kWeld - "Next level" DIY battery spot welder

[Fraser]

I have ordered those XT90 plug sets for my unit. I got a good deal on 10 pairs of connectors, so plenty to experiment with.

I intend to use a pair of the contacts in parallel for each power cable from the battery.  I can split the cable strands into two smaller bunches to suit the connector.

Fraser

[fuzzoli]

Will the bare boards be available for purchase?  I noticed only the fully assembled boards are listed in the shop.  Thanks!

[tatus1969]

Will the bare boards be available for purchase?  I noticed only the fully assembled boards are listed in the shop.  Thanks!

The bare boards would be no problem, but I decided to protect the firmware code against Chinese copying (encryption + special bootloader), because it has a number of novel features and I don't want to make it too easy for them. I don't have a QFP48 ZIF socket here [yet], so I cannot pre-program the bare microcontrollers.

[Fraser]

I ordered some 12.7mm x 12.7mm x 300mm square copper bar today. In the UK it cost me £12 delivered. I think this excellent project deserves decent buss bars though I am considering tapping the holes M4 to increased the contact area between the bolt head and the copper buss bar. That may be overkill though. At least copper is easy to work with

I also ordered some nice large 5mm RC EC5 connector pairs that are rated at 120A continuous and they take 10awg cables

Fraser

[tatus1969]

Will the bare boards be available for purchase?  I noticed only the fully assembled boards are listed in the shop.  Thanks!

The bare boards would be no problem, but I decided to protect the firmware code against Chinese copying (encryption + special bootloader), because it has a number of novel features and I don't want to make it too easy for them. I don't have a QFP48 ZIF socket here [yet], so I cannot pre-program the bare microcontrollers.

I'm also not sure if it is worth it. I just uploaded the part list (except 0603 resistors / capacitors) to Digikey, the calculated total is 60.36€ plus VAT. I would sell the programmed microprocessor for 10€, and my price for a bare board would be another 10€, so there is ~20€ to save, which would be quickly eaten up by shipping costs.

[tatus1969]

I ordered some 12.7mm x 12.7mm x 300mm square copper bar today. In the UK it cost me £12 delivered. I think this excellent project deserves decent buss bars though I am considering tapping the holes M4 to increased the contact area between the bolt head and the copper buss bar. That may be overkill though. At least copper is easy to work with

I think it's safe to say that they will not be the weakest link of the chain 

I also ordered some nice large 5mm RC EC5 connector pairs that are rated at 120A continuous and they take 10awg cables

I just did a measurement with the XT90 connectors that I forgot in the first place. The voltage drop across one contact pair at 1205A is just 46mV, which results in a contact resistance of 38 micro ohms, or 76 micro ohms round trip. The power's dissipation during the pulse is then 55W.

I'll repeat that with the XT150, but I think that this value is already good enough, considered that the system will have a duty cycle of maybe 20% max. The only disadvantage is that it only fits cables up to 6.5mm diameter, and that the soldering terminal is too small for AWG8 cross section either. I will stick with XT150, that is perfect for this application, although a bit expensive. And maybe add an XT90 option if there is enough interest.

[Fraser]

Tatus1969,

In case I have not said it before, thank you very much for not only making your design available to us, but also for selling it at such a reasonable cost. I know that the profit margin is very slim and it is little reward for all your hard work, both in developing the design and the work required to sell us a kit.

I know it does not pay the bills, but you have my sincerest thanks for this act of kindness 

Fraser

[tatus1969]

Tatus1969,

In case I have not said it before, thank you very much for not only making your design available to us, but also for selling it at such a reasonable cost. I know that the profit margin is very slim and it is little reward for all your hard work, both in developing the design and the work required to sell us a kit.

I know it does not pay the bills, but you have my sincerest thanks for this act of kindness 

Fraser

My reward will not (only) be the paid bills, but the (hopefully positive) feedback from you folks making use of it, talking about it, and maybe doing the one or other review. I just had a job interview today (I had quit my last job because I didn't enter into a supporting environment when taking it two years ago), and one of the things that I said was that I am looking for a job where I can do something meaningful. (They do custom electric vehicle conversions [their first project was a DeLoran - yes it can do 88mph], and I'm very excited.) I keep driving this project because I have the same feeling here as well.

p.s. and if I'm lucky it will get me the Rohde&Schwarz oscilloscope that Dave is giving away - I participated and of course showed this project among others

[Kjelt]

Tatus1969,
In case I have not said it before, thank you very much for not only making your design available to us, but also for selling it at such a reasonable cost. I know that the profit margin is very slim and it is little reward for all your hard work, both in developing the design and the work required to sell us a kit.
I know it does not pay the bills, but you have my sincerest thanks for this act of kindness 
Fraser

+1 very well worded,
Also already my big thanks to you Frank for sharing this great project     

[tatus1969]

Also already my big thanks to you Frank for sharing this great project     

Thanks alot, it is a bit overwhelming I accept rewards in the form of honest public feedback and reviews 

[tatus1969]

I am busy with the next product video, and realized that the new probe design is now thermally the limiting factor. Too bad as it is the only part that you need to hold. Therefore I started a third iteration, the idea is to use 6mm copper rod and thread that down to M5. The holder will be threaded as well, which means that you then can screw in the electrodes. That should significantly reduce ohmic losses there. Parts ordered, I'll keep you updated.

[Kjelt]

I just stumbled upon a box with 140 elcos 2200uF/63V
I guess even if I place them all parallel it still would not hold sufficient energy to do a proper weld?

[tatus1969]

I just stumbled upon a box with 140 elcos 2200uF/63V
I guess even if I place them all parallel it still would not hold sufficient energy to do a proper weld?

Could be worth a try, resistance will most probably be low enough when paralleling all of them. Do you have that data? It may even be necessary to add more cable length to keep total resistance high enough to not exceed the 2000A limit.

You should use as much voltage as possible, my design should be good for up to 30V, although not tested yet.

The stored energy at 30V is 140J, the system efficiency with my new battery is 15% to have a number to start with. That would allow weld energies up to 20J. Close, but not enough I think.

I am concentrating on the Turnigy nanotech 3S 5Ah 130C. I stress tested it two days ago while shooting video, and then accidentially deleted the entire footage, 30Gb... :?

What I can already say that a complete discharge cycle resulted in approx 900 pulses of 30J into 1mOhm, and an accumulated pulse time of 15 seconds, at a current of 1400A average. And most importantly, absolutely no visible battery swell.

That is enough for me to say that I can recommend them for my welder. The only limiting factor is that they have only AWG10 wires, which limits pulse repetition rate. The battery temperature needs to be checked as well, I could not run my rapid firing test in one go, there were quite a few cooling breaks.

[Kjelt]

Thanks for the info, i don't have the esr for the elco unfortunately but can measure it. But since it is a normal series (cheap) the esr will probably not be low enough. I will check the turnigy out, tips where to buy them? Also going to need a charger. I am not running any production just a few cells here and then and perhaps some other application.

[tatus1969]

Thanks for the info, i don't have the esr for the elco unfortunately but can measure it. But since it is a normal series (cheap) the esr will probably not be low enough. I will check the turnigy out, tips where to buy them? Also going to need a charger. I am not running any production just a few cells here and then and perhaps some other application.

You can get them here:
https://hobbyking.com/en_us/turnigy-nano-tech-5000mah-3s-65-130c-lipo-pack.html

The charger that I use is SkyRC RS16, available from many sources. That can do up to 6S, which is not needed here.

[tatus1969]

The manufacturing of the pre-ordered kWeld electronics is running well, and I expect to have received everything by end of next week. I will start testing and programming them right after arrival and expect to be able to ship in week 34. If you have pre-ordered one, then I hope that you like working with the system as much as I do, and please give me your honest feedback.

In the meantime, there have been a few more additions and improvements. I am showing them on Youtube:

https://youtu.be/rQnODV4VQjU

- the system firmware has been extended by a bootloader that allows firmware updates via a simple serial connection. All you need is a terminal program that supports YMODEM file transfers, and an adapter that is available in the shop: https://www.kicksurfer.de/index.php/product/kweld-firmware-update-tool/


- a first sample of a 3D printed housing has arrived, and I am currently testing it. I plan to make it available in the shop, and also to provide the 3D data for free to enable you printing your own.


- I have tested a new type of Lipo battery for use with the welder: Turnigy nano-tech 3S 5Ah 130C (https://hobbyking.com/en_us/turnigy-nano-tech-5000mah-3s-65-130c-lipo-pack.html). This is the first battery that has survived more than one hour of continuous torture testing without any measurable swell or other damage. This is so far the only Lipo that I can definitely recommend to you. You can download the torture test results here: https://www.kicksurfer.de/wp-content/uploads/2017/08/kWeld-torture-test.zip

- the electrode system needed to go through a third iteration. The battery torture test showed that the previous system had a too high ohmic resistance and was heating up too quickly. As a side effect, I could again significantly reduce the involved machining effort. The new system only requires crimping the cable to a brass tube, thread-tapping tube and copper electrode, and heat shrinking an isolation. This also makes electrode replacement easier than ever before.


As the previous electrode design did not meet my quality requirements, I will inform all pre-order customers that they will receive the new design instead at no additional cost.

[Kjelt]

  very nice improvements again.

[mzzj]

I just stumbled upon a box with 140 elcos 2200uF/63V
I guess even if I place them all parallel it still would not hold sufficient energy to do a proper weld?

Sufficient energy but at wrong voltage level. (unless you couple them with transformer)
60 volts give you more likely loud bang and sparks but poor or no weld at all. And charged to ~20 volts you don't have enough energy.

[Marco]

Why would 60v not work? Too much energy in too little time?

[Kjelt]

Why would 60v not work? Too much energy in too little time?

the kWeld was designed for max. 30V , I think the Mosfets are the limitation ?
But it does not matter, a carbattery would go nicely also. The batterypacks are out of stock so I have to look at another alternative.

[Fraser]

I am going to try a large capacity sealed lead axis 12V battery with my kWeld. Hopefully it will be up to the job and less likely to complain than a LiPo.

From memory I have brand new 15Ah and 36Ah SLA 12V Gel batteries.

Fraser

[mzzj]

Why would 60v not work? Too much energy in too little time?

Something like that. Not sure about the exact physic behind it but for sure it didn't seem to work. (tried some years ago)

[tatus1969]

Why would 60v not work? Too much energy in too little time?

Something like that. Not sure about the exact physic behind it but for sure it didn't seem to work. (tried some years ago)

I can imagine that too much current can cause problems because the pulse time control of existing systems may be too coarse to achieve the ultra short pulses that will be necessary for this. The kWeld firmware pulse control loop runs at a 10 microsecond interval and should be able to handle this as well. The other problem that I can imagine is that a current that is too high may instantly blow away some material and ignite an electric arc instead, eating more material.

brand new 15Ah and 36Ah SLA 12V Gel batteries.

Do you have the equipment to measure those's ESR? An oscilloscope and a power resistor of 0.5  ~ 1.0 ohm that you can sacrifice would be enough.

I just ordered six of these: Maxwell BCAP0310 P270 T10 supercapacitors. They have an ESR of 2.2 milliOhm, and I calculated that four of them in series should already do the job for pulses up to 50 joules. That would be another cost effective solution, as these cost only 10€ each.

[tatus1969]

I've got some work on my desk   

[Marco]

Something like that. Not sure about the exact physic behind it but for sure it didn't seem to work. (tried some years ago)

Too short a pulse might evaporate metal before the melt can spread I guess.