kWeld - "Next level" DIY battery spot welder

Gosh it gets better and better. Now I wish I hadn't been in a hurry to have a backup and ordered so early! I have yet to use, and it's 2-3(maybe more) revisions behind todays progress! Congratulations on the development! The photo with caps looks great!
 
tatus1969 said:
Alcus said:
Once I get my two kCaps, I'll have the power and will be running some tests on these thin copper wires. If anyone has already done this, please feel welcome to share your settings and tips.
If kWeld does not work out with these copper wires, I plan to test using pure nickel wire. Again, if anyone has already done this, please feel welcome to share.
Are you saying that you plan to put two kCap in parallel for enough current to weld fuse wires? Is that to make the pulse duration as short as possible in order to concentrate the energy?

We've started shipping kCap modules and enclosures today, here's a first impression of how that assembly looks:

IMG_20180607_160429.jpg

p.s. no update from that customer yet. But I won't insist, I don't want to annoy my customers.

By "having the power", I mean that I did not bother buying a decent battery. I was not comfortable using "prismatic" cells, I didn't want a big car battery on my desk and chose to wait for kCap. I'll be testing one kCap and two kCaps in parallel.

If you recall a few months ago you shared your spreadsheet with me and pointed out the small "joule loss in capacitor per pulse" when adding additional capacitors in parallel rather than in series. The thought is: smaller loss should result in lower demand on the power supply speeding up the cycle time even with a smaller power supply. That should get me up an running until you release your HP bench power supply board.

If 2 kCpap also happens to assist in welding 30 and 40AWG copper "fuse" wire by concentrating the energy, that will be a bonus.
 
tomjasz said:
Gosh it gets better and better. Now I wish I hadn't been in a hurry to have a backup and ordered so early! I have yet to use, and it's 2-3(maybe more) revisions behind todays progress! Congratulations on the development! The photo with caps looks great!
Thanks :wink: and sorry 8)

Alcus said:
By "having the power", I mean that I did not bother buying a decent battery. I was not comfortable using "prismatic" cells, I didn't want a big car battery on my desk and chose to wait for kCap. I'll be testing one kCap and two kCaps in parallel.

If you recall a few months ago you shared your spreadsheet with me and pointed out the small "joule loss in capacitor per pulse" when adding additional capacitors in parallel rather than in series. The thought is: smaller loss should result in lower demand on the power supply speeding up the cycle time even with a smaller power supply. That should get me up an running until you release your HP bench power supply board.

If 2 kCpap also happens to assist in welding 30 and 40AWG copper "fuse" wire by concentrating the energy, that will be a bonus.
Okay, I understand your rationale. One kCap will deliver 1300A to the welder, which is it's "comfort zone". Two in parallel will definitely improve system efficiency and reduce the demand from the charger, but you may get close to the 2kA current limit of the welder.

The HP charger is second next on the project list, I hope to be able to start a first small series batch by the end of this month.
 
tatus1969 said:

Okay, I understand your rationale. One kCap will deliver 1300A to the welder, which is it's "comfort zone". Two in parallel will definitely improve system efficiency and reduce the demand from the charger, but you may get close to the 2kA current limit of the welder.

The HP charger is second next on the project list, I hope to be able to start a first small series batch by the end of this month.

Great! Close was my goal :)

From the calculations using your spreadsheet we are looking at +-1493 amps at 3S 4P. Allowing pulses of up to the kWeld maximum of 500 joules while still under the 2000A limit and within the 250ms limit.

Please let me know if I have made any mistakes or if you disagree.
 
Using the kWeld for a few builds recently, but the cables/electrode holders can get too hot to touch after 5/6 welds. It takes 10 minutes or so to get this point.

After calibration I am getting 3.9ohm using a 12v 90ah car battery. Is this figure too high?

I am also using the previous version of the electrode holder. The bullet connectors...
 
sacko said:
Using the kWeld for a few builds recently, but the cables/electrode holders can get too hot to touch after 5/6 welds. It takes 10 minutes or so to get this point.

After calibration I am getting 3.9ohm using a 12v 90ah car battery. Is this figure too high?

I am also using the previous version of the electrode holder. The bullet connectors...
The bullet connector electrodes should give you roughly 2.5 ~ 3mOhms during calibration. I suspect that your battery isn't delivering enough current, it should be at least 1000A for an accurate calibration. When you keep the foot switch down after making a weld, what current level does the display read?

What material are you welding? If that is 0.3mm nickel, then you will need regular cooling breaks. This is one of the reasons why Ron and I have been working on an beefier electrode system. The problem is that the welding heat is partially transferred to the electrode tip. As copper is a good thermal conductor, it is distributing that into the electrode holder as well. Our solution to that is to make everything bigger which increases both thermal mass and heat dissipation capacity. The new kWeld stock electrode system also attempts to use a longer electrode rod connected only at the rear end, so that heat has to travel a longer way.
 
I couldn't tell you the welding current from memory, but I will have a look this evening.

I am using 0.3mm nickel strip.

Are the new electrodes available for customers with the previous model?
 
sacko said:
Are the new electrodes available for customers with the previous model?
Yes they are, but as I have been outsourcing purchasing/manufacturing of them and have only ordered full kits with and without assembly, I don't have individual electrode holders to sell. I have made a quick calculation though, and the cost of electrodes and holders make up for 83% of the cost of the stock articles (https://www.keenlab.de/index.php/product/kweld-cable-electrode-holders/ https://www.keenlab.de/index.php/product/kweld-electrode-set/ ) I am asking my existing customers to purchase those for an upgrade. In parallel I am working on suitable shipping option, so these can be shipped more economically in a bag.
 
Many thanks. I will purchase the electrode holder if/when you have them in stock.

Regarding the longer electrodes, are you able to crimp them into a bullet connector for me; so I can do a straight swap with my existing ones?

Many thanks.
 
sacko said:
Test weld this morning.

Welding 0.3mm nickel strips together.

120j manual- 1163a.
Current is a bit on the low side for 0.3mm nickel. It is better to get this up to 1500A (more voltage and/or less battery ESR), because the pulse duration becomes shorter then and less heat can spread out during the pulse, lowering both energy demand and losses. This will keep the elctrodes cooler as well.
 
so looking to use a kWeld to do either Nickel Plated copper or just straight copper to batteries (tba which option)

somewhere between the 0.1 to 0.25mm mark. though i suspect 0.25mm will be impossible without nickel plating + a channel cut.

what power source would you recommend? currently looking at dual nanotech 3S lipos 65C or should I go to 5/6s?

i noted there have been some users who have used 6S with and without success, and there was talk that high voltages doesn't play so well with copper?
 
Alcus said:
Great! Close was my goal :)

From the calculations using your spreadsheet we are looking at +-1493 amps at 3S 4P. Allowing pulses of up to the kWeld maximum of 500 joules while still under the 2000A limit and within the 250ms limit.

Please let me know if I have made any mistakes or if you disagree.

Received the two kCaps yesterday.
Looks like 4 cables at 225mm each should allow enough clearance for two kCaps wired in parallel feeding one kWeld.
+-1,250mm total cable length using one kCap and +-1,700mm when using two in parallel. I seem to recall the limit is somewhere around 2M due to high kick back but I am having trouble locating that spec.

1 - Do I use the figures above during calibration or 1,250 under both conditions?
Planning to test each kCap individually prior to combining them.

2 - I'm not including the kCap input cable lengths and will likely use 10AWG as I will not be exceeding 20A. Makes sense?

3 - Originally my thought was to also power the two kCaps in parallel at 8.1V at up to 20A (162 watts).
Last night, I thought, why not wire the kCap inputs in series and feed them 16.2V at up to 20A (324 watts).
Do you see any issues with that Frank?

4 - The kit is very intuitive. I've already pre assembled most of one unit and love the no glue enclosure. Couldn't be any easier to assemble. Having said, I do not see a manual online. If there is one, please point me in the right direction. Thanks.
 
sacko said:
Many thanks. I will purchase the electrode holder if/when you have them in stock.

Regarding the longer electrodes, are you able to crimp them into a bullet connector for me; so I can do a straight swap with my existing ones?

Many thanks.

It looks to me like the bullet connector is no longer used with the revised system.
The following is from the online shop, "...requires access to a suitable crimping tool with inserts for 6.5mm diameters...".
I'm thinking that you cut off the old bullet connector and crimp on this new copper electrode holder. I'm thinking that the set screws are the new way to swap out the new longer electrodes.
If I'm wrong, I'm sure Frank will correct me when he has some free time.

Unless you have access to a 6.5mm crimp die, it might be best to choose the assembled option so the holder comes crimped onto a cable.
 
soylentgreen said:
so looking to use a kWeld to do either Nickel Plated copper or just straight copper to batteries (tba which option). somewhere between the 0.1 to 0.25mm mark. though i suspect 0.25mm will be impossible without nickel plating + a channel cut.
I've made tests with 0.1mm soft copper, both without and with channel cut, but without any success. I still have to repeat this at a higher current level like 1700A, but I'm not optimistic because both electrical as well as thermal conductivity is just so much higher compared to nickel or steel. In any case, this will most certainly require tungsten electrodes.

soylentgreen said:
what power source would you recommend? currently looking at dual nanotech 3S lipos 65C or should I go to 5/6s?
I recommend to use a Turnigy nanotech 3S / 5AH / 130C, or a 65AH car starter battery. When more cells are availale, i recommend to put them in parallel and not in series to reduce stress and increase current.

soylentgreen said:
i noted there have been some users who have used 6S with and without success, and there was talk that high voltages doesn't play so well with copper?
Higher voltage means higher risk of arcing, and higher losses in the battery (by Ohm's Law, the battery ESR must rise to keep the current in the sweet spot of ~1500A).

Alcus said:
Received the two kCaps yesterday.
Looks like 4 cables at 225mm each should allow enough clearance for two kCaps wired in parallel feeding one kWeld.
+-1,250mm total cable length using one kCap and +-1,700mm when using two in parallel. I seem to recall the limit is somewhere around 2M due to high kick back but I am having trouble locating that spec.
The [strike]2m[/strike] (EDIT)1m is round trip and including the output leads. Running wires in parallel to two capacitors reduces the effective inductance. I have to write down the calculation, but using 0.7 times would be my 'educated' guess. If your wires are longer, then adjust this in the menu which reduces the current limit trip value.

Alcus said:
1 - Do I use the figures above during calibration or 1,250 under both conditions?
Planning to test each kCap individually prior to combining them.
Calibration is independent from the power source (it only 'looks' at the output), but current should be 1000A or more for good accuracy.

Alcus said:
2 - I'm not including the kCap input cable lengths and will likely use 10AWG as I will not be exceeding 20A. Makes sense?
That's correct, they don't count from the inductive kickback perspective.

Alcus said:
3 - Originally my thought was to also power the two kCaps in parallel at 8.1V at up to 20A (162 watts).
Last night, I thought, why not wire the kCap inputs in series and feed them 16.2V at up to 20A (324 watts).
Do you see any issues with that Frank?
The modules can easily be paralleled but not easily put in series. This is because of the way the balancers work. They make sure that the voltages of all three capacitor pairs stay the same, but they are not limiting to a max voltage. When being charged to 15V, drift could cause the capacitors of module 1 to have 3x2V, but module 2 gets 3x3V. To wire these modules in series, you'd need some king of module-wide balancer, for example a power resistor across each module.

Alcus said:
4 - The kit is very intuitive. I've already pre assembled most of one unit and love the no glue enclosure. Couldn't be any easier to assemble. Having said, I do not see a manual online. If there is one, please point me in the right direction. Thanks.
Sorry that I haven't finished it yet, I'm a bit drowned in work at the moment.

Alcus said:
Regarding the longer electrodes, are you able to crimp them into a bullet connector for me; so I can do a straight swap with my existing ones?
I wouldn't recommend doing that, because the bullet connector's material overlap is just a few millimeters and not mechanically stable enough. I have already built up new stock of those bullet connector style tips to be able to supply replacement parts for existing customers. It's not online in the shop yet, if any of you has immediate demand please let me know.

Alcus said:
It looks to me like the bullet connector is no longer used with the revised system.
The following is from the online shop, "...requires access to a suitable crimping tool with inserts for 6.5mm diameters...".
I'm thinking that you cut off the old bullet connector and crimp on this new copper electrode holder. I'm thinking that the set screws are the new way to swap out the new longer electrodes.
If I'm wrong, I'm sure Frank will correct me when he has some free time.
Exactly! The tool that I am using here is something like this: https://www.jaycar.com.au/heavy-duty-terminal-crimper/p/TH1849 in the (smallest) 6mm^2 setting. The brass electrode holders were a bit too brittle and could break if not centered well, but the new copper holders are very easy to do.

I am sorry but I made the mistake of not having ordered some individual new electrode holders from my supplier (only complete sets, plus spare electrodes), which would have allowed me to also sell them individually as an upgrade part for those already owning a kWeld. I made a quick calculation, and the set of electrode holders and electrodes make up for more than 80% of the price for the unassembled cable kit, so I hope for your understanding.
 
last weekend I finished the assembly of the battery, this is the definitive setup I used to spot weld the 280 cells of my 20S14P battery.
Currents went from about 1600A at the beginning of each section to around 1750A after around 100-200 spot welds, I guess my welder battery has warmed up and and reduced it's IR.
20180602_161440.jpg

20180602_161430.jpg


In my opinion this is the best shape of the electrode tip. It allows to maintain a constant contact surface while consuming the electrodes. The tips are turned down to 1,5mm diameter.
20180602_175139.jpg

20180602_175821.jpg

20180602_180200.jpg
 
bigbore said:
last weekend I finished the assembly of the battery, this is the definitive setup I used to spot weld the 280 cells of my 20S14P battery.
Currents went from about 1600A at the beginning of each section to around 1750A after around 100-200 spot welds, I guess my welder battery has warmed up and and reduced it's IR.
I also experience this, the chemistry likes it hot. But the major part of this is the current reading itself. Current is measured from the voltage across the MOSFETs (the most cost effective principle), which has a temperature dependency. A typical "rough use" MOSFET temperature is 25K above ambient, which accounts for an error of approximately +10%. You shouldn't notice this when welding, the variation from material escaping the weld from arcing is much bigger that that. In fact the current production batch (that arrived TODAY! [still waiting for cables]) has a temperature sensor close to the MOSFETs to correct that.

Great pictures, thanks for that :)
 
When I first received my two kCaps, my results were very poor. Around 3.1mOhms at calibration with original cable dimensions (1M) and very low amperage when welding. Even with the setup below, the results were bad and inconsistent.
Two kCap side by side .jpg


With good advice from Frank, I quickly soldered the copper tips to the holder to assist the poor crimp and imediately dropped to 2.76mOhms.
Electodes soldered.jpg


I thought that was good until I got my new crimper.
My old Anderson crimper works great on normal amperage but for over 1,000 Amps, it simply did not do the job. That is using the smallest die available (6mm2).
Crimper - bad job.jpg


The new hydraulic crimper was a fraction of the price. It is not of great quality and may not survive hundreds of crimps but while it lasts, it certainly provides the required crimping power.
View attachment 4
Knowing what I know now, I should have ordered the cables assembled.

Ultimately, the majority of my issues were crimp related.
Here are the results after a couple of weeks of fiddling around:

One kCap 1.1M cabling and Anderson connectors:
At calibration almost 1,300A at 2.51mOhms
Simulated weld into chunk of copper 1,200+A at 0.48mOhms
Actual weld .1 nickel plated steel +-1,050A at 1.54mOhms
Anderson kCap.jpg


Two kCap 1.3M cabling and Anderson connectors:
Different layout to shorten cables.
At calibration 1,572A at 2.48mOhms
Simulated weld into chunk of copper 1,416A at 0.58mOhms
Actual weld .1 nickel plated steel +-1,200A at 1.78mOhms
Actual weld .1 pure nickel 1,280A
Please note that as advised, the output of the first kCap is connected to the output (not input) of the second kCap.
Frank - "... as long as you route the wires between the two caps to their output posts. Don’t use the input terminals on any of them for this connection, they have less copper to carry the current."
Output to output.jpg


Two kCap (+-1.2M adjusted) cabling (2 X 225mm cables to simulate +-6AWG):
At calibration 1,694A at 2.48mOhms
Simulated weld into chunk of copper 1,510A at 0.58mOhms
Actual weld .1 nickel plated steel over 1,300A at 1.50mOhms
View attachment 1


I'm very happy with the results but I'm not quite done yet.
Once I receive my 6AWG cable, larger lugs and Anderson terminals, I'll shorten the cables to replace the 4 between the kCaps and kWeld.
By then hopefully everything will fit properly into the enclosures. My goal is not to exceed 1,800A at calibration but get as close as possible.
If anyone has any interest in those results, let me know and I'll post that.

In the meantime, this is my current setup:
18650 battery and 20A DC to DC converter all connected with Anderson connectors for quick setup and dismantling.
PS - It also works very well with a lithium power tool battery.
 
Thanks Alcus, and I can direclty pick up on that because I just got the message from my cable assembly firm yesterday that they will be delivering early next week. Here are a few pictures from their work. This is what customers having ordered after April will receive. You can notice the upgraded electrode system, which was partly because of the electrode system crimping problems that the previous version showed.

1.jpg

View attachment 1

3.jpg
 
tatus1969 said:
... upgraded electrode system...

The clear shrink wrap is a great choice. Looks really good and very practical.
Looks like a much greater crimping area. How many mm does the cable insert into the revised holder?

PS - I had no difficulties welding 40AWG and 30AWG thinned copper strands onto 18650 cells. It was actually surprising how easy it was to do.

Nothing like copper cell level fuses!
IMG_20180714_153224~01.jpg


Note that pulling the cells apart with force did not result in breakage at the welds.
IMG_20180714_153640~01.jpg


Same kWeld settings on the positive and negative cell ends.
IMG_20180714_154133~01.jpg


More testing to come but I'm really impressed so far. Note the tiny 40AWG strands on the lower cell.
IMG_20180714_155851~01.jpg
 
Alcus said:
The clear shrink wrap is a great choice. Looks really good and very practical.
Looks like a much greater crimping area. How many mm does the cable insert into the revised holder?
That hasn't changed actually (10mm). It just works much better with copper compared to brass.

Alcus said:
PS - I had no difficulties welding 40AWG and 30AWG thinned copper strands onto 18650 cells. It was actually surprising how easy it was to do.
Awesome pictures!
 
Hi, any update for those eagerly awaiting arrival of their welder.

There be a stack of cells that need zapping!
 
Back
Top