Ok, a little background: I have been following this thread for quite a while. I am an electrical engineer with over 40 years experience and a zillion bucks of sophisto test equipment. I have designed power systems that handle over 7.5 megawatts. I have built a FET based capacitive discharge welder/cutter that runs rings around commercial models costing over $20,000.
Here is the straight dope on car audio capacitors: over 90 percent are well and truly fraudulent crapassators. Big names are absolutely no guarantee of a proper capacitor. I have purchased caps from over a dozen different brands and tested them for capacity and ESR. Only three met their published specs. A couple others were usable. All the others had between 1/15 and 1/4 their advertised capacity! Some ESRs were 2-6 times what was specified (if specified at all). Buyer beware... Caveat emptor... You're oh, so screwed...
The good:
Rockford Fosgate RFC1 - measures 1.019F, 1.17 milliohm ESR @ 20Hz
Scosche ECAP1 E2 - measures .932F, 1.52 milliohms ESR @ 20Hz
Monster Cable 1F - measures 1.015F, 1.16 milliohms ESR @ 20Hz (note: no part number, thought to be the same as their current model MPC P300 MCAP-1F)
I have tested multiple samples of each of these caps and they were all very consistent.
The three caps listed have a voltage rating of 20V... The Rockford cap is specified at 16V/20V surge. Unknown if that is a working value or "surge" value on the other two. I have occasionally run my Rockfords at 20V and did not have problems. Actually, I let one sit at 20V for a week without any problems (my welder has leakage sensors over the vent holes and shuts down and screams bloody murder if it detects any electrolyte leakage). YMMV. Actually YM probably WV.
No telling if any cap actually meets its published voltage specs... it takes a destructive test to find out. I suspect that if a manufacturer meets their capacitance spec, then their voltage spec will be OK. If they are cheating you on capacitance, expect the same on the voltage rating. Try to avoid using capacitors at over 4V less than their surge value and you should be OK. If you have too, you can run them at the full voltage, but you might wind up with electrolyte spewage and a nifty paperweight...
Very few people need a 3F welder if their caps are good. A single decent 1F cap with a good ESR will weld just about any battery tab you want at 16V or less... and probably much less. I have blown holes in 18 gauge steel with that cap. People having problems with their welders are most likely using crapassators.
If you connect two caps in series, you will get double the voltage rating, double the ESR, and half the capacitance. You do need to connect a charge balancing resistor across each cap. Try something like 200-500 ohms. The charge balancing resistors will also act like safety "bleeder" resistors that will slowly drain the capacitors if you shut off the charge supply... a good thing that should be there on any capacitor. A mother bear welder cap (if you really must run at over 16-20V) would be to connect two (or more) pairs of two series connected caps in parallel. I would not recommend series connecting caps that are not identical...
4 milliohms is a terrible ESR for a welder cap. You can easily lose well over 2/3 your welding power in that ESR alone. Avoid so called "hybrid" and carbon capacitors like the plague. They are utterly useless for this application. Avoid single large value capacitors. If you need more capacitance you are MUCH better off paralleling three 1F caps than using a single 3 farad cap... besides, you won't find a real 3 farad cap. The ESR of three 1F caps in parallel will be less than 1/3 the ESR value of a 3F cap.
No matter what you do, strip off those stupid voltmeters and do not use the power terminals/threaded studs, etc shipped with the caps. Bolt directly to the capacitor terminals with real single piece screws. The ESR at the top of those "platinum" and "gold" terminals can be over three times what you get bolting directly to the capacitor. Be careful not to strip the threads. Use a good load spreading washer and lock washer. Watch your polarity. Connect a cap up backwards and it is toast.
Welding to that aluminum A123 battery case is another matter altogether... expect problems no matter what you do... particularly with an SCR based welder. It takes much more advanced monkey technology to make that weld.
And on the subject of bogus values, watch out for "car audio" cables. One brand advertised as 100% pure oxygen free copper was actually plated aluminum with a resistance 1.62 times that of copper. Another was some mystery metal with a resistance over three times that of copper. The Scosche brand of "Flux" cable tested particularly well. Also it is ridiculously flexible. You can wrap 4 ga cable in a two inch diameter coil.
Otherwise, you might want to buy real welder cable... but no telling if there is also hanky-panky going on there. I suspect that you will have less chance of buying bogus welder cable than car audio cable... particularly if it is made in the US. All the bogus cables came out of China. All my welder cable was made in the US and tested just fine.