kWeld - "Next level" DIY battery spot welder

[tatus1969]

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.

It is not the voltage that needs to be limited but the current. I got good results at 1000A with the old Lipo, and it is also good with the new Lipo and 1400A. The kWeld power switch is software-limited to 2000A. The current is determined only by the combined system ESR, and the weld spot resistance. We'll have to live with zero ohm weld spot resistance. When using a 12V power supply, and limiting the current to 1700A, this yields a minimum system resistance of 7 milliOhms. Double that resistance, and you can use a 24V supply. But there is no real benefit in going to higher voltages, you'll only reduce system efficiency further. The 12V system delivers approx 15% of the energy drawn from the battery into a 1 milliOhm weld spot, the other 85% are wasted in battery, power switch and cabling.

[Fraser]

Another piece of my 'kWeld parts collection' arrived today..... some 5mm copper rod to make the new threaded version of the probes. I thought I would psot some pictures of my parts collection for this project. I bought my own parts rather than the official kWeld parts kits as I want to experiment and will need more material. For many people without the bits in their parts bins, the kWeld kits of parts make better financial sense.

I do like the foot pedal I found though. It was actually bought to add foot pedal control to my Chinese Tab welder but it can now serve both that and the kWeld. The brass hollow rod was being sold by a chap who had many in stock. They look like long stand-off's, maybe left overs from a project. They are all 100mm long and have finished ends. I will be having 100mm long probes for my kWeld.

I also purchased a selection of high current connectors for various uses. I followed the OP's recommendation and bought the XT150 types as well as the EC5's.

The battery will be a decent capacity Yuasa 12V sealed lead acid gel type that I already have as that is quite a saving.

Fraser

[tatus1969]

I will be having 100mm long probes for my kWeld.

but make sure to fill the brass tube entirely with the electrode copper, or it may have excessive resistance / heating.

[SirJMD]

I'd love to see a "shot counter". This would make it great for evaluating the state of the battery (or other power source). If the number of shots starts to decrease dramatically, we know the battery is dying.

[tatus1969]

I'd love to see a "shot counter". This would make it great for evaluating the state of the battery (or other power source). If the number of shots starts to decrease dramatically, we know the battery is dying.

Thanks, very good idea! I've directly put this on my todo list. I'll make it an input energy meter, and I'll also implement a function that measures the battery ESR.

[SirJMD]

I'd love to see a "shot counter". This would make it great for evaluating the state of the battery (or other power source). If the number of shots starts to decrease dramatically, we know the battery is dying.

Thanks, very good idea! I've directly put this on my todo list. I'll make it an input energy meter, and I'll also implement a function that measures the battery ESR.

Sounds great! I'm looking forward to receving the kWeld and start testing.

I was thinking of ordering some supercapacitors and use those as a power source. I'm unsure how the capacitors will cope with the high discharge rates. The Maxwell ones I've looked at is rated at 250A maximum, with an internal resistance of 2,2 mOhm and 2,5V nominel.

I was thinking of using 6 in series, which would result in just under 60F at 15V, for a total of ~6.5kJ. I'm not sure if the current capabilities of these are sufficient.

[tatus1969]

I was thinking of ordering some supercapacitors and use those as a power source. I'm unsure how the capacitors will cope with the high discharge rates. The Maxwell ones I've looked at is rated at 250A maximum, with an internal resistance of 2,2 mOhm and 2,5V nominel.

I was thinking of using 6 in series, which would result in just under 60F at 15V, for a total of ~6.5kJ. I'm not sure if the current capabilities of these are sufficient.

That must be the previous version of their Boostcap parts, the current ones have 2.7V. It appears that I have almost the same type of cells here: BCAP0310 P270 T10, these have 2.2mR as well.

I am trying to use as few of them as possible, and will start with four. They have ridiculously strong solder tabs, and I had to first get a suitable soldering iron, the 80W Weller didn't blend...

I realized though that, if it works good enough at all, these will only be able to deliver a single pulse. And as you don't really want to wait for several seconds for them to recharge, we will need quite a decent power supply. I think a 400-600W ATX power supply will do, but we need a current limiter betwen PSU and capacitors, because these normally operate in hiccup mode and that is not what we want here. I am thinking of a boost converter with input current measurement, but it will need to be quite beefy. Do you have another idea of how to accomplish this? I think any dissipative solution is out at these power levels.

[tatus1969]

It was quite some work, but I have managed to ship all orders in time (week 34) as promised! Thank you for your patient waiting, and I am curious about your feedback.

[Kjelt]

For the supercaps, how about those carbattery psu's they are quite beefy and with a good voltage (13-14V).

I am eagerly waiting for the postman to show up with the kWeld although I do not have a suitable powersource yet.

[Fraser]

Regarding power sources, one of the cheapest is a small car battery as fitted in say a Classic Mini. In the UK, a decent one is only about £25. They are built for abuse as cars are not the kindest loads or chargers ! Car battery chargers are also cheap and available. Basic solid technology, if a little agricultural

Wet batteries are risky though as there is the acid spill and hydrogen accumulation explosion risk.... especially when heavily loading them with a spark producing welder ! I am going to use heavy duty 12V Gel batteries in a well ventilated area.

Fraser

[tatus1969]

For the supercaps, how about those carbattery psu's they are quite beefy and with a good voltage (13-14V).

I just checked ebay for that, you only get 6A output current models when limiting the price to 20€. At that current level recharging would last 139 seconds (5 x 310F charged to 13.5V)...

If we want the system to be able to fire a 50J weld once a second, and the system efficiency if 15% (based on measurements with my new battery), then we need to suppy 50 / 0.15 = 333W of continuous power. For my solution with a current limiting step-down converter with lets's say 90% efficiency, this meand that a 12V power source needs to be able to deliver 333 / 12 / 0.9 = 30.8A. A >= 650W ATX PSU should be able deliver this, these start at 30-40€.

[Kjelt]

I am not sure if a pc psu likes these giant currentjumps, they might shut down on OCP.
 Perhaps some form of current limiter between the psu and supercaps ?

[tatus1969]

Perhaps some form of current limiter between the psu and supercaps ?

That's exactly what I am talking about in the previous posts 

[Marco]

It is not the voltage that needs to be limited but the current.

With this circuit, a low dynamic resistance thyristor isn't terribly expensive and could handle the current.

I am not sure if a pc psu likes these giant currentjumps, they might shut down on OCP.
 Perhaps some form of current limiter between the psu and supercaps ?

Easier to use a PSU with a current limiter rather than an ATX PSU. If you want to charge at max power with an ATX supply you would have to use a buck converter with an input filter to smooth the current.

[tatus1969]

With this circuit, a low dynamic resistance thyristor isn't terribly expensive and could handle the current.

You cannot switch DC with a thyristor.

Easier to use a PSU with a current limiter rather than an ATX PSU. If you want to charge at max power with an ATX supply you would have to use a buck converter with an input filter to smooth the current.

I am looking for the solution at lowest possible price. The buck converter will cost maybe 20-30€, the ATX PSU another 30-40€, totaling 50-70€. Can you can recommend a 12V / 40A PSU that can easily be sourced and meets this price.

[Marco]

You cannot switch DC with a thyristor.

That's not how capacitive discharge works. You charge the capacitor with a low current switching converter, then you dump all the energy from the capacitor with the thyristor. When the current drops to zero the thyristor turns off.

I am looking for the solution at lowest possible price. The buck converter will cost maybe 20-30€, the ATX PSU another 30-40€, totaling 50-70€. Can you can recommend a 12V / 40A PSU that can easily be sourced and meets this price.

Hmm, looking around, it's hit or miss to find a Mean Well clone which can be easily modded into constant current. Some have current limiting, some shut down when they detect a fault ... not like they have accurate data sheets :/

[tatus1969]

That's not how capacitive discharge works. You charge the capacitor with a low current switching converter, then you dump all the energy from the capacitor with the thyristor. When the current drops to zero the thyristor turns off.

The main idea behind my welder is that it measures the amount of energy that is dumped into the weld spot, and stops the current when a user adjustable threshold is exceeded. This provides very consistent welds. This can only be achieved with MOSFETs.

[SirJMD]

That's not how capacitive discharge works. You charge the capacitor with a low current switching converter, then you dump all the energy from the capacitor with the thyristor. When the current drops to zero the thyristor turns off.

The main idea behind my welder is that it measures the amount of energy that is dumped into the weld spot, and stops the current when a user adjustable threshold is exceeded. This provides very consistent welds. This can only be achieved with MOSFETs.

Aha. I was wondering why you didn't use thyristors instead of MOSFETs, since they come in rather large packages that can handle several thousands of amps, while being relatively cheap.

Have you looked into forced commutation? Possibly a class D commutation setup might do it.

[tatus1969]

Aha. I was wondering why you didn't use thyristors instead of MOSFETs, since they come in rather large packages that can handle several thousands of amps, while being relatively cheap.

Have you looked into forced commutation? Possibly a class D commutation setup might do it.

I had considered them, and there were several reasons why I decided against them:
- a 300A SCR would probably have done the job, these are ~30€ which is same cost as I have now
- I was concerned if I would be able to design a reliable forced communitation circuit - it it fails to turn off the SCR, then the fuse is the last hope to prevent something from catching fire
- the voltage drop of the SCR would probably be 2V @ 1000A, which is more than 10 times of what it is now
- they are much bigger and I wanted to integrate everything on a circuit board
- last but not least, my knowledge of MOSFETs is several decades higher than that of SCRs

[Kjelt]

Awaiting the package this week so I need to do some scouting for powersources.

Does the wire need to be silicon?
I can find 10 and 16mm² awg7 and awg5 i believe in flexible wire for in the power cabinet.
I probably gonna buy some anyway for the next item:

Yesterday evening I met someone who might be interested in swapping my R&S HF generator that is eating dust since I do nothing with hf, against a 1500W 45A  digital VC and CC controlled power supply 

If that deal is happening then I am seriously looking for some 5 or 6 supercapacitors, I am only not sure how many F's they should have ?  Any link besides Ebay or suggestion is welcome 

I also see some protection boards with loadbalancing and current limiting, what is that about? Since we do not want current limiting in this application is it safe to use supercapacitors this way or could they also explode as LiPos?

[JohnG]

Aha. I was wondering why you didn't use thyristors instead of MOSFETs, since they come in rather large packages that can handle several thousands of amps, while being relatively cheap.

Have you looked into forced commutation? Possibly a class D commutation setup might do it.

I had considered them, and there were several reasons why I decided against them:
- a 300A SCR would probably have done the job, these are ~30€ which is same cost as I have now
- I was concerned if I would be able to design a reliable forced communitation circuit - it it fails to turn off the SCR, then the fuse is the last hope to prevent something from catching fire
- the voltage drop of the SCR would probably be 2V @ 1000A, which is more than 10 times of what it is now
- they are much bigger and I wanted to integrate everything on a circuit board
- last but not least, my knowledge of MOSFETs is several decades higher than that of SCRs

At low voltage, it is very tough to beat a power MOSFET. Cost is way down from where it used to be, they are robust, and fully controllable. Any bipolar device (SCRs included) have a saturation voltage of a few tenths of a volt, at least. Since this is a pretty slow application, the large capacitance of multiple high-current MOSFETs should not be a problem.

I don't have a ton of experience with thyristors, either, but what little I have had made me think that they are a lot more finicky and less straightforward than they seem.

John

[anishkgt]

I've designed one with a Microwave Oven Transformer and and another version on its way. There the switch on time is always controlled  with a dual SCR in a back to back configuration.

SCR or thyristors are basically used with AC voltages as they switch OFF at zero cross but when dealing with dc supply MOSFTETS do a better job.

[tatus1969]

Does the wire need to be silicon?
I can find 10 and 16mm² awg7 and awg5 i believe in flexible wire for in the power cabinet.

Haven't heard of silicon wires yet  Are they any good? That reminds me of a scene from Terminator 2: in the German translation, Arnie is referring to his high performance silicone CPU. The audience in that movie theater didn't catch why I had burst into laughter 

I chose silicone high flex because a) it eases electrode handling and b) it is 200°C rated and won't immediately burst into flames when something unwanted happens.

a 1500W 45A  digital VC and CC controlled power supply 

I dream of one of these 

If that deal is happening then I am seriously looking for some 5 or 6 supercapacitors, I am only not sure how many F's they should have ?  Any link besides Ebay or suggestion is welcome 

I have some Maxwell BCAP0310 P270 T10 here, but I haven't found the time to test them yet. Their DC ESR is 2.2 mOhm, I first want to see if that is enough or not, before I can recommend them. I'll keep you updated!

I also see some protection boards with loadbalancing and current limiting, what is that about? Since we do not want current limiting in this application is it safe to use supercapacitors this way or could they also explode as LiPos?

Supercapacitors don't like being overcharged, similar to Lipos. In a similar way, they need balancers when connected in series. For low current applications, that can be a simple resistor, but in our cases we need an active circuit. Typically that is a (e.g. 2.6V) comparator that switches a transistor which connects a resistor in parallel with the cell.

What they like even less is being operated at high temperatures. It is the same as for regular electrolytic capacitors: every 10°C more halves their lifetime.

I haven't heard yet that they can explode, and I have never made this experience. They rather tend to put other things on fire because of their extreme power density 

[Kjelt]

I chose silicone high flex because a) it eases electrode handling and b) it is 200°C rated and won't immediately burst into flames when something unwanted happens.

Ah yes and you solder that wire to the electrodes that will also need the insulation of the wire to withstand the higher temperature.
Well I found bigger silicone wire (16mm² upto 120mm²) from HELU or LAPP and othere manufacturers at www.tme.eu , a bit pricy but available if needed to lower the resistance even further 

I dream of one of these 

me too, I so hope that the deal goes through, a dream would come true.
He even has one with 120A    but that one he wants to keep to jumpstart his wifes car, he says he puts it at 16V and can start the car without battery just on that one power supply.
A bit rediculous because the listprice of these supplies are >5k€ but he has gotten them cheap so yeah if he has a practical use for it who am I to judge but I know some people that read this will burst in tears.

I have some Maxwell BCAP0310 P270 T10 here,I'll keep you updated!

Great! I am looking forward of hearing the experience.

Supercapacitors don't like being overcharged, similar to Lipos. In a similar way, they need balancers when connected in series. For low current applications, that can be a simple resistor, but in our cases we need an active circuit. Typically that is a (e.g. 2.6V) comparator that switches a transistor which connects a resistor in parallel with the cell.

If I get this correctly if you have 4 supercaps in series and charge it with a controlled Voltage of 10V, still one of the caps might get more than 2.6V due to imbalances in the capacitors?
Than a balancer circuit is also on the shopping list 

[ozwolf]

G'day,

I've been watching this for a while now.  I had a look at your shop and find "out of stock".

Any plans to assemble more kits?

Ozwolf