset when welding with high current power supplies on the battery.

having a large power supply also adds another factor for me as i am limited in power in my shop, i only got 16A to share between all my stuff and that includes heating/aircon.
i am working on a in line 2x 5kW battery supply i can put between the grid and my devices so i am not power limited as much as i am now and it should solve any transients that might get back into the grid from welding and other things i might be doing but that is for the (not so) near future.

i am probably going to use a meanwell HRP series, i have a couple dozen of those on the shelf, they are medical grade so pretty robust and have lots of filtering and isolation and should be able to keep the battery topped up enough so i dont lose a noticable amount of capacity/volts/IR.

still a 40Ah (or 80Ah if a parrallel up the cells) 8S bank has 1kWh ( or 2kWh at 80Ah) in storage. that should be enough to absorb fast runs of welding between pauses of reorienting the packs, mounting the strips and cooling/redressing the tips and other tasks that slow down large welding jobs.

the main question remains: can the welder actually handle a 28V high current battery bank? the power levels als considerably higher then with a kCap with such a voltage.
 
flippy said:
the main question remains: can the welder actually handle a 28V high current battery bank? the power levels als considerably higher then with a kCap with such a voltage.
It can but that is not advisable, I have explained the reasoning here: https://endless-sphere.com/forums/viewtopic.php?f=14&t=89039&start=600#p1461705
 
If i want to search about a PSU which can deliver 8.1v...
whats the miniumum amps the psu have to deliver to fill the caps?

in other words: if i have a 8.1v source with 1A how long does it take to fill the caps?
and how i know that they are full and ready to give me joules i need?
LED indicator would be helpfull to be sure all welds are great.

i want to make the decision about using a 50a PSU that screams like hell to provide the desired amps
or using a lower psu with maybe only 5amps and just wait a "certain time" between the welds but without noise.
 
to fill the caps i am looking at a mean well HRP-300-7.5 or a SP-150-7.5 wich is passivly cooled.
either should be able to fill caps in decent time and considering they are proper mean wells they should be fairly silent.
only thing i am wondering is how much crap it lets back into the grid.
that is something i have to test.
 
Merlin said:
If i want to search about a PSU which can deliver 8.1v...
whats the miniumum amps the psu have to deliver to fill the caps?
There is a graph in the kCap manual: https://www.keenlab.de/index.php/product/kweld-ultracapacitor-module/ that gives an estimation.

flippy said:
to fill the caps i am looking at a mean well HRP-300-7.5 or a SP-150-7.5 wich is passivly cooled.
either should be able to fill caps in decent time and considering they are proper mean wells they should be fairly silent.
I have an RSP-150-7.5 here, and it does a decent job and can be adjusted to 8.1V. I haven't been recommending it though because it instantly blows up in smoke when you pull the AC cord while leaving the charged capacitor connected. If you use it, then you'll either need an ideal diode circuit at its output (couldn't find suitable boards on ebay though, most of them need too much voltage to operate, and/or the Chinese sellers make unrealistic current ratings), or at least a switch. A regular Schottky diode is not desirable, because you want to keep kCap at 8.1V precisely to get maximum current.
 
thank you. graph was helpfull.
10a means 10s on a 100J pulse. thats hard. but anyway i will not weld 0.3 anymore ;)


is there any optical indicator to be sure that the caps are "charged"?

in case of slow working and using a fanless psu for example. Like a LED or maybe possible to put a tiny Voltage display parallel?!
 
Merlin said:
is there any optical indicator to be sure that the caps are "charged"?
in case of slow working and using a fanless psu for example. Like a LED or maybe possible to put a tiny Voltage display parallel?!
The welder itself has an input voltage monitor, you can adjust the threshold in the menu to 7.9V and it will then refuse pulsing while the caps are being topped up.
 
tatus1969 said:
I'd suggest running two kCap in parallel then, both of them equipped with fans. The kCap power supply that I have been working is also on its way, the first production batch is ordered and I'm about to put it in the webshop for preorder shortly.
Here I can't help but ask the same boring question: any plans for 5S2P kCap?
:wink:
 
I just got my Kweld assembled. Has anyone tried using two 12V Lead acid batteries in series for 24V with this? If so, do you see any advantages of using the 24V?
 
I have read that the last two versions (V3/V4) can use up to 30V but since this device measures the power in joules, it will read the flow-through energy in real time, and cut off the pulse when the amount of joules are reached. I believe that higher volts would mean fewer amps.

You might consider trying several types of power sources to see which one provides the best results.
 
Thanks ....I will try just one battery ...then maybe in parallel. I just wondered since it is supposed to be able to use the extra voltage ...it might actually be an advantage.
 
Let us look at something, For a moment: and I wish for Tatas to correct me if he could, for I am an amateur and know very little where it comes to the laws of electricity: I wish for a math lesson.

This machine is capable of input in the area <30v Dc. Presuming we are diehard spot weldors, and have to shoot as much battery or supply at this machine as possible: we chose 6s for convenience and simplicity... Let us assume 25v input.

So, @ 25V input, you have the option of anywhere from 10 joules, to 500 joules.

The time available to weld, I believe I heard somewhere is at a maximum of 250 milliseconds, or .25 Seconds. This is the maximum pulse duration.

Now, with this example, lets use my battery ( or any battery / power supply ) as a source for our weld current. We calibrate at a healthy 1700A, because we are diehards who must have the best of currents for out battery building. Over the course of our test welds on our best nickel we see 1500A, with good connection, some a little higher and some a little lower:

The welder operates at a setting of either 10 joules or 100 joules, and lets say for the sake of diehard power, 500 joules.

The weld time is 15ms, 50ms, and 250ms ( just guessing at these figures, for simplicity... )

Would the corresponding WATTAGE invoked by the switch be in this range, for our three welds at this current? 10 J @ 15ms = 666.6w , 100J@50ms = 2000W , 500J@250ms = 2000W,

..... However, let us say that the time to weld our super duper conductive media for our battery building is so conductive, the time to fuse is greatly reduced, and now only takes .125 Seconds... NOW the wattage increased to 4000W! Is this true, given this equation :

P(W) = E(J) / t(s) ......?

Also, in theory, should our 500 Joule burst now take even less time, for the path has so little resistance, and you have such current available, the weld time is only 50 ms....... The resulting power is an astonishing 10,000W!!!!

I see the advantage of the lower voltage and the massive paralleling, along with a good sized battery ( or capacitor bank?) to absorb the burst.... I believe I understand it this way.

Is that how that works? When relating to high current power supply and the conductivity of the metals of differend natures? Ty for the answer, simple yes or no will suffice. I am learning you see.

I do wish some others would post the results of Low v, large Ah and paralleled lipo data. I would love the comparison. I do suppose the power supply and Kcap is one of the most painless and reliable ways to go.


( ALSAS How does the machine efficiency figure into the equation, too?... )

If the answer is to complicated, I understand.
 
Or even better: a massive battery with a kCap between the battery and kWeld....
 
Megavvolt said:
Here I can't help but ask the same boring question: any plans for 5S2P kCap?
:wink:
Yeah I know, sorry but no plans here at the moment. I have to focus on a new project that I am on, which is a novelty battery tester / charger (spoiler alert). I want it to revolutionarize our ability to assess and compare all kinds of batteries. With that device, you will know instantly if your overseas purchase is in fact a genuine brand cell, or a backyard build filled up with sand. You don't even have to discharge the cell for that. A first functional model has already proven that the device can hold up with expensive professional gear (the stuff that companies like Tesla use for their research) in terms of accuracy. And the first prototype design has also already proven that I will be able to manufacture this at a "DIY compatible" price.

Merlin said:
using them parallel would be better...
had a talk with Frank....all versions should do up to 30v.
Yes, but I should mention that older firmware revisions (2.5 or lower) had a silly bug (numeric overflow) that prevented it from being used with >26V.

I wrote about the disadvantage of using higher voltages a few posts earlier: https://endless-sphere.com/forums/viewtopic.php?f=14&t=89039&start=600#p1461705

DogDipstick said:
and I wish for Tatas to correct me if
It's "Tatus" (by the way, that is the name of the horse in the book "Medicus") 8)
I'm afraid but your calculations aren't correct, you are somehow putting the cart before the horse.

kWeld is a resistance spot welder, as such it pushes current through the electrical resistance at the weld spot itself. That spot has a given resistance, a typical value is 1 milliohm. The power that we can deliver into it is then directly determined by the flowing current: P = I^2 * R. At 1500A, the spot receives 2250W of power. kWeld will do exactly this calculation in realtime and sum up power over time, to derive energy: E = integral( P, dt). As we can assume that P is more or less constant during the process, we get t = E / P. For a set pulse energy of 100J, we then get t = 100 / 2250 = 44ms.

The job of the supply battery is just to deliver the right amount of current into the given total resistance. This is the mentioned 1 milliohm of the weld spot, plus a total of approx. 3 milliohms for the complete kWeld system, from batter terminals to electrode tips. The battery's own ESR is of course also in the loop, which results in this formula: I = U / (R_spot + R_kweld + R_battery). When you resolve this formula to R_battery, then you see that you need to raise the battery's resistance when you go to a higher voltage. This is why I don't recommend doing that, because all you do is to waste more power in the battery.
 
I tried the 12V lead acid batteries(several that I have on hand) yesterday, never trying them as a 24 volt setup. I did try them in parallel. And my tests came up with undercurrent every time. This was not that surprising to me as I did not buy them specifically to use with the Kweld. I just had them so I gave them a try.

However, I did try a 24V Lifepo4 battery that I had built previously with Headway 38120 25C cells in an 8S configuration and all worked fine. So with reading your comment Tatus about higher voltages, should I still pursue a different battery? Or is this "if it ain't broke don't fix it" and as long as I am getting good results I am good to go? This battery serves other purposes for me so it would be great if I do not need to pursue another solution.

Thanks..Richard
 
Headrc said:
worked fine. So with reading your comment Tatus about higher voltages, should I still pursue a different battery? Or is this "if it ain't broke don't fix it" and as long as I am getting good results I am good to go?
These were all considerations for users who plan their new setup. If your battery delivers enough current and does not overheat or break during use, then there's no need to replace it.
 
tatus1969 said:
Yeah I know, sorry but no plans here at the moment. I have to focus on a new project that I am on, which is a novelty battery tester / charger (spoiler alert). I want it to revolutionarize our ability to assess and compare all kinds of batteries. With that device, you will know instantly if your overseas purchase is in fact a genuine brand cell, or a backyard build filled up with sand. You don't even have to discharge the cell for that. A first functional model has already proven that the device can hold up with expensive professional gear (the stuff that companies like Tesla use for their research) in terms of accuracy. And the first prototype design has also already proven that I will be able to manufacture this at a "DIY compatible" price.

let me know if you need/want a tester, i am very interested in that product.
it is limited to lipo or can it do different chemistries?
 
flippy said:
let me know if you need/want a tester, i am very interested in that product.
it is limited to lipo or can it do different chemistries?
I'll start a separate thread here as soon as I feel being close enough to give away first pre-series units. I'll be heavily relying on you guys here to measure all sorts of batteries (no, it's not limited to a particular one) in different conditions (age, cycle count) with it, and upload the results to an online database. Only this will enable the main feature of comparing something against a known baseline. I have something similar in mind like companies like Passmark have shown for computer CPUs. Here's a picture of the current protoype. I had an eye on versatility, and it can also be used as a regular 6S / 5A balance charger/discharger (once I get it working 8)).

IMG_20190505_120046.jpg
 
i got dozens of old and new cell models, so i got plenty of data to make.
right now i am mostly focused on wear profiles but if i can get more usable data and can make correlations that would be a winner for me.

ps: nice PCB design, i love gold plating. i see it's your job or you have too much free time. :mrgreen:
 
Beautiful, Tatus'69.

Thankyou for the input, I do "grasp at straws" from time to time, and correction is learning.

As for the new project, Wunderbare Neuigkeiten!
I will not have to try give my home-made batteries away anymore to unsuspecting friends for testing...! ( anybody wanna battery to test? lol)
You cn reserve the answer for the new project thread, should you wish: I implore though, please though include data logging for the simple minded graph lovers in the world like myself!

-DogDip
 
flippy said:
ps: nice PCB design, i love gold plating. i see it's your job or you have too much free time. :mrgreen:
It's my job :) Screen printed solder paste, hand placed components, soldering on hot plate. Looks like from the factory 8)

DogDipstick said:
include data logging for the simple minded graph lovers in the world like myself!
That's one of the core features on top of the list! Data collection is everything when you want comparability. One other planned feature is that it will tell you how to best arrange your cells in a large pack, so that cell capacitances / resistances spread most uniformly.
 
Back
Top