Wiring for Lighting 7

[shlim]

Hi there,

I have a case where a small fire had occurred at a wall switch for lightings in one office space. My electrician suspected too much loads routed to one single wall switch, up to 12 fluorescent lights (28Watt each), and that the norm is maximum 10 lights per switch.

Please share some lights if there is such a guideline or recommended practise? Have not come across any reference in the 17th Wiring Regulation or the Internet.

Thank you.

 

[JG2828]

If you are in the USA, the NEC, National Electrical Code, would be a good place to start. But just looking at the numbers you posted, it doesn't appear to be an overload issue. 12 x 28 = 336 watts. Assuming they are running on 120vac, 336 / 120 = 2.8 amps. A standard light switch should be able to handle 15 amps. Something is not right.

 

[waross]

Agree with JG2828.
The initial cause may have been bad workmanship such as loose connections in the initial installation, a very poor quality switch or a faulty switch.
Such a small load would not generate enough heat to do damage under normal circumstances.
If a loose connection or a faulty switch started arcing then that could start a fire.
If the norm in your jurisdiction is 10 fixtures and the initial electrician installed 12, that may cast doubht on the qualifications of the electrician, depending on the local level of adherence to, and enforcement of any local standards.


Bill
--------------------
"Why not the best?"
Jimmy Carter

 

[shlim]

Hi, JG2828 and waross.

I am from Malaysia, and we follow British or generally IEC standards. I am looking for any reference to my technician's comments, just to confirm whether it is a written guideline, or just a norm that people follow in this part of the world.

I would certainly agree with your comments above that the light switch should be able to handle that amount of load, and that the fire was likely attributable to wiring connection or the switching itself.

 

[HCBFlash]

Yes, there certainly is a written guideline, and it will use current to determine capacity of wiring and switches, but there may be a maximum quantity of connected devices specification as well. That said, you may well find that many technicians speak or think in kilowatts, which are proportional to amperes (the unit of Electrical current).

A 28 watt florescent lamp is quite small, as far as lighting a room in a public commercial occupancy. I certainly wouldn't expect less than 10 watts per square meter, and 2 or 3 times that even for general illumination using florescent luminaries in some applications. I say this because I'm afraid that you don't have good figures for us. In the US we typically have 15 ampere lighting circuits, and the most common cause of overheating and failure of our local, wall mounted, manually operated light switches is poor connections in the switch device itself, particularly "push-wired", spring style terminals.

Also suspect is the size and placement of overcurrent protection related to your switch. For lighting, the "upstream" overcurrent protection should protect the switch as well as the luminaries and wiring on the load side of the switch.

.

(Me,,,wrong? ...aw, just fine-tuning my sarcasm!)

 

[shlim]

HCBFlash,

I am not familiar about the specifications for W/sqft for lightings purposes, but your point is taken.

We typically used 10Amp MCB for lighting circuit.

Nevertheless, we managed to detect more hotspots at other similar light switch installations using infrared thermo-camera. The wire terminations were all good, so yes, the source of overheating would be from the inside, switching mechanism itself.

 

[IRstuff]

Who did the thermo survey? What was the actual temperature? 0.1 degrees looks the same as 10 degrees if the scaling is changed. One is irrelevant, while the latter might be a possible concern.

TTFN (ta ta for now)
I can do absolutely anything. I'm an expert!

https://www.youtube.com/watch?v=BKorP55Aqvg

faq731-376 forum1529 Entire Forum list

http://www.eng-tips.com/forumlist.cfm

 

[shlim]

IRstuff, our team did. There was about 2-3 degree C difference between the normal switches and the faulty one (one wall-mounted light switch has up to 5 switches).

 

[IRstuff]

How hot were the "normal" switches?

I get something around 8 milliohm higher resistance for the "faulty" one.



TTFN (ta ta for now)
I can do absolutely anything. I'm an expert!

https://www.youtube.com/watch?v=BKorP55Aqvg

faq731-376 forum1529 Entire Forum list

http://www.eng-tips.com/forumlist.cfm

 

[itsmoked]

If these are electronic drivers I can easily see a problem with the inrush current frying the switch into a resistive self-heating additional load. Most electronic ballasts or drivers have hideous inrush currents in the 30A range in 120V service. In 230V service it is CONSIDERABLY worse.. Tag 12 of those together and the switch sees a 360A in-rush on closing. If the switch bounces even a tiny bit the contacts will be... very un-pleased. Seemingly small load ballasts often have printed "maximum number in parallel" numbers that make no sense in the 'running current' realm but is very important in the inrush region. If you exceed the limit you toast the switches and have lots of nuisance head-scratching random switch-ON circuit breaker trips.

Distortion power factor is also important since lots of electronic ballasts have no power factor correction because their input sections are simply rectifiers feeding capacitors. This means they can only get useful power from the mains during just the very peaks of the power sinewave. While pointed out above, a total of only 336W of load are present it doesn't mean the current drawn is a steady 336W/230V = 1.4A but that the current is delivered in 25A pulses during only about %3 of each powerline cycle. This means switch contacts are seeing approximately 10 million 25A pulses a day. Remembering that current related power losses are I[sup]2[/sup]R the 25A pulses across the contact resistance in the switch are squared losses and capable of considerable heating - which of course hastens further oxidation and steadily increasing contact resistance always ending in a cooked switch. I have seen this exact scenario play out several times.

It's a bit aggravating because the electronic ballast maker takes the side of "This is only a 28W load and so it deserves the absolute cheapest (garbage) rectifier front-end". This logic, while not great, may be reasonable for a user who puts one or two in a lighting circuit. But then often they may be used en-mass for an application and then only active front-end power factor correcting ballasts should be used. They cost a couple of dollars more each and so building designers only reach of the crap rectifier front-end units and we see the results of this in this thread. By the way big facilities like libraries and office buildings with floors and floors of electronic ballasts and drivers can have really serious electrical gear issues with the service entrance equipment due to this.

Keith Cress
kcress -

http://www.flaminsystems.com

 

[shlim]

IRstuff, for example the highest recorded is 30.8 degreeC vs the colder region of about 27.6 degreeC, at the same light switch.

itsmoked, thank you for your detailed explanations on how in-rush current and heating (and increased contact resistance) caused cyclic deterioration to the light switch mechanism.

 

[John2025]

Keith, Great explanation. It's something I've kind of understood, but never really gave enough thought to. Is there anything practical that can be added to a string of these kinds of ballasts after the fact?

 

[itsmoked]

Hi John.
I've thought about that and it's a bit of a problem. There is really nothing you can add to Fix It, all you can do is sort of hide it. By putting a ON-DELAY relay in series with a ballast you could prevent single units or whole banks from all coming on at the same time. Problem is those relays cost a serious percentage of the ballasts and would also have to be able to handle the needed inrush. There are inrush limiters as used on televisions and such but just adding them in the open would be a UL nightmare since they are usually a board level component. All these bandaids end up costing a significant percentage of a new power factor corrected ballast so it probably makes more sense to flesh out a plan to replace all the functionally crippled ballasts with PF ones.

In the OP's shlim's case I'd replace the switch and four of the 12 ballasts in that group with PF ballasts.

Any place else, I'd just set up a replacement schedule that's palatable and slowly work to bring a facility into normalcy over the next year or two depending on how often burnt switches and mystery breaker trips are occurring. In some places there could be a significant power bill reduction by having the distortion power factor eliminated.

If I recall correctly, in the EU I believe they passed a rule requiring any devices but certainly lighting equipment to all be PF corrected if they're over a fairly small power level. That's to prevent exactly the problems caused as noted in my earlier post.

By the way, switching a PC power supply in my old-iron gaming rig from stock to PF-corrected allowed my UPS to run it just about twice as long since the uncorrected distortion power factor pulses applied a horrid efficiency hit on the UPS inverter.

Base line: Every facility eventually re-fixtures. Make absolutely sure that when that happens its to PF-corrected systems and likely LEDs anyway as who likes the repetitive fluorescent tube replacements. :}

Keith Cress
kcress -

http://www.flaminsystems.com

 

[waross]

And don't forget about the times three factor for shared neutrals on three phase installations.

Bill
--------------------
"Why not the best?"
Jimmy Carter

 

[itsmoked]

Please remind us further Bill?

Keith Cress
kcress -

http://www.flaminsystems.com

 

[waross]

Will those peaks cancel on the neutral, and more importantly, does it matter?
The harmonics from discharge lighting ballasts are a known issue causing unexpectedly high currents on neutrals. Breaker panels with 200% rated neutral buses are available for installations where a lot of harmonic current is anticipated.
Can we expect similar issues with these types of ballasts?
Bill
--------------------
"Why not the best?"
Jimmy Carter

 

[TugboatEng]

The inrush for a standard incandescent light bulb is also very high, close to 10x the bulb's rated current. I discovered this after I had a 15 amp toggle switch catch fire switching on 1000 watts worth of halogen lights. If you're not using a lighting specific switch, check the datasheets for a lighting rating. It's usually quite a bit lower than the switch's printed rating.

 

[itsmoked]

Yeah Tug! Typically listed as "tungsten".

Bill.. I can't really answer that with any assurance but since this is essentially using only that smidgen of every cycle, I don't see the three phases countering each other very well at the neutral. That is if all the strings are phase to neutral.

I would expect three 25A pulses a half cycle on the neutral.. so yeah maybe the neutral gets thrice as hammered as the phases. Not sure..

Keith Cress
kcress -

http://www.flaminsystems.com

 

[waross]

I'm not sure either.
I suspect that a poor connection (higher resistance, loose and susceptible to heat cycling) may have a shorter service life.

Bill
--------------------
"Why not the best?"
Jimmy Carter

 

[bacon4life]

I saw something similar in bedroom due to an air conditioner motor. The air conditioner was plugged into the switched outlet typically used for a portable lamp. Eventually the wall switch overheated.

In targeting replacements, you might also consider prioritizing devices that are switched most often.