 |
 | DIY Capacitive Discharge Battery Spot Welder |  | Ian Hooper, 6 January 2012 | |
Many batteries come in a cylindrical format with no inherent means of interconnection (such as screw terminals). Two common examples are the 18650 and 26650 sizes of lithium cell. The best way to connect such cells together permanently is to spot weld nickel strips between the terminals. (Soldering to battery terminals is not recommended due to the amount of heat that gets into the battery - potentially damaging the internal chemistry.)
A common method for doing this is with Capacitive Discharge (CD) spot welding, which basically involves dumping a pulse of energy stored in a capacitor through the nickel strip, causing localised melting of the nickel and welding it to the battery. (Critically, with battery spot welding the weld current must pass through two points on the same side of the battery terminal - not through the battery!)
Commercial CD welders typically cost several thousand dollars. But with a little ingenuity, you can build your own for a fraction of the cost. Here are a couple of examples of people who did it before me (and even more economically!)
Basically you just need a large bank of capacitors, a method of charging them up, and a method of discharging rapidly through heavy duty cabling and electrodes into the nickel strip. |
Example of a spot-welded battery pack
Close-up of spot welds
|
With a quantity of K2 26650EV cells sitting in my workshop, orphans from a cancelled project, it seemed like a good idea to build myself a CD welder such that I could assembly these cells into packs suitable for use in my various electric vehicle projects.
Completed CD spot welder |
My completed CD spot welder is pictured to the left. A box for it might be nice, but the naked electronics has it's own charm too. I won't go into explicit detail about the design because everyone is likely to do things a little differently, but will review my own component choices.
The heart of the welder is a bank of six Maxwell Ultracapacitors , each rated to a whopping 650 Farads, at 2.7V. Running six in series results in 108 Farads at up to ~16V. These capacitors feature very low internal resistance, so can deliver very high discharge current. Based on research, I learned that a typical battery spot weld requires approximately 200 Joules of energy. With this figure you can work out how much voltage and capacitance you will need for each weld, based on the formula for energy stored in a capacitor:
E = 1/2 CV² |
Hence at 15V, I would need about 1.8 Farad per weld. Clearly 108F is much more than necessary, and unlike the other designs above I need to be able to start and stop my pulse (not discharge the entire capacitor energy in a single weld), but having such a large capacitance does allow for many welds in quick succession, if desired. (I got these capacitors cheap off a friend. If I had to buy new, I'd probably opt for something smaller.)
Right is a labelled diagram showing the layout and control board on top (click to view larger).
The pulse timer is based on an Atmel ATtiny13A microcontroller. The same function could also be done with something like a 555 timer, but I like the precision of a microcontroller, and it actually involves a lower total part count. This fires an IXYS IXDD4141PI gate driver, to switch a bank of MOSFETs.
I also installed a 100ohm bypass resistor to slowly discharge the capacitors. It's a bit of a waste of power while running, but avoids any surprises lest you come back a day later and the capacitors are still charged. |
|
 |
To switch the pulse current on and off, I'm using four IXYS IXFX120N20 MOSFETs, pictured left. They're actually part of an old motor controll assembly I had lying around. They are rated to 120A continuous or 480A peak current each, so blowing them up is fairly unlikely. I imagine it's overkill, but sometimes it's nice to have a bit of safety margin.
All the high power wiring is 8AWG, which has a resistance of around 2mohm per metre. Including electrode leads, I have about a metre worth in the welding circuit. The electrodes are made from solid copper rod with tapered ends. |
To charge the capacitors and power the logic circuit, I use a dual-output laboratory power supply. Separate power supplies for capacitors and logic circuit must be used because the ultracaps charge very slowly (about a minute to reach welding voltage), and the AVRs don't particularly like such a slow ramp up of supply voltage.
As can be seen from the pictures at the top, the welds are quite small but hold on tenaciously and should be good for about 10A per spot. A good way to verify weld penetration (on a test weld) is to try tearing off the nickel strip. It should leave part behind on the battery, and a hole in the nickel strip where the weld occurred.
|
|
Nickel strip can be purchased from Powerstream or Sunstone Engineering. For those outside the United States, shipping is pretty expensive (if available at all). I'd be interested to hear from anyone who knows of a good source of nickel strip in Australia. And feel free to contact me if you have any requirement for battery pack assembly!
~Ian Hooper, 6 January 2012
June 2012 Update: Schematics, Repairs, Upgrades and Musings
Schematics: Firstly, I've had a few requests for the schematics, so here they are (click for enlargements, or here for the Eagle file). On the left is the logic board, and on the right the power board (or, one of the MOSFET/diode pairs - need many in parallel). These are reverse engineered from the real thing so I hope I didn't forget anything! The power board layout is pretty much as per my most recent controller designs. Not something you can make on normal breadboard unfortunately due to the power involved! But mounting the MOSFETs on a heatsink and soldering thick wires directly to the legs is probably the easiest option.
|
|
|
I've also uploaded a copy of the microcontroller C code. It's pretty straight forward. Grab a copy here.
Repairs and Upgrades: The original power stage actually blew a MOSFET recently, after perhaps a few thousand welds. Fixing/upgrading it was quite a learning experienced, speckled with several other blown MOSFETs.
When one of the IXFX120N20s blew, I replaced the lot with IRFP2907s, a lower voltage FET with much lower on-resistance (about 5mohm instead of 20mohm). I also replaced the thin copper interconnectors between ultracapacitors with proper, clean busbars. It seemed like a good idea, but without realising the consequence at the time, I had greatly reduced the overall resistance of the weld circuit, which greatly increased weld current, and blew the new MOSFETs!
Turns out the cumulative resistance of the equipment is kind of relevant in getting the correct weld current. (In hindsight, I could have just welded at a much lower voltage to keep current lower.) Anyway, not having any more IRFP2907s on hand, I did another rebuild using IRFP4668 FETs, and upgraded to six instead of four devices in parallel, to give a bit more headroom. So far so good. I now weld at about 11V instead of 13V for the same result, since the lower resistance (compared with IXFX120N20) means less voltage is needed for the same current, and the heatsink stays a lot cooler.
As for weld times and energy, using a 10A power supply, it takes about 10 seconds to recharge to ~10V between each weld (after dropping about a volt). So, about 1000 joules of energy per weld. I have read that the spot weld itself needs about 200 joules - not sure if that's per spot or for a pair, but at any rate it's clear that the equipment is soaking up most of the energy rather than delivering it to the weld. |
|
Musings: I think if I built another spot welder, I'd go with 1 Farad of normal capacitors instead of ultracapacitors. The ultras are intimidatingly powerful. I worry what would happen if my timer circuit failed and the capacitors discharged completely! I imagine a shower of sparks, and/or things getting burned. Also the ultracapacitors don't really speed up welding because for the best welds you still have to wait for the power supply to top them up to optimum voltage each time - so it's P/S limited. May as well be using a capacitor bank with a single weld charge.
I did a bit of research on the options for a 1F high power capacitor bank, and for my money I'd go with 20x Panasonic 16V 47000uF. Digikey has them for about $5ea, so quite a bit cheaper than the ultracapacitor route. And, probably safer. As for the car stereo capacitors others have used.. I don't trust the cheaper ones to be either up to spec in capacitance or ESR. You need a lot of amps for good spot welds (ballpark a thousand, by my estimation), and their internal resistance may be too high (depending on brand, I'm sure - YMMV).
And a final note to anyone else out there building your own CD spot welder - take care! There are some pretty serious power levels involved which can make a mess if things aren't working right.
Comments| | Andrew on 19th Jun 2012
Can you please post a schematic of your control board assembly? Also, the part number for the gate drive should be IXDD414PI. Thanks for the great write up on your CD welder. | | |  | | | Ian Hooper on 22nd Jun 2012
Done! See update section above. |
|
| | Alan on 16th Jul 2012
I have 6 pieces of 47000uF 80V computer grade capacitors. Will these capacitors sufficient for welding battery tab?
As I do not have facility to compile the C code, could you please upload a copy of the 'HEX' code?
Thank you for the project. | | | | | | Ian Hooper on 17th Jul 2012
I'm not sure.. In parallel all six would only give you 0.28F which is a quarter of what most people use when welding at 12V. If you worked at say 50V the caps could hold enough energy to weld, but it'd be a much shorter pulse at much higher current (several thousand amps). And higher current is harder to control/switch, of course. |
|
| | Penny Imes on 28th Jul 2012
This is precisely the important information I'd been searching for. Incredible blog. Very inspirational! Your posts are so good and also detailed. The links you come with are also very beneficial as well. Many thanks :) |
| | Sylvia on 19th Aug 2012
Putting the ultra caps in series raises the problem that they might not be well matched in capacity. The nominal capacitance is only the initial mininimum, so the voltages across the capacitors will typically vary, leading to the possibility that one reaches its limit while others are still charging to bring the total voltage to the level intended.
| | | | | | Ian Hooper on 19th Aug 2012
Hi Sylvia, you are correct! (More info for the benefit of others: capacitor voltage is proportional to its charge divided by capacitance according to V = Q/C, and while series capacitors will necessarily all have the same charge in/out, variations in capacitance will cause variations in voltage as they charge up.) The easy solution is probably to monitor all the individual capacitor voltages while they're charging up and make sure none are exceeding their voltage limit. Generally you won't need any of them to approach max voltage. (The ones I'm using are 2.7V each so 16.2V theoretical max, but I only needed to charge them to 10-12V for welding; plenty of headroom for variations in voltage.) |
|
| | azzythehillbilly on 27th Dec 2012
Very nice blog. You have done agreat job. |
| | azzythehillbilly on 27th Dec 2012
I didn't understand the circuit well. ( too dumb) I tried discharging some capacitors through a dead short once. The inductance of the shorting bar caused the capacitors to be reverse charged almost to the smae voltage. With long leads and thus hugher inductance for sure. I don't understand how you prevented that from happening as I understand that the electrolytics are not too happy if reverse charged. Is there a diode across the caps to prevent that? | | | | | | Ian Hooper on 27th Dec 2012
Hi Azzy, hmm I'm surprised to hear you observed reverse voltages happening on the capacitor. I suppose it depends on the ratio of capacitance to inductance. A diode in reverse parallel with the capacitor would probably protect it, but shouldn't be necessary.. With CD spot welding, the capacitance should be far higher than the inductance (so the RLC circuit is "overdamped" and the caps don't go below zero). I guess my suggestion would be to try a larger capacitance, and maybe lower voltage? |
|
| | Mark on 2nd Jan 2013
If multiple ultra capacitors in series produce too much power that is difficult to control then maybe we can build a spot welder with only 1 ultra capacitor + step down charger ??? | | | | | | Ian Hooper on 2nd Jan 2013
Hi Mark, the snag would be the lower voltage, as most ultracaps are only rated to a few volts which probably couldn't push enough current through the circuit for a good weld. Typically the total circuit for CD spot welding may have say 10mΩ resistance (including capacitor internal resistance), which then needs 10V to achieve 1000A weld current (according to ohm's law, V=IR). 3V would only get 300A which may not be enough for a good weld. Just ballpark figures but you get the idea. Perhaps the same number of ultracaps but 1/6th the capacitance would be fine though, and should be cheaper than the ones I used. |
| | Mark on 2nd Jan 2013
thank you for fast response.
I found the specs for Maxwell BCAP3000 P270, and it seems that maximum peak current (1 second) is 2200A and 0.29 mΩ resistance. |
| | Mark on 2nd Jan 2013
I already have the Maxwell caps and I would like to build a very simple spot welder like Robert Thompson did, cos being honest I'm not sure if I could handle something more complicated. Do you think I have any chance to succeed? If yes what SRC do i need, 50A 100A, 200A? |
| | Ian Hooper on 2nd Jan 2013
That capacitor internal resistance should work well, but it's hard to keep the rest of the circuit low resistance, e.g 1 metre of 8 gauge wire for the probes has about 2mΩ resistance, and each connection adds a bit, etc.
As for SCRs, the problem is that you can't turn them off once they start conducting, the current has to stop flowing by itself, so would have no choice but to discharge the ultracap entirely - which is a lot of energy! Hence why I used MOSFETs, which can be turned off on command and do a timed pulse, whereas R.Thompson and P.Pemberton's designs have smaller capacitance which discharges fully so the SCR can turn off. Basically, unfortunately I can't see a way to use ultracaps with an SCR switch. Easiest option may be to build or buy a 1 Farad capacitor as per their designs! |
|
| | electric diagram on 4th Jan 2013
Can i have the electric diagram and the bill of the component of this spot welder? Thanks
can you send it to my email address xxxxxx@libero.it | | | | | | Ian Hooper on 5th Jan 2013
Hi, the electric diagram (including part codes) and AVR C code can be found in the "June 2012" update section. I don't have a bill of materials for the build because it was just a one-off. The hardest part to duplicate would probably be the power stage as you probably won't have suitable PCBs on hand, but can just solder large wires straight to the legs of the MOSFETs. |
|
| | John on 23rd Mar 2013
Hi! I have eight BCAP1500 P270 Maxwell Ultracapacitors! How many capacitors should I use for battery pack welding? If parameters for each are 1500F and 2.7V. I thought 4 capacitors may work fine if delivered energy as follows is calculated: 2.7V*4=10.8V; 1500F/4=375F and in the end E = 1/2 CV˛ = 0.5*375F*10.8^2= 21 870J. Is total voltage and capatiance acceptable? What is your opinion? Ty already! | | | | | | Ian Hooper on 24th Mar 2013
Hi John, they would have plenty of energy and low enough resistance for the job, but you may need more than four for sufficient voltage.. I usually weld between about 10-12V, but it depends quite a bit on the resistance of the entire welding circuit (caps, probe wires, connections, MOSFETs, and the thing you're welding). Six might be a better number to work with so you can crank the voltage up a little if necessary. (As discussed with Sylvia in an earlier comment, because the capacitors vary by ~10% or so, you need to only charge to 10% lower than the theoretical maximum to ensure none get damaged.) |
|
| | Lionelb on 23rd Mar 2013
Have you tried stud welding with say 2 or 3mm dia. studs ?
I'm looking for suitable studs and a stud-holder/gun. | | | | | | Ian Hooper on 24th Mar 2013
Hi Lionel, I haven't tried it with stud welding, no. As such I'm not really familiar with how much weld current or energy is required for that! But I suspect it'd be a fair bit more than we use for welding 0.01" nickel strips onto batteries.. |
|
| | bombastinator on 18th Apr 2013
I am interested in something even smaller:
I need only to weld very small wires together. 32-28awg. Specifically kanthal a to any old non-resistance wire of similar diameter.
I should be able to power the thing with nothing more than an AA battery I think. Do you have any suggestions on how to go about figuring out how to do it?
| | | | | | Ian Hooper on 19th Apr 2013
Not too sure about this one sorry! You might still need 10V to get the current flow but probably a much smaller charge/capacitance for a smaller weld. I'm not sure a AA battery would be able to put out the current required, but there's one way to find out! |
|
Please note that due to spam problems all comments are moderated
and comments advertising other websites are automatically rejected!
|
