Strange.

In my school today, I was sitting in homeroom waiting for class. To be exact, gym and homeroom are my first periods. THe gymnasium has Metal Halide HID high bay lights, and during a thunderstorm this morning, the power flickered, and of course, all of them turned off for a few minutes, cool down, and restarted. But after the 2nd power surge, 13 out of 20 of them, "popped", and died. Gym went really dull, and the breakers tripped in the electrical room. The breakers control half the lights each, so 2 breakers, 10 lights each.

Teacher went to reset them, and none but the 7 that were "unharmed" came on. Teacher left the lights off for 10 minutes, restarted, and still the same.

The whole gym period, 2 hours about, the lights never came on. Each bulb in the fixture was pure black, and surprisingly, they didn't explode on anyone.

Fast forward to the end of the day. They had 4 fixtures down off the ceiling, and they were opened up. Each one was filled with blackness and burnt inside. They tried unscrewing the bulbs, and the bulbs just broke off inside the base from heat damage!!!

Obviously the power surge killed most of the lights!  

I think it might be a bit more than just a surge. A big one, or some other problems involved



Example guess story :

The 2 circuits are on 2 lives, and use a shared neutral wire

There is weak connection somewhere in the neutral (connection is glowing, arcing and melting away somewhere) (most known to happen with aluminum wiring but happens with copper too)

When the lamps warm up or work normally the load on the neutral wire is very small (as the slight difference between the 10 fixtures and the 10 other fixtures), so the bad connection does not "show up"

When the lamps were restriking, some of them allready restriked and some didnt, so the load on the neutral was about as the difference in amount of working fixtures on the lives

Now the bad connection could not withstand the higher load, and gave up (disconnected), The voltage on each live is no longer 120V, but 240V*X and 240V*(1-X) respectively, where X and 1-X roughly depend on the amount of lit lamps on either live. So if 1 live had 2 lamps that allready restriked and 1 live had 5, the 2 would get much different voltage than the 5

Now the ballasts can fail from overvoltage. If one of the failing ballasts failed shorted, it'd now "move" the allmost-240V voltage to the other live, blowing up ballasts in there too

Finally, after at least 2 ballasts shorted (at least 1 on either live), a short circuit formed between the 2 lives and tripped the breakers, but not before a lot of ballasts blew up on both lives, and the blown ballasts blew the lamps

Meanhile the bad connection might have welded itself back together, and won;t show up untill future incident

If there are 10 lamps per circuit, there must be a lot of current passing, more than the 15 amps normal lighting circuits have on 120V. If the lamps are 400 W each, that means 4000 W total plus ballast load. That is way more than the 1800 that is the normal maximum current on a 15 amp circuit. At my school, the gymnasium is lit with 16 lamps of 400 W each, but only two lamps are on the same circuit (8 circuits total, 15 amp breakers). Even with 175 watt lamps, that totals 1750 watts lamp load (plus ballast load) and it is very close to or exceeding the limit of 1800 watts. Anyways, it was indeed a catastrophic failure from the arctube point of view.

is it possible that the system is 277v?  That could support 10 400w lamps if it was a 20A circuit or 10 250w lamps on a 15A circuit. Most high bay MH systems here are 277v, which is one leg of a 480v 3 phase system. What would happen in that case of a bad neutral if two legs are used?

One place I know of runs all the lights off a 3 pole breaker and another runs individual breakers to each bank with 6 separate banks.

As the lamps died instantly and ballasts were damaged, it was not the 120V supply (120V need autotransformer and I could not imagine a ballast failure killing lamp instantly).

A different story would be on the 277V, where simple series reactor would be sufficient. Then the surge from the nearby lighting could have caused the ballast insulation to break down, sending direct mains to the lamp, destroying the lamp instantly and tripping the breaker.

The loose Neutral on two legs of three phase system yield to the individual phase loads running in series on the 480V. With unbalanced loads (and if the ballasts were CWA, their balanced state is not stable) the voltage split unevenly, what may cause the big part pf the 480V to be on the one side. If that exceed too much the rated 277V, i could damage the ballast. But I don't know, how this would lead to an instant failure of the lamps: The capacitor in CWA is voltage-wise the most sensitive component, so it could fail. But that would yield to moderate or no increase in lamp current (so the lamp would need some time to fail) and huge increase of the primary current, so the voltage would drop on that leg and/or the transformer overheat and burn out, so stop delivering any current to the lamp at all - again no reason for lamp sudden failure.

Though it is sometimes posable to spec fixtures with specific ballasts for specific voltages, commercial HID fixtures here often use multiple-tap autotransformer ballasts, due to the different voltages that could be encountered in such buildings.  I have several that have taps for 120, 208, 240 and 277 volts.

The "120V ballasts" in the previous post apply actually behavior of all autotransformer based ballasts, regardless of the rated voltage.

BTW could you use the primary of the autotransformer ballast to power something else from the same supply ? (like adding 120V socket on 240V powered fixture by using the 120V tap as output) ?

How much VA would be safe to load it (as frction of the lamp power) when a HID lamp is there too, or without th lamp (so using the ballast just as voltage converter) without overheating the primary, ie what is the design margin ?

@Ash: It is routinely done - the 120V tap is used to supply the "emergency" halogen capsule (intended to at least partially replace the nonworking tube either during the "hot restrike" cool down or when the lamp had failed), on HX ballast the power factor correction capacitor is connected to some higher voltage tap, as there the capacitor would be physically smaller and the rather high reactive power to be handled is spread over the majority of the primary winding, what lower ballast losses.