kWeld - "Next level" DIY battery spot welder

[Fraser]

Frank,

What a very kind offer.

I personally will happily pay for the parts.

PM on its way

Best Wishes

Fraser

[Leo Bodnar]

This is one funny looking j-link.  Did Segger change the design recently?
Leo

That is an older version I got long ago from ebay. Hopefully not a clone?

Looks totally legit.

[eKretz]

Pre-orders are already all sold out, eh?

[tatus1969]

Pre-orders are already all sold out, eh?

I apparently underestimated the demand, and ran out of stock by just pre-orders in less than a week. I plan another batch right after having received the current one. I'll inform again through the newsletter. But in order help me estimate the demand, I have just implemented an "inform me on availability" button for the sold out products that puts you on a waiting list when you click it.

[Derriell]

What is the recommended internal resistance of a 2S LiPo battery pack for max 1500A?

Is my following calculation correct?

Given 2S LiPo, lets say at 4V per cell, giving room for 2*1V voltage decrease during load in order not to undervolt the cells.

2V/1500A = 1.3mOhm for the whole pack  => 0.66mOhm per cell

Or do the LiPos behave differently than a normal resistor under load?

Thank you

[tatus1969]

What is the recommended internal resistance of a 2S LiPo battery pack for max 1500A?

Is my following calculation correct?

Given 2S LiPo, lets say at 4V per cell, giving room for 2*1V voltage decrease during load in order not to undervolt the cells.

2V/1500A = 1.3mOhm for the whole pack  => 0.66mOhm per cell

Or do the LiPos behave differently than a normal resistor under load?

Thank you

The calculation goes a different way. You calculate the current that the battery delivers when it is being short circuited by the welder. Welder and (stock) cables contribute approx 2.4 milliOhm, a typical resistance of the weld spot itself is 1 milliOhm. current = lipo_pack_open_circuit_voltage / (3.4 milliOhm + lipo_pack_resistance).

You don't need to worry about undervolting the battery when using it like this, that voltage is related to the chemistry and not to the output voltage under load. You should respect the output current capacity of the pack though, as standard 50C packs get damaged from the high current. So far only the ultra high current packs like Turnigy nanotech or graphite have been tested to be considerably robust. But respect the three golden Lipo rules anyway: a) only operate them with great caution, b) store them in a fire safe place when not using them, and c) charge them under supervision only.

[Derriell]

Ah yes, thank you! This means, if I want 1500A I need a 2S battery  pack (8V) with:

8V/1500A - 3.4mOhm = ~ 2mOhm

[tatus1969]

Ah yes, thank you! This means, if I want 1500A I need a 2S battery  pack (8V) with:

8V/1500A - 3.4mOhm = ~ 2mOhm

exactly!

[Derriell]

A 2S6P pack with the new Samsung SDI INR21700-30T cells with just 10mOhm is good for around 1000A welding power. I find them even more interesting to make nice and power full battery packs  If you get them for 5$ per cell that makes it 60$.

Test:
http://www.dampfakkus.de/akkutest.php?id=673
https://www.e-cigarette-forum.com/threads/samsung-inr21700-30t-35a-3000mah-21700-bench-test-results-an-incredible-40a-3100mah-battery.798683/

I'm not sure about the power source yet though. I might as well just use several 3S RC-LiPos with 4mm gold connector in parallel for the welding. When I weld I can't fly anyway  4-6 3S packs should do the jobs just fine with an adapter cable. But a problem with this seems to be that every time I want to weld I need to charge the batteries to the same voltage and if I want to store them later I need to discharge them a bit.

[tatus1969]

A 2S6P pack with the new Samsung SDI INR21700-30T cells with just 10mOhm is good for around 1000A welding power. I find them even more interesting to make nice and power full battery packs  If you get them for 5$ per cell that makes it 60$.

Test:
http://www.dampfakkus.de/akkutest.php?id=673
https://www.e-cigarette-forum.com/threads/samsung-inr21700-30t-35a-3000mah-21700-bench-test-results-an-incredible-40a-3100mah-battery.798683/

I'm not sure about the power source yet though. I might as well just use several 3S RC-LiPos with 4mm gold connector in parallel for the welding. When I weld I can't fly anyway  4-6 3S packs should do the jobs just fine with an adapter cable. But a problem with this seems to be that every time I want to weld I need to charge the batteries to the same voltage and if I want to store them later I need to discharge them a bit.

In the calculations, remember to take the total resistance of the welder into account as well. For kWeld, that is around 2.5mOhms. And the weld spot itself accounts for another 0-2mOhms.

[tatus1969]

I'm happy to announce the release of kWeld firmware rev2.0. It is backwards compatible with the existing rev2 modules as well the new rev3 ones that are on the way to me.

New features:
- added support for hardware revision 3
- added support for rotary encoder
- added configuration menu
- current limit is now calculated from adjustable cable length
- made auto mode firing delay adjustable
- added adjustable battery undervoltage detection
- improved open circuit detection when welding

The binary can be downloaded from this page:
https://www.kicksurfer.de/index.php/kweld/

For those of you who already have the firmware update tool in hands: I'm sorry but I haven't found the time to include the  update procedure in the operating manual yet. I am busy shipping out the current production batch by the end of this week, and I will update the manuals thereafter.

For now, I only have this short description, if something is not obvious please let me know:
1. Connect the small adapter board to the USB adapter so that the connectors on the adapter are facing up, and the LEDs on the USB adapter are visible (i.e. "natural" position, components on both boards facing up)
2. Connect the assembly to kWeld with the short cable (goes out to the right), use a mini USB cable (not included) to connect to PC, if necessary install drivers (http://www.ftdichip.com/Drivers/VCP.htm)
3. Find the new serial port via Windows Device Manager
4. In extraputty (http://www.extraputty.com/), establish a serial connection with this COM port and 115200 / 8N1 / no handshake
5. Power up kWeld, then type in a small "k" in the console
6. Bootloader must respond now
7. Now type a small "y"
8. Now make a YMODEM transfer with the firmware file via the extraputty menu (Files Transfer -> YMODEM -> Send)
9. Wait for the transfer to complete, it should then read an "OK"
10. Now trigger kWeld restart by typing a small "a"

For those who have added the encoder to their kWeld rev2: you need to type "v3" into the console when kWeld is running normal (not in firmware update mode). (There will be no echo.) This enables encoder use. Typing "v2" instead switches back to potentiometer mode. The firmware does a reset in each case and restarts.

Here is a short description of the new configuration menu:
1. for rev2 (potentiometer) systems, turn dial all the way left and press trigger switch to enter the menu // for rev3 (encoder) systems, push the dial know instead
2. turn the knob to choose from one of the available options:
"CAL": enter calibration function (as known)
"Mode": select between manual and auto trigger modes
"TrgDelay": [default 0.5s] adjust trigger delay in auto mode
"DefaultE": [default 10.0J] adjust set energy level after power-on (only when using encoder)
"CableLen": [default 1.0m] enter the total length of the power wiring (add up all conductor segments), the firmware calculates the correct current limit from that (using the inductive kickback formula from the operating manual). This provides correct protection in case you are using longer wires than standard
"LoBatt": [default 0.0V] sets battery warning voltage level (set to 0 if you don't want to use it)
"Exit": leave the menu

[Fraser]

Excellent work Frank 

I now have a huge incentive to find time to construct my KWeld and upgrade it to V3 

A very big 'Thank You'

Fraser

[tatus1969]

Excellent work Frank 

I now have a huge incentive to find time to construct my KWeld and upgrade it to V3 

A very big 'Thank You'

Fraser

you're welcome :-) Did you receive your upgrade kit already?

[Fraser]

Hi Frank,

Yes it has just arrived safely and was added to my KWeld project parts box Thank you 

Fraser

[TopQuark]

Hi Frank, newbie here I have looked at the schematics you've posted and are very curious about how you measure current through the welder. I see your circuit measuring voltage at the input, out+ and out-. The input fuse and the parallel mosfets does have resistance and allows current to be derived, but are those resistance alone precise and stable enough to be used as a shunt? Or do you measure current with other clever methods? Hope this is not too much of a commercial secret for you to share. 

Alex

[anishkgt]

He has answered it earlier for the same question I had asked. Scroll to page from 1-6 or something if I can recall.


Www.Georgehobby.wordpress.com

Equipments: DSO104z, Hakko FX888D, Brymen BM869s

[TopQuark]

Not sure I got it correctly or not. So the kWeld determines current draw by measuring voltage drop across the welding cables and electrodes? Does the welder rely on the length and thickness of welding cables used to be the same as those provided by Frank? I apologize if I am asking too much.

To be really honest, I am trying to replicate one for myself to learn electronics. Being a undergraduate freshman who don't have a major yet, I definitely dare not sell my potential product to others, I might even blow up my dorm room in the process of trying to build one Also I am too broke to buy the real deal from Frank.

[tatus1969]

Not sure I got it correctly or not. So the kWeld determines current draw by measuring voltage drop across the welding cables and electrodes? Does the welder rely on the length and thickness of welding cables used to be the same as those provided by Frank? I apologize if I am asking too much.

To be really honest, I am trying to replicate one for myself to learn electronics. Being a undergraduate freshman who don't have a major yet, I definitely dare not sell my potential product to others, I might even blow up my dorm room in the process of trying to build one Also I am too broke to buy the real deal from Frank.

kWeld is measuring current through the on-state resistance of the MOSFETs, using them as a shunt. This is not very accurate as you suspected, you end up with something in the order of +-30%. And there is quite a temperature coefficient. But absolute accuracy is not that important here, the idea is to have repeatable results. The user will adjust the pulse energy to his needs anyway. The temperature coefficient is more important, and that is one of the reasons why I have chosen the most powerful MOSFETs available (at quite a cost), because they do not heat up much. Even during my torture tests, I never managed to heat them above 50°C.

[anishkgt]

Well if you want accurate current measurement, one way would be to measure the resistance of the cable via 4 wire method and measure voltage on portion of the cable like 10cm apart then use ohms law to calculate current.

You could add these to the code also.
Hope this helps.


Www.Georgehobby.wordpress.com

Equipments: DSO104z, Hakko FX888D, Brymen BM869s

[Fraser]

On the matter of current measurement accuracy, I concur with Frank.

It is not necessary to have absolute accuracy of current measurement in a spot welder as there are many variables that can effect such, including the performance and contact area of the probes combined with battery tag strips. What is important is repeatability of weld. Franks design is excellent in this respect because of his novel approach of using Energy measurement rather than just a timer, as found in some other designs.

It would be possible to create a highly accurate current measurement for a spot welder, but one has to ask, is it needed and what advantages does it offer to offset the added complexity of the design ? I would argue that Franks approach is both novel and relatively minimalist for the required task (the KISS principle). The operating principle is well thought through and produces the desired repeatable results    A far better design than my £80 Chinese spot welder that just uses a transformer and timer controlled triac to set the weld current duration.

Fraser

[tatus1969]

Well if you want accurate current measurement, one way would be to measure the resistance of the cable via 4 wire method and measure voltage on portion of the cable like 10cm apart then use ohms law to calculate current.

You could add these to the code also.
Hope this helps.


Www.Georgehobby.wordpress.com

Equipments: DSO104z, Hakko FX888D, Brymen BM869s

The tempco of copper is a in the same order as that of a MOSFET (4m / K), but the wiring also heats up much more. So there is no advantage with that respect. The absolute accuracy could be better, but at additional cost and mechanical complexity. This is why i decided against that in the beginning. I also thought of adding a dedicated shunt resistor, but didn't like the additional cost and complexity either.

[anishkgt]

I wouldn’t argue about franks design but just sharing idea for current measurement.


Www.Georgehobby.wordpress.com

Equipments: DSO104z, Hakko FX888D, Brymen BM869s

[anishkgt]

The tempco of copper is a in the same order as that of a MOSFET (4m / K), but the wiring also heats up much more. So there is no advantage with that respect. The absolute accuracy could be better, but at additional cost and mechanical complexity. This is why i decided against that in the beginning. I also thought of adding a dedicated shunt resistor, but didn't like the additional cost and complexity either.

Yes a bit of work too but reading voltage “ 10cm apart” is like using a shunt. Since you also provide the cable too the resistance would be a constant but as mentioned by others “how important is current accuracy?”


Www.Georgehobby.wordpress.com

Equipments: DSO104z, Hakko FX888D, Brymen BM869s

[tatus1969]

the resistance would be a constant

No it's not. It is copper, which has a huge temperature coefficient 

[anishkgt]

Hmmm yea. Oooopsy[emoji16] again that wouldn’t be accurate but would give an idea of it.


Www.Georgehobby.wordpress.com

Equipments: DSO104z, Hakko FX888D, Brymen BM869s