Does anybody know if it would be safe to run a magnetic ballast (pre-heat, rapid start, etc.) on a UPS without burning the ballast out? or should a get an electronic ballast?

Ive ran magnetic HID ballasts before on Tripp-Lite brand battery back ups, the lamps flicker far more noticeably, im sure due to the modified sine wave inverter. If operated on pure sine wave inverter, it would be no different than running on mains. I would assume magnetic fluorescent ballasts would exhibit the same slight flicker on modified sine wave 

Pure sine wave UPS units are available but probably cost more than modified sine wave units... I was curious about this as well, thinking about maintained emergency lighting with switch start 8 watt lamps.

I have an APC Back-UPS RS 900, I'm not totally sure if its a pure sine wave one or not though.  I tested a 4 watt lamp on pre-heat, it seemed to dim slightly and the ballast seems a little more louder but it didn't seem to get any hotter than normal.

I've got an older APC Back-UPS ES 700. When I powered a 11W (magnetic/pre-heat) PL lamp with it, it took about 10 seconds of blinking until the lamp started. But I didn't notice any side effects afterwards.

Yeah, magnetic ballasts are fine on modified sine wave inverters. If running one on such an inverter long-term, lamp life might be slightly shortened from the rougher waveform, but I don't know for sure.

I would, however, try to refrain from running electronic ballasts on them. I imagine those are far less tolerant.

That's strange, I always thought electronic ballasts would tolerate something like that.

I ran a 4 watt fluorescent on the UPS for 3 hours and the ballast ran cooler than if it was on normal power

Electronic ballasts will run on pure DC

Yeah, magnetic ballasts are fine on modified sine wave inverters. If running one on such an inverter long-term, lamp life might be slightly shortened from the rougher waveform, but I don't know for sure.

I would, however, try to refrain from running electronic ballasts on them. I imagine those are far less tolerant.

Well, the fact some ballasts seemed to work well could be just luck. Either it haven't fried anything, or the ballasts were of a more tolerant type.


The "modified sinewave" is designed to have the same rms voltage and peak voltages as the real sinewave, while sufficing with just switches in the output stage. That means three things:

- The peak voltage is the same, so the electronic ballasts have no reason "to complain" at all. In fact those UPS inverters are actually designed to operate mainly the rectifier load (all the computers are like that), so actually the electronic ballasts are the type of load these inverters are primarily designed for.

- The magnetic saturation in a magnetic device like transformer is about 10% lower than on the real sinewave. With the regular mains transformer with the rectifier on it's secondary (most low power electronic), that is no problem, the transformer just has more margin towards saturation. But with a lag ballast (series choke, HX autotransformer,...) that means 10% lower magnetic flux, so more than 10% lower lamp current. Fluorescents usually tolerate that well, HID's will run cooler than designed.

- The waveform has sharp edges, what means quite severe problems with circuits, that contains significant capacitors on the AC circuit (so several uF and above; not just the EMC filters of few 100's nF).
Directly connected PFC or similar capacitors (uF's across the mains input) would most likely fry the inverter inside of the UPS (or the transformer).
Auxiliary capacitive dropper circuits (LED's, DC supply sources for photo or movement detectors or cimilar small electronic devices) would get fried by the high peak currents
Circuits with capacitors inside some "out of resonance" LC circuits, like CWA or any other lead ballast, will form rather high 3'rd harmonic content in the currents, with the ability to either cause higher than designed load currents or cause an erratic behavior (multiple current zero crosses at wrong phases may cause difficulties with arc reignitions after that)

I'll chime in here...with personal experience from living off-grid. 
Not sure about HIDs so much (Though I have run a 175w MV yardblaster on MSW) but that was years ago) but fluorescents WILL work on MSW, magnetic or electronic.  Performance varies.  Cheap 2-lamp LPF rapid start ballasts (25/34/40w and 14/15/20w perform the worst, buzzing noticeably louder and "flickering" rather randomly.  Also they run very noticeably dimmer.  Single-lamp 14/15/20w LPF trigger start works well but hums loudly on MSW.  2X40w HPF magnetic RS also works well, no noticeable change in light output though it may buzz a bit louder.  8ft slimlines...it depends in my case.  An 80s 1.45a Advance doesn't like MSW so much, likes to only light one lamp and/or "Rapid start" (So weird seeing that on a slimline!).  An Advance Mark III I have, however, fires right up on MSW though it buzzes slightly more.  Both of the latter situations are with 60w lamps, no idea how 75s are as I don't have any!  Preheat chokes...they will buzz louder and can "flicker" as well.  (22w circline adapters flicker the worst).
Even CFLs do OK in my experience, as do electronic F32T8s.  The latter buzzes much more though, for an electronic ballast.  Weird thing is, sometimes if there's a really short power interruption (less than a second) they'll only relight at partial brightness and flicker like crazy though.
I have a .77a Advance Kool-Koil 40/35/34w rapid start ballast that runs cooler on MSW than "normal" power.
Even the "LOA" shoplight type ballasts work just fine!

The buzz in the electronic F32T8 comes from the input EMC filter choke, when exposed to the fast edge currents.
The dim glow after the brief interruption may be the lamp have extinguished, but the ballast haven't reset, so it didn't swept through the resonance to start the lamp properly (without that it is not able to feed the cold electrode discharge by enough current to heat up the electrodes)

Otherwise the NPF shoplights are practically the same as series inductor preheat (once the lamp burns).

With the HPF ballast I would really expect some components being overloaded. It does not have to cause temperature increase on the ballast case, but it will mean shorter life

For these types of loads and mainly for permanent installation is way better to install the true-sinewave inverter, then you could be sure there are mo problems at all.

I've ran an instant start ballast with 4 f32t8 tubes on one. The ballast barely received enough power, causing one of the bulbs to be terribly underpowered. That bulb went EOL once I plugged the fixture into a regular outlet and left it for an hour.

Yeah, electronic ballasts don't seem to enjoy UPS's

Yeah, electronic ballasts don't seem to enjoy UPS's

This I do not believe, when both the ballast and UPS are OK (but that's a must). If anything (not only the ballast) tolerate the modified sinewaves or such nastiness, it would be only the electronic, with just a rectifier and DC block filter on it's input. And if even the electronic does not tolerate some UPS, then that UPS is just faulty.

The electronic ballast is just the same load as any computer or any other electronic device, so the type of load the UPS are specifically designed for.

For a magnetic ballast, relying on the 60Hz quite decent sinewave, there is no wonder some designs have problems with the shape. But really the electronic should not have any problems at all.