I was just curious.  What would happen if I added a capacitor (or capacitors) to an old Westinghouse 400 watt 120 volt reactor ballast??

I have eight capacitors:

2@ 12uF/480 volt (1975 Westinghouse caps)

2@ 15uF/360 volt (1971 GE caps)

2@ 15uF/370 volt (New caps)

1@ 3.5uF/480 volt (1971 GE cap)

1@ 4uF/440 volt. (New cap)

To improve the power factor of a simple reactor ballast or high-reactance autotransformer, a cap is sometimes connected across the mains wires to the ballast/lamp circuit. In other words, it is to reduce the line load even though the lamp wattage draw is the same.

As to the correct value of capacitance, it depends on the line voltage, ballast type, and lamp wattage. When I experiment to find a good value, I simply connect a true RMS AC ammeter in series with the line input wire going to ballast and fire up lamp and let it warm up to get a good current reading. After that, I carefully connect caps parallel to the ballast input (AFTER the connection point for meter) until line current drops to at least 2/3 of it's normal value.

I have a 175 watt Merc that I put on a 55 uF cap...the line current went from 3.1 amp down to 1.77 amp. when fully warmed up. But with a cap value that high, it makes a nasty "pop" when I connect the power. Too much cap value can make excessive arcing or possibly voltage spikes in the line circuit.

As for connecting the cap in series with the lamp, I've tried it several times on HX autotransformer ballasts and it doesn't seem to work very well. The lamp has trouble warming up...even with a correct value cap.

Series cap are using only CWA ballasts. And you can not create a CWA from HX by only the capacitor, the whole ballast has to be designed so.

And when connecting capacitor to multitap HX ballast, you should always follow the wiring diagram, as connecting (even equivalent) capacitor to improper tap cause the primary to overheat (as the current flow somewhere else then designed).

I have a question about adding capacitors too.  I have a 100w HPS yard light with an OEM 120v S54 reactor ballast.  I have a 36uF/120v cap from an Advance 100w HPS 120v S54 reactor ballast kit.  Being the same type and wattage of ballast, I would think there would be no problem using the Advance cap on the OEM ballast, but could there be any differences that would mean it would need a different cap?  The label on the OEM ballast shows an optional cap but doesn't list values for it.

I have a question about adding capacitors too.  I have a 100w HPS yard light with an OEM 120v S54 reactor ballast.  I have a 36uF/120v cap from an Advance 100w HPS 120v S54 reactor ballast kit.  Being the same type and wattage of ballast, I would think there would be no problem using the Advance cap on the OEM ballast, but could there be any differences that would mean it would need a different cap?  The label on the OEM ballast shows an optional cap but doesn't list values for it.

If the cap is directly connected to the mains, it is only to compensate for the phase shift coming from the ballast (so improve the power factor) of the current drawn from mains, but it has no influence on the ballast functionality, so such capacitor you might connect. It is weird, then the manufacturer does not state the value, as virtually only he know, what inductive current has to be compensated out. But as all series choke should have equivalent behavior, the capacitor for correct compensation would be equal.
But the capacitor charge current causes power surges upon power ON (unless you "hit" the moment, when the voltage is near the zero crossing), so it would cause high current sparks on switch contacts. So i would use the capacitor only on longer wires, where it is really needed (as the uncompensated lamp current would cause too high voltage drop) and where the wire resistance limit the power-up inrush current.

Lately, I have discovered that Japan apparently has some probe start metal halide ballasts that utilize a series capacitor in the circuit but have very poor power regulation. Interestingly enough, in one of the ballasts that I imported into my collection, it is designed for 1000W probe start metal halide lamps with 250V 4.5A specifications, yet it has a 300V OCV!! When I tried running the ballast with a North American 1000W M47 probe start metal halide lamp, a North American 1000W H36 mercury vapor lamp, an Iwasaki Sunlux Ultra Ace 940W high pressure sodium lamp designed to retrofit 1000W H36 mercury vapor lamps and 1000W M47 probe start metal halide lamps, and a Philips AllStart 860W ceramic metal halide retrofit lamp for 1000W M47 probe start metal halide ballasts and 1000W M141 pulse start metal halide ballasts, they mostly seemed to run very well without cycling whatsoever. However, whenever the ballast was subjected to low line voltage, it easily underdrove the lamps in a similar manner to HX and reactor ballasts. This seems to tell me that in these ballasts, the series capacitor appears to mainly function as a lamp starting aid rather than a current regulator like you see with CWA and CWI ballasts.

Here are the test results of that experiment in case anyone is curious:

https://www.lighting-gallery.net/gallery/displayimage.php?album=lastcom&cat=0&pos=3&pid=270998