low-loss voltage regulator

[[email protected]]

I've got a bike helmet light (40 LED's 11,000 mcd's each, 7 AAA NiMH
batteries in series) that works great in terms of photonic spray. But
because the LED's need a precise 3.6 volts supply, I have added an
LM317 variable voltage regulator. This works fine, almost. The 317
works so inefficiently that the 2 square inch heat sink screwed onto
its tab gets almost too hot to touch. I'm looking for a variable 5-10
volts input, constant 3.6 volts output power supply that has very low
heat loss.
Bruce Ratcliffe,
Fresno

 

[mike]

I've got a bike helmet light (40 LED's 11,000 mcd's each, 7 AAA NiMH
batteries in series) that works great in terms of photonic spray. But
because the LED's need a precise 3.6 volts supply, I have added an
LM317 variable voltage regulator. This works fine, almost. The 317
works so inefficiently that the 2 square inch heat sink screwed onto
its tab gets almost too hot to touch. I'm looking for a variable 5-10
volts input, constant 3.6 volts output power supply that has very low
heat loss.
Bruce Ratcliffe,
Fresno

Look at some of the maxim and linear technology parts.
mike

--
Return address is VALID.
Wanted, PCMCIA SCSI Card for HP m820 CDRW.
FS 500MHz Tek DSOscilloscope TDS540 Make Offer
http://nm7u.tripod.com/homepage/te.html
Wanted, 12.1" LCD for Gateway Solo 5300. Samsung LT121SU-121
Bunch of stuff For Sale and Wanted at the link below.
http://www.geocities.com/SiliconValley/Monitor/4710/

 

[John Larkin]

I've got a bike helmet light (40 LED's 11,000 mcd's each, 7 AAA NiMH
batteries in series) that works great in terms of photonic spray. But
because the LED's need a precise 3.6 volts supply, I have added an
LM317 variable voltage regulator. This works fine, almost. The 317
works so inefficiently that the 2 square inch heat sink screwed onto
its tab gets almost too hot to touch. I'm looking for a variable 5-10
volts input, constant 3.6 volts output power supply that has very low
heat loss.
Bruce Ratcliffe,
Fresno

Switching regulator is the only way. National, Maxim, LTC.

What's the total LED current?

John

 

[[email protected]]

I've got a bike helmet light (40 LED's 11,000 mcd's each, 7 AAA NiMH
batteries in series) that works great in terms of photonic spray. But
because the LED's need a precise 3.6 volts supply, I have added an
LM317 variable voltage regulator. This works fine, almost. The 317
works so inefficiently that the 2 square inch heat sink screwed onto
its tab gets almost too hot to touch. I'm looking for a variable 5-10
volts input, constant 3.6 volts output power supply that has very low
heat loss.
Bruce Ratcliffe,
Fresno

Two things, I hope you're current limiting the LED with a resistor
(unless they are already regulated.) Two, do what the other guy said
and look for a more efficient regulator. If you do that, you can have
really low dropout. So you may be able to loose two or three of your
batteries to save weight. Since you have them all in series now, you'd
have the same battery life with 3/4s the weight. 5 cells even at the
lowest you should take them (.95V) would still put out 4.75 volts,
which is easy to design around.

 

[Mac]

I've got a bike helmet light (40 LED's 11,000 mcd's each, 7 AAA NiMH
batteries in series) that works great in terms of photonic spray. But
because the LED's need a precise 3.6 volts supply, I have added an
LM317 variable voltage regulator. This works fine, almost. The 317
works so inefficiently that the 2 square inch heat sink screwed onto
its tab gets almost too hot to touch. I'm looking for a variable 5-10
volts input, constant 3.6 volts output power supply that has very low
heat loss.
Bruce Ratcliffe,
Fresno

What kind of LED's need a precise 3.6 Volt supply? Usually the I/V
relationship of LED's is such that it makes more sense to drive them with
a current source than a voltage source.

Anyway, if you want to design a step-down regulator with an input range
of 5-10 Volts and an output of 3.6, you can probably find a good solution
in a TI or National or Linear Technologies chip. The only major external
components will be filter inductors and filter capacitors.

What you are looking for is a step-down regulator or buck
regulator. The first one I found at National's website, which may or may
not be suitable for your application is the LM2734Y.

Good luck!

--Mac

 

[John Woodgate]

I read in sci.electronics.design that [email protected] wrote

I've got a bike helmet light (40 LED's 11,000 mcd's each, 7 AAA NiMH
batteries in series) that works great in terms of photonic spray. But
because the LED's need a precise 3.6 volts supply, I have added an
LM317 variable voltage regulator. This works fine, almost. The 317
works so inefficiently that the 2 square inch heat sink screwed onto
its tab gets almost too hot to touch. I'm looking for a variable 5-10
volts input, constant 3.6 volts output power supply that has very low
heat loss.
Bruce Ratcliffe,
Fresno

You should NOT run LEDs from a constant-voltage supply. The current is
an exponential function of voltage and temperature. You need a constant-
current supply.

 

[Paul Burke]

The 317
works so inefficiently that the 2 square inch heat sink screwed onto
its tab gets almost too hot to touch. I'm looking for a variable 5-10
volts input, constant 3.6 volts output power supply that has very low
heat loss.

Apart from fairly small differences in quiescent current, ANY linear
voltage regulator will dissipate the same power, and consequently
produce the same heat. The only way to reduce the heat waste is to use a
switching regulator - google for "simple switcher". What current are you
running at?

Paul Burke

 

[Anders F]

Apart from fairly small differences in quiescent current, ANY linear
voltage regulator will dissipate the same power, and consequently
produce the same heat. The only way to reduce the heat waste is to use a
switching regulator

Well - in the current design just loosing some of the batteries would
improve the efficiency with close to no effect to the battery life... 3.6V +
LDO drop (say 200mV) should not take more than maximum 5 cells!

Besides that I second using a switcher dedicated for LED-driving. Meaning:
Outputting constant current so the power isn't wasted in series resistors...

Cheers,
Anders

 

[Robert Baer]

I've got a bike helmet light (40 LED's 11,000 mcd's each, 7 AAA NiMH
batteries in series) that works great in terms of photonic spray. But
because the LED's need a precise 3.6 volts supply, I have added an
LM317 variable voltage regulator. This works fine, almost. The 317
works so inefficiently that the 2 square inch heat sink screwed onto
its tab gets almost too hot to touch. I'm looking for a variable 5-10
volts input, constant 3.6 volts output power supply that has very low
heat loss.
Bruce Ratcliffe,
Fresno

You cannot get around P = I * E in a linear regulator.
So, at 10V in, the regulator must drop 6.4V; multiply by the current
to the load (guess 20mA per LED times 40 = 800mA) --> about 5 watts
dissipated ===> need goodly heatsink.
Solution: a switching buck regulator.

 

[Leon Heller]

I've got a bike helmet light (40 LED's 11,000 mcd's each, 7 AAA NiMH
batteries in series) that works great in terms of photonic spray. But
because the LED's need a precise 3.6 volts supply, I have added an
LM317 variable voltage regulator. This works fine, almost. The 317
works so inefficiently that the 2 square inch heat sink screwed onto
its tab gets almost too hot to touch. I'm looking for a variable 5-10
volts input, constant 3.6 volts output power supply that has very low
heat loss.

You'd be better off redesigning the thing and using a switching technique to
drive the LEDs, like this:

<http://www.microchip.com/stellent/idcplg?IdcService=SS_GET_PAGE&nodeId=1824&appnote=en012040>

Leon

 

[Tam/WB2TT]

I read in sci.electronics.design that [email protected] wrote

You should NOT run LEDs from a constant-voltage supply. The current is
an exponential function of voltage and temperature. You need a constant-
current supply.
--
Regards, John Woodgate, OOO - Own Opinions Only.
The good news is that nothing is compulsory.
The bad news is that everything is prohibited.
http://www.jmwa.demon.co.uk Also see http://www.isce.org.uk

If he uses the right part, he should be able to run a buck regulator in
current limiting mode. Also, and I haven't worked this out, he should be
able to do something by using a very small output capacitor.

Tam

 

[Mark]

Do you have all the LEDs in parallel?

Connect them in groups of two in series and with a small dropping
resistor in series with each pair.

Mark


Mark

 

[Rich Grise]

I've got a bike helmet light (40 LED's 11,000 mcd's each, 7 AAA NiMH
batteries in series) that works great in terms of photonic spray. But
because the LED's need a precise 3.6 volts supply, I have added an
LM317 variable voltage regulator. This works fine, almost. The 317
works so inefficiently that the 2 square inch heat sink screwed onto
its tab gets almost too hot to touch. I'm looking for a variable 5-10
volts input, constant 3.6 volts output power supply that has very low
heat loss.

Your fundamental mistake is trying to run LEDs in parallel. It doesn't
matter how precise you make your voltage supply, they simply do not
current-share. They are not resistors. Their current increases
exponentially with voltage, and they all have different offsets.
Lessee: 8.4V, but the leds want 3.6; two leds in series would be 7.2, that
leaves 1.2V for the dropping resistor. Or current source. You don't say
how much current the leds are rated for - that's the target. At "3.6V",
the curve could be very steep indeed.

Good Luck!
Rich

 

[[email protected]]

No, the LED's are not regulated. I'm not current limiting them at this
point. My (simplistic) thinking is that any resistor is going to cause
I^2R power loss. Also, you mention "really low dropout". Pardon my
ignorance, but what is "dropout?"
Thanks,
Bruce

 

[[email protected]]

Thanks for the note, Mac. I looked at the LM2734Y you mention. It
says it is "configured to convert 5Vinput to 1.8V output at 1 Amp".
Since the spec sheet on the LED's says their optimum voltage is 3.6, it
sounds as if the LM2734Y will come in shy of the mark.
Bruce

 

[[email protected]]

HI, Paul,
I'm running at about 1.2 A when the battery pack (7 cells) is fully
charged. i.e. when the voltage regulator is delivering its nominal 3.6
V.
Bruce

 

[[email protected]]

I third the switcher solution. Do you have a vendor/product number to
recommend?
Thanks,
Bruce

 

[Mac]

Thanks for the note, Mac. I looked at the LM2734Y you mention. It
says it is "configured to convert 5Vinput to 1.8V output at 1 Amp".
Since the spec sheet on the LED's says their optimum voltage is 3.6, it
sounds as if the LM2734Y will come in shy of the mark.
Bruce

Where did you read that?

The LM2734Y output Voltage can be set by a resistor divider to any Voltage
down to 0.8 V. Of course, since it is a step-down regulator, the input
Voltage must be higher than the output by some amount (not sure how much).

Here is a URL for the datasheet:
http://www.national.com/ds.cgi/LM/LM2734.pdf


Good luck with your project. ;-)

--Mac

 

[Mac]

No, the LED's are not regulated. I'm not current limiting them at this
point. My (simplistic) thinking is that any resistor is going to cause
I^2R power loss. Also, you mention "really low dropout". Pardon my
ignorance, but what is "dropout?"
Thanks,
Bruce

All linear regulators require that the input Voltage be somewhat higher
than the output Voltage. Usually there is a minimum difference below which
the regulator is not guaranteed to work.

For example, 2 Volts might be a typical value for older linear regulators.
So if you want to regulate down to 3.6, you need to make sure you keep the
input above 5.6. And the 2 Volts is called the dropout Voltage, because
the regulator will (or might) drop out of regulation if the difference
ever gets that low.

Some newer regulators might allow you to operate with the input Voltage
down to Vout + 0.8 Volts or so. This would then be called a Low Drop Out
Voltage regulator, or LDO.

--Mac

 

[[email protected]]

Dear John, I've got several questions, if you've got the time.
I've contacted Maxim, and thbey have an evaluation kit that appears to
do the job for. . . $60 (ouch!) Am I going to be able to get what I
want for less? (The kit contains a PC board, SMT, a MAX1685 switching
buck regulator and associated components to give me output
characteristics I need). I've thought about contacting other ventors.
Will anyone have the equivalent of the Maxim 1685 in a 14-pin DIP
package? You suggest LTC. What company is that?
Thanks, Bruce