Because CWA ballasts act as more or less constant current supplies within a certain range, the voltage rise of HPS lamps over their lifespan causes a runaway condition with voltage rising higher and higher, and therefore power dissipation as well. This leads to lamp life problems. This is a well known characteristic, and is why saturated HPS retrofit lamps cannot operate on CWA ballasts. This characteristic is avoided in choke/HX ballasts since as the lamp voltage rises the current will substantially decrease, meaning that over the lamp lifespan the power dissipation will still increase, but not nearly to the extent that it would increase on a CWA ballast. This avoids the runaway condition.

This begs the question which I have asked before, which is the following:

"If standard saturated HPS retrofit lamps cannot operate on CWA MV ballasts, why can normal saturated HPS lamps operate on CWA HPS ballasts?"

This led me to believe that the characteristics of CWA HPS ballasts were somehow electrically different from those of CWA MV ballasts. I believe I have concluded that this is not the case. For this explanation I will be using the example of a 100W European 100V 1.2A lamp just for simplicity. I believe the reason that the 100V standard was chosen for many HPS lamps is so that they could have some headroom to have reasonable voltage increase over their life without cycling off on ~240VOC. This means that with a ballast providing 220VOC at 148 ohms (standard reference ballast for 100W lamp), we have room for the voltage to increase significantly from 100V, let's just say 130V (a made up number just for explanation), before cycling off.

On a choke/HX ballast having 148 ohm impedance at 220VOC, this voltage increase from 100V to 130V should reduce the current draw from 1.2A to just 1.09A (according to my caluclator which has a semi-accurate history). This yields a new power dissipation of 118W, assuming the original .83 total power factor of the lamp is reasonably preserved (100V*1.2A*.83=100W, 130V*1.09A*.83=118W, I am sure this is a simplification but you get it). Obviously 118W is more than 100W, but not super duper unreasonably so. Let's try the same situation with CWA, assuming the same OCV.

Now lets try the same for CWA. Of course the original specs are preserved, 100V, 1.2A, .83PF, 220VOC, max 130V before cycling (again a simplificaiton I am sure). Now if we were to increase the lamp voltage to 130V, the current will not substantially decrease. Assuming the current remains constant (not gonna be true, but close enough), we get 129W (130V*1.2A*.83=129W). This is decently higher, and might start causing a problem, which is why we have problems with CWA sometimes.

However, the OCV of CWA HPS ballasts seems to never actually be 220V. The ones I have seen range from 165-205V. That is significantly lower than the 220V reference ballast, and of course lower than the much more common 240-250VOC that is usually used in other situations (MV and MH) where the reference ballast says 220V. Now having a capacitor in series with the winding can help with keeping the lamp struck with lower OCVs, and I am unsure how much that contributes here. But in other situations like PSMH, CWA ballasts typically still have OCVs in the 1.9-2x lamp voltage range instead of the 1.85x lamp voltage range, so I am thinking the series L-C is not doing a whole lot for us here, though that is of course speculation (like all of this).

Have they just lowered the OCV of CWA HPS ballast so that the lamp cycles off before destruction occurs? Maybe this lifespan decrease due to lower OCV is offset by the much more regulated lamp current being fairly independent of mains voltage, but that depends on whether the mains voltage actually fluctuates or not. And this is not taking into account electrode life effects of having a capacitor in the ballast circuit which results in high current pulses.

HX/choke HPS ballasts (Only used for 55V 35-150W lamps here) don't have this OCV characteristic. Choke ballasts will of course have ~120VOC (a healthy 2.2x lamp voltage), and HX ballasts can often have 125-130VOC (even more generous, up to 2.36x lamp voltage). This leaves tons of room for voltage increase, which as said before is relatively safe for the lamp given the current decrease associated with it when not using a dreaded CWA circuit.

Does any of this make sense? Is this true? What do you think? TL;DR: CWA HPS ballasts have cruddier OCV so the lamps cycle off before they destroy themselves, maybe.

Thanks!