Twin flourescents.

All LEDs lose brightness with age.

That's why they have a "10,000 hour" or "50,000 hour" rating. The rating is "hours until unit operates at 70% intensity". Having only 70% of light output is defined as "failure".

LED bulbs have failed in the past, because of premature PSU failures. (In that class of failure, the light output drops to zero.)

The "specs" are not taking such behaviors into account. On the newer bulbs, if you go right to the Philips web site, there is a spec for "switching cycles", which tells you what they think the PSU can withstand in terms of ON/OFF cycle count.

The battens that Murmansk asked about the other day, those employ a wall wart, and don't have an inrush limitation. That's because the DC produced by the adapter, is present even when the "light" is off. That means there is one less thing to go wrong (wearout by switching). The DC switching is inside the unit.

Philips 70W 72in T8 Cool White EUROPEAN Fluorescent Tube (Pack of 5)

Medium Bipin Base 6200 Lumens ---------+ Cool White Bulb Color | 4000K Color Temperature | 85 CRI +--- These two numbers give lumens per watt 1" Diameter | 70W Energy Used ---------+ 72" Length T-8 Shape

Philips LED B22 Frosted Light Bulbs, 8 W (60 W) - Warm White, Pack of 6 [Energy Class F]

Type of Bulb ?LED Cap Type ?B22 EU Energy Efficiency Label ?F Luminous Flux ?806 Lumen \___ These two numbers give lumens per watt Wattage ? 8 watts / Incandescent Equivalent ?60 Watts Colour Temperature ?2700 Kelvin Colour Rendering Index (CRI) ?80 Average Life ?15000 Hours Average Life ?15000 Hours Lamp Power Factor ?0.7 [...phase angle between V and I, on mains]

The B22 lamps are slightly more efficient in terms of lumens per watt. You would need about 8 of them for the same light output.

Generally, fluorescent and LEDs are in the same ballpark on efficiency. Both techs are better than incandescent.

The T8 in the example, has a cooler-white (4000K) light. But it's not looking significantly bluish yet. 9000K would be annoyingly blue. You use

The CRI of that fluorescent is better. That's if you trust CRI as a metric (a useful metric).

Lighting devices also have "cycle limits". Maybe the B22 bulbs can be switched on and off 10,000 times, before the internal PSU fails on inrush. Sometimes, you can find a spec for this. On the T8, the electronic ballast may have a spec like that before the e.ballast fails. I don't know if the T8 tube itself is cycle dependent.

If you use a LED batten, it will share some of the B22 properties.

The higher the colour temperature, the more efficient the LED. When a LED advertises 120 lumen per watt, that means the light is going to be bluish in colour. It gets this way, because the phosphor coating on the LED is thinner. This allows more of the blue light to leak through. (While lab LEDs have hit 120 lumen per watt, that does not mean every production LED matches it, which is why I can estimate the 120 number is achieved by cheating on colouration.)

Remote phosphor bulbs (no longer made) were an attempt to prevent phosphor smothering on the small LEDs. Some of those bulbs incorporated a couple red LEDs to blend with the other light, to make the colour temperature lower and in the "warm" category. LED lighting does not have to be made from only blue LEDs with white phosphor painted over them.

LEDs work best as arrays, as that blends out the sharp shadows that happen if you use a single powerful LED. I tried a single LED once at around 500mA, and you could throw the most sharp sock puppets on the walls. But that gets boring after a few minutes and I don't consider such lighting to be all that pleasant. Array lighting and bouncing it off a white finish, reduces the irritation of LED lighting. I don't like particularly staring into LED lights.

Be particularly careful with "LED floods" for outside, as they are borderline eye-unsafe. You can burn a retina with those. Be careful where you point them. When testing them before deployment, use the same degree of care in your setup, so you don't damage your sight. The 360 degree bulbs (like that B22) don't tend to be quite as bad as a "focused" flood. Like a flood with a Fresnel lens on the front - nasty.

Paul

When you fill an empty mains-level capacitor with juice, it produces a *huge* spike. What this tends to do, is not impact conductors, but it can damage the plating on switching elements. It is the room light switch I would be worried about in such a case.

Computer ATX power supplies, some of them, the inrush is 40A to 80A on mains. The inrush lasts for one or two mains cycles in duration. But this can pit the switch contacts if you're not careful.

We had a switch fail on a product at work, due to inrush. I had one fail in my fingers one morning, the plastic toggle just disintegrated. These failures also happened in the field.

Only occasionally does the I^2T exceed the threshold on a breaker, causing a breaker to trip. But the behavior can damage things like switches.

In this design, NTCR1 is the inrush limiter. The two 470u caps are the thing causing the inrush problem.

Paul

Hard to believe, but it's a thing. I guess the LED devices are too cheap to have their own, and besides, the thermal profile isn't all that good in the light itself (it's a little warm in there).

This is also one of the reasons, you should not rapidly toggle an ATX PSU switch on the back. When switched off at the back, wait a bit before turning it on again, for best inrush limiting behavior. This is not necessary, if you know the protection is done with an Active PFC circuit instead of an NTC (such info will not be printed on the box).

If not properly engineered, NTCs can burn out. So if the lights don't work some day, you know where to start looking. Look for a crispy critter where the NTC used to be.

Paul

OK Paul. I'm converting some twin 5foot fluorescent fittings to LED. Fortunately they are switchstart so I can leave the ballast in circuit. I'll probably connect 3 x 5' twin lamps to a switch.

That skates over the question whether protection is required (and the link to guidelines on the Australian site didn't work for me). While it looks unlikely to matter for Tim I'd suggest a better read is for UK installations is BEAMA's

I'm not understanding that guidance.

If you have added an extra series element (MOSFET) into the path for the express purpose of altering the inrush time, you could use the series element to flatten the pulse entirely (that's what dual usage Active PFC was doing, is using the series element the Active PFC uses, as a series resistor during capacitor inrush). Setting it up to use zero crossing, while clever, imposes a requirement for an additional power device in the path. I can just imagine the squeals of delight at the light bulb plant.

While an NTC is pretty low tech by modern standards, it at least affords the option of comparing the price of the NTC to the price of a three-legged power device to alter the power dynamics.

I also notice the transient this is causing, is extremely short. Maybe this could be modified by doing something to the capacitor design.

On things like ATX supplies, the duration of the inrush might be a cycle or two, rather than 250us. The ATX supply also has a scary number of joules stored in it. If one of those fails, I don't want to be in the room when it happens. I lost hearing in one ear for ten minutes, the last time a multi-joule capacitor did its thing. The one in an ATX is no joke either.

Paul

Thanks for that. I take comfort from the suggestion that luminaire numbers are restricted to 100 circuit watts.

snip.

Up to 10% they say.

Thanks, Paul. And for the rest of your piece.

When CFLs first arrived, so called spec started to mean nothing. A so called 100 watt equivalent was anything but. Same with early LEDs.

And even more so when you take light quality into account.

Plenty seem to claim LEDs are much more efficient than florries too. But never seem to give all the parameters they're measuring. Of both the florrie and LED.

Of course, if all you want is illumination to find your way along a corridor, the figures may well be accurate. But, to me, there's more to lighting than that.

That is very obvious, given how many liked early LEDs.