"Frosted" LED light bulbs ?

[NY]

When we first replaced our tungsten bulbs with daylight CFLs, they seemed very blue. The difference between the bedrooms with daylight and the bathroom with tungsten GU10s was very noticeable - tungsten looked horribly yellow.

I don't notice the difference between the two as much now. What I do like about daylight bulbs is that the inside of my study is now the same colour as the daylight coming through the window, whereas before the difference was very noticeable. The only problem with that is that occasionally I forget that I've left the light on during the day and go out with it still on!

It's quite an eye-opener to see the spectrum of a typical CFL and how many peaks it has - not exactly black-body radiation :-) It's amazing that the colour rendition is as good as it is with peaks and gaps like that.

I've always wondered why daylight slide film made fluorescent lights look a horrible green colour (even when they were probably warm-white around

3500-4000K) whereas you don't get that with a digital camera. The green emulsion of the film must have been unusually sensitive to one of the lines in a fluorescent tube, in a way that the green pixels of a digital camera aren't to the same extent.

[The Natural Philosopher]

Actually the digital is just as green BUT some cameras seem to recognise this and do an overall 'white balance' as part of taking the picture

[Martin Brown]

The eye is easily fooled and does auto white balance.

And if you used daylight film indoor with incadescent lights it would have a strong orange cast. Basically the film records reality.

Not at all. In fact through a curious coincidence colour panchromatic film is actually less sensitive to green light than it should be with an insensitive zone around 500nm which is used as a safelight.

The consequence of this was that all the old Palomar slides of deep sky objects are coral pink and powder blue even when visually the objects are in fact apple green from the OIII line at 501 & 496nm. The film was completely blind to this wavelength and it wasn't until the mid 1970's that true colour images of nebulae were obtained (front page of SciAm)

Slide film:

CCD:

Visually it looks an oily green in the very brightest parts, but a few of the bright planetary nebulae are a bright apple green visually but still appear to be pink and blue on slide film.

[dennis]

Fluorescent lights have an emission line spectrum that has a strong green.

The eye doesn't see this.

Digital cameras do see it but the camera can correct for it quite well.

[NY]

I realise that. I was meaning, though I didn't say so explicitly, that a slide film and a digital camera, each set to the *same* white balance (eg daylight) will record fluorescent lights differently. On daylight film they will be greenish, on daylight-balanced digital they will be amber if they are warm-white tubes or reasonably neutral if they are daylight tubes. Conversely if you use tungsten-balanced film (or an 80A blue filter and daylight film) and a digital camera balanced for tungsten, then the fluorescents will still have a strong green component on film but will be either neutral or blue (depending on whether they are warm white or daylight tubes) on digital.

So it seems that slide film is unusually sensitive (despite what you say about "by curious coincidence colour panchromatic film is actually less sensitive to green light than it should be") to a component in the fluorescent tubes, whereas digital cameras aren't.

Actually, thinking about it, that's not completely true. Film is sensitive to *some* component of fluorescents (not necessarily green light) which

*manifests itself* as a green cast - ie it triggers the green-sensitive grain.

It's less noticeable with negative film because a green cast will probably be corrected (either manually or automatically) at the printing stage, even if that causes other colours then to be rendered incorrectly.

[Martin Brown]

No you have fundamentally failed to grasp the effect of the digital camera having an automatic white balance. It sees the same weird green colour cast as the panchromatic film (only more so) in the raw data but it is corrected before being presented as an image.

On a more advanced camera you can force it to manual white balance and save the raw sensor data but most default to automatic white balance. (and have done almost since the advent of digital cameras)

Yes - the very strong mercury green line at 546nm to which it is quite sensitive away from the 500nm safelight wavelength.

Auto white balance in the printing stage will compensate for a multitude of sins (although often leaving some trace of colour cast).

[Dave Plowman (News)]

Depends on the phosphors used.

[Dave Plowman (News)]

You do realise there is no such uniform thing as daylight? The colour temperature of that varies according to time of day, time of year, part of the world and weather conditions. And likely more I've missed out.

[dennis]

Not really, more on the gas.

[Martin Brown]

No. The green and purple lines are from the mercury emission lines and are intrinsically very strong sharp spikes at specific wavelengths. See

The lower lumpy emissions are from the phosphors (and have inferior colour rendering characteristics to all but the worst LEDs).

"Daylight" ones tend to have a phosphor 500 +/-50nm FWHM

"Pink" for meat/plants have a phosphor 580 +/- 50nm FWHM

Other phosphors allow a better approximation to white light but there are always very strong emission lines from any CFL or fluoro tube.

Trivial to see this with a CD reflection spectroscope.

[NY]

I thought I was being clear enough this time by referring to "daylight-balanced digital". I meant with the digital camera set to a

*fixed* "daylight" white balance (eg round about 5500K for sunlight) and *not* with auto white-balance enabled.

Auto-white balance, either in a digital camera or of the printing stage with negative film, can correct a multitude of sins, so all bets are off if that's used.

[NY]

Yes, I meant the camera/film settings rather than the colour of the light. I think direct sun in a mostly blue sky has a colour temperature of *around*

5500K (subject to time of day etc variations). And it is this colour temperature (AFAIK) that "daylight" colour slide film and the "sunlight" white balance setting of a digital camera are balanced for.

Likewise tungsten film and the "incandescent" setting of a digital camera are balanced for tungsten light which is about 2500-3000K depending on whether it's normal household bulbs or Photoflood photographic lights.

My digital camera has a variety of different "fluorescent" white balance settings for warm-white, white, daylight, low/high-pressure sodium and mercury lights (I know sodium and mercury are discharge without fluorescent phosphors, but the camera lumps them all together). None of these settings, when used with sunlight, gives a pink cast which would be necessary to counter a green cast from fluorescent lights; instead all these settings seem to give varying proportions of amber and blue, which seems to suggest that the digital sensor is less affected by the green line in fluorescent tubes and therefore does not have settings that correct for it.

I do realise that light from a 100% cloudy sky is bluer (8000K or beyond), and that shade could be all sorts of weird colours depending on what the light is reflecting from (grass, buildings of various colours, people's clothes etc) but is predominantly bluer than direct sunlight because of the blue from the sky.

[~misfit~]

My opinion for what it's worth is that the biggest problem with LED lighting is people trying to retro-fit it into fittings deigned for incandescant bulbs. The LEDS and their drivers tend to overheat and their lives are consequently shortened considerably.

If you're getting the place rewired anyway I'd suggest getting LED lamps fitted into the ceiling that have remote drivers and have finned heatsinks in the ceiling space. I've seen them at a local hardware place and. while they're not exactly cheap they should last 10 to 20 years and use ~20% of the electric that other options would.

FWIW.

[Johnny B Good]

That's about the size of it with the current crop of 70 to 90 Lm per watt lamps and the not so well ventilated luminaries (typically closed off lamp shades that allow a warm blanket of air to surround the lamp).

Well, quite!

Otoh, you could simply sit tight and use up your current stock of bulk purchased CFLs until those 200Lm/W LEDs finally land on shop shelves as promised 2 years ago by both Cree and Philips after they'd demonstrated respectively 303Lm/W and 240Lm/W lamps in their laboratories claiming that experience suggested it took some 18 to 24 months to go from 'Lab Specimen' to 'Shop Product'.

At that time (2 years ago), the likes of Asda were routinely selling

81Lm/W LED lamps. The highest efficiency lamps on sale so far have barely broken the 100Lm/W barrier so very little progress on the efficiency front has taken place in all of that time.

I'm still hanging on for those 200Lm (and better) per watt LEDs simply because such high efficiency lamps will then be able to match or exceed lighting requirements from existing GLS luminaries with nowhere near the same risk of overheating as the current crop of GLS LED lamps.

Thinking about it will reveal that doubling the efficiency from 90Lm/w to just 180Lm/w not only halves the 16 watts consumed by a "100W equivalent filament GLS lamp" down to 8 watts but also, of that 8 watts, an even greater proportion is emitted as useful light, leaving even less to be dissipated as unwanted heat.

Hanging on for the promised higher efficiency LED lamps not only offers a very modest additional saving on the total household electricity bills but, more importantly, it'll save you on the costs of fitting 'special' luminaries (a win win scenario).

[ss]

For what it is worth I have just refurbed the shower room which had 4 x halogen 60w bulbs, I replaced this with 2 LED 12 watt 6" panels as other half wanted bright. They are flush fitting to the ceiling and she is delighted with them, I bought cheapo off ebay around £8 each. And so far so good. I am now cosidering similar for the kitchen (24 feet) whereby I will replace 9 halogen (540watts) with 5 LED panels (60watts) I will do a temp hook up to ensure she likes before committing to cutting holes in the ceiling.

[Johnny B Good]

It looks like T/B has broken the attributions. None of the text you're replying to belongs to me, it all belongs to "~misfit".

However, it does remind me that the four 35W halogen lamps flush fitted into the ceiling of our shower room when it was refurbished about 5 years ago have yet to fail. I suspect this longevity is due to my insistence upon 12v lamps (with, as it happens, 60W rated smpsu "Transformers" which exhibit current limiting to completely take the sting out of the switch on surge). It's not the warmest room in the house so the 140W of radiant heat is a welcome bonus. :-)

[Dave Plowman (News)]

;-) Several times a day. Every day of the week.

[Johnny B Good]

And, they're definitely 240v filament lamps? If that's the case, it bodes very well for my 12v lamps (extrapolation would suggest I'll never have to replace a single lamp in *my* lifetime :-).

[Dave Plowman (News)]

They are a mixture of GU10 and PAR 75 watt, all in downlighters. They are fitted into a roof void where the insulation is well clear of them, so probably run as cool as any. And no vibration from above.

Round the mirror I have golfball bulbs - dressing room style - and although they are only switched on when needed, like shaving, have a very short life. But do look pretty. ;-)

[bert]

In article , ~misfit~ writes

I have 4 PAR38s mounted in the kitchen ceiling. Currently I run 2 incandescent and 2 CFL. That way I save a bit of juice but still get instant light when I switch on. I've been looking at LED alternatives and noticed the better (More expensive) ones seemed to have air vents around the periphery. (Not the Diall ones in B&Q!!) So presumably that is why and so do you think these would be OK to use in these fittings?