An Experiment With L E D Lighting

[david96]

The arrival of my tools and test kit from the UK (including my Lux-meter) means I now have some quantitative measurements:-

Two measurements were taken:-

a. 1 Metre below the centre of the fitting

b. 1 Metre below the centre of the fitting and off-axis by 1m (to get an idea of beam width).

Results.

...................................... On Axis Off axis

34W circular fluorescent:- 360 Lux 190 Lux

21W warm white LED:-.... 300 Lux 194 Lux

21W white LED:-............. 384 Lux 165 lux

9W white LED:-............... 260 Lux 88 Lux

What would the readings be taken at 1 metre above floor level and a ceiling height of 2.5metres, for this is what one would want in practice.

[Crossy]

What would the readings be taken at 1 metre above floor level and a ceiling height of 2.5metres, for this is what one would want in practice.

Some are never satisfied

OK a repeat of some measurements, 1m above the floor 2.5m ceiling

Results.

..................................... On Axis Off axis

34W circular fluorescent:- 135 Lux 90 Lux

21W white LED:-............. 128 Lux 88 Lux

[david96]

What would the readings be taken at 1 metre above floor level and a ceiling height of 2.5metres, for this is what one would want in practice.

Some are never satisfied

OK a repeat of some measurements, 1m above the floor 2.5m ceiling

Results.

..................................... On Axis Off axis

34W circular fluorescent:- 135 Lux 90 Lux

21W white LED:-............. 128 Lux 88 Lux

Thanks for the information.

This height above floor level is the level where you require the level of illumination to work.

From your figures the illumination level has fallen 50%.

[dingdongrb]

Crossy seems to have done his homework and is starting to become quite knowledgeable about LED lighting, my hats off to him as I have always thought him to be a rather wise gent....

Although I am not currently in the field of LED lighting anymore I do have about 3 years experience working with LED lighting. I was employed with one of the very first companies that began the journey of designing lights using LEDs as the light source. At that time there were only a few players in the game. However today with everyone going 'green' there are more players than one can imagine. Any many more copycats from the Asian markets. The company I worked for was purchased a few years back by even a larger company. The larger company wanted the patents and technical rights of the smaller company.

Many can make LEDs lights like I mentioned but the difficult part is two-fold; making it inexpensive so one will buy it and the making it where it will last and not burn out. There is however another important factor with LED lights that the basic consumer isn't so aware of and thats the wavelength of the light being emitted. Many will say their white light is within a certain wave-band of the light spectrum. This is noted because if one would string 20 LED lights together some many seem a little more yellow, pink, or blue. Very important when using in a commercial environment and not a big factor when only one is being displayed.

When I first started mfg LED lights the cost of a MR-16 LED (7W usage and nearly a 60W light output) was around $80 USD each. Since that time the cost has come down considerably (maybe $15-20). One because the production processes have gotten better, two because more volume and that means larger price breaks on materials, and three because of competition. LED lights, if properly mfg and installed will last, as Crossy stated 50k hrs+... The savings in electricity is 70% from using traditional lighting. The LED lights are GREEN thus no mercury as in fluorescents... They run cooler, and if you drop one there is no glass to clean up as well as the light will still work. The commercial market was first targeted because one figured they could identify spending $70 on a light where the consumer would laugh. In the commercial field, as the old joke goes, it takes two to change a light....One to hold the ladder and one to change the light. You may joke but with OSHA it is true... and in the USA and European markets there are labor unions. So paying two guys $40/hr or so to change a light is expensive. And with a traditional light only lasting 6mths to a year this can add up very quickly when your factory has thousands of lights. Then as I mentioned there is the savings in electricity using LEDs. $$$$$$$

The little LEDs on the metal (I assume copper) pcs that Crossy first showed are what the industry calls 'Light Engines'. Yes it takes a Driver to drive these LEs and as Tywais stated usually it is a PWM driver as a constant current source. Another factor which Crossy pointed out is the need to heatsink the LEDs. They are small but very bright and powerful so they do need to be kept cool in order to last.

Anyone can make an LED light but the real key is making the LED. There have been many who try and were not successful. Most LED lighting mfgs usually buy the LEs from someone who has success at making them and then use them to mfg their trademark light. OSRAM and Sylvania bought both the LEs and lights from the company I worked for and just had us put their name and logo on the light and package. To make a good LE and be successful and making them binned to the correct (or same) wavelength one needs to properly understand the process and applications for applying phosphorus to the encapsulation material over the LEDs. Certain colors of phosphorous and mixes of will produce WHITE light. Not too many are successful for doing that.

If you would really like to know more about the details of LED lights, how to use them, how to make them, or where to go to buy the good ones then PM me and I'll see what I can to provide a helping hand....

BTW, the company I worked for actually designed the LEDs used in the NYC ball that drops every year (Times Square)..... Although it was somewhat of a joint effort with Phillips it was our LEDs being used.

[Crossy]

A useful effect which I've not really verified is that the lack of UV emissions from these lights means they don't attract insects. Bad for the geckos good for us

The LED fittings don't seem to be filling up with dead bugs (as the fluorescent ones did) so there is definitely something about the light that they don't like.

The company I mentioned earlier has supplied LED floodlighting to the new theme-park / casino in Singapore as well a LOTS of bulbs to a certain mouse (Mickey that is). The bulbs for Disney were actually designed to have the same yellow colour and beam pattern as the under-run incandescents used on the rides, you'd be amazed how fussy mice can be when it comes to the look of the rides (we have a nice collection of 'reject' samples that all look the same to me).

Rest assured, I'm encouraged by the success of my home-brew lighting and once our new home is completed it will be entirely LED lit, I'll never buy another conventional bulb again

[David006]

all good stuff guys...just one question...it seems rechargeable LED flash lights are quite inexpensive...anyone know if permanently wired to 240v ...would the charger/battery likely to self destruct? Was thinking of building emergency and yard lighting with a couple as here in the boonies we are forever having outages.

Did have one charger for rubber cutter head lights start to smoke and melt after a week or so.. gas emission was awful! So maybe I am answering my own question...any thoughts on modification?

[Crossy]

Did have one charger for rubber cutter head lights start to smoke and melt after a week or so.. gas emission was awful! So maybe I am answering my own question...any thoughts on modification?

I've had a couple apart after they stopped charging. To my horror (but not surprise) I discovered the charger consisted of a 1N4004 rectifier, a small LED charge indicator and a capacitive dropper / current limiter (which had gone pop), absolutely no mains isolation!!!!

Years ago I had a 'battery eliminator' which was the same size as a 9V battery which worked on the same principle, used it for years with my baby trannie and headphones. Only in later life did I come to realise exactly how close I must have been to frying (literally) my brain (some would say I wasn't close and did actually fry it).

As a minimum I'd rip out the charging circuit and build something simple and mains isolated, you can then over-design it so it doesn't fry

I've seen lots of baby switching PSUs for a few Baht (look in the assorted-bits shops in Pantip and Fortune), they are intended to run iPods etc via the USB. The current-limited (and isolated) 5V output could be just what you're looking for (maybe with a charge current limiting resistor)

[Tywais]

I designed a UV lightbox for PCB (printed circuit board) film exposure for the laboratory due to the rainy season not given us enough sunlight most of the time. It consisted of 120 UV LEDs in a grid pattern. I purchased 3 packs of 50 UV Leds (150) for about $45.00 total. Since light exposure for PCB lithography needs a very even distribution the LEDs need to be closely matched. The simplest way was to use a constant current source and monitor the voltage drop across the LED. It ended up with 2 batches with close Vdrop with a smaller batch way off.

Anway, I took a couple of minutes and designed a constant current source LED/Diode test box and as you can see the circuit is very simple and runs off a 9V battery. Removing the zener and using a simple PWM such as a PIC micro it can also be a constant current LED dimmer.

[dingdongrb]

I designed a UV lightbox for PCB (printed circuit board) film exposure for the laboratory due to the rainy season not given us enough sunlight most of the time. It consisted of 120 UV LEDs in a grid pattern. I purchased 3 packs of 50 UV Leds (150) for about $45.00 total. Since light exposure for PCB lithography needs a very even distribution the LEDs need to be closely matched. The simplest way was to use a constant current source and monitor the voltage drop across the LED. It ended up with 2 batches with close Vdrop with a smaller batch way off.

Anway, I took a couple of minutes and designed a constant current source LED/Diode test box and as you can see the circuit is very simple and runs off a 9V battery. Removing the zener and using a simple PWM such as a PIC micro it can also be a constant current LED dimmer.

You must have bought the cheap one as I pay approx $6 USD each for the ones I currently use to UV cure adhesives....

[Tywais]

You must have bought the cheap one as I pay approx $6 USD each for the ones I currently use to UV cure adhesives....

Yep, could have paid an arm and a leg as some of them are even much more expensive then $6. It was a hobby shop overseas that apparently buys them by the thousands direct from manufacturer. Hmm, $6.00 = ~23,000 Baht for 120. Could nearly buy a commercial unit for that.

Actually in another project, lumino-fluorescence, we need a sub 200nM high intensity narrow band UV led - price around $300.00 for one.

[GungaDin]

Are these bulbs you are discussing the same LEDs as used in 12V flexible strips on motorcycles?

[Crossy]

Are these bulbs you are discussing the same LEDs as used in 12V flexible strips on motorcycles?

Sort of

Those strips have low power (>0.1 Watt) LEDS along with simple electronics to stop them frying.

We are really discussing power LEDS (>1 Watt) and more complex electronics to drive them efficiently from the mains (you can drive them from 12V too).

What's your application GD?

[GungaDin]

^^^^ http://www.youtube.com/user/TwistedThrottl...u/2/xLwkx0Ysy5A

I have a Chinese Givi (top box) in Customs that I hope to clear on Friday.

The demo is only 45 secs long but better turn the sound down.

[Crossy]

What would the readings be taken at 1 metre above floor level and a ceiling height of 2.5metres, for this is what one would want in practice.

Some are never satisfied

OK a repeat of some measurements, 1m above the floor 2.5m ceiling

Results.

..................................... On Axis Off axis

34W circular fluorescent:- 135 Lux 90 Lux

21W white LED:-............. 128 Lux 88 Lux

Thanks for the information.

This height above floor level is the level where you require the level of illumination to work.

From your figures the illumination level has fallen 50%.

That figures as the sensor is now 1.5m from the source.

Since illumination follows an inverse-square law (only really true for a point source) the new level should be 360 / (1.5^2) = 360 / 2.25 = 160 . OK not exact but within the realms of experimental (in the bedroom after beer) error. The LED or fluorescent arrangement cannot be considered (even remotely) as a point-source.

Certainly the LED and fluorescent are producing comparable illumination levels with a significant power saving for the LED.

[Crossy]

Anway, I took a couple of minutes and designed a constant current source LED/Diode test box and as you can see the circuit is very simple and runs off a 9V battery. Removing the zener and using a simple PWM such as a PIC micro it can also be a constant current LED dimmer.

Hi Tywais.

May I humbly suggest that for more accurate operation the LED should actually be in the emitter circuit between the emitter and the sense resistor rather than in the collector.

With your arrangement (which is fine for your purposes) the base current (between 1/30 and 1/300 of the collector current with a 2N3904 h_FE of between 30 and 300) does not go through the LED but it does go through the sense resistor introducing a measurable error.

For PWM, I'd leave the zener in circuit (to set the maximum current), and pull the zener cathode down to 0V via an open-collector drive to provide PWM (0V on the cathode gives zero LED current).

Of course, we're both showing our age by using bi-polar transistors

[Tywais]

Anway, I took a couple of minutes and designed a constant current source LED/Diode test box and as you can see the circuit is very simple and runs off a 9V battery. Removing the zener and using a simple PWM such as a PIC micro it can also be a constant current LED dimmer.

With your arrangement (which is fine for your purposes) the base current (between 1/30 and 1/300 of the collector current with a 2N3904 h_FE of between 30 and 300) does not go through the LED but it does go through the sense resistor introducing a measurable error.

Actually there will be no error difference between driving high side versus low side loads as the voltage at the input of the OPAMP is controlled directly by the voltage drop of the current sense resistor. All three elements in series means current is the same and the feedback voltage the same regardless. Of course there are whole chapters on high versus low side loads. I use both depending on circumstances. In other words, the gain isn't an issue as the opamp will pump the necessary base current to push the current through the feedback resistor pushing the voltage input to the opamp to match the set point. If I used a darlington with a gain of a thousand, just means the opamp requires less current to drive the base to get the same load current. In other words, this is effectively a closed loop feedback system.

Just finished designing and building a 20Amp version of the above for a focusing magnet system. Tracks perfectly. Of course the design is just a bit more sophisticated, but the fundamental block above is identical. In that design, I use parallel darlington transistors (4 of them for power dissipation) a .05 ohm sense resistor for the feedback and high side drive to the coil. In this case, low side would be an issue as it is an inductive load.

Of course, we're both showing our age by using bi-polar transistors

Actually I have a MOSFET version but this was cheaper and parts in our stock and built (pcb included) in a couple of hours. All my power switching designs I use power MOSFETs (motor control and similar).

[Crossy]

Actually there will be no error difference between driving high side versus low side loads as the voltage at the input of the OPAMP is controlled directly by the voltage drop of the current sense resistor. All three elements in series means current is the same and the feedback voltage the same regardless.

It's been a LONG time since I did any of this, but don't forget the base current, consider this:-

In your design.

I_LED = I_C

I_SENSE = I_E

I_E = I_C + I_B

Therefore I_LED = I_SENSE - I_B

Moving the LED into the emitter circuit:-

I_LED = I_E

I_SENSE = I_E

Therefore I_LED = I_SENSE

I'd left Uni. before we were even allowed to LOOK at a FET (and that was a JFET).

The apprentice school had huge bags of Mullard AC128 (Germanium PNP) transistors, these were production rejects donated by the factory so a quick check with the AVO to check for two junctions followed by a visit to the curve tracer (to check it had at least some gain) was needed before any meaningful design could be commenced.

Second year project was a medium-wave superhet, BC108s were handed out like gold (the AC128 wasn't fast enough).

Ah the memories, how about a Class A power (ish) amp using 5 AC128s in parallel (with sharing resistors or the whole thing ran away in a puff of smoke), huge fun.

Did you ever make a uni-junction transistor do anything useful? (I didn't).

[Tywais]

Actually there will be no error difference between driving high side versus low side loads as the voltage at the input of the OPAMP is controlled directly by the voltage drop of the current sense resistor. All three elements in series means current is the same and the feedback voltage the same regardless.

It's been a LONG time since I did any of this, but don't forget the base current, consider this:-

Ah yes, got to clean that rust off - most of my designs are digital now, probably 80% designs using microcontrollers. This is a case where using a MOSFET would not effect it as it is a voltage driven device as opposed to the bipolar's current device. I'm slowly switching over to primarily MOSFET in linear drive now. For my 20A power design the error due to base current summing comes out to about .05% at full power which is acceptable.

For a clear picture of it, a simulation.

[dingdongrb]

Actually there will be no error difference between driving high side versus low side loads as the voltage at the input of the OPAMP is controlled directly by the voltage drop of the current sense resistor. All three elements in series means current is the same and the feedback voltage the same regardless.

It's been a LONG time since I did any of this, but don't forget the base current, consider this:-

Ah yes, got to clean that rust off - most of my designs are digital now, probably 80% designs using microcontrollers. This is a case where using a MOSFET would not effect it as it is a voltage driven device as opposed to the bipolar's current device. I'm slowly switching over to primarily MOSFET in linear drive now. For my 20A power design the error due to base current summing comes out to about .05% at full power which is acceptable.

For a clear picture of it, a simulation.

What simulation program do you use, spice?

[Tywais]

What simulation program do you use, spice?

National Instruments Multisim previous known as Electronics Workbench. I do most of my analog design with it before putting it on a pcb. Excellent software.

[dingdongrb]

What simulation program do you use, spice?

National Instruments Multisim previous known as Electronics Workbench. I do most of my analog design with it before putting it on a pcb. Excellent software.

Never tried that one, maybe I'll have to check it out since I am very familiar with LabView. Thx

[cloudhopper]

Great info everyone.

Would it improve the cost/reliability if the lighting circuits were 12 VDC and thereby get rid of all the separate little power supplies? Would be easy to do in new construction.

[Crossy]

Great info everyone.

Would it improve the cost/reliability if the lighting circuits were 12 VDC and thereby get rid of all the separate little power supplies? Would be easy to do in new construction.

Not really, you still need electronics to drive the LEDs with a constant current, plus you need the 12V supply. White LEDs drop about 3.5V so you can run 3 in series off 12V (plus 1.5V for the current source). You could get away with using a simple resistor as a current limiter, but that would be inefficient, we're looking at improving power usage

The other lights I built have a single power unit that drives 20 LEDs in series and the same company do a 50W (that's 50 1Watt LEDs) unit.

Obviously if you're off-grid 12V would be the way to go.

[cloudhopper]

^^^^^Yes I will be off grid with 12VDC. So is this right? - 3 LEDs in series each have 3.5 V drop so with nominal 14VDC input and 0.35A current a 10 Ohm resistor would provide the necessary remaining voltage drop and would dissipate 1.2W (about the same as each of the LEDs). Then a series-parallel arrangement to reach the required illumination? TIA

[Crossy]

^^^^^Yes I will be off grid with 12VDC. So is this right? - 3 LEDs in series each have 3.5 V drop so with nominal 14VDC input and 0.35A current a 10 Ohm resistor would provide the necessary remaining voltage drop and would dissipate 1.2W (about the same as each of the LEDs). Then a series-parallel arrangement to reach the required illumination? TIA

If you use a circuit like the one Tywais shows it will compensate for falling battery voltage, a little more complex and just as inefficient power wise.

A couple of ebay sellers have 12V versions of the 220V drivers I used, they are much more efficient, PM me if you want details (forum rules preclude my publishing the link here).

[Tywais]

^^^^^Yes I will be off grid with 12VDC. So is this right? - 3 LEDs in series each have 3.5 V drop so with nominal 14VDC input and 0.35A current a 10 Ohm resistor would provide the necessary remaining voltage drop and would dissipate 1.2W (about the same as each of the LEDs). Then a series-parallel arrangement to reach the required illumination? TIA

When I designed my 120 UV Led light box I found this calculator to be useful.

http://led.linear1.org/led.wiz

Some more info > white leds

[Crossy]

Some interesting and useful stuff in here too:-

AV02_0532EN.pdf

[cloudhopper]

Thanks for the very useful info on this topic guys.

[canopy]

Does HomePro have a quality LED lighting available? Are there other places that do?

[Crossy]

Does HomePro have a quality LED lighting available? Are there other places that do?

Sod all in HomePro (no surprises there).

IIRC there are a couple of specialist LED shops in BKK (others may remember names) with, sadly, specialist prices