Discharge Lights Overheating

[Sidewinder]

Now, I have seen a few fluorescent lights overheat starters & ballasts if left energised for a long time (say perhaps even continuously) which had failed to light.

I have seen a couple of double fittings that had overheated when being in use for say around 10 hours per day average for say 5-8 years with one of the tubes deliberately removed to "save energy".

I have also been told by old hands that this can and does happen, with Fluoro's and other types of discharge lamps such as Metal Halide, Sodium etc.

I can also recall a street lamp near home (SON) which had been out for a few weeks one winter, spontaneously catching fire one night.

Is it due to the failure to light?

The one with the missing tube obviously could not light the tube, so no current flow through that side of the fitting, but, it still overheated.

Now I cannot be sure, though if the damage was done prior to the tube removal.

I did get a call to a large well known Builders Merchants a few years ago where there had been a Fire Brigade attendance to a fluorescent light catching fire.  That again had been used for say 10 hours per day average, and had been out for a few months, don't know if it was due to lamp, ballast or starter though, and I can't recall if it was HF either, so perhaps no starter even!

Is this just another urban myth of the industry or is there some science or proof behind it?

Does anyone have any thoughts on this please?

 

[ProDave]

I've known fluorescent ballast chokes to get exceedingly hot in a fitting that's been working fine for years. Then it starts to hum loudly then shortly afterwards start smelling. almost certainly shorted turns in the choke due to insulation breakdown.

And then there was the metal halides at out community hall. I was called to re lamp all the failed fittings, some I found had no lamp in at all.  A few weeks later one that I had fitted a new lamp to (lamp was missing) started smoking and the FB were called. The igniter had done just that and melted and let all it's smoke out.  Then two more did the same leading them to scrap all the remaining fittings of that type and replace them with new.

The advice from the wisdom of this forum at the time is metal halide igniters don't like running without a lamp (or presumably with a failed lamp) 

 

[Evans Electric]

Good question Sidey . 

I have no technical proof but in my experience , fluorescents , especially older ones , don't like having a dead lamp ,  I've found many cases of a red hot choke because of a dead tube , resulting in the varnishes melting and choke fails.   Replace the tube , choke cools down again.  

I found the same with low-bay sodium & Metal halides .    Manufactorer's  leaflets always said , Replace lamps ASAP .   Also Metal Halides  that are on constantly should be switched off once a week for 15 mins .    I can't remember the reason for that TBH .  No one ever does it.  

Electronic ballasts & HF crap are not affected AFAIK  .   HF ballasts controlling two tubes will shut off if one tube goes.  

As an aside , for office & studio  Cat 2 fittings  costing around £35 - £40 ea  I found it cheaper & easier to buy a Crompton Batten fitting and remove the ballast ...than order a choke @ £15 ,

Could we get an expert  answer  from a technical rep. perhaps ?

 

[Marvo]

The way I understand it is a ballast circuit is basically a constant current driver. When the cathode breaks down a higher voltage becomes applied to keep the current constant which means more overall power in the circuit. This causes more heat and if the circuit isn't designed to self limit correctly it can be a fire hazard.

Electronic ballasts have a usually have a safety circuit that completely shuts the ballast down under rising voltage conditions, this is often only reset by power down and reapply power, changing the lamps whilst the fitting is live usually won't restart the ballast.

**edit** I think it's more than an old wives tale, there's a written test spec in the IEC regs that dictates the behaviour required of a ballast under filament failure and asymmetric power draw. I'll run a search and see if I can find it.

**edit2** I think it's IEC61347-2-3 http://www.tridonic.com/se/download/Standards.pdf

 

[Marvo]

From IEC61347-2-3 (In the interests of not breaking UK copyright regs here's a draft from the Saudi Standards Authority )

17 Behaviour of the ballast at end of lamp life17.1 End of lamp life effectsAt the end of lamp life the ballast shall behave in such a way that no overheating of lampcap(s occurs at any voltage between 90% and 110% of the rated supply voltage.For the test simulating end of lamp life effects, three tests are described:a) asymmetric pulse test (described in 17.2);b) asymmetric power dissipation test (described in 17.3);c) open filament test (described in 17.4).Any of the three tests may be used to qualify electronic ballasts. The ballast manufacturershall determine which of the three tests will be used to test a given ballast based on thedesign of that particular ballast circuit. The chosen test method shall be indicated in theballast manufacturer’s literature.NOTE Checking ballasts against their capability to cope with the partial rectifying effect is recommended byIEC 61195, Annex E, and IEC 61199, Annex H.Lamps used in the ballast test circuits shall be new lamps seasoned for 100 h.[b]17.2 Asymmetric pulse test[/b]The ballast shall have adequate protection to prevent lamp cap overheating at the end of thelamp life cycle. Compliance is checked by the following test.The following values of maximum cathode power Pmax apply:– for 13 mm (T4) lamps, Pmax = 5,0 W;– for 16 mm (T5) lamps, Pmax = 7,5 W.(Other diameters are under study.)Test procedureRefer to the schematic diagram in Figure 3.If only one connection per electrode is available at the ballast and/or lamp, T1 shall beremoved and then the ballast shall be connected to J2 and the lamp to J4. The ballastmanufacturer should be asked which of the output terminals has to be connected to J4 and, incase two output terminals per electrode exist, whether they can be short-circuited or bebridged with a resistor.(1) Close switches S1 and S4, and set switch S2 to position A.(2) Turn on the ballast under test and allow lamp(s) to warm up for 5 min.(3) Close S3, open S1, and wait for 15 s. Open S4 and wait for 15 s.(4) Measure the sum of the average power dissipated in the power resistors, R1A toR1C and R2A and R2B, and the Zener diodes, D5 to D8.NOTE The power should be measured as the average value of the product of the voltage betweenterminals J5 and J6 times the current flowing from J8 to J7. The voltage should be measured with adifferential voltage probe, and the current should be measured with a dc current probe. A digitaloscilloscope can be used for the multiplication and averaging functions. If the ballast operates in a cyclingmode, the averaging interval should be set to cover an integer number of cycles. (Each cycle is typicallygreater than 1 s.) The sampling rate and number of samples included in the calculations should be sufficientto avoid aliasing errors.The power dissipation shall be below Pmax.If the power dissipation is greater than Pmax, the ballast has failed and the test isdiscontinued.(5) Close S1 and S4.(6) Set S2 to position B.(7) Repeat steps (2), (3) and (4).The ballast shall pass both position “A” and position “B” tests.(8) For multi-lamp ballasts, repeat steps (1) to (7) for each lamp position.A multi-lamp ballast shall pass the tests for each lamp position.(9) For ballasts that operate multiple lamp types (e.g 26W, 32W, 42W), each lamp typespecified shall be tested. Repeat steps (1) to (8) for each lamp type.[b]17.3 Asymmetric power test[/b]The ballast shall have adequate protection to prevent lamp cap overheating at the end of thelamp life cycle. Compliance is checked by the following test.The following values of maximum cathode power Pmax apply:– for 13 mm (T4) lamps, Pmax = 5,0 W;– for 16 mm (T5) lamps, Pmax = 7,5 W.(Other diameters are under study.)Test procedureRefer to the schematic diagram in Figure 4.(1) Set switch S1 to position A.(2) Set resistance of resistor R1 to 0 (zero) Ω.(3) Start lamp(s) by turning on power to ballast under test and allow lamp(s) to warm up for5 min.(4) Increase the resistance of R1 rapidly, (within 15 s) until the power dissipated by resistorR1 equals the test wattage value of 10 W for a T4 lamp or 15 W for a T5 lamp. If theballast limits the power in R1 to a value less than the test wattage, set R1 at the valuewhich produces the maximum wattage. If the ballast switches off before reaching thetest wattage, continue with (5). If the ballast does not switch off and limits the power inR1 to a value less than the test wattage, set R1 at the value which produces themaximum wattage.(5) If the test wattage value was reached in step (4), wait for an additional 15 s. If the testwattage value was not reached in step (4), wait for an additional 30 s. Measure thepower in R1.The power dissipation in resistor R1 shall be below or equal to Pmax.If the power dissipation in resistor R1 is greater than Pmax, The ballast has failed andthe test is discontinued.(6) Turn off power to ballast. Set switch S1 to position B.(7) Repeat test procedure steps (3) to (5) above.The ballast shall pass both position “A” and position “B” tests.(8) For multi-lamp ballasts, repeat test procedure steps (1) to (7) for each lamp position.A multi-lamp ballast shall pass the tests for each lamp position.(9) For ballasts that operate multiple lamp types (e.g. 26W, 32W, 42W) each lamp typespecified shall be tested. Repeat steps (1) to (8) for each lamp type.

Given the recent changes to the Ex regs as well I'd say fluorescent fire due to tube failures was 'myth plausible'.

 

[Sidewinder]

Thanks Marvo,

I think I will have to look up the Standards myself and look into this.

That test procedure is quite complex to read off screen like that.

I am still in the overheating believer camp, that is believer, not Belieber!

IF there was ever an antithesis for a Belieber, I'm possibly it!

 

[kerching]

Never heard of Belieber...so now i am wishing i had not googled it!

However i have found out i must be a BeRileyer...........

 

[Marvo]

I'm not sure what happened with the parsing of the text in the codebox above, the forum software added random funky colours throught even when there were no bbcode tags to cause it.

 

[steptoe]

arent the Saudi regs simply a rehash of 7671 anyway?  :C

 

[Evans Electric]

Marvo , you're a Diamond Geezer :Salute

Anyone remember these strange fluorescent fittings .  (  I don't think I dreamed them up)

One had no ballast other than a tungsten GLS lamp in the fitting that didn't light up .

The other , I'm sure was real,  was again a ballastless fitting but the spine of the fitting became hot during use .

 

[binky]

Most florries will survive a lamp failure, at least once, but cheapo fittings seem to be taking the fire-risk to new levels. We've come across old Fitzgerald lighting with original ballasts still working fine after 20-30years, but the newer cheapo Chinese stuff just seems to be failing all too easily. It's not a good situation, but one that I'm sure we will see more of in coming years. The only good thing is that it may encourage more regular maintenance, which is half the problem anyway. Even the well designed old stuff gets very hot trying to start dead/ nearly dead lamps.

 

[NozSpark]

I have a couple of shops that I do work for, they only have half the tubes fitted to save energy. I have tied to sell them on LED panels but they're having none of it

The only thing that we did have regarding failure was burning out fuse holders. In the end the rep got someone in to sort it as they had quite a few failing

 

[Marvo]

Most of the newer generation electronic ballasts I've seen have enough intelligence built into them to shut down their output completely if a tube is removed which prevents runaway voltage due to low current draw so deliberately removing tubes would not have any adverse affects although there may be a few rogue manufacturers where this is not the case. If their fittings are the older type or inductive ballasted then I think it wouldn't be a good idea to run them with tubes missing.

 

[Phoenix]

Certainly seen metal halides go up, sw. Believe the failure mode is that the lamp wont strike but the (probably badly designed) gear keeps trying until it goes on fire. One department store had it happen more than once, trumpton would come out, knock off the lighting dist board, put a set of steps up and empty one of the stores own CO2 extinguishers into it, cut the flex, dump the thing outside and comment to the manager that its not the first one.

Seen my fair share of cooked chokes on florries but can't see how it can go from having lamp missing, that should break the circuit and nothing should happen. Had one once where it was failed in such a way that with a dead lamp and starter in it wouldn't do anything, no tripped breaker, didn't look or smell overheated, etc. but put a new tube and starter in and it would destroy them both (blow the heater filaments at end of tube) with a right flash as the breaker tripped

 

[revjames]

Thanks Marvo,

I think I will have to look up the Standards myself and look into this.

That test procedure is quite complex to read off screen like that.

I am still in the overheating believer camp, that is believer, not Belieber!

IF there was ever an antithesis for a Belieber, I'm possibly it!

My wife says I have become obsessed with The Monkees

I thought she was kidding....

Then I saw her face

 

[Sidewinder]

A development on the overheating.

Removed the starter in the attached pic from a non-functioning fitting yesterday, and the ballast was VERY hot.

Will be back in a few weeks, and will be able to see if the ballast has cooled down.

 

[Andy™]

seen a few of them... stays closed, more power than normal flows, stuff gets hot. eventually sometime fails. usually choke lets its smoke out or 1 of the filaments melts

 

[binky]

seen quite few of those on cheap fittings where the starter has been located directly above the lamp