Electrolytic vs supercap for CD spotwelder?

I need to make a cheap spotwelder to rebuild some battery packs
(e.g,. welding thin nickel strips).

My options are:
(a) Try to find most of a farad at up to 25V or so on eBay. This could
take a while. (Buying new electrolytics in the 0.1F range and above
at these voltages is pretty expensive.)
(b) Go for supercaps from Digikey (e.g). They've got some 5V 1F models
for only a few bucks, and some "fast-discharge" ones for 10x more.
Of course, if I go for more than 10V or so, I'll have to build
some sort of charge-balancing circuit around the series caps,
and I'd rather not unless it's really the best way to go---
though it might be as simple as a resistor in parallel to each cap.
(c) Try one of these weirdo "digital supercapacitors" that are all
over eBay for the crowd who fill their entire trunk with stereo
amps. These look like they're rated at 1-5F at 13.8V and (I would
hope) include some sort of charge balancer. Dunno if they're
suitable for the sort of instant-discharge I want for a welder,

I'm having difficulty figuring out:
(a) How fast a discharge I need for a spotwelder anyway; I'm going to
be using an SCR to trigger it and presumably want the lowest-inductance
arrangement I can get, but are we talking milliseconds or microseconds?
(b) What can I get out of a supercap? Milliseconds or microseconds?
I haven't worked with supercaps before, and most people who seem
to be designing circuits with them are using them as batteries.

Other random questions:

My reference design had 0.5 to 1F at 0-25V through a 600V 50A TO-65 SCR; what
worries me are things like http://www.avxcorp.com/docs/Catalogs/bestcap.pdf,
whose page 3 shows that non-ultra-low-ESR supercaps (e.g., the
affordable ones) seem to have virtually no capacitance for pulse
widths of 10ms or less. (The BestCaps in that datasheet claim
something like 60% of nominal instead, which is pretty good!)
I dunno how narrow the pulse width needs to be to get good welds,
but I'm guessing pretty narrow---for example, one guy claimed in
http://www.philpem.me.uk/elec/welder/ that just the difference in
putting the SCR on the low side and not the high side (so the trigger
for it didn't have to go through the wires, electrodes, and workpiece)
worked a whole lot better, which implies to me that he might be
talking microseconds. But I just don't know.]

Of course, I also don't know whether I need 0.5 F at 10V or 1F at
25V for the things I'm considering.

Not to mention---the 50F caps at http://dkc3.digikey.com/PDF/T072/P1354.pdf
(some of which claim ESR's of 0.025 ohm at 1kHz) have these skinny little
0.5mm leads on them. How in hell would these leads not be vaporized if I
charged up the cap and then shorted it, as a spot-welder would be doing?
(And sure, I could try to attach #4 copper wire to them---somehow---and
send that to the sharpened tips on my welding electrodes, but boy I feel
silly doing that---seems like whatever part of the cap's leads aren't
attached to the heavy wire will just evaporate when shorted, if they can
really dump that kind of current, even if it -is- just for milli-to-

The Instructables site has some rambling about using supercaps for CD
spotwelders, but it's just rambling---nobody who's actually built one
using them, for example, or who seems to talk about discharge rates,

sort by: active | newest | oldest
Zaen9 years ago
I don't know very much on welders, but if you're looking for cheap, fast discharge capacitors and don't mind a little bit of work, I'd suggest using disposable cameras. A target near my house just gives them away, and they usually hold one 330V, 120 microfarad capacitor each. To give you an example of how strong these are, I built a 5 capacitor array, and if you just touched the two wires together, they would form a tiny arc weld. You literally had to break off a part of the wire to get them apart. I think it came out to be around 20J of energy. You would have to build your own power supply because the prebuilt circuitry doesn't charge fast enough, but It seems to me if you are tackling a project to build your own spot welder, this shouldn't be hard. Good luck!
westfw9 years ago
Don't forget that energy goes up with V2, so using a 25V electrolytic instead of a 5V supercap gives you 25 times the energy for the same capacitance, in addition to putting less strain on your wires (which is why long transmission lines are high voltage.)
I don't think you actually WANT very fast discharge for a spot welder; don't you have to allow some time for thermal conductivity and physical melting? "real" spot welders are DC transformers with quite macroscopic "ON" times, aren't they?
HilaryGage (author)  westfw9 years ago
I am absolutely not forgetting the V-squared dependency, which is one of the reason why I'm asking the question in the first place---must I deal with stringing 5 5V's in series (and dealing with charge-balancing) in order to get enough energy? Unfortunately, having neither at the moment, I'm trying not to waste a bunch of money buying supercaps only to find out that I need the fast-discharge capability of an electrolytic---or that, for thing nickel strips, 1F at 5V was adequate in the first place. [Buying one 1F5V cap at a time is expensive because of minimum-order and shipping charges, which complicates the whole thing.] (Chase those URLs---check out the discharge curves on one, and the UK guy's comments about high-side SCR not yielding good welds. The latter sure -seems- to imply that in -this- case, you need the fast dischange. "Real" (transformer-based) welders may have macroscopic on-times, but they can at least maintain their output voltage across that; I'm worried that, without a fast risetime, a supercap will merely get things warm all over without making a good weld anywhere.) In any event, I'm hoping that someone will chime in either with experience re fast-discharge of supercaps, or some knowledge about what sort of risetime I need for a cap-based resistance welder. [One advantage of supercaps might be that I can put them and the SCR directly -at- the electrodes, since they're so small. That would certainly reduce losses and inductance in the wires, but if they can't dump charge very fast, it won't make much of a difference.]
HilaryGage (author)  HilaryGage9 years ago
Err, sorry for the weird formatting---the "preview" button does absolutely nothing for me, and apparently using square brackets was being interpreted as markup. But since I can neither preview beforehand nor edit afterwards, oh well...
NachoMahma9 years ago
. Not sure about the rise time, but what westfw says makes sense to me. I would think that a slower rise time would reduce the flying sparks that spot welders are famous for (I assume those are caused by vaporization at hot spots).
. Since there is no major transfer of metal (ie, it doesn't have to jump from a stick/wire to the puddle) AC should work very well. But that does nothing to answer your question. :(