LEDs … The Good, The Bad, and The Ugly

[Michael GilliganParticipant @michaelgilligan61133]

Fellow Experimenters … A request for information:

• My 10W LED, on it's small test heatsink, runs quite hot [details to follow, when the infra-red thermometer arrives] … The heatsink, of course, needs to dissipate most of the nominal wattage.
• BUT … The 9W Diall bulb that I mentioned earlier, has no visible heatsink … it's a sealed unit of plastic and glass [the only visible metal being the bayonet cap]
• YET … The Diall bulb runs barely warm to the touch, even after several hours.

Question: How do they do it

Photographic evidence, or a published description, would be particularly welcome; but I will accept wild hypotheses … I am bewildered.

MichaelG.

..

Edit: For info, the  B&Q Diall LP-A65-1S 9W bulb seems to be no longer available.

… I don't know why this is.  … I hope it's not a bad sign.

Edited By Michael Gilligan on 01/10/2014 20:58:31

[Michael GilliganParticipant @michaelgilligan61133]

Here's one for the "Good" category:

Four of these arrived from China, today [ordered on 19-Sep]

Silly-Cheap, considering all the work that's in them !!

The ebay description is a bit odd, but look at the pictures: they are designed to push into the standard holes for reversing sensors, but the spring-clips are removeable if you want a different mounting.

Excellent light output and pattern.

MichaelG.

[MuzzerParticipant @muzzer]

Michael – apparently, these LEDs are in the range 15-35% and in the US are required to have a min 25% conversion efficiency from electrical to light energy. Incandescent lamps are 5-10%. I hadn't realised they were as good as that, partic incandescent. Even so, the majority of the power is going to end up as heat in the LED and LEDs don't like to get hot.

Generally they are mounted on Insulated Metal Substrate (IMS) to provide the circuit tracking and heatsinking together. One simple test would be to measure the volts and amps going in to the LED (multiply them together to get Watts) and see how that compares with the claimed spec. It would be interesting to know what you find but it sounds unlikely to be close to 9W. Do they rate these lamps according to the input power or the LED power?. You probably can't change anything, as it's regulating the current in the LEDs regardless of input voltage. The example IMS in the link has UL approvals for operation at 150C and claims a dielectric strength of 2500VAC, so good for mains voltages. It's possible the original LED lamp used something of a similar rating but it should have had a cover to prevent any risk of touching live parts.

As LEDs age, their output falls. Ageing is highly dependent on temperature, so providing adequate heatsinking is key to getting a good life.

None of these LED lamps seem to have any energy storage and there's little reason why they would need them. They are probably conducting for 80-90% of the time anyway, giving low flicker. The simplest and cheapest (and safe) source for completely flicker-free lighting might be a 12V laptop power supply. However, I've never been able to see any signs of even mild strobing or flicker when using LED lights with machine tools and that's with El Cheapo IKEA bendy lights and VFDs on both the lathe and mill.

The legally binding safety regulations governing construction and protection of electrical products are broadly the same regardless of whether the power source is a battery charger, a power supply or a light driver, whether switched mode or linear, so the biggest risk is likely to be the use of products that don't have meaningful CE marking and/or from unknown sources. Stick to the names we know.

Murray

Edited By Muzzer on 03/10/2014 03:39:51

[Michael GilliganParticipant @michaelgilligan61133]

Posted by Muzzer on 03/10/2014 01:18:47:

(a)… the majority of the power is going to end up as heat in the LED and LEDs don't like to get hot.

(b) … One simple test would be to measure the volts and amps going in to the LED (multiply them together to get Watts) and see how that compares with the claimed spec. It would be interesting to know what you find but it sounds unlikely to be close to 9W. Do they rate these lamps according to the input power or the LED power?

(c) … As LEDs age, their output falls. Ageing is highly dependent on temperature, so providing adequate heatsinking is key to getting a good life.

(d) … None of these LED lamps seem to have any energy storage and there's little reason why they would need them. They are probably conducting for 80-90% of the time anyway, giving low flicker. The simplest and cheapest (and safe) source for completely flicker-free lighting might be a 12V laptop power supply. However, I've never been able to see any signs of even mild strobing or flicker when using LED lights with machine tools and that's with El Cheapo IKEA bendy lights and VFDs on both the lathe and mill.

.

Murray,

Thanks for joining-in

There are some useful points in your post, and I have extracted a few for easy reference.

(a) … True; and the big 'geometric' difference, between LED and incandescent, is that LEDs produce heat "out the back" [their light is cold], whereas incandescents lose a lot of their heat within the spectrum of the radiated light "out of the front" [their light is hot].

(b) … Yes; that's what I'm doing with my test 10W LED … it needs about 11v to consume 10W and then, within 10minutes, the heatsink gets too hot to hold. I will document this properly when I have the thermometer. I have no idea what the internal contruction of the Diall 9W Mains Bulb is [which is why I asked if anyone knows "how they do it"]; you may well be correct in hinting that they are under-running a high-rated LED.

(c) … TRUE

(d) … The laptop power supplies do seem to be very good, but for the ultimate stability I have settled on using battery as a source [wherever practical]. Flicker has only been a real problem with some [dedicated microscope] cameras, and will probably not concern most members of this forum. … Note: IKEA bendy lights have a Constant Current power supply; which is a GOOD way to drive LEDs.

Grateful for any further input … especially on the heat dissipation side.

MichaelG.

[Michael GilliganParticipant @michaelgilligan61133]

For ease of reference: This page has a photo of the Diall 9W lamps

… but it's the internal detail we need.

MichaelG.

[Neil WyattModerator @neilwyatt]

This page on Wikipedia doesn't support Murray's 10% figure for incandescent lamps, the figures are much closer to what I have long understood to be the case:

Luminous_efficacy

"Sources that depend on thermal emission from a solid filament, such as incandescent light bulbs, tend to have low overall efficacy because, as explained by Donald L. Klipstein, “An ideal thermal radiator produces visible light most efficiently at temperatures around 6300 °C (6600 K or 11,500 °F). Even at this high temperature, a lot of the radiation is either infrared or ultraviolet, and the theoretical luminous [efficacy] is 95 lumens per watt. No substance is solid and usable as a light bulb filament at temperatures anywhere close to this. The surface of the sun is not quite that hot.”^[20] At temperatures where the tungsten filament of an ordinary light bulb remains solid (below 3683 kelvins), most of its emission is in the infrared.^[20]"

 

Neil

Edited By Neil Wyatt on 03/10/2014 10:00:23

[Michael GilliganParticipant @michaelgilligan61133]

With the obvious caveat, that all light-bulbs are not created equal:

Here is a useful comparison table, covering the various types.

MichaelG.

[Michael GilliganParticipant @michaelgilligan61133]

Posted by Muzzer on 03/10/2014 01:18:47:

Do they rate these lamps according to the input power or the LED power?.

.

Hi, Murray

Here are a couple of photos, showing the "rating plates" for 6W and 9W Diall bulbs.

… Clearly rated by input power [which is after all, what all the eco-legislation is about]

MichaelG.

[Michael GilliganParticipant @michaelgilligan61133]

Here is a picture of my 10W LED on its Test HeatSink.

There is nothing special about the HeatSink design; it was just a convenient scrap of channel that will easily accept additional fins.

As it stands, it is woefully inadequate for this LED running at the full 10W.

Compare it with the size of the 9W Mains bulb and you will see why I want to know how they keep those bulbs cool. [for scale: the metal plate of the LED is about 20mm square]

MichaelG.

.

[Neil WyattModerator @neilwyatt]

You may find THIS useful Michael.

Neil

[frank brownParticipant @frankbrown22225]

Neil, I thought the link was going to a picture of a spray can of matt black paint.

Frank

[Michael GilliganParticipant @michaelgilligan61133]

Posted by Neil Wyatt on 03/10/2014 19:35:57:

You may find THIS useful Michael.

.

Thanks, Neil

I actually rather like this one

MichaelG.

.

P.S. This is one of the best pieces I know on HeatSink design.

… But it would be … he's an Audio legend.

Edited By Michael Gilligan on 03/10/2014 20:20:24

[Neil WyattModerator @neilwyatt]

> he's an Audio legend.

Why was I expecting Ivor Catt?

Neil

[Anonymous]

Posted by Neil Wyatt on 03/10/2014 20:51:59:

> he's an Audio legend.

Why was I expecting Ivor Catt?

Neil

I've no idea! Most of his rants in Wireless World were to do with EM theory and how everybody else, including Maxwell, had got it all wrong. He was also involved with Anamartic and wafer scale integration, until it went bust. During my period working in the mad world of motor racing (with Pi Research) we moved into the old Anamartic offices in Milton Hall, near Cambridge.

Regards,

Andrew

 

Edited By Andrew Johnston on 03/10/2014 21:45:14

Edited By Andrew Johnston on 03/10/2014 21:46:26

[Michael GilliganParticipant @michaelgilligan61133]

The mind is boggled …

Here is a mains bulb with LED filaments built on sapphire substrate.

… and check the price !!

MichaelG.

.

Edit: You can even buy the filaments

Edited By Michael Gilligan on 03/10/2014 22:48:48

[Michael GilliganParticipant @michaelgilligan61133]

For info. here is a quick check on the 10W LED pictured earlier.

Applied Voltage vs Current Drawn [you can work out the Wattage for yourselves]

I used two digital meters, each of which has only two decimal places; so it's fairly rough … but it shows the general characteristic.

MichaelG.

 

 

P.S.  the image was posted into my Album as a .png file [and presumably the forum software has converted it  to .jpg] … which shows that you can't believe everything you read in the intructions.

Edited By Michael Gilligan on 04/10/2014 10:01:25

Edited By Michael Gilligan on 04/10/2014 10:04:42

Edited By Michael Gilligan on 04/10/2014 10:09:13

[Neil WyattModerator @neilwyatt]

This is interesting, you can see the hint of the non-linearity as the diode starts to conduct.

Ttwo examples I looked at in detail, both show (but don't mention) the interesting effect that as current increases the voltage across a green LED increases and then drops. I wonder if other colours do this is over-driven?

Neil

[Michael GilliganParticipant @michaelgilligan61133]

Posted by Neil Wyatt on 04/10/2014 10:35:03:

This is interesting, you can see the hint of the non-linearity as the diode starts to conduct.

.

… and also [I believe] the roll-off as the LED saturates beyond about 11v.

MichaelG.

[Michael HornerParticipant @michaelhorner54327]

Hi

Anybody used the 5M LED strips to make up lighting strips. I desparately need some extra lighting, I have 1 CFL in a single garage. I was going to go down the flourescent tube route and saw this thread.

TIA

Michael.

[Michael GilliganParticipant @michaelgilligan61133]

Posted by Michael Horner on 05/10/2014 11:56:49:

Anybody used the 5M LED strips to make up lighting strips. I desparately need some extra lighting, I have 1 CFL in a single garage. I was going to go down the flourescent tube route and saw this thread

.

Michael,

I haven't used them myself, but I have seen them [cut to length] stuck inside a cylinder, as a ring-light for a stereo-microscope … which worked very nicely.

I think you may be on to a winner there !!

MichaelG.

[Oompa LumpaParticipant @oompalumpa34302]

Posted by Michael Horner on 05/10/2014 11:56:49:

Hi

Anybody used the 5M LED strips to make up lighting strips. I desparately need some extra lighting, I have 1 CFL in a single garage. I was going to go down the flourescent tube route and saw this thread.

TIA

Michael.

I have used them extensively in my office and workshop with great success. I will try to dig some pictures up. What I did was nail plasterers J Bead up towards the top of the walls and then blend it in with plaster. I then ran LED strip around the walls so it gives an uplight effect.

graham.

[Oompa LumpaParticipant @oompalumpa34302]

Here you go, this is all LED strip.

Gives you a general idea anyhow. Works well.

graham.

[Michael HornerParticipant @michaelhorner54327]

Posted by Oompa Lumpa on 05/10/2014 16:33:09:

Here you go, this is all LED strip.

Gives you a general idea anyhow. Works well.

graham.

Hi Graham

Looks great. Are they the IP65 variant of the strip? The reason I ask I have used the IP65 type before for under cupboard lighting and for the first 3-4 months it was great but then the light output would fall off. Checking the strip I noted the resistors had burnt! This happened on more than one occasion so I am not keen to repeat. The house was unaturaly warm (30deg) (SHMBO) may have had something to do with it. The garage wont be that hot.

Cheers Michael.

[Oompa LumpaParticipant @oompalumpa34302]

Frankly Michael, I cannot remember. They have been there for three years now and they are in daily use.

The issue with LED's I feel is they are developing so fast some people are reluctant to dip their toe in, but my question for those people is – When do you want to start saving money? I have a LOT of LED lights in the office and workshop, even the fluorescents are now LED and it has made a difference.

graham.

[Michael HornerParticipant @michaelhorner54327]

Thanks for that Graham.

I will try a reel of none IP65 because they will get rid of the heat better. I don't need the IP65 rating because they are going to be ceiling mounted for general lighting.

Cheers Michael.

[Author]

Posts