LED backlight dimmer for battery operation

What output do you need?

I assume you are driving white LEDs and there are quite a few specific solutions for that (See Maxim, LTC and TI) that will handle up to 16V in. If you want an ordinary converter, there are large numbers that will handle your input with no problem.

As to the inductor and diodes - in a very space sensitive design I use a coilcraft DO1605T-472MLB for a 4.7uH (on the output of a LT1767) with a MMDL770 for boost (SOD323) and MBRM120 (2.54 x 3.6mm) for the catch diode. Although they are larger than the chip, it's a lot smaller than most. You need a catch diode for this converter as there is no internal FET to ground, and you are warned not to forward bias the internal parasitic diodes.

I'll admit the inductor is a little higher series resistance than I would like, but it's still low enough to be acceptable.

Cheers

PeteS

Thanks for your reply.

I said the diodes were 2 volts each. I need an output voltage less than the 7 volts minimum battery voltage. I am assuming I can series up the keypad LEDs in 3 chains of 3 each for 6 volts. The LCD backlight will be the wild hair and I'll have to add a diode or resistor as needed to make a fourth chain. So the current will be 80 mA max in this configuration.

All of the converters I have found are either boost converters or are very high current, >= 1.5 Amps. This requires a larger inductor to handle the current in case an LED or two shorts and draws too much current.

This is exactly the part the rep suggested for the power converter. I prefer the TPS62110 from TI for the power converter. It is very small and does not need the diodes. The cost is too high for the LED driver. The DO1605 may be the best fit I have seen for the coil, but I don't think it is sheilded which is required.

I am giving some consideration to using the 3.3 volt rail on the output of the converter to power one of the boost LED drivers. But I do have the discrete circuit and will give that a go before I go with a boost driver.

You would need two adjustable switching regulators.

$2 is more reasonable.

You can scale down the L&C if you don't need the maximum current.

They can't dissipate all the heat internally.

Good solution if $4 to $5 is acceptable.

The cheapest solution is a switching controller and external MOSFET. Can probably do $2 in Ks.

Thanks for your reply. I only need one voltage regulator. The other application is an LED driver/dimmer. They sell for about $1.

If I were selling them I would say so, but I am buying and I can make the circuit for under $1 using three or four transistors. We will be building 100,000 of these, so a dollar saved is $100,000 earned. :)

What heat? The TI part has no external diodes. The power is not the issue. LTI has told me they keep their die smaller (and cheaper) by leaving the diodes off. Problem is that TI includes the diodes (or uses a PFET which doesn't need the boost circuit) and goes toe to toe with them on price while saving me the space.

I don't see $4 for any of these solutions. The only pricey part is the chip and they are running $2 to $2.50 at 1k qty for the voltage regulators that can put out 1.5 Amps. The sum of the rest of the cost is under a buck.

Or an LED driver for $1. I just need to find one that runs with 14 volt input which is where this thread started.

Any suggestions?

You still need coils & caps for higher efficiency, or pay for the battery cost.

Perhaps $1.5 per 100K.

The regulators are build for 1.5A, whether you need it or not. They have to allow for worst case condition.

I mean total cost. So, you end up with $3.50/K.

To store the energy in voltage differentials, you need external L&C. There's no way to build a single chip or to build a module for less than $1. We are ordering some regulator modules for $2/K. The module is tall, but the PCB footprint is small. All we care about is the PCB footprint.

Yes, there will be a coil and caps. What is your point?

The diodes add very little power to the part. LIke I said, TI does not use external diodes and the package is just as small, so clearly power is not the reason LTI leaves the diodes out. Could it be that LTI was being honest with me when they said they thought ti was cheaper to make the die smaller and let you buy separate diodes?

Why do you add a dollar for a couple of caps and a coil? I can get all that for about $0.50.

You need to talk to Enpirion. If they had a unit that would work with

14 volts input, I would be paying under $2 for the total solution and would only be adding a pair of $0.15 caps. :)

If your regulator has a 5 volt input, you can save some money and board space with a 4x5 mm chip from Enpirion, EN3512, IIRC. $1.60 plus two

0603 caps and you are done!

I can buy a coil, three transistors and a small number of caps and resistors and get a solution for under $1. The down side is the design time which is always too short. If I can find a chip for $1, I will use it instead of spending my time on the transistor based solution.

volts.

range

So I

circuit

Most

and

have

high

The 4.7uH coil alone would cost around $0.50. We use a 10uH coil for same price but larger.

No, we need 20V input, 1.5V, 2.5V, 3.3V and 5V outputs between 1A to

3A.

You still need a controller for another $1.

Linear regulator? That would be outlawed in California soon, for certain range of power supplies.

that

volts.

range

expect

cost.

the

So I

circuit

Most

south.

and

have

the

this

they

high

switchers?

use

I can get a coil for $0.35.

20 volt input makes it much tricker. Basically you have to go to a 1.5 Amp converter and all the parts get bigger and more expensive. But then you know that.

No this is not a voltage regulator. This is an LED dimmer and I don't need a separate controller. I can use the PWM output from the MCU. The MCU measures the battery voltage with the ADC and combines that with the brightness setting to come up with the pulse with and period. The circuit does not regulate a constant voltage, it puts out pulses that ramp up to the 60 mA max of the LEDs before turning off and ramping back down. Full brightness of 20 mA average is at a duty cycle of 1 on and 2 off. The battery voltage determines how fast the current ramps up so it sets the pulse width. The only hard part is setting up the RC and transistor switches to protect the circuit if the PWM output goes haywire.

Literally outlawed??? You have to love it when they start legislating engineering. That is like the time one of the states passed a bill that said PI was equal to 3.14 or something to that effect. They didn't like the fact that it did not have a finite number of digits.

But for my app, linear is what they are using now. Very wasteful of the battery power. If I have to I can use a boost circuit on the 3.3 volt rail. They are only a buck plus the passives.

Thanks, that certainly comes close. But it requires an external FET and it is not rated down to 7 volts.

I have found a potential Intersil part. The salesman said the price is $1.80, but they might be able to get down to $1.20. This part will do either buck or boost and is rated for up to 3 Amps. I will give it a hard look although it will add a product life cost of around $80,000 and I can do a lot of engineering for that. Too bad I can't use it for the power converter... or maybe I can!

Paul Mathews wrote: