kWeld - "Next level" DIY battery spot welder

[tatus1969]

I updated the FW to r2.8. No problems!

I also tried to weld with 0.0J energy! It is not 0.0J! It was very big, the electrodes got very hot and burned hole to nickel strip! When setting to 0.1J the energy level is correct.

You've actually found a bug that I have introduced in the latest firmwares. I was using E=0 to encode the "OPEN" calibration step (in need to disable the undercurrent test) and forgot about the consequenses. Just fixed this, new firmware attached to this post.

[tatus1969]

"Encodergate" is finally over. I switched production to a new type (after having disassembled one to see the internal construction): EN11-HSB1AF15. We have also replaced these parts in the entire existing stock.

[bobmutch]

EDIT2: Okay, I'll have to take back everything that I said about using flux. The below welds are with 0.2mm nickel strips, and you can see the coloring which cannot be prevented when doing it as usual. Except both leftmost welds per cell, where I added a drop of flux! Awesome, I can definitely recommend that to everyone! The used flux was Chipquik SMD291.

Hi Frank, I posted your test on a FB group and some people think it may be that the better looking welds are because they are further apart (both far left welds on each cell are electrodes appear to be further apart). Do you have a picture showing this results where the electrodes are not further appear on the no-discoloured welds when you put the flux on top of the Nickel between the Nickel and the electrodes?


Thanks Bob.

[tatus1969]

Hi Frank, I posted your test on a FB group and some people think it may be that the better looking welds are because they are further apart (both far left welds on each cell are electrodes appear to be further apart). Do you have a picture showing this results where the electrodes are not further appear on the no-discoloured welds when you put the flux on top of the Nickel between the Nickel and the electrodes?

Sorry, I wanted to repeat this but ... too busy... I see what you mean though, but I can reassure you that the results with and without flux were very different. It's not the larger distance that makes the spots shiny.

[lothar77]

Can someone point me to page 2 of the rev3 schematic? I have pg1 of rev3 from post 449 but nothing since. A parts list would do too. I have pg2 of rev1 but it's not matching up with my new-ish rev3.

I was moving right along on my second pack when I had to adjust my p/s battery. On the next weld sparks and smoke ... it seems one jaw of my small bent-nose pliers snuck through the hole in the case on the output side. D8 now shorts the output. I noticed that D19 has the same marking and I'm thinking D8 is the LF SMDJ28A (two on order). Also, it looks like D18 was populated - a part number for that would be appreciated too.

Many thanks T. for a great device. I appreciate all of the thought and care given to the project. Truly impressive

[tatus1969]

Can someone point me to page 2 of the rev3 schematic? I have pg1 of rev3 from post 449 but nothing since. A parts list would do too. I have pg2 of rev1 but it's not matching up with my new-ish rev3.

That component blows when a quite common mistake happens, and that is feeding voltage into the electrodes from your battery under construction. In most cases, this only causes this component to turn into a short, which protects the rest of the circuit from this voltage.

That diode (it is actually D18, this is a misleading print) is in fact a SMDJ28A, but to improve protection you can also solder an SLD8S28A (= D8) to the bigger pads instead. That one can take more power (but will still eventually fail if that accident takes too long).

[lothar77]

Thanks so much for the speedy reply and confirmation of the components. I'm eager to get the kWeld unit back into action!

[lothar77]

A quick update ... I got the parts yesterday, finished the kWeld board repair, finished the last weld on a new 26650 LFP pack (the one where things went south), and repaired an 18V NiCd tool pack. Many thanks for the support and a unique product that I enjoy using.

[tatus1969]

A quick update ... I got the parts yesterday, finished the kWeld board repair, finished the last weld on a new 26650 LFP pack (the one where things went south), and repaired an 18V NiCd tool pack. Many thanks for the support and a unique product that I enjoy using.

Happy to hear that your unit survived!

[Fraser]

Hi Frank,

I bought one of your first generation kits with the potentiometer control 

Sadly ill health got in the way of putting it all together 

I am revisiting the world of battery tab welding and see that the KWeld has been developing nicely  I have taken a look at what fully built tab welders are coming out of China these days and they look as bad as ever. The KWeld is so much more refined 

I think I will just buy the complete set of parts with the latest board revision and make life easy for myself this time. The MK1 PCB is languishing in a box with all the bits and pieces needed to construct probes etc.

You can expect an order from me very soon 

Best Wishes

Fraser

[tatus1969]

Sadly ill health got in the way of putting it all together 

I hope you are doing well again?

You should only need the new electronics module and the cable assembly, you can order them separately from the shop. Plus maybe the busbars, I don't remember anymore if they were different in rev1 (I don't think so). Can you maybe post a picture?

[tatus1969]

I finally made it! "kSupply" power supply module 0-12V 0-70A to charge the ultracapacitor modules are ordered and available in my shop for preorder. https://www.keenlab.de/index.php/product/ksupply/ 

[Will_SG]

I finally made it! "kSupply" power supply module 0-12V 0-70A to charge the ultracapacitor modules are ordered and available in my shop for preorder. https://www.keenlab.de/index.php/product/ksupply/ 

Ordered!

Understand that you had recommended the DPS-800GB to be used with kSupply. I am assuming that DPS-1200FB (12v 100A) will work as well. If it doesn't, what are the criteria for selecting the power supply? thx!

[tatus1969]

Ordered!

Thanks!

Understand that you had recommended the DPS-800GB to be used with kSupply. I am assuming that DPS-1200FB (12v 100A) will work as well. If it doesn't, what are the criteria for selecting the power supply? thx!

You can use the mounting holes to connect any kind of 12V power source, but the card edge connector on the module will only fit to certain PSU models. I've started a list of suitable models, so far these are on that:
Fujitsu DPS-800 GB
HP HSTNS-PD05 (connector layout identical to DPS-800 GB, but yet to be confirmed)
HP HSTNS-PR01

Here are pictures from the two different card edge layouts that I am aware of, and that are supported:

[Will_SG]

guess I have to try my luck for the 1200FB. Although the pins layout are similar but the width are different.

[tatus1969]

guess I have to try my luck for the 1200FB. Although the pins layout are similar but the width are different.

Yes, the connector won't fit, but you still can solder the wires directly to the PSU and follow the info online how to enable it.

[mzzj]

Took less energy than I expected to weld 0.10 and 0.13mm tabs and I was able to get sound welds with only 0.132F total capacitance.
0.10mm tabs were welding best at about 21 volts =  29 Joules
0.13mm tabs were ok at 25 volts = 40 Joules.

Weld pulse is pretty damn short and if the contact pressure is not sufficient I get loud bang and blackened terminals.
Need to re-write the program to actually use dual-pulse functionality, now the pre-pulse is already doing all the welding and draining the caps.

Took only a bit over a year to cobble up the 0.7F 35v cap bank.

Welds the 0.13mm tabs fine at 11 volts =40 joules

Same energy input but without the loud bang the 25v charge voltage previously was making 
Still throws shover of sparks if the contact is bad but without loud bangs.

[tatus1969]

Just wanted to share a piece of work with you that I am doing for a customer. These teflon handle will provide additional heat insulation and should allow significantly faster welding speed. The assembly also uses 6AWG leads, which should help keeping them cooler as well.

[dirtcooker]

I'm using, an HP DPS-1200FB with kweld, kcap, and ksupply. I can confirm that it works well, no problems unplugging it with ksupply and kcap attached. This power supply seems to be bullet proof. However, I had to make the following modifications to ksupply: remove R23, solder a 12k resistor between pins 36 and 37 (to turn the power supply on), and remove the connection of pins 13 and 52 to the circuit board. +12v ouput fills positions 1-13 and the corresponding positions on the opposite side, and pins 14-26 are ground (see photo). Thus, the unmodified ksupply when attached places a short circuit directly across the power supply output. Fortunately, the supply is well protected against such things, refusing to power up at all with an unmodified ksupply connected.
I dug up some information about this power supply on the internet, which I summarize here:

DPS-1200FB Power Supply
http://colintd.blogspot.com/2016/10/hacking-hp-common-slot-power-supplies.html
Pinout:
1-13: +12V Power out
14-26: GND Power out
27, 28, 29: I2C slave address selection pins.
30: signal GND (for I2C and I2C slave address selection, I guess)
31: I2C SCL
32: I2C SDA
33: ENABLE#
34: LOAD SHARE
35: STATUS
36: PRESENT
37: +12V stand-by, good for 2.5A
38: PSALARM, normally have an active low alarm output
I can confirm that pin 37 is +12v when the power supply is plugged in, and connecting pin 37 to 36 and grounding pin 33 turn the supply on. I have not used the other pins.
I found on ebay a circuit board for sale with edge connector for these supplies, which list the models that it works with (listed below). DPS-800GB is one of them, which is mentioned in the manual that ksupply will work with. Since these power supplies obviously have different pinouts, I am curious to see how they solved the compatibility problem so I ordered one. When I receive it I'll post again.

Common Slot" Power Supply Models:
DPS-800GB A
DPS-1200FB A
DPS-1200QB A
PS-2751-5Q
PS-2751-LF-1F
HSTSN-PL12
DPS-700LB D
PS-2112-5L
DPS-750RB A
https://www.ebay.com/itm/DPS-1200FB-Power-Supply-Adapter-Board-10-Pcs-6-2P-Cables-For-Ethereum-Mining/292358226805
28 nov 2019 update: attached are pics of the ebay power supply adapter. Just as I suspected, pins 13 and 52 of the edge connector are not attached.

​

[dirtcooker]

I built kweld, ksupply, kcap, and HP DPS-1200FB power supply into an old shuttle computer case.

[tatus1969]

I built kweld, ksupply, kcap, and HP DPS-1200FB power supply into an old shuttle computer case.

Looks awesome! One note: if you rotate the kCap by 180 degrees then you can make the wires between it and kWeld shorter. Long cables like in your picture will add too much resistance.

[anishkgt]

How did you manage to balance the Caps ? Are you using SAB MOSFETS because i can see two SOIC-16 in here -> https://endless-sphere.com/forums/download/file.php?id=235316 ? then one IC can balance 4 you would just need one. Isn't it 3s3p rather than 3s2p mentioned in your webpage ?

[tatus1969]

How did you manage to balance the Caps ? Are you using SAB MOSFETS because i can see two SOIC-16 in here -> https://endless-sphere.com/forums/download/file.php?id=235316 ? then one IC can balance 4 you would just need one. Isn't it 3s3p rather than 3s2p mentioned in your webpage ?

No, and I'm also not really a fan of this technique. Doable as a one-off, but if you look closer at the specs of these devices then you'll see that this is not suitable for larger scale manufcaturing. The balancer on kCap uses a solution based on opamps and comparators and is accurate within +/- 75mV.

Where did you read that 3S3P? I checked the docs and can't spot it. kCap is 3S2P.

[anishkgt]

Not Suitable in Large scale manufacturing? they can be daisy chained as mentioned here -> https://www.aldinc.com/pdf/SABMB16.pdf to have more caps.

I meant isn't it a 3s3p rather than a 3s2p 6caps of which 3 in series and other 3 in parallel. How does it work with 3s2p one would be left out of the six caps.

[tatus1969]

Not Suitable in Large scale manufacturing?

I had the reason wrong in memory for not using them. Rechecked the datasheets - it's the V/I curve that I didn't like. The device with 2.7V threshold (ALD810027) draws 1uA at 2.7V, 1mA at 3.14V, and 10mA at 3.7V. Thats far too slopply for big caps like the ones that I am using, which can require 100mA+ of balancing current when worn out. The parts are run at their full rated voltage, and I don't want to exceed that.

It is a 3S2P. Two caps in parallel at each level, and three of these pairs in series.