Now that I have a small array of capacitors to play around with, I've been experimenting with seeing how they affect lighting setups in various ways. The main thing I've observed is the results of putting them in series with ballasts. With almost all of my experiments, I've found that when a capacitor is placed in series with a ballast, it causes lamp current to go up. This makes sense because total reactance equals the reactance of the inductive component of the system, minus the reactance of the capacitive component. So, inserting the capacitor reduces the total reactance.

However, I've found that with my 175W MV CWA ballast, when the capacitor is removed, current (both line and lamp) nearly doubles. So in this case the capacitor increases the reactance. Why is this? The only thing I can think of is that the reactance of the capacitor is (approximately) quadruple the reactance of the ballast itself, so when inserted, the reactance becomes negative, but the magnitude of the reactance is double the magnitude of the reactance of the ballast alone, so current is cut in half. But wouldn't that mean that the ballasting impedance is primarily capacitive, which is bad for the lamp?

Another example is another member's experiment, in which he was able to reduce the lamp current of a standard HPF ballast by putting a capacitor in series with the lamp. I tried this on one of my own HPF ballasts, but again, the cap just raised the lamp current. What am I missing here?

Thanks.

It depend on what will be the overall impedance of the series LC vs just the inductive part. Adding a capacitor could increase, as well as increase the lamp current, it really depend on the actual combination of the L and C.

With the CWA the ballast is designed to have the capacitive component indeed as the dominant one.
But how that will behave when the capacitor is removed depends on how is the impedance constructed, how big is the inductive part.
Many have rather large inductance, so the series LC operates closer to it's resonance. That means the impedaance responds more strongly to any inductance changes, mainly imposed by the saturation of the magnetic shunts in the ballast. With those the removal of the capacitor means the current dropping.
 Some have just minimum leakage inductance, so the capacitor does not have to be designed to such high voltage rating (however needs higher capacitance). As the leakage inductance is rather low, the lamp current without the capacitor becomes higher.

Interesting info; it makes sense. Thanks. So the capacitor being the dominant reactive component in a CWA system doesn't make the current crest factor too high? I guess because the inductive component is there?

Interesting info; it makes sense. Thanks. So the capacitor being the dominant reactive component in a CWA system doesn't make the current crest factor too high? I guess because the inductive component is there?

Exactly. The inductive component has two main functions in a CWA:
First provide filtering, so lowers the current crest factor and allows the generation of high enough voltage for the arc reignition after the current zero cross.

Second function is to regulate the lamp current and so keep the lamp power constant even when the mains voltage differs or even vary (there the name "constant wattage" came from, even when in reality it means just constant lamp current):
When the mains voltage increases, with just the capacitor it would mean increase of the lamp current. But for the inductive component in series it means it gets more saturated. The saturation means lower inductance. And lower inductance in series with the dominant capacitance means the overall impedance increases, so counteract the current increase.
So when the lamp current tolerance is +/-10%, with HX or series choke it means the allowed mains voltage tolerance just about +/-5%,
the CWA regulation feature allows the mains to vary even +10%/-30% (some models), while the lamp current still stays within the +/-10% limits.

I know this is old, but could changing the cap value in a cwa ballast make for a different wattage? Eg. 10uf cap on a 175e ballast instead of a 17.5uf cap.

Yep. Whether increasing/decreasing the capacitance will increase or decrease lamp current is a factor of the ballast design, but for most of them, more capacitance = more current. When the cap is removed from my 175W MV/MH ballast (which is like having infinite capacitance), lamp increases from 1.4A to over 2A.

This also goes for HPF fluorescent lamp ballasts since they are in a CWA configuration. They're usually the opposite - more capacitance = less current. don93s has modified many ballasts to run different lamps this way.

Of course, if you choose to increase current, you need to beware of ballast heat. Also, the lower capacitance you go, the more voltage the capacitor sees, and with a really tiny cap it could be potentially over 1000V, so you need to beware of that.

Both ballast kits the caps came from are with 280v caps, while I have motor start caps for 330 to replace the plastic crapacitor. I also noticed that the ballast itself on the ground fault 100w CWA is the same as the newer 175w CWA, both advance, different eras. So, I could use the 10uf cap with the 175e core&coil ballast to safely and properly run a 100w merc?

Edit: 7uf for 50w and 8uf for 75w. Also, 100w is 10uf. This info was used from Greg's gallery (75) and tmcdllr (50). It should also be noted that each ballast is Philips advance. Would a 7.5uf power a 50 or 75 watt bulb best?

I have no idea. Different ballasts have different inductances, so there's no telling what a particular value of capacitor will do. The only way to find out is to set up a safe circuit that's fused, use a dummy lamp you don't care about like a 34 watt F40T12, and measure the lamp current and capacitor voltage.

By altering the capacitor you may alter the current, but there are two problems with it:
First when you increase the current in any way, it means higher load on the winding and that means higher losses, so more heat. And that could become above what the thing is able to handle.
Second problem is, normally the saturating characteristic of the magnetic shunt is there to stabilize the designed current even when the mains voltage  changes. This feature will then "fight" against any attempt to change the current by changing the capacitance in certain range (normally this feature is supposed to cover certain capacitor tolerance as well). So to really change the current, you will have to change the capacitance further and that would mean you move the operating point outside of the regulation range, so the ballast won't have any room to act against the variations, so the current will be very sensitive on the mains and tolerances of all components. So you will have to "tune" each such piece, as all components have certain tolerance.

The magnetic part of different wattages may look the same (same size, same wire gauge, same number of turns), but what is important here is the air gap in the magnetic shunt. And this small part, inserted after the thing is assembled, is what steers the lamp current. And that shunt may well be the only difference between the two wattage ballasts.

On the topic of fluorescent capacitors, I am replacing the wiring in my Jefferson Electric two lamp ballast. I figure that while I have the ballast open, I should replace the old capacitor. It is a Cornell Dubilier 3mfd 330VAC. Is it okay if I use a 3mfd cap with a higher voltage rating? I can find 3mfd 370VAC caps relatively easily.

It should be fine, 40 volt difference shouldn't harm it.

Is it okay if I use a 3mfd cap with a higher voltage rating?

Yes, it will work just the same.

It should be fine, 40 volt difference shouldn't harm it.

Any difference is fine - the capacitor could be rated 10000V and it will work exactly the same. It's just the maximum voltage the capacitor is rated to handle without breaking down. Just as long as the new capacitor is rated at least as much as the original one.

With the capacitor rating it is important to differentiate
- "DC rating": Usually a figure without any detail (e.g. 630V). It means the maximum DC, but as well peak voltage the capacitor can be subjected without degrading faster than rated life.
- "AC rating": either in the form of "275VAC 50/60Hz" (capacitors rated for mains frequency operation) or a specific graph in the datasheet (usually ion the form of a graph of the voltage as a function of frequency). Takes into account not only the peak voltage, but the stresses related to the repetitive charging/discharging and related losses. This rating is usually 30..50% of the DC rating when speaking about 50/60Hz, for HF operation it is really a fraction (e.g. a resonant capacitor in a HF ballast seeing 60V during operation and 400V during start leads to 1000V DC rating or more).
Both ratings are usually a function of the expected life, details are again in their datasheets

A special category are "motor start" capacitors: These are AC voltage rated, but the exposure time is limited to few second per cycle and the cumulated time over the life is in the 10h range. These capacitors are really usable only for starting the induction motors and are supposed to be disconnected when the motor has started. They tend to be small, frequentlyu bipolar electrolytes, but not able to handle any permanent load.


So for the lighting you always need capacitors AC rated for "lighting" and/or "motor run" (so permanently connected to AC).

In any case, for whatever use, using higher voltage rated capacitor (for applicable rating) means just it's slower degradation. The only bad things are, these capacitors are usually bigger and more expensive. The later you don't mind (if you have them already), the first may mean it just may not physically fit... And take care of the required clearances around the component (some capacitors are designed to bulge when they fail and in that way break the internal connections, so safely disconnect; if you block that bulging, you may compromise that safety disconnect feature)

I appreciate the information! Now I have to decide on a plan of action. 

Here is the capacitor in the ballast.

If the capacitor is good (no corrosion, no leak, no visible cracks in the soldered seams,...), just keep it there.
The thing is, the PCB's allowed to push the size really down for the rating and still maintain very long life.
The PCB-free oils are not that good in handling the eventual internal discharges when it comes to some low energy (short spike,...) overvoltage events.
And the dry film types just degrade over time, don't expect much more than 100khours MTBF from those (well, unless you overrate them by a significant factor, so e.g. use 600VAC rated for a 300AC use; but then the size would suffer).