Powering white LED headlamps

Mike said:

Has anybody done this before? Will try to figure a way to
modify them for constant current.

The Silicon Chip Luxeon driver about a year ago used the same IC as in
a lot of mobile phone car chargers. I've got a couple sitting at home
waiting for me to finish them off.

I'm fairly sure the Oatley Electronics module uses something similar,
and they may have schematics up.

Dave

On Sun, 06 Feb 2005 at 14:11 GMT, Mike (aka Bruce)
was almost, but not quite, entirely unlike tea:

Quoted message said:

Following recent threads,
I'd like to add an LED headlamp to my bike. Maybe a Luxeon, or a few of
the cheap 1/2W LEDs at oatley. I'd be nice to have a switch-mode supply,
so it can run efficiently off my 12V batteries.
My electronics knowledge is minimal, but I looked on the web for some
sample circuits. The problem is they use ICs that don't seem to be
easily available. Then it hit me - I have some similar power regulators

BTW - I've found that http://www.rs-components.com.au stock a lot of
ICs that I've needed in the past for power regulation.

--
TimC -- http://astronomy.swin.edu.au/staff/tconnors/
My cats are forbidden from walking on my computer keyboard on the desk
when I'm asdfjjhhkl;ljfd.;oier' puyykmm4hbdm9lo9j USING IT. --unk

Random Data said:

The Silicon Chip Luxeon driver about a year ago used the same IC as in
a lot of mobile phone car chargers. I've got a couple sitting at home
waiting for me to finish them off.

Ok, thanks. I looked in a few "car-chargers", and found the MC34063
chip. There is plenty of info on-line, and its easy to modify these
phone chargers for different voltage. Better than buying the chip.

Quoted message said:

I'm fairly sure the Oatley Electronics module uses something similar,
and they may have schematics up.

Thanks - i found it. Its a bit dissapointing though. It drives the LED
with a voltage regulator and series resistor. It uses a trimpot to
adjust the voltage for the particular LED used. I was hoping for better
ideas on a current regulator!
The design has around a 0.4V drop on the series resistor, so
that a drop of 0.1V on the LED would increase current by 25% .
Are all white LEDs stable enough over time and temperature for that
to work well? Will current go up or down as it gets hotter?

The best alternative I can think of is to loose 1.25V across a
resistor in series with the LEDs, and feed that to the comparator.
That would waste 10% of the power with 3 x 3.6V LEDs (12V output),
but give better current regulation. Or 25% with 1 LED.

Any better ideas?

Mike said:

Thanks - i found it. Its a bit dissapointing though. It drives the LED
with a voltage regulator and series resistor. It uses a trimpot to
adjust the voltage for the particular LED used. I was hoping for better
ideas on a current regulator!

You can still use it as a current regulated supply providing it has an
external voltage feedback loop (which it should have).

Use the current limiting resistor in series with the led on the negative
side, off the ground rail.

Feed the hot side of that resistor into the feedback loop, and adjust the
resistor value to suit the feedback voltage reference, and to adjust to a
suitable current for the led.

This way, that resitor serves two purposes, current limiting and current sensing.

If you can get a switcher with a low referece voltage, you can use a lower
value sense resistor, since it's limiting function is rather redundant in this
configuration.

If you use a more flexible switcher, you can even configure it as a voltage
boost regulator, and string all your output leds in series. With this method,
you can easily string leds with a total drive voltage higher than your supply.

Both current regulated, and more efficient than the equivalent simple series
resistor arrangement. More expensive though.
--
Linux Registered User # 302622 <http://counter.li.org>

Mike said:

Random Data said:

The Silicon Chip Luxeon driver about a year ago used the same IC as in
a lot of mobile phone car chargers. I've got a couple sitting at home
waiting for me to finish them off.

Ok, thanks. I looked in a few "car-chargers", and found the MC34063
chip. There is plenty of info on-line, and its easy to modify these
phone chargers for different voltage. Better than buying the chip.

Quoted message said:

I'm fairly sure the Oatley Electronics module uses something similar,
and they may have schematics up.

Thanks - i found it. Its a bit dissapointing though. It drives the LED
with a voltage regulator and series resistor. It uses a trimpot to
adjust the voltage for the particular LED used. I was hoping for better
ideas on a current regulator!
The design has around a 0.4V drop on the series resistor, so
that a drop of 0.1V on the LED would increase current by 25% .
Are all white LEDs stable enough over time and temperature for that
to work well? Will current go up or down as it gets hotter?

The best alternative I can think of is to loose 1.25V across a
resistor in series with the LEDs, and feed that to the comparator.
That would waste 10% of the power with 3 x 3.6V LEDs (12V output),
but give better current regulation. Or 25% with 1 LED.

Any better ideas?

The "pucks" sold by Luxeon are sealed in a nice solid epoxy block. Very
reliable and easy. Not much difference in price by the time you've
finished stuffing around.

Mike said:

The best alternative I can think of is to loose 1.25V across a
resistor in series with the LEDs, and feed that to the comparator.
That would waste 10% of the power with 3 x 3.6V LEDs (12V output),
but give better current regulation. Or 25% with 1 LED.

You could always bung in an op-amp (say an LM386) between the current sense resistor and the switcher to multiply the current sense voltage. Setting gain to 10 would mean you only drop 125mV across the current sense resistor - much more efficient. The efficiency would then be dictated by switching and inductor losses in the switcher itself.

OTOH, you could just spring for a Luxdrive. They're not cheap, but they're small, efficient, and work well (especially when you bung a 100n ceramic cap across the input so they don't interfere with wireless computers).

Regards,

Suzy

Oscar said:

The "pucks" sold by Luxeon are sealed in a nice solid epoxy block. Very
reliable and easy. Not much difference in price by the time you've
finished stuffing around.

The phone chargers are $5, and I have a few spare.
Power-puck is US$20 + post. And bulkier than I want/need.
And i _like_ "stuffing around" with DIY :-)

Oscar said:

The "pucks" sold by Luxeon are sealed in a nice solid epoxy block. Very
reliable and easy. Not much difference in price by the time you've
finished stuffing around.

The phone chargers are $5, and I have a few spare.
Power-puck is US$20 + post. And bulkier than I want/need.
And i _like_ "stuffing around" with DIY :-)

John Tserkezis said:

Mike said:

Thanks - i found it. Its a bit dissapointing though. It drives the LED
with a voltage regulator and series resistor. It uses a trimpot to
adjust the voltage for the particular LED used. I was hoping for better
ideas on a current regulator!

You can still use it as a current regulated supply providing it has an
external voltage feedback loop (which it should have).

Use the current limiting resistor in series with the led on the
negative side, off the ground rail.

Thats what I meant later in the post, but wasn't so clear.
The problem is that the reference is 1.25V, and I don't know an easy
way to avoid "wasting" that 1.25V at the full ouput current.
But its better if used in a step-up configuration, with a few LEDs in
series.

Quoted message said:

Feed the hot side of that resistor into the feedback loop, and adjust
the resistor value to suit the feedback voltage reference, and to adjust
to a suitable current for the led.

So R = 1.25V/0.15A = 8.3 ohm.

Quoted message said:

This way, that resitor serves two purposes, current limiting and
current sensing.

If you can get a switcher with a low referece voltage, you can use a
lower value sense resistor, since it's limiting function is rather
redundant in this configuration.

If you use a more flexible switcher, you can even configure it as a
voltage boost regulator, and string all your output leds in series.
With this method, you can easily string leds with a total drive voltage
higher than your supply.

OK. The common MC34063 chip will do that. It seems the best way for a
bike-lamp. One long string of 4 to 10 x 1/2W LEDs, and it'll run on
any battery I have handy. Maybe even a single LiIon cell.

But the sample circuits seem to be either step-up, or step-down.
So no good for running a single LED from a single LiIon cell.
But thats another project.

thanks for the replies. I've ordered the LEDs.

John Tserkezis said:

Mike said:

Thanks - i found it. Its a bit dissapointing though. It drives the LED
with a voltage regulator and series resistor. It uses a trimpot to
adjust the voltage for the particular LED used. I was hoping for better
ideas on a current regulator!

You can still use it as a current regulated supply providing it has an
external voltage feedback loop (which it should have).

Use the current limiting resistor in series with the led on the
negative side, off the ground rail.

Thats what I meant later in the post, but wasn't so clear.
The problem is that the reference is 1.25V, and I don't know an easy
way to avoid "wasting" that 1.25V at the full ouput current.
But its better if used in a step-up configuration, with a few LEDs in
series.

Quoted message said:

Feed the hot side of that resistor into the feedback loop, and adjust
the resistor value to suit the feedback voltage reference, and to adjust
to a suitable current for the led.

So R = 1.25V/0.15A = 8.3 ohm.

Quoted message said:

This way, that resitor serves two purposes, current limiting and
current sensing.

If you can get a switcher with a low referece voltage, you can use a
lower value sense resistor, since it's limiting function is rather
redundant in this configuration.

If you use a more flexible switcher, you can even configure it as a
voltage boost regulator, and string all your output leds in series.
With this method, you can easily string leds with a total drive voltage
higher than your supply.

OK. The common MC34063 chip will do that. It seems the best way for a
bike-lamp. One long string of 4 to 10 x 1/2W LEDs, and it'll run on
any battery I have handy. Maybe even a single LiIon cell.

But the sample circuits seem to be either step-up, or step-down.
So no good for running a single LED from a single LiIon cell.
But thats another project.

thanks for the replies. I've ordered the LEDs.

suzyj said:

Mike said:

The best alternative I can think of is to loose 1.25V across a
resistor in series with the LEDs, and feed that to the comparator.
That would waste 10% of the power with 3 x 3.6V LEDs (12V output),
but give better current regulation. Or 25% with 1 LED.

You could always bung in an op-amp (say an LM386) between the current
sense resistor and the switcher to multiply the current sense voltage

Ah, clunk! Memories of Electronics 100 coming back. Is there anything
that op-amps cannot do?
But now I'm thinking that a bipolar transistor is a current amplifier.
So just use a transistor with beta=100, and feed 1.5mA to the base.
Wire the LED between the regulator output (4V?) and collector. E to GND.
There has to be a catch. Looking at a spec sheet, I see gain
has a huge range. Are there common transistors with accurately fixed
gain? Or back to plan A, as in the Oatley kit.

Quoted message said:

OTOH, you could just spring for a Luxdrive. They're not cheap, but
they're small, efficient, and work well

And miss out on the electronics refresher course? :-)

Quoted message said:

(especially when you bung a
100n ceramic cap across the input so they don't interfere with wireless
computers).

Thanks for the warning.

Mike said:

But now I'm thinking that a bipolar transistor is a
current amplifier. So just use a transistor with beta=100,
and feed 1.5mA to the base.

Not a good idea. Transistor beta varies by lots, both with process variations and temperature. In any case, it's a voltage (as sensed by the current sense resistor) that you want to multiply.

Regards,

Suzy

suzyj said:

Not a good idea. Transistor beta varies by lots, both with process
variations and temperature.

Damn! Was afraid of that.

Quoted message said:

In any case, it's a voltage (as sensed by
the current sense resistor) that you want to multiply.

But with a constant-gain transistor, you could eliminate that resistor.
A pity no such thing exists.

BTW, here is a circuit for a 1.5V white LED driver. Might be good
for a helmet light:

http://www.elecdesign.com/Articles/Index.cfm?AD=1&ArticleID=5886