kWeld - "Next level" DIY battery spot welder

[anishkgt]

I actually meant to show the size of the MOT and you could use it as power source for the kweld to charge the caps.

The secondary wiring should be thinner to get the desired voltage and thick enough to get the desired amperes.


Sent from my iPhone using Tapatalk

[tatus1969]

I actually meant to show the size of the MOT and you could use it as power source for the kweld to charge the caps.

The secondary wiring should be thinner to get the desired voltage and thick enough to get the desired amperes.


Sent from my iPhone using Tapatalk

Ah, great, I didn't realize that idea. Okay, that means that you use the MOT to make up a decent power supply. You'd also need a high power current limiting voltage regulator then, preferably a buck converter. Or would you try to see how much current the MOT delivers, and just let if flow? Or, at a second thought, we could use primary side phase angle control to set the current.

[anishkgt]

What is your desired voltage and current ? Since your using a lipo I am guessing it would be 12v and mayb 700A ? A 20 turns of 20 AWG silicone wire would give the desired voltage and amp. The turn is just a guess, you may have to increase or decrease it as needed.


Sent from my iPhone using Tapatalk

[SirJMD]

I actually meant to show the size of the MOT and you could use it as power source for the kweld to charge the caps.

The secondary wiring should be thinner to get the desired voltage and thick enough to get the desired amperes.


Sent from my iPhone using Tapatalk

Ah, great, I didn't realize that idea. Okay, that means that you use the MOT to make up a decent power supply. You'd also need a high power current limiting voltage regulator then, preferably a buck converter. Or would you try to see how much current the MOT delivers, and just let if flow? Or, at a second thought, we could use primary side phase angle control to set the current.

I would just "let it flow".

I did a quick look for a rectifier bridge and found one capable of 1.8 kA surge for ~45 euro. I haven't checked the datasheet but it might be suitable.

[tatus1969]

What is your desired voltage and current ? Since your using a lipo I am guessing it would be 12v and mayb 700A ? A 20 turns of 20 AWG silicone wire would give the desired voltage and amp. The turn is just a guess, you may have to increase or decrease it as needed.


Sent from my iPhone using Tapatalk

Desired voltage is 12V if you want to run kWeld directly from your main supply. You can also use lower voltages, but then you need provide an additional 12V supply for the logic. The current should be between 1kA and 1.5kA. Lower currents are also possible, but that will limit the welding capabilities.

[anishkgt]

Oh well then the MOT would be a perfect fit for 1kA and 2AWG wire with 3 windings at the secondary and then filtering and rectification follows.

A secondary supply to power the pcb.



Www.Georgehobby.wordpress.com
Equipments: DSO104z, Hakko FX888D

[Kjelt]

The current should be between 1kA and 1.5kA.

The main fuse was 300A I believe but can withstand these currents for short peaks or how is that designed?

[tatus1969]

The current should be between 1kA and 1.5kA.

The main fuse was 300A I believe but can withstand these currents for short peaks or how is that designed?

The time-current curve of the used fuse shows 1 second max at 1kA, which would translate to 0.25 seconds at 2kA. The hardware-limited maximum pulse duration is 0.2 seconds. The fuse will certainly not survive 2kA / 200ms pulses with high repetition rates, but the rest of the system would neither. During my 1.5kA torture / battery discharge test with the new Turnigy nanotech, I continuously monitored several temperatures, and the fuse was by far not the limiting factor - less than 50°C.

[Kjelt]

Ok thanks clear!
Colleague sent this, no supercaps or batteries required
https://www.telonic.co.uk/Lab-Power-LAB-DCH20-1500-p/lab-dch20-1500.htm

[tatus1969]

Ok thanks clear!
Colleague sent this, no supercaps or batteries required
https://www.telonic.co.uk/Lab-Power-LAB-DCH20-1500-p/lab-dch20-1500.htm

if that is an option for anyone, then I guess I'll have to rethink the price of my welder

[SvanGool]

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 tried to make a simulation in LTSpice for your Spotwelder based on capacitors. Requirements:

• Cheap capacitors
• Normal capacitor voltage above 10V
• Around 50 Joule in 20 ms
• Less than 1500A maximum current for 1-4 mOhm material
• Short circuit current below 2000A
• Able to weld 10 consecutive welds in 10 seconds

I first tried your idea of BCAP0310:


The graphs:


Some comments:

• The circuit "Charge_Cntrl" is an additional MOSfet that sees to it that the power supply is not connected when a pulse is fired.
• The CC_PSU is one of those $20 12V/54A server supplies that can be adjusted to 10.8V
• The circuit "Weld_Cntrl" is your electronics
• Do check the parameters, there might still be a mistake in there

The result would be, if no mistakes made , that with 4xBCAP0310 app. 40 Joule welds can be made with 2 mOhm material in 25 ms. The voltages are on the low side for the circuit so it might be necessary to use a Schottky diode and a larger capacitor on the supply of the electronics.
I also did the simulation for 4x BCAP0650: with increase of the wires to 70cm each, all conditions are met and a 75 Joule pulse is possible in 25 ms.

Edit: I updated the screenshots with the latest versions that match the LTSpice files attached in a later post, some cosmetic changes and corrected the Chrg_Cntrl parameters

[SirJMD]

Ok thanks clear!
Colleague sent this, no supercaps or batteries required
https://www.telonic.co.uk/Lab-Power-LAB-DCH20-1500-p/lab-dch20-1500.htm

I think I can top that..

Might need something to step it down.. 70 Mvar at 132 kV.

[Kjelt]

Cheap capacitors
BCAP0310:
BCAP0650:

Wow very nice simulation thanks
Pricing from Maxwell at Mouser is pretty weird.
The BCAP0310s are affordable at around €10 ex VAT a piece but the 0650 is already €35 then the 3000 is "only" €50. So if the 310s would suffice that would be good news otherwise I will probably get the 3000 or even 3400.
Anybody have seen better (lower) prices for those new supercaps except dubious sources such as Ebay?

[SirJMD]

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 tried to make a simulation in LTSpice for your Spotwelder based on capacitors. Requirements:

• Cheap capacitors
• Normal capacitor voltage above 10V
• Around 50 Joule in 20 ms
• Less than 1500A maximum current for 1-4 mOhm material
• Short circuit current below 2000A
• Able to weld 10 consecutive welds in 10 seconds

I first tried your idea of BCAP0310:
http://homit.nl/EEVblog/screenshot.130.jpg

The graphs:
http://homit.nl/EEVblog/screenshot.129.jpg

Some comments:

• The circuit "Charge_Cntrl" is an additional MOSfet that sees to it that the power supply is not connected when a pulse is fired.
• The CC_PSU is one of those $20 12V/54A server supplies that can be adjusted to 10.8V
• The circuit "Weld_Cntrl" is your electronics
• Do check the parameters, their might still be a mistake in there

The result would be, if no mistakes made , that with 4xBCAP0310 app. 40 Joule welds can be made with 2 mOhm material in 25 ms. The voltages are on the low side for the circuit so it might be necessary to use a Schottky diode and a larger capacitor on the supply of the electronics.
I also did the simulation for 4x BCAP0650: with increase of the wires to 70cm each, all conditions are met and a 75 Joule pulse is possible in 25 ms.

Looks promising. Shouldn't it be a CC/CV source for charging the caps?

We most likely need to balance the caps or individually manage their voltage, with something like a BMS. For example charge them individually, like how it's done with lipo packs.

[SvanGool]

@Tatus1969:
I forgot to include the 300A fuse (which is also for capacitors still a good idea) in the internal resistance parameters, it is estimated to have a DC resistance of 375 uOhm, as far as I can find information about that. But I don't see it back in your resistance estimates either or is it already included in one of the parameters?

@Kjelt:
I also simulated 4x BCAP3000s (which I own myself also), they need at least 2x 1 meter AWG8 to not pass the current limits and are then capable of supplying 100 Joule in 25 ms. So they are real "beasts", but a lot of power is "burned" in the cabling. If a higher voltage than 10.8V is required, this spot-welder would be unsuitable for more than 4x BCAP3000, at this moment, for the currents produced.

@SirJMD:
You are correct, the power supply should be the "CC first then CV" type of power supply, that is also what the circuit is simulating.
I didn't simulate a balancer, that would normally operate in parallel of the capacitors shown. But yes, I would at least recommend a "top balancer"/protection for these capacitors. It would need quite some "MOSFet switching" to be able to charge the capacitors individually, I think the current proposed solution with the CC/CV power supply with one additional MOSFet would be a more practical solution.

[tatus1969]

I tried to make a simulation in LTSpice for your Spotwelder based on capacitors.

Thanks a lot for that, I'll check carefully and give feedback. Can you provide the simulation files?

Your idea to take a used server PSU is brilliant. Do you know which models are the most common? It seems that almost every model and brand uses a different output connector.

I forgot to include the 300A fuse (which is also for capacitors still a good idea) in the internal resistance parameters, it is estimated to have a DC resistance of 375 uOhm, as far as I can find information about that. But I don't see it back in your resistance estimates either or is it already included in one of the parameters?

I was busy with the operation manual during the last days, and a precision shunt resistor just arrived. I will use that to provide more accurate figures. I already know that I was off by ~10% with my crude current measurement by using a section of AWG8 as a shunt. The cable resistance appears to be higher than expected.

I also simulated 4x BCAP3000s (which I own myself also), they need at least 2x 1 meter AWG8 to not pass the current limits and are then capable of supplying 100 Joule in 25 ms. So they are real "beasts", but a lot of power is "burned" in the cabling. If a higher voltage than 10.8V is required, this spot-welder would be unsuitable for more than 4x BCAP3000, at this moment, for the currents produced.

I included a section in the operation manual that discusses the inductive kickback situation. As mentioned there, I'll try to find the absolute limit through experiments. For now, I am using an estimation (2J is the energy that the TVS diode can take), and that already suggests that the current needs to be limited further when going to long cable lengths. And it would be wise to control the short circuit current (0Ohm weld spot resistance) to stay on the safe side.

[anishkgt]

A server psu such as https://www.ebay.com/itm/121957144997would help.

I use two of these similar ones to power my lipo charger - iCharger 406DUO


Www.Georgehobby.wordpress.com

Equipments: DSO104z, Hakko FX888D

[SvanGool]

Can you provide the simulation files?

Attached.
I used a standard free LTSpiceIV / LTSpiceXVII version, without any modification (except for changing the options for "background" colors, enabling grid and increasing line thickness for plots).

I modified the "Charge_Cntrl" pulse-timing, because there was an error in there (increased number of pulses) and I modified that timing to be always fitting, using the CV behavior of the power supply (both modifications are corrected in the screenshots of my earlier post).
Here is a detailed picture showing the charging between the two first welding pulses:

Do you know which models are the most common?

I purchased a second-hand HP DPS-600PB to test, it can be modified to adjust voltage between 8-13.8v.

I was busy with the operation manual during the next days, and a precision shunt resistor just arrived. I will use that to provide more accurate figures. I already know that I was off by ~10% with my crude current measurement by using a section of AWG8 as a shunt. The cable resistance appears to be higher than expected.
I included a section in the operation manual that discusses the inductive kickback situation. As mentioned there, I'll try to find the absolute limit through experiments. For now, I am using an estimation (2J is the energy that the TVS diode can take), and that already suggests that the current needs to be limited further when going to long cable lengths. And it would be wise to control the short circuit current (0Ohm weld spot resistance) to stay on the safe side.

When you have more information about the parameters and inductive kickback, I can adjust the model to have a better match with the real-life behavior 

A server psu such as https://www.ebay.com/itm/121957144997would help.
I use two of these similar ones to power my lipo charger - iCharger 406DUO

Looking at the price, that indeed might be a good candidate, do you know what the CC/CV behavior is and if the voltage is adjustable?

[anishkgt]

Nope I've not tested it to that extreme.  But I can if one can guide me.


Www.Georgehobby.wordpress.com

Equipments: DSO104z, Hakko FX888D

[plazma]

I bought some Sony 18650 cells for my RC car. The specs got me thinking if they would be good for this application. Max. 100A per cell for <6s. 35A max. continuous. The model is Sony US18650VTCA. In a 2s10p setup it would give max. 1000A 7.4V.

[anishkgt]

Oh don't bother 18650 cells. You'd need a lot of them compared to just one lipo.


Www.Georgehobby.wordpress.com

Equipments: DSO104z, Hakko FX888D

[plazma]

Oh don't bother 18650 cells. You'd need a lot of them compared to just one lipo.


Www.Georgehobby.wordpress.com

Equipments: DSO104z, Hakko FX888D

18650 cells can have very different chemistries and current ratings. Show me a similar volume LiPo with better current rating than the Sony cell.

[anishkgt]

Well I was talking about the size to capacity ratio. They can have different chemistries but the burst amps is what makes the difference. A lipo is good at it.


Www.Georgehobby.wordpress.com

Equipments: DSO104z, Hakko FX888D

[plazma]

Looks like my first message got truncated to half. The Sony cell can give 35A continuous and 100A max. for 6s. A 2s10p pack would give 1000A 7.2V max. I also wrote the cell type Sony US18650VTC5A.

[SvanGool]

Nope I've not tested it to that extreme.  But I can if one can guide me.

You would need a normal large electrolytic capacitor of e.g. 22000uF/16V, like http://www.mouser.com/ProductDetail/Cornell-Dubilier/229TTA016M and a storage oscilloscope to show you a few milliseconds when you connect the power supply you proposed (rating 12V/100A). Connect the oscilloscope to the capacitor and set it to trigger just above 0V and then connect the power-supply to the capacitor (matching polarity of PSU and cap). Do you use thick wiring between PSU and cap.

The picture on the oscilloscope should look similar to this, when your power supply would operate, as desired, for ultra-capacitor usage: