On 4/16/2018 12:54 AM, Jeff Liebermann wrote:

What do you like for a small spot welder? I've been thinking of getting
one.

I looked at what was available and decided that I didn't like anything
that was also affordable. So, I'm building my own. Nothing fancy.
The biggest automotive starting battery I can find (I want cold
cranking amps), a large high current relay, a less than 1 second timer
board to drive the relay, a foot switch to start the timer, some heavy
gauge wire, and some machined parts for the copper electrodes. It
didn't take much testing to determine that if I wanted to weld
something, I want DC, and to use AC if I wanted to blow a hole in
something. There are several YouTube videos describing how to throw
something together. For example:
"Simple and dirt cheap Spot Welder for lithium cells"
https://www.youtube.com/watch?v=UU7QC5Uby6M
It needs 2 electrodes, no ground clamp, and heavier gauge wire.

Stable/controlled pressure is very important.
You can make a serviceable electrode pair from a small arbor press
and some hobby store brass. I use #6 solid copper wire for the
electrode and a file
to shape the end. The trick is to make sure that the current doesn't
go thru any part of the probe actuation assembly.
I could put up some pix if anybody cares.

You need a lot of energy in a short time to melt the nickel without
overheating the bottom of the cell. At the risk of repeating myself,
the secret is to use an energy pulse, not a current or voltage pulse.

I started with a microwave transformer and some logic to count full
cycles of 60 Hz. to keep the transformer from saturating arbitrarily.
I used 4-6 cycles for a weld. If you try to do it without the full
cycle counter, the transformer ends up at some arbitrary point on
the B-H curve and saturation effects can raise havoc with your welds.
Repeatability was horrible.
I could make some very nice welds, but not enough in a row to make
a battery pack.
Starting the pulse at a zero crossing and counting cycles improved
repeatability dramatically, but it was still crap.
Problem is the low voltage is very intolerant of contact resistance
variation.

I tried discharging a cap, but at the time, I didn't have anything
that would switch the required current. And I didn't have a
storage scope so I could see what I was doing. Turns out that it's
harder than I thought to instrument a welder. When the pulse fires,
everything on the bench jumps. Getting a scope to trigger on
a specific part of the waveform was just wishful thinking.

I tripped over a Unitek CD welder for $17 shipped on EBAY and snapped it up.
It can put 7V across a milliohm. Welds got instantly very much more
repeatable. Since it's an energy pulse rather than a current pulse,
it's much more tolerant of resistance variations.
It's only 125 Watt-Seconds, so it works great on thin nickel for low
current stuff. It won't weld the thick links for high current stuff.

Turns out that hobby store brass sheet can be cut into battery tabs.
It welds very easily. I wouldn't use it for anything high current tho.