LED lighting...what's the best way?

[fungus]

Probably a stupid question but here goes...

I'm redecorating a room and I want to dabble with LED lighting to see
how it goes. If it works out OK I might do some more.

There seem to be two ways to wire things up:

a) Use little current-regulated power supply boards to go directly
from mains A/C to LED power.

b) Set up a 12V DC power supply then put in little current regulator
boards as necessary for the LED arrangement.

Option (b) costs a little bit more (which is fine) and seems to be the
more 'professional' way to do it.

As electrical engineers, is there any real reason to prefer one method
over the other?

eg. Option (b) seems to introduce more points of failure but it might
distribute the heat better. Also the power supplies can be a bit
bigger. OTOH, option (a) uses basically the same switch-mode circuitry
as the power supply in (b) so why bother with the extra step?


Is it just a case of "don't buy cheap-ass stuff and you'll be fine"?

(Easier said than done - price is no guarantee of quality...)


Note: I'm probably going to buy the parts on eBay. I'm good at eBay
though - I live in a foreign place so I get lots of practice.

 

[Jamie]

fungus wrote:

Probably a stupid question but here goes...

I'm redecorating a room and I want to dabble with LED lighting to see
how it goes. If it works out OK I might do some more.

There seem to be two ways to wire things up:

a) Use little current-regulated power supply boards to go directly
from mains A/C to LED power.

b) Set up a 12V DC power supply then put in little current regulator
boards as necessary for the LED arrangement.

Option (b) costs a little bit more (which is fine) and seems to be the
more 'professional' way to do it.

As electrical engineers, is there any real reason to prefer one method
over the other?

eg. Option (b) seems to introduce more points of failure but it might
distribute the heat better. Also the power supplies can be a bit
bigger. OTOH, option (a) uses basically the same switch-mode circuitry
as the power supply in (b) so why bother with the extra step?


Is it just a case of "don't buy cheap-ass stuff and you'll be fine"?

(Easier said than done - price is no guarantee of quality...)


Note: I'm probably going to buy the parts on eBay. I'm good at eBay
though - I live in a foreign place so I get lots of practice.
The option (B) as you say is the way to go.. There are a couple of

reasons I can think of.

1. The final string can operate from a conventional battery supply if
needed.

2. If you have a failed LED or R in one of the rows, the rest will
still work.


3. It's safer to handle.

Jamie

 

[ehsjr]

fungus wrote:

Probably a stupid question but here goes...

I'm redecorating a room and I want to dabble with LED lighting to see
how it goes. If it works out OK I might do some more.

There seem to be two ways to wire things up:

a) Use little current-regulated power supply boards to go directly
from mains A/C to LED power.

b) Set up a 12V DC power supply then put in little current regulator
boards as necessary for the LED arrangement.

Option (b) costs a little bit more (which is fine) and seems to be the
more 'professional' way to do it.

As electrical engineers, is there any real reason to prefer one method
over the other?

eg. Option (b) seems to introduce more points of failure but it might
distribute the heat better. Also the power supplies can be a bit
bigger. OTOH, option (a) uses basically the same switch-mode circuitry
as the power supply in (b) so why bother with the extra step?


Is it just a case of "don't buy cheap-ass stuff and you'll be fine"?

(Easier said than done - price is no guarantee of quality...)


Note: I'm probably going to buy the parts on eBay. I'm good at eBay
though - I live in a foreign place so I get lots of practice.

With option A, if you have multiple LEDs, they have to be in series,
and can't be individually controlled. For individual control you'd
need a power supply for each LED.

With option B, the LEDs can be individually controlled. You
don't need current regulator boards, per se. You could use a
dropping resistor in series with each LED.

Ed

 

[Rich Grise]

fungus wrote:

Probably a stupid question but here goes...

The only stupid question is the one you don't ask.

I'm redecorating a room and I want to dabble with LED lighting to see
how it goes. If it works out OK I might do some more.
snip

If it were me, I'd go to the local home hobbyist store (Lowe's, Home Depot,
etc. or your local equivalent) and see what's available off-the-shelf, and
go from there. Being able to get parts goes a _long_ way toward getting a
project finished!

Actually, I'm rather curious as to what's available in LEDs for
illumination.

Good Luck!
Rich

 

[Ecnerwal]

In article <j58bi5$vri$1@news.eternal-september.org>,
ehsjr <ehsjr@nospamverizon.net> wrote:

fungus wrote:
snippage
a) Use little current-regulated power supply boards to go directly
from mains A/C to LED power.

b) Set up a 12V DC power supply then put in little current regulator
boards as necessary for the LED arrangement.
snippage
With option A, if you have multiple LEDs, they have to be in series,
and can't be individually controlled. For individual control you'd
need a power supply for each LED.

Rarely an issue - hardly ever want to dim just one - simply set up
logical circuits as you would with any other lighting. However, it takes
a LOT of LEDs to use up the available voltage, and if you don't use up
most of the available voltage it's inefficient, as you are drawing (say)
20mA at whatever line voltage, regardless of how many LEDs you are
actually lighting, at least with most designs. May also be more prone to
expire in the face of transients on the power line, depending how it's
designed.

With option B, the LEDs can be individually controlled. You
don't need current regulator boards, per se. You could use a
dropping resistor in series with each LED.

....thus making them as inefficient as possible. Heating up resistors is
such a good use of power. Also, an actual current regulator allows you
to have an adjustable current (and one that's tolerant of and resistant
to slight voltage variations - important if trying to get the maximum
amount of light out of the LED but not fry it.) Heck, even white LEDs
can be run 3 (in series) from a 12V supply for the same exact power as
running one from a 12V supply, and produce 3 times the light for the
same power input. Reds you can get 5-6, depending on the LEDs. Finally,
backup power for backup/emergency lighting (using the primary lights) is
relatively easy with 12V (or 24V) lighting - and there are a lot of good
items from the car/RV market you can repurpose, if you roll your own
stuff.

24V is a somewhat better point just in terms of number of current
regulators / number of LEDs they can run, while still being "low
voltage" from the safety/regulatory point of view - unless you happen to
have some desire to control each individual LED's brightness, rather
than groups of them.

--
Cats, coffee, chocolate...vices to live by
Please don't feed the trolls. Killfile and ignore them so they will go away.

 

[fungus]

On Sep 20, 12:31 am, Rich Grise <ri...@example.net.invalid> wrote:

If it were me, I'd go to the local home hobbyist store (Lowe's, Home Depot,
etc. or your local equivalent) and see what's available off-the-shelf, and
go from there. Being able to get parts goes a _long_ way toward getting a
project finished!

(Haven't got either of those around here)

I already bought a couple of test LEDs on eBay to
play with.

One was a set of three 3W LEDs with a 12V->LED controller.
I hooked them up with an old wall-wart and they work fine.

The other was a 10W LED with a 240VAC->LED controller.

Both are too bright to look at directly, but even the 10W
wouldn't be enough to light up a room brightly. You'd need
three or four of them.

Both of them get HOT - much too hot to touch. The 10W
LED is mounted on a pretty big heatsink, too. I'm thinking
of running them under-current to keep them cooler.

The curve of current to brightness isn't linear, I connected
the set of three in parallel instead of series and they still
give plenty of light without getting much more than room
temperature.

The datasheet says they should be about half brightness
at one-third current. A two-by-two grid of them (eg.) would
run at half current and 70% of full brightness.

If I go with bigger LEDs I might buy 20W and run them at
half current. The difference between a 20W LED and a 10W
is only a couple of dollars so it seems worth it to not run
them on the ragged edge.

Actually, I'm rather curious as to what's available in LEDs for
illumination.

I'm quite impressed so far, there's still a lot of experimenting
to do though. (Isn't that why we're here...?)

 

[fungus]

On Sep 20, 2:30 am, Ecnerwal
<MyNameForw...@ReplaceWithMyVices.Com.invalid> wrote:

Rarely an issue - hardly ever want to dim just one - simply set up
logical circuits as you would with any other lighting.

Re: Dimming... I just notice that most of the 12V regulator
boards have PWM connections for dimming.

The A/C line voltage->LED boards don't.

 

[ehsjr]

Ecnerwal wrote:

In article <j58bi5$vri$1@news.eternal-september.org>,
ehsjr <ehsjr@nospamverizon.net> wrote:


fungus wrote:

snippage

a) Use little current-regulated power supply boards to go directly
from mains A/C to LED power.

b) Set up a 12V DC power supply then put in little current regulator
boards as necessary for the LED arrangement.

snippage

With option A, if you have multiple LEDs, they have to be in series,
and can't be individually controlled. For individual control you'd
need a power supply for each LED.


Rarely an issue - hardly ever want to dim just one - simply set up
logical circuits as you would with any other lighting.

So what? That statement was, and is true. It is a factor the op
should consider when making his decision.

However, it takes
a LOT of LEDs to use up the available voltage, and if you don't use up
most of the available voltage it's inefficient, as you are drawing (say)
20mA at whatever line voltage, regardless of how many LEDs you are
actually lighting, at least with most designs. May also be more prone to
expire in the face of transients on the power line, depending how it's
designed.


With option B, the LEDs can be individually controlled. You
don't need current regulator boards, per se. You could use a
dropping resistor in series with each LED.


...thus making them as inefficient as possible. Heating up resistors is
such a good use of power.

So what? Those statements were, and are true. They are factors the op
should consider when making his decision.

Ed

Also, an actual current regulator allows you
to have an adjustable current (and one that's tolerant of and resistant
to slight voltage variations - important if trying to get the maximum
amount of light out of the LED but not fry it.) Heck, even white LEDs
can be run 3 (in series) from a 12V supply for the same exact power as
running one from a 12V supply, and produce 3 times the light for the
same power input. Reds you can get 5-6, depending on the LEDs. Finally,
backup power for backup/emergency lighting (using the primary lights) is
relatively easy with 12V (or 24V) lighting - and there are a lot of good
items from the car/RV market you can repurpose, if you roll your own
stuff.

24V is a somewhat better point just in terms of number of current
regulators / number of LEDs they can run, while still being "low
voltage" from the safety/regulatory point of view - unless you happen to
have some desire to control each individual LED's brightness, rather
than groups of them.

 

[fungus]

On Sep 20, 2:30 am, Ecnerwal
<MyNameForw...@ReplaceWithMyVices.Com.invalid> wrote:

...thus making them as inefficient as possible.

Re: Efficiency...

I just did a quick measurement of Volts and
Amps to get an idea of efficiency with the
regulator boards.

With a 12V supply and one LED connected
(worst case scenario) I measured 78% efficiency.

nb. This is almost exactly what the datasheet
says I should get ... rising to 90-odd percent
efficiency with more LEDs.
..
PS: The PSU has a little adjustment potentiometer
to tweak the voltage. I can dial it down to about
10V. Would that be worth doing if I don't need 12V
for the LEDs?

 

On Mon, 19 Sep 2011 15:31:37 -0700, Rich Grise
<richg@example.net.invalid> wrote:

fungus wrote:

Probably a stupid question but here goes...

The only stupid question is the one you don't ask.

I'm redecorating a room and I want to dabble with LED lighting to see
how it goes. If it works out OK I might do some more.
snip

If it were me, I'd go to the local home hobbyist store (Lowe's, Home Depot,
etc. or your local equivalent) and see what's available off-the-shelf, and
go from there. Being able to get parts goes a _long_ way toward getting a
project finished!

Actually, I'm rather curious as to what's available in LEDs for
illumination.

Good Luck!
Rich

check out

http://stores.ebay.com/Shop4leds?_trksid=p4340.l2563

Up to 100 watt modules...

The 1-5 watt modules need a lot of heatsink if they are to survive
summer outdoors, the 100 watt beast is probably hard to cool. The
"star" mounted leds aren't good to go - that little aluminum chunk is
just a heat spreader not a heat sink.

I had some 10 mm "straw hat" leds (from their shape) that were rated
at 100 ma - with a constant current source they were dying as soon as
the weather turned warm. 50 ma and they are fine (for 1-1/2 summers
now) Each has 4 chips embedded in a chunk of plastic, the dying ones
were visibly damaged - they turn brown. My holder is a 3/8" thick
rectangular aluminum bar with the holes reamed to friction fit the
leds (theoretically these don't need a heatsink) In a flashlight
maybe...

My next project is under counter lights with a light strip ~$40 for 5
meters of warm white smd leds mounted and run from 12 VDC.

 

[Rich Grise]

Ecnerwal wrote:

In article <j58bi5$vri$1@news.eternal-september.org>,

With option B, the LEDs can be individually controlled. You
don't need current regulator boards, per se. You could use a
dropping resistor in series with each LED.

...thus making them as inefficient as possible. Heating up resistors is
such a good use of power.

So, look up the peak current spec, and design a chopper to get the right
duty cycle so you don't make them go phut!

Cheers!
Rich

 

[fungus]

On Sep 21, 5:37 pm, default wrote:

Up to 100 watt modules...

Try searching for "150W LED"...

http://www.ebay.com/sch/i.html?_nkw=150w+led

The 1-5 watt modules need a lot of heatsink if they are to survive
summer outdoors, the 100 watt beast is probably hard to cool.  The
"star" mounted leds aren't good to go - that little aluminum chunk is
just a heat spreader not a heat sink.

The big ones seem to use PC CPU coolers for cooling
(i.e. with heatpipes and fans).

My 10W came bolted to a heatsink so I'm guessing
it's supposed to be able to cope. Still too hot to touch
though...

 

[fungus]

On Sep 21, 5:37 pm, default wrote:

The "star" mounted leds aren't good to go - that little
aluminum chunk is just a heat spreader not a heat sink.

Yep, I only ran mine for 30 seconds at a time
at full power. They get real hot.

This is why I'm thinking of underpowering
them. The output curve is very non-linear
and there's a point when increasing the
current generates far more heat than light.

Going back to the original question, I've
decided to go the 12V route. It just gives
me a fuzzier feeling than messing with
mains A/C.

 

[fungus]

On Sep 22, 12:04 am, Rich Grise <ri...@example.net.invalid> wrote:

So, look up the peak current spec, and design a chopper to get the right
duty cycle so you don't make them go phut!

The controller boards are about $6 on eBay
(free shipping).

How much is a 10W resistor?

Anybody want to work out how long it will
take for the controller board to pay for
itself? ;-)

 

On Wed, 21 Sep 2011 13:47:47 -0700 (PDT), fungus
<openglMYSOCKS@artlum.com> wrote:

On Sep 21, 5:37 pm, default wrote:
The "star" mounted leds aren't good to go - that little
aluminum chunk is just a heat spreader not a heat sink.


Yep, I only ran mine for 30 seconds at a time
at full power. They get real hot.

This is why I'm thinking of underpowering
them. The output curve is very non-linear
and there's a point when increasing the
current generates far more heat than light.

Going back to the original question, I've
decided to go the 12V route. It just gives
me a fuzzier feeling than messing with
mains A/C.

I tend to run them under current in everything. There seems to be no
standardized specifications, and even if there were, it is not easy to
figure out how much heat sinking one needs.

Well I did "mess with mains AC" when the white "stars" came out. I
made a proper heat sink and mounted three of them and powered them
using a capacitor to drop voltage (which I've been doing for many
years for small night lights) They worked very well for a few weeks
then died. The circuit board that insulates them from the aluminum
slug they are mounted on shorted out - so I'm thinking they aren't
tested for high potential insulating properties.

BTW I've had pretty good success using RTV silicone rubber to glue
stars to heat sinks. I just keep pressure on them until the silicone
sets up. You can use screws to mount them (like a 2-56 size) but
there's a good chance of shorting the thin circuit board to ground
since there may be exposed copper where they punch the holes.

I did a thermometer with LEDs to signal the temperature outside. I
used a tri color high flux (piranha style square package) with a
picaxe to do the logic and a DS18B20 for a (digital) sensor. It
flashes the temperature tens then a pause and units. Very cold is
tens blue units blue, then tens blue units green, green - green, green
- red, red -red. It's easy to read through a glass block wall and
mounted on a pole outside.

 

On Wed, 21 Sep 2011 13:26:48 -0700 (PDT), fungus
<openglMYSOCKS@artlum.com> wrote:

On Sep 21, 5:37 pm, default wrote:

Up to 100 watt modules...


Try searching for "150W LED"...

http://www.ebay.com/sch/i.html?_nkw=150w+led


The 1-5 watt modules need a lot of heatsink if they are to survive
summer outdoors, the 100 watt beast is probably hard to cool.  The
"star" mounted leds aren't good to go - that little aluminum chunk is
just a heat spreader not a heat sink.


The big ones seem to use PC CPU coolers for cooling
(i.e. with heatpipes and fans).

My 10W came bolted to a heatsink so I'm guessing
it's supposed to be able to cope. Still too hot to touch
though...
Out of my league perhaps...

If I have to get that fancy a fluorescent light makes more sense.

 

On Thu, 22 Sep 2011 12:36:03 -0700 (PDT), fungus
<openglMYSOCKS@artlum.com> wrote:

On Sep 22, 8:39 pm, default wrote:

I tend to run them under current in everything.  There seems to be no
standardized specifications, and even if there were, it is not easy to
figure out how much heat sinking one needs.


I figure the ebay sellers are making wild claims
about wattage to up the prices ("20W" sells
for more than "10W").

Most Ebay sellers. Particularly China.

I did a thermometer with LEDs to signal the temperature outside.  I
used a tri color high flux (piranha style square package) with a
picaxe to do the logic and a DS18B20 for a (digital) sensor.

The next project in my head project involves high power
RGB leds and an Arduino...

It flashes the temperature tens then a pause and units.

Let's hope it's Celsius...

Actually no, Fahrenheit has better resolution, and I'm in the US and
used to it. The reason for using colors is that it makes it fairly
easy to count. Tens flash more slowly than units and most of the time
I just have to count units. The chip has raw resolution of something
like .065 degrees C...

It sounds unwieldy to describe it - but in a day or two it is second
nature.

 

[fungus]

On Sep 22, 8:39 pm, default wrote:

I tend to run them under current in everything.  There seems to be no
standardized specifications, and even if there were, it is not easy to
figure out how much heat sinking one needs.

I figure the ebay sellers are making wild claims
about wattage to up the prices ("20W" sells
for more than "10W").

I did a thermometer with LEDs to signal the temperature outside.  I
used a tri color high flux (piranha style square package) with a
picaxe to do the logic and a DS18B20 for a (digital) sensor.

The next project in my head project involves high power
RGB leds and an Arduino...

It flashes the temperature tens then a pause and units.

Let's hope it's Celsius...

 

[fungus]

On Sep 22, 10:53 pm, default wrote:

On Thu, 22 Sep 2011 12:36:03 -0700 (PDT), fungus
I figure the ebay sellers are making wild claims
about wattage to up the prices ("20W" sells
for more than "10W").

Most Ebay sellers.  Particularly China.

I just received another LED, it's 20W and
comes attached to a heatsink.

I wanted to see what happens when you run
a 20W LED with a 10W driver (which is what
I'm planning to do...)

The heatsink looked quite big in the photos but
it's really no bigger then the heatsink on my
10W LED.

It gets too hot to touch at 10W, not quite as hot
as the 10W LED but still pretty damn hot. There's
NO WAY that's going to run at 20W and survive.

On the plus side it definitely puts out more light
at that power than the 10W LED does. I think it
will be workable with a bit of tweaking (different
heatsinks or maybe a different controller board).

Let's hope it's Celsius...

Actually no, Fahrenheit has better resolution

I was just thinking Celsius would be less flashes
to count.

 

On Fri, 23 Sep 2011 05:06:46 -0700 (PDT), fungus
<openglMYSOCKS@artlum.com> wrote:

On Sep 22, 10:53 pm, default wrote:
On Thu, 22 Sep 2011 12:36:03 -0700 (PDT), fungus
I figure the ebay sellers are making wild claims
about wattage to up the prices ("20W" sells
for more than "10W").

Most Ebay sellers.  Particularly China.


I just received another LED, it's 20W and
comes attached to a heatsink.

I wanted to see what happens when you run
a 20W LED with a 10W driver (which is what
I'm planning to do...)

The heatsink looked quite big in the photos but
it's really no bigger then the heatsink on my
10W LED.

It gets too hot to touch at 10W, not quite as hot
as the 10W LED but still pretty damn hot. There's
NO WAY that's going to run at 20W and survive.

On the plus side it definitely puts out more light
at that power than the 10W LED does. I think it
will be workable with a bit of tweaking (different
heatsinks or maybe a different controller board).

Err on the side of caution, particularly when they are expensive.

I have a 120 VAC 7.5 watt Led light bulb. Figuring that the
refrigerator would benefit most I spent the $15 it cost. Alas the
heat sink at the base precludes it from fitting in the socket and most
of the light comes out the end. Great for a table lamp though.

It gets hot to the touch after being on a few hours but not scalding
hot, maybe ~120F.

Leds don't usually die completely just get dimmer with time. That's
where some standardization in testing needs to be done. So they last
50,000 hours - but they are down to half the light output at 10,000
hours.

Let's hope it's Celsius...

Actually no, Fahrenheit has better resolution

I was just thinking Celsius would be less flashes
to count.

Yes but... I seldom have to count the tens since the temperature

doesn't change that dramatically and the color changes make it a lot
easier to "count." Tens flashes slower than units - all those things
make reading it almost automatic after a short while. (and it looks
good coming through a glass block wall)