e -Spot : DIY Spot Welder
When it comes to making 18650 cell packs, it seems to me that there remains some debate as to whether to solder or spot weld. For my part I believe heating a cell (by soldering) is NOT good for its immediate health or future longevity - no matter the size / temperature of the iron used and how short one can make the heating period.
I looked at buying a suitable spot welder but couldn't find one that I felt would be flexible enough whilst being reasonably priced... In the end I felt I could probably make one that, whilst not necessarily being 'better' than a purchased item, would suit me and my perceived use of it better ....
In the end I adapted an old microwave transformer, added a SSR on/off switch, a programmable timer to control pulse duration and various other parts to make the final item flexible enough and useable for me...
This is what I came up with and how I did it :
Initially, my thought process went along the lines of a car battery, a starter solenoid, a momentary switch and some connecting wire and nails as the welding tips/electrodes. Yeah, I know... but I had to start somewhere ! That idea didn't even get passed the "what would the final 'thing' be like to use and what would it look like ?" phase !
I'd have to charge the battery before I could even use it (and keep it charged) ...and, a car battery, being a car battery, would ensure any final 'solution' would always be inelegant, ugly, ungainly, probably awkward to setup, not to mention heavy to handle and large to store someplace between uses...
I then thought of a smaller (Li-Po ? perhaps) battery... that would still be able to deliver large intermittent / pulsed currents without suffering unduly... but then how to regulate the 'on' / weld time so that constant results could be maintained ? and the battery would still need to be charged / managed...
As always, I never want my primary project (whatever that is / would be at the time) to get hijacked by a different project to make a piece of kit necessary for the former. That's a great way to not achieve much and ensure the primary project never gets finished !
During my apprenticeship we'd learned and played about with hi-power radio transformers (they could really bite ! ). I remembered how we'd been able to change a transformer's output by changing the number of turns and wire gauge of the secondary winding... which got me to thinking and Googling...
To be fair, I found quite a few people who had had the same idea and put old microwave transformers to good use. Thinking about things further, it would not have to be a microwave transformer, just a relatively 'chunky' transformer with independent primary and secondary coils (like the one I remembered from my apprenticeship). Obviously, any transformer would have to be rated at say 1500 / 2000watts BUT if I could remove the secondary winding (without damaging the primary one) I should be able to replace it with something suitable for spot welding 18650 cells.
I figured I'd need an output of around 2 or 3Volts at somewhere in the region of 250 to 400Amps. Some basic math, based upon V=IxR and Power=IxV (where V=Volts, I=Amps and R = Resistance), indicated that my
'starter for ten' would be something like :
transformer power capability (stated on the box / unit I'd scavenge it from) : 1500W
desired output voltage (depending upon the number of windings) was to be, say : 3V
desired output current (depending upon the cross-sectional area of my cable) was to be, say : 250A
The Power = IxV function produces I(250amps) x V(3volts) = 750watts; in an ideal world (optimistically assuming a 100% efficiency); around 50% of the transformer rating... hmmm... seemed perfectly feasible... (in reality, transformers typically have a >95% efficiency rating at designated - full load).
NOTE : These were really basic / ballpark figures; no more than very simple / base 'expectations'. I didn't, ideally, want the voltage to be higher than any individual battery cell Voltage and the 'required' current would really depend upon the weight / gauge / thickness of metal I would be trying to spot weld and how quickly that metal would dissipate the heat I intended to produce with the power (function of I x V) that I would be passing between the two electrodes and the duration of the weld period.
To the tip / dump / decheterie...
I found an old washing machine at the local tip and looked at the transformer. It appeared to be an integrated type (both windings wired together). Not much good because I needed to be able to separate the two relatively easily. I also found a cast-off microwave (well, there was a choice of at least a dozen !). 10 minutes later, with the case removed, I noted this particular unit DID have the separate windings I wanted. So... back to the workshop and, an hour or so later, I had myself a potential donor transformer.
Another hour or so after that and, with judicious use of a hacksaw, heavy screwdriver and hammer, I found myself with a primary-only wound transformer - the original secondary winding now being in pieces on the floor and in the bin. No accidental / unintended damage done, it was time to play about with a suitable secondary.
Parts / Components
The transformer, SSR (solid State Relay, 12V Fan, Timer board, Connectors (later replaced with M8 machine screws & nuts (to reduce heat / losses), IEC mains input Socket, Timer board, On/Off Switch, 3D printed buttons (for timer board), 12V PSU (runs the Fan and the Timer board)...
I also needed various lengths of cable, some heatshrink and some fast-on connectors, plus the 3D printed Enclosure
I turned to Sketchup to design my enclosure. The transformer is pretty heavy and would, therefore, need a pretty strong design and probably a cooling fan (for those larger battery packs I foresaw myself making). I came up with a split unit (top and bottom) where all components are mounted in the, well-ventilated, bottom half (including display and 3D printed timer configuration switches) and the top is 'just' a simple cover that adds rigidity. This is what I came up with...
Cabling (Transformer secondary & Electrodes)
To pass between 250A and 400A to a pair of electrodes, even for a relatively short period of time (<1second) could generate an inordinate amount of heat (due to cross-sectional area resistance) unless suitable conductors are used. I didn't want to 'waste' power in the cabling or find myself having to deal with the generated heat but, similarly, I didn't want excessively heavy cables that would make selecting the spot to be welded difficult... My solution was to two-fold :
1: Keep the cables 'heavy' / large cross-sectional area
Heavy 10mm diameter / 85mm cross-sectional area / 000AWG / 300A cable for the secondary winding.
Lighter 4mm diameter / 13mm cross-sectional area / 6AWG / 100A cable for the electrodes.
2: Split the cables at the 3D printed case / enclosure I was going to design and build. Use M8 machine screws for the +ve / -ve connections with some 3D printed parts.
Mains Connection & Cooling
I don't like cables hanging from kit when the kit's not in use so I thought a simple 3 pin IEC (Kettle-style) socket would be ideal as the mains / 240V input.
I also think the unit may get warm with extended / larger battery pack usage, so I added an 'always On' fan for ventilation.
My idea here was to use a readily available (sBay / Amazon) variable pulse delay unit that could be configured to provide a pre-set, stable, pulse duration and externally triggered by a momentary switch I would mount on an electrode. I designed the front panel with 4x mounting posts for the Timer board and holes to take 4 x 3D printed timer 'configuration' buttons. The timer board is powered from the 12V PSU and simply directs a 12V pulse of configured duration (e.g : 0.1s to 5s) to the Solid State Relay which directs 240V power to the primary of the transformer for the selected duration. The secondary cabling was to be attached to 'banana' plugs (nice & neat but not sure as to losses) or a pair of M8 machine screw connection 'posts'.
I wanted a lit switch on the front of the box to switch the unit On / Off (as per the image, above).
The switch would simply pass the 240V to :
1: the input side of the SSR and
2: the 12V PSU for the 'always On' fan.
I didn't particularly like the thought of a standard click-click relay doing hundreds of hi-powered On/Offs... so decided upon a solid state unit - a SSR switched with a low DC voltage, switching a higher (240v) AC voltage at up to 25A (to allow for inrush etc). I chose the SSR DA-25 (Amazon / eBay).
3: the 12V PSU for the Timer board and the 'firing' circuit... I wanted to 'standardise' a flexible approach so my idea for the 'firing' circuit was to have a momentary switch on each electrode that would initiate a pulse when depressed. These are the little yellow buttons that can be seen on the electrode bodies (below). Alternatively I could then design and build (or buy) a foot switch. In any case I decided to use a pair of barrel jacks to enable whatever firing circuit I devised to be connected to the enclosure and initiate a spot-weld pulse. This way I would need to press a switch when I was ready for the weld... seemed a lot simpler than auto switching or height activated switching etc...
I couldn't make up my mind as to whether I would connect the individual cells with nickel strip or fuse-wire for greater pack protection / safety. Additionally, I was at a loss as to whether I'd want a twin electrode housing or a pair of individual electrodes. The connection ideas ('banana' plugs & sockets or M8 connection posts) were to facilitate simple connection to whatever electrodes I finally decided to use for any particular job...
Doing some research on whether to fuse individual cells or NOT to fuse individual cells, it became obvious that NOT fusing was far simpler - just straight forward nickel strip between the cells BUT ideally, if possible, I wanted to make the safest pack possible - after-all the battery pack is the 'heart' and probably the most dangerous and most expensive single part of any future system I would build. More on this in a later section... The point was to maintain flexibility for future modifications / additions hence the banana / M8 connection posts that could connect suitable electrodes for the job in hand...
To cut a long story short, I designed and built 3 types of electrodes :
1: a twin, individually sprung (and switched) electrode pair which was to be mounted on top of the enclosure thereby providing enough space below it to pass the various cell packs I envisaged myself making. I designed this unit with replaceable electrode shrouds (the heat after 50+ spot-welds begins to melt the ABS (and I don't have the patience to let things cool down - I want the job finished !) so it seemed sensible to have easily replaced / throw-away (yellow) shrouds housed in the red electrode guides (image) which are also replaceable.
2: a pair of single copper pin 'pen' electrodes with a switch on each (either one to be able to start a pulse).
3: a pair of ferrite electrodes. Only one would ever be required at any one time so there is no switch on these. One 'ferrite' electrode would be used with one 'pen' electrode (using switch on the 'pen' electrode to fire the weld pulse). The second of the pair was a spare / to be used when the first simply became too hot...
All bases covered... or so it seemed...
In reality it took a lot of messing about / trial and error with the electrodes. I found that unless I split the paired electrode into two individually sprung electrodes I would, after a while (with heat build up etc) inevitably end up with the two electrode pins at very slightly different heights - which provided a horrible spark, damaging the nickel strip and messing up a spot-weld - VERY annoying. The solution, as explained, was to split the electrode pair into two, each one being individually spring loaded. Excellent results every time ! This was great for large-ish cell packs BUT for the odd weld the pens worked better because they let me get closer to the cell and select the individual pin points for a weld. They worked great too.
However, when it came to fuse-wire spot welding... well, that was a different story altogether... I found the pin-style electrode to be rather difficult to place directly on a wire (without rolling off and ending up resting on the cell surface). In addition, I found, the copper pin itself dissipated the heat (required for the weld to occur) away from the wire too quickly resulting in burned - but not welded / electrically un-joined - nickel / wire...
I experimented with chiselled tips, larger diameter (3mm) tungsten electrodes and various other options / configurations but couldn't get reliable results (and tungsten is rather brittle - often breaking when tightening up the connector screws on the electrodes).
In the end I sourced 5mm diameter ferrite rod, and filed a chisel tip on one end whilst 'screwing' the other end 30mm up into a copper pipe which had already been soldered to its 100A interconnect cable. BRILLIANT ! The chisel tip enabled a good, solid connection to the fuse-wire and the pen-style electrode simply completed the circuit whilst providing the 'firing' switch. Occasionally, I needed to 'touch-up' the chisel end on the ferrite but that's very easy (modeller's file) and very quick. And, once a tip is filed too far it can simply be unscrewed and a new length of ferrite rod screwed back in, in its place.
I have made many, many small and large packs with this method.
Initially, I tried 'banana' plugs and sockets to connect the required electrodes to the spot-welder but I found they became warm quite quickly, so I switched to M8 bolt connection posts with soldered cable ring terminals and 3D printed nut shrouds (orange in the images above) to make tightening and releasing the nuts easier... With this method i was able to cut each spot-weld pulse duration by around 30%.
I also added a pair of barrel jack sockets (beneath the mains On/Off switch) to connect the firing pulse switches (yellow in the images). Either of the barrel jack pair may be used and, in practice, this works great too - depress the button on an electrode and the timer circuit fires which activates the SSR, passing 240V into the primary of the transformer. The heavy (300A cable) secondary coil then directs the output voltage to the two M8 stud connector posts which, in-turn, are connected to the material to be spot-welded via the selected pair of electrodes.
I also figured that, if I needed to, I could buy or make a suitable foot-switch at a later date. I could then use the foot-switch to complete the firing circuit (via either of the barrel jack sockets). As of yet I haven't needed the foot-switch, preferring to use one of the momentary switches on the electrodes instead. I am keeping my eyes open though for an old sewing machine (and it's foot switch) ;)
Voltage & Current
Once I had all the components connected and in the enclosure, I was able to begin testing and configuring the transformer output.
The output VOLTAGE is directly proportional to the number of turns of the secondary winding (assuming, as is the case here, the primary coil is fixed). I found between 3 and 5 turns provided between 3 and 8V.
The output CURRENT is directly proportional to the thickness / diameter / gauge / number of cores in the wire being used. The heavier / thicker the wire, the great the current handling capacity.
In the end, after some experimentation, I found 3 turns of 000AWG (300A) cable provided 3.8V at 380A (circa 1Kw) on a short circuited secondary. (Unfortunately I don't have an image of the clamp meter showing the current - I'll try and find one / take one and update this section).
Once I finished testing and had determined that 3 turns was sufficient for my requirements I shortened the spare cable length you can see in the images in order to reduce system losses.
Some images of the results...
Spot welding nickel to nickel,
fuse-wire to nickel (I used 1/4W resistor tails)
nickel to batteries
fuse-wire to batteries
and... a spot-welded 20S4P LiFePo4 NCR18650 (72V, 13Ahr) battery pack for the electric (ex. 50cc) scooter conversion...