My 2 cents on the issue:
1. Since the mix is heterogeneous, sparks in different places will ignite the mix differently. Also, the bigger or more sparks, the faster the mix will ignite.
However, faster does not mean more powerful. The physics involved are complex, factors including the volume, geometry and temperature of the mix, projectile mass, fit in the barrel, materials etc., not to mention voltage and current of the spark.
Since spuds have a lower mass to caliber ratio than bullets in a gun, faster propellants will work better. Then again, since the barrels are longer and also have more volume (ergo the resulting gas have more time to expand and cool), the combustion could become insufficient to further accelerate the projectile before it exits the barrel. So a slower-burning fuel would be needed to keep the gas hot, thus expanding, thus propelling the projectile.
So basically different cannons need different conditions. There is no one spark, one fuel or one projectile perfect for any design. Only through trial and error, and based upon past experiences, can we tune a design to perfection.
2. An arc welder has a lot of current, yet very little voltage. So there won't be a spark if the electrodes are spaced more than like an atom apart
For best results (high voltage and high current), a cascade such as the Marx Generator would work best. I've seen (online, not in person) sparks as long as 30 cm generated by those things so - there you have it - that's pretty much your best ignition source. (Unless, of course. you want to talk about Z-pinch machines or nuclear fusion).