Increasing the Battery Life of an LED Circuit

[Ozen]

TL;DR Summary

How can you increase the battery life of a circuit with a specific minimum LED mA?

Hello Everybody!

I am new to electrical engineering and only had 1 formal intro class on the subject (albeit I did very well in it) and have found myself in a position where I need to design a circuit for a micro reflex sight I have been designing. So let me first describe how my competitors battery information I could dig up: they use a CR2032 which has a current flow of .2mA, capacity of 220mAh, and voltage of 3.0V. They claim on their highest setting the battery lasts for ~25 days (~600 hours), which I plugged this data into a battery life calculator and that yields their current for that as ~.36mA. My understanding of this part is the battery only runs optimally at the specified .2mA but the circuit can demand more current or less current than that. On their #4 setting they claim the battery can last 4 years (~35000 hours), which would be around .0063mA on continual usage. How the on Earth do they achieve that?

Looking at how my circuit is coming along, it will be a battery, likely the CR2032, which leads to a push button to toggle power on/off, then going to a voltage regulator (maybe a capacitor too to even the flow out) into a micro-controller with 2 push buttons in the loop to control the brightness levels, out to the LED with maybe a resistor in front of it, and that completes it. Looking at the micro LED chips, the only suitable chip I could find was the L128-DRD1003500000 which has a Vf of 2.15V and If of 120mA. Which seems like it is quite power hungry. So I expanded out into looking at just the bare dies and having them integrated into a circuit board itself, but I can't find any under 20mA with a suitable emission area. Which brings up another point, let's take the LA-HR20WP3 bare die as example (here is the link: https://www.chips4light.com/assets/files/products/LA-HR20WP3.pdf): notice the N contact in the middle of the emission area? Isn't that interrupting the light path and not going to form the light image I want, even with the usage of a mask? Or does it somehow not effect it? I head about solid state LED's but can't find any that aren't power hungry green wavelength ones.

So with all that said, I have no clue how to get this battery life anywhere near what my competitors have. The brightness is controlled by a PWM so it will be hitting that current rate very briefly, and then the micro-controller will be continually using current, along with the voltage regulator, so it doesn't seem possible to get consumption so low it is around the .0063mA region. And to top it off, say I use one of the bare dies I have found, isn't it going to be insanely expensive to have a PCB made with a bare die integrated to it, especially when trying to get a prototype made?

Any insight you can offer to this newbie in this field struggling with getting things rolling is much appreciated!

 

[Baluncore]

Build an efficient current regulator by;

Specify the current required through the LED.
Avoid an LED current limiting series resistor, use a series inductor, L.
Install a minimum resistance current sensing resistor, Rsense, in the LED-L chain.
The half-bridge that drives the LED-L-R is made from two MOSFETs.
One is a switch, the other is a synchronous rectifier (= perfect flyback diode).
PWM the voltage across the LED-L-R chain to regulate the Rsense voltage.

 

[Ozen]

Build an efficient current regulator by;

Specify the current required through the LED.
Avoid an LED current limiting series resistor, use a series inductor, L.
Install a minimum resistance current sensing resistor, Rsense, in the LED-L chain.
The half-bridge that drives the LED-L-R is made from two MOSFETs.
One is a switch, the other is a synchronous rectifier (= perfect flyback diode).
PWM the voltage across the LED-L-R chain to regulate the Rsense voltage.

Thank you for the reply!

Why would a current regulator be needed? I had heard of them before in an example where somebody had multiple LED's in a series, but it doesn't seem necessary to my circuit to either increase battery life or make the system work. As for the steps you posted, unfortunately that is quite above my head for what I currently know; for instance I have never heard of a series inductor or what a half bridge is.

To better describe my circuit here is a basic schematic (poorly done in paint) to show what I am thinking for the circuit:

 

[Baluncore]

Your proposed circuit will be about 2.15V / 3.00V = 72% efficient.

All current including LED current flows through the linear regulator. That wastes energy in the regulator voltage drop. LED current then flows through R1 which wastes more energy.

LED and battery voltage will vary with temperature. The fixed value of R1 has little voltage drop after the regulator, so variations in LED forward voltage will cause variations in LED current.

A single MOSFET could protect your components from reversed battery polarity without significant voltage drop. Why do you need a voltage regulator?

As for the steps you posted, unfortunately that is quite above my head for what I currently know; for instance I have never heard of a series inductor or what a half bridge is.

The circuit I suggested is a switching converter that will do better than 95% efficiency.

If 70% is sufficient, then ignore the switching regulator. But if you want to screw the last 1% from the battery I would suggest a switching current regulator.

 

[Ozen]

LED and battery voltage will vary with temperature. The fixed value of R1 has little voltage drop after the regulator, so variations in LED forward voltage will cause variations in LED current.

If I am understanding this part correctly, as the temperature rises or decreases, the LED Vf will change with it, which results in the circuit not being optimal?

A single MOSFET could protect your components from reversed battery polarity without significant voltage drop. Why do you need a voltage regulator?

Is reversed battery polarity even a concern for this circuit since it is DC only? I selected a voltage regulator after reading that micro-controllers tend to want a specific constant voltage, and so the voltage regulator would stabilize it and also drop the voltage towards the LED's needed 2.15V. As for a MOSFET...I looked up some tutorials on YouTube and some articles and I can safely say I have no idea how they are supposed to work, so until I figure out what they do and how they are setup beneficially, using it is not a good idea.

I am all for getting the most out the battery, don't get me wrong, I didn't even know any other way existed to set-up this circuit. The issue is I don't really understand how that stuff works and the "tutorials" explain it as if you already know it from an inside perspective. Micro-controllers were simple enough to figure out, it is mostly coding it to get the outputs you want, which is simple. But like how you describe after the LED-L-Rsense chain, a halfbridge with two MOSFETS, I can't even imagine how that is setup or even works. And then you describe to use a PWM to regulate the Rsense voltage, why would that be beneficial? In the current circuit, the PWM turns the LED off and on rapidly to achieve the different brightness levels I will need. In the system you described, what exactly is the PWM doing?

If you could explain this as if you are talking to somebody with no prior knowledge of electrical engineering (even though I have a tiny bit of knowledge in it), it would greatly help! Everything else on this sight is of the best quality where I've been able to achieve the competitors claims or do better than them, getting the electronics there too would be fantastic.

 

[DaveE]

How do you know that your competitor is telling the truth? If you really want to know, how about buying one and testing or reverse engineering it? In any case, the best you can do is the best you can do, aim for that. This may be more of a sales/marketing problem than an engineering problem.

 

[DaveE]

If I am understanding this part correctly, as the temperature rises or decreases, the LED Vf will change with it, which results in the circuit not being optimal?
Is reversed battery polarity even a concern for this circuit since it is DC only? I selected a voltage regulator after reading that micro-controllers tend to want a specific constant voltage, and so the voltage regulator would stabilize it and also drop the voltage towards the LED's needed 2.15V. As for a MOSFET...I looked up some tutorials on YouTube and some articles and I can safely say I have no idea how they are supposed to work, so until I figure out what they do and how they are setup beneficially, using it is not a good idea.

I am all for getting the most out the battery, don't get me wrong, I didn't even know any other way existed to set-up this circuit. The issue is I don't really understand how that stuff works and the "tutorials" explain it as if you already know it from an inside perspective. Micro-controllers were simple enough to figure out, it is mostly coding it to get the outputs you want, which is simple. But like how you describe after the LED-L-Rsense chain, a halfbridge with two MOSFETS, I can't even imagine how that is setup or even works. And then you describe to use a PWM to regulate the Rsense voltage, why would that be beneficial? In the current circuit, the PWM turns the LED off and on rapidly to achieve the different brightness levels I will need. In the system you described, what exactly is the PWM doing?

If you could explain this as if you are talking to somebody with no prior knowledge of electrical engineering (even though I have a tiny bit of knowledge in it), it would greatly help! Everything else on this sight is of the best quality where I've been able to achieve the competitors claims or do better than them, getting the electronics there too would be fantastic.

It sounds to me like you are trying to design a circuit that you aren't really ready for. There are lots of online resources available on this subject, it is a common application, and frankly, the circuits aren't too hard compared to other analog/power circuits. @Baluncore has given you a good general plan to follow, but it is unlikely that he, I, or anyone else here is going to do the detailed design for you. There is data missing and some choices that you will have to make, I'm not sure that we can do that for you.

I think you have a few choices: Farm out this design to a EE that knows what to do, buy a pre-built circuit (eBay, for example), or study some more until you do understand LEDs and their drive circuits. There are many good search results available to learn more. Personally I would try to look at written descriptions, like from IC manufacturers, for example, instead of youtube videos.

 

[Ozen]

How do you know that your competitor is telling the truth? If you really want to know, how about buying one and testing or reverse engineering it? In any case, the best you can do is the best you can do, aim for that. This may be more of a sales/marketing problem than an engineering problem.

Thanks for the reply!

I have soft confirmed their battery life, multiple of my friends have the Trijicon RMR on their pistols and they only leave them on, its been over 2 years since they bought it and have yet to change the battery. As to buying one and dissecting it, I would love to do it that way, except I don't have any of the equipment, the RMR costs $400, and I am considered very poor. To put into perspective, I have enough saved to build a prototype, which the prototype along with the market research and business plan put together will be enough for my city to pay $5k to get my company into an incubator with a 95% chance of successful funding on crowdfunding. I have no doubt that their device can what they say, they aren't ones to lie about a device literally in the hands of law enforcement and soldiers.

It sounds to me like you are trying to design a circuit that you aren't really ready for. There are lots of online resources available on this subject, it is a common application, and frankly, the circuits aren't too hard compared to other analog/power circuits. @Baluncore has given you a good general plan to follow, but it is unlikely that he, I, or anyone else here is going to do the detailed design for you. There is data missing and some choices that you will have to make, I'm not sure that we can do that for you.

I think you have a few choices: Farm out this design to a EE that knows what to do, buy a pre-built circuit (eBay, for example), or study some more until you do understand LEDs and their drive circuits. There are many good search results available to learn more. Personally I would try to look at written descriptions, like from IC manufacturers, for example, instead of youtube videos.

I would agree that this is a circuit above what my current knowledge is, hands down, that's why I reached out here. I don't expect anybody to design this for me, it is a job solely for me, I just don't know how to get it how I want it in the end (long battery life on medium + adjustable brightness levels). There are so many electrical components that I don't even know of, when somebody replies with one of them, I now have a spot to start researching.

The issue with what's happened thus far is I don't understand how a MOSFET works nor have I been able to figure out how @Baluncore's suggestion would work, even with digging around on the topics. I know I will be able to get to that point eventually, but asking here really does help reduce how much blind stumbling I have to do to learn all this.

A good example is how I made the optical system. When researching it, I asked around all over Reddit and even here about the optical physics of it. In the end my optical system is better than the Doctor/Noblex optics, they have ~.3mm spherical aberration, while mine is well under .1mm without having to use a tiny aspheric surface on a 3rd lens. Objectively it is superior to it in all aspects of it (except maybe cost due to the material I want), and I have no formal education in optics at all.

With all that said, I can still learn so much from his suggestion and any suggestions you or other users may have. Dumbing it down is certainly very helpful to me, since I have to lookup pretty much everything, reach leads to a never ending tangent of other topics that factor in and all that stuff.

I did look at hiring somebody on the side, but the money I could put up for it all the electrical engineers scoffed at it (even saying they wouldn't even read the requirements for it), so that is out of the picture. Could you elaborate a bit more on buying a pre-built circuit? What would you suggest I study to up on LED's? I currently understand that the dies are made of whatever material to get the desired wavelength, and it has a p and n contact where it will be joined to the circuit. To prevent burnout, the voltage in has to be the voltage drop of the LED (at least for a system with 1 LED), so a resistor is often placed in front and the value can easily be calculated.

 

[DaveE]

You want to think of the LED as a current driven device. The LED voltage is not stable (mostly temperature dependent). Also the I-V curve is quite non-linear at low currents. That is why @Baluncore was talking about a current source power supply and questioning your choices about voltage regulators. That is also why people often put a resistor in series with the LED if they are driving it from a constant voltage source. Look up the I-V relationship for diodes for a start. Also search for adjustable (dimmable) LED driver circuits. I'm sure there is an IC you can buy that will do most of the work you need.

 

[Ozen]

You want to think of the LED as a current driven device. The LED voltage is not stable (mostly temperature dependent). Also the I-V curve is quite non-linear at low currents. That is why @Baluncore was talking about a current source power supply and questioning your choices about voltage regulators. That is also why people often put a resistor in series with the LED if they are driving it from a constant voltage source. Look up the I-V relationship for diodes for a start. Also search for adjustable (dimmable) LED driver circuits. I'm sure there is an IC you can buy that will do most of the work you need.

Thank you for the starting point to further research this topic! I think I am getting a better understanding of all this and now understand how important the current regulation is to the LED rather than the voltage.

I do have a question I can't seem to find the answer to. Say I put in a LED driver IC (https://www.infineon.com/cms/en/product/power/lighting-ics/linear-led-driver-ic/bcr402w/) and let's just focus on the battery, this, and the LED; the voltage drop across the IC is 0.75V and has a preset I to 20mA (which we will assume is fine for this example), that leaves the the voltage as 2.25V still when passing thru the LED. What will happen since that is above Vf but the current is correct? Would a resistor have to be added before the IC to drop the voltage 0.1V, and then the outcome will be the correct voltage and current?

In the above example, that is just an example LED driver IC I found, I am still researching whether a switching one or linear will be better since I only have 1 LED in this circuit and all the examples I see the switcher used is with multiple strings of LEDs.

I also found this one (https://www.onsemi.com/pub/Collateral/CAT3649-D.PDF) that looks quite promising for what I want but I have some questions that involve these switching drivers in general. This one is preset to 25mA for one LED on each output; what if I have a different current required than the preset? My other question is about the PWM control. I know how to set that up using a microcontroller with the push buttons I want, but how does this change things? In my rudimentary circuit I drew before, the output of the microcontroller is going directly to the resistor and LED. I imagine that wouldn't work for this case, or if that output is indeed able to go into the PWM input, wouldn't I still have to use a voltage regulator to get a suitable voltage for the microcontroller?

 

[Tom.G]

I do have a question I can't seem to find the answer to. Say I put in a LED driver IC (https://www.infineon.com/cms/en/product/power/lighting-ics/linear-led-driver-ic/bcr402w/) and let's just focus on the battery, this, and the LED; the voltage drop across the IC is 0.75V and has a preset I to 20mA (which we will assume is fine for this example), that leaves the the voltage as 2.25V still when passing thru the LED. What will happen since that is above Vf but the current is correct? Would a resistor have to be added before the IC to drop the voltage 0.1V, and then the outcome will be the correct voltage and current?

Most of the stuff we work with (CPUs, audio amplifiers, etc) either need or work best with a fixed voltage and draw whatever amount of current they need.

LEDs are the opposite, they need a given (or at least controlled) current and will allow whatever voltage across them they need to obtain that current. In other words, the current is the independent variable (the one you set) and the voltage is the dependent variable (value depends on the current and the LED characteristics).

The Infineon LED driver IC you mentioned is designed to allow a fixed current to flow but allows the output voltage to vary as needed to reach that current. Any excess voltage from the supply appears across the driver IC, no extra resistor needed. You can, however, change the current setting with an external resistor across pins 3&4. See fig 3-6 on pg 13 of the datasheet.

I also found this one (https://www.onsemi.com/pub/Collateral/CAT3649-D.PDF) that looks quite promising for what I want but I have some questions that involve these switching drivers in general. This one is preset to 25mA for one LED on each output; what if I have a different current required than the preset?

Pin 12, signal ADIM, can decrease the LED current in 3% increments. See pgs 9&10 of the datasheet.

...the output of the microcontroller is going directly to the resistor and LED. I imagine that wouldn't work for this case, or if that output is indeed able to go into the PWM input...

That would work. Again datasheet pg9 pin 13, signal PWM is typically fed a PWM signal to switch the LEDs On and Off, high is On and low is Off.

To vary the visual brightness you can feed a PWM signal to pin 13, or you can tie it to logic high and vary the LED currents with pin 12, signal ADIM, or do both. That let's you try the ADIM signal using a pushbutton by tying PWM high. (saves some programming that way while you are experimenting)

Yes, you will still need a voltage regulator for the microcontroller, and probably any other logic you may use.

Overall, I suggest you start with a simple analog driver chip like the Infineon for getting a bit used to LED circuitry and characteristics. Once you are comfortable with that, go on to the fancier microcontroller PWM and current control aspects.

Hope this helps, and enjoy!

Cheers,
Tom

 

[Baluncore]

I do have a question I can't seem to find the answer to. Say I put in a LED driver IC (https://www.infineon.com/cms/en/product/power/lighting-ics/linear-led-driver-ic/bcr402w/) and let's just focus on the battery, this, and the LED; the voltage drop across the IC is 0.75V and has a preset I to 20mA (which we will assume is fine for this example), that leaves the the voltage as 2.25V still when passing thru the LED. What will happen since that is above Vf but the current is correct?

The BCR402W is a "high-side current mirror" with a supply voltage overhead; VSmin = 1.2 volts. That says the supply must be 1.2 volts above the LED forward voltage, in order for the mirror to regulate the current.
With LED Vf of 2.15V, the supply must be at least; 2.15 + 1.2 = 3.35 volt.
The CR2032 has a voltage of 3.0 V, so a BCR402W will NOT be able to regulate the LED current.

 

[DaveE]

Since you started this thread by asking about high efficiency, but then you were looking at linear current controller ICs, I think you need to understand a bit more about linear power supplies. In a linear power supply you define a current that you want to operate your LED at. Then you provide a power supply which provides enough voltage (let's call that Vcc) to operate the LED (Vd) plus some extra to allow the regulator to stabilized the current (this is a series voltage regulator, BTW. We'll ignore linear shut topologies since that's not a great fit here). The current through the regulator is the same as the LED current. Regardless of the type of linear current regulator, the efficiency of this circuit is Vd/Vcc. It doesn't matter if the regulator is a simple resistor or a transistor that is controlled by an IC. In a linear current regulator, it is the regulator's job to drop the excess voltage and create waste heat in the process. For low power LEDs this is almost always done with a simple resistor because it works well enough for most applications.

You have already determined that you don't want this inefficiency, that is why we have been talking about a switching power supply implementation. This is pretty uncommon for low power devices, so you may find it hard to find an IC that is advertised for just this niche. Most designers will use one of the many generic SMPS ICs to design their own SMPS current regulator. If you want to procced with a generic IC SMPS implementation, I would look at "current mode buck" control ICs. However, you will also want to study a bit about buck DC-DC converters to do this on your own.

Also be aware that the control IC will also consume some power. This can be negligible, or it could be significant depending on the IC you choose.

 

[hutchphd]

I don't think anyone has explicitly mentioned that the voltage output of any battery will drop below nominal long before all the energy is used up. The Lithium cells are pretty good actors but always look at the discharge curve of the battery or you will be disappointed in the lifetime. The switcher will also do better for this of course. I would be shocked if there was not a dedicated IC optimized for this application and this ubiquitous battery!

 

[anorlunda]

Don't forget the more obvious ways to extend the life that can be simpler than an efficient power supply.

1. Use a very low current LED. For the micro reflex sight application, not much light should be needed.
2. Make sure that the LED is on only when needed. When mounted on a rifle, you could sense motion. If no motion is detected for say 90 seconds, turn the LED off. 4 years of micro reflex sight lifetime could mean as little as 100 minutes of LED on time.

 

[DaveE]

I would be shocked if there was not a dedicated IC optimized for this application and this ubiquitous battery!

Yes, that's what I expected too. But I did a cursory search (TI, and Analog Devices mostly) and didn't see anything obvious. They have the ICs but they are all targeted at >1A or so. I'm sure some of those would also work at lower currents. However, for low power, it's really easy to make your own circuit. For example you might use the CMOS gate driver output intended for a MOSFET to directly power the LED.

These ICs are typically targeted at large users (Auto, Lighting, etc.) that want special functions, integration etc. I'm not sure there's a huge market for a special purpose IC that does the same thing as the general purpose SMPS chips, but at lower volume/higher price. Also, at low power, most designs don't require very high efficiency.

It's not that they don't exist, it's that they might not be called "LED controllers".

 

[Baluncore]

There is no requirement for the LED current to flow through the voltage regulator for the controller. With only a 3 volt supply, finding a low Ron MOSFET will be interesting.

Since battery current is limited, any solution will probably benefit from having an LC reservoir supplying current to the LED.

Maybe a minimalist controller can be programmed to produce a short digital pulse, sufficient to initiate a current through the LED and series inductor, then turn off after a microsecond or so, allowing flywheel current to continue. Brightness is then controlled by the number of pulses issued per 50 milliseconds, and by the 'on' time of the pulse. That gives high efficiency and brightness control with only an inductor and a freewheel diode.

 

[DaveE]

With only a 3 volt supply, finding a low Ron MOSFET will be interesting.

Not really a problem at low currents. VN2222L for example, with Vgth < 2.5V.

 

[DaveE]

Maybe a minimalist controller can be programmed to produce a short digital pulse, sufficient to initiate a current through the LED and series inductor, then turn off after a microsecond or so, allowing flywheel current to continue. Brightness is then controlled by the number of pulses issued per 50 milliseconds, and by the 'on' time of the pulse. That gives high efficiency and brightness control with only an inductor and a freewheel diode.

Yes. Since the current isn't too big this could even be a uP output, if it's a strong one. Certainly with a buffer if needed. But this is topologically the same as applying a control voltage across the LED, so the current may vary quite a bit with temperature and such. There wouldn't be any "ballast" function or current regulation without some sort of current sensing. This might be pretty easy to provide with a sense resistor, but you would probably need some sort of amplifier before a uP ADC.

A linear SMPS IC like the LT1619 provides a lot of these functions (there are lots of other choices too).

 

[Baluncore]

There wouldn't be any "ballast" function or current regulation without some sort of current sensing.

You are making the minimilist solution too complex. The microcontroller would need an open drain output. Maybe a CMOS output could act like a switch and a synchronous freewheeling diode, but it would worry me.

We know the voltage difference between the LED and battery ±20%, we know the inductance, and the pulse Ton, so we can compute the charge per pulse. It does not need to be precise since a human is looking at it against a variable background, and brightness is selected by the operator.

VN2222L does not meet the specs since with Vgs = 2.5 V, it only sinks 1 mA with Vds = Vgs, so there is no voltage output difference to appear across the LED and inductor. The problem is getting low Ron with a low gate voltage. Low Ron usually requires gate voltage overdrive. Maybe a saturated BJT is a better choice here.

 

[DaveE]

You are making the minimilist solution too complex. The microcontroller would need an open drain output. Maybe a CMOS output could act like a switch and a synchronous freewheeling diode, but it would worry me.

Lots of CMOS uC's have strong outputs, but you would need to parallel a few to do this job. For example this MSP430 (pg. 24) can source or sink 6mA at <0.3V saturation, 8 in parallel could do 48mA at 14mW. This gets into how conservative you want to be. I would probably put a discrete buffer in as you suggest, but I've seen lots of designs that run low power SMPS direct from CMOS logic. I would also add the commutating diode Schottky, in this case), but I wouldn't be surprised if you could get away with the body diode of the output stage for low frequencies. I wouldn't do this because the common ground with the rest of a very complex uC chip would scare me.

We know the voltage difference between the LED and battery ±20%, we know the inductance, and the pulse Ton, so we can compute the charge per pulse. It does not need to be precise since a human is looking at it against a variable background, and brightness is selected by the operator.

Yes, I agree it would probably work ok. I'm not sure I would do it. You don't want a dim light on a bright cold winter day with an old battery or extra power dissipation on a hot day in the desert with a new battery. But I was never any good at low cost designs; everyone I worked for wanted the other extreme.

However, it is true that in this topology the LED voltage will equal the duty cycle times the supply voltage, there is no regulation of current (except parasitic resistance in the components). The LED current will be determined by the diode equation.

VN2222L does not meet the specs since with Vgs = 2.5 V, it only sinks 1 mA with Vds = Vgs, so there is no voltage output difference to appear across the LED and inductor.

That is the maximum value for worst case analysis, not typical performance. The data sheet requires interpretation because test circuits for screening and specification aren't always the same as typical applications. They had to choose a fixed value of Vds to use when measuring the gate characteristics, it's not how they are actually used.

Of course Vds is mostly unrelated to Vgth (first order anyway). Vds in saturation can be much lower than Vgs. I'm not claiming that that part is the best choice (it's actually much bigger than necessary), it's just a common part that proves it's possible.

Also look at 74HC04, for example, which can operate with 2V power supplies. MOSFETs can be made to operate at low thresholds, which is pretty common for ICs that operate from batteries.

 

[Ozen]

Most of the stuff we work with (CPUs, audio amplifiers, etc) either need or work best with a fixed voltage and draw whatever amount of current they need.

LEDs are the opposite, they need a given (or at least controlled) current and will allow whatever voltage across them they need to obtain that current. In other words, the current is the independent variable (the one you set) and the voltage is the dependent variable (value depends on the current and the LED characteristics).

The Infineon LED driver IC you mentioned is designed to allow a fixed current to flow but allows the output voltage to vary as needed to reach that current. Any excess voltage from the supply appears across the driver IC, no extra resistor needed. You can, however, change the current setting with an external resistor across pins 3&4. See fig 3-6 on pg 13 of the datasheet.Pin 12, signal ADIM, can decrease the LED current in 3% increments. See pgs 9&10 of the datasheet.That would work. Again datasheet pg9 pin 13, signal PWM is typically fed a PWM signal to switch the LEDs On and Off, high is On and low is Off.

To vary the visual brightness you can feed a PWM signal to pin 13, or you can tie it to logic high and vary the LED currents with pin 12, signal ADIM, or do both. That let's you try the ADIM signal using a pushbutton by tying PWM high. (saves some programming that way while you are experimenting)

Yes, you will still need a voltage regulator for the microcontroller, and probably any other logic you may use.

Overall, I suggest you start with a simple analog driver chip like the Infineon for getting a bit used to LED circuitry and characteristics. Once you are comfortable with that, go on to the fancier microcontroller PWM and current control aspects.

Hope this helps, and enjoy!

Cheers,
Tom

Thank you for the very detailed reply!

I think I have an understanding now of how to set this up. Just a few questions remain, mostly with setting up and operating the ADIM pin of the CAT3649. To reach the required current, I would need the pulses (n) to equal 7, putting it at a 19.4mA current. My first question is, does the timing reset after so long? Say I have a microcontroller or something send 7 pulses and then stops, will the current stay at 19.4mA until the power button is toggled off? Or does it reset after so long and I would need to keep sending that pulse; but if it is that case, wouldn't that damage the LED quickly since it will be receiving current, albeit briefly, above the stated limit?

My second question with the ADIM pin, is what is it called setting up a microcontroller to send 7 pulses upon activation but cuts off after the 7th? I would like to research this and get a better idea of how to code that and where this microcontroller would be placed (likely with the PWM microcontroller that already has a voltage regulator).

And my last question for now is again with the LED and current relationship. Say the current is the desired 20mA, but the voltage coming out of a driver is 4V and the drop across the LED is 2.15V. Would the wire immediately after the LED have a voltage of 1.85V? The LED does not have any harmful effects from having the excess voltage pass thru it because the current determines the voltage drop, correct? Lastly, say after that LED it goes to ground, there is no problem having 1.85V going to ground other than being waste?

Don't forget the more obvious ways to extend the life that can be simpler than an efficient power supply.

1. Use a very low current LED. For the micro reflex sight application, not much light should be needed.
2. Make sure that the LED is on only when needed. When mounted on a rifle, you could sense motion. If no motion is detected for say 90 seconds, turn the LED off. 4 years of micro reflex sight lifetime could mean as little as 100 minutes of LED on time.

The LED that is the likely candidate is: https://www.chips4light.com/assets/files/products/LA-HR09WP3.pdf. That only lower current LED's with that wavelength hit 5mA but has a tiny (30um) emission area and I am researching whether the light spread could even spread out fast enough (needs to be 114.7um for a reticle mask) to be usable. They smaller ones could work if I wanted say a 2.5MOA dot instead of the " ^ " shape I am aiming for.

The motion senser is an interesting idea, although a longer turn off time would be needed (25 minutes or so) because what if somebody decides to hold a crouched or prone angle for a long time? You never want the reticle going out during combat and I think 90 seconds is too low. This though could replace a power button and I could see consumers liking this feature (a lot of them like to leave theirs on at all times to be ready whenever); thanks for the idea!

 

[Baluncore]

I would also add the commutating diode Schottky, in this case), but I wouldn't be surprised if you could get away with the body diode of the output stage for low frequencies.

Yes, an external schottky for protection, but the upper cmos transistor will be on when voltage is high, so it will conduct the freewheel current.

The LED current will be determined by the diode equation.

The short on time, with the series inductor, limits the current via; V = L ⋅ di/dt;

That is the maximum value for worst case analysis, not typical performance.

I always design for the worst case, or fully test components and select.
How else can I know the product will work every time.

 

[DaveE]

Yes, an external schottky for protection, but the upper cmos transistor will be on when voltage is high, so it will conduct the freewheel current.

I think we are thinking about different configurations. I assumed that the commutation occurs when the high-side FET is off and would flow (backwards) through the low-side FET.

The short on time, with the series inductor, limits the current via; V = L ⋅ di/dt

No, that's not how the volt-second balance works on an inductor in a buck PS. For V=L⋅(di/dt) the voltage V is across the inductor V=Vcc-Vo, so the current slope di/dt=(Vcc-Vo)/L. The on time duration for fixed frequency PWM only determines how much the inductor current will change during that cycle. di/dt=0 , steady state, only occurs when D⋅Vcc=Vo (D is the duty cycle). The average voltage across the inductor must equal 0 in steady state or the average inductor current will change. This (to first order) has no dependence on the DC inductor current.

I always design for the worst case, or fully test components and select.
How else can I know the product will work every time.

Yes, me too. That's what the people I worked for wanted. But we didn't sell our stuff on eBay or in Walmart. The truth is, lots of consumer products won't meet our design review standards. In our capitalist, consumer world there is a market niche for both.

 

[DaveE]

Just a few questions remain, mostly with setting up and operating the ADIM pin of the CAT3649.

No, I won't read and interpret the function of a complex IC for you (mostly because I don't know). That is why On Semi writes datasheets and other customer support documentation. Perhaps, after you search their web site, you could call one of their application engineers. I would be shocked if anyone here has experience with this special purpose IC. To use ICs like this you must do some research. Read every part of the data sheet first, then look for other supporting documentation.

BTW, in my (limited) experience ADIM is not a common digital interface.

Say the current is the desired 20mA, but the voltage coming out of a driver is 4V and the drop across the LED is 2.15V. Would the wire immediately after the LED have a voltage of 1.85V? The LED does not have any harmful effects from having the excess voltage pass thru it because the current determines the voltage drop, correct?

The LEDs for this IC will be connected from the power supply to the IC output. So, for example, if your LED voltage drop is 1.6V and your power supply is 3.0V the voltage from the IC output to ground has to be 3.0-1.6=1.4V. It is the IC's job (based on your programming) to control the LED current as you asked by setting the correct voltage at it's output.

Another way of saying this is that the LED current will be whatever current is necessary to make the LED voltage drop equal the voltage that is applied across it by your circuit. Or, from a current driven perspective, the LED voltage drop is determined by the amount of LED current that your circuit makes flow through it. These are the same thing from two different perspectives.

 

[Baluncore]

No, that's not how the volt-second balance works on an inductor in a buck PS. For V=L⋅(di/dt) the voltage V is across the inductor V=Vcc-Vo, so the current slope di/dt=(Vcc-Vo)/L. The on time duration for fixed frequency PWM

I am certainly NOT suggesting fixed frequency PWM. I am saying a fixed width pulse will dump a standard charge through the inductor and LED. The average LED current will be proportional to the frequency of the software controlled pulse rate. That way you get efficient LED current conversion, and can adjust the brightness by changing the pulse rate. It eliminates current sense overhead inefficiency, and all those extra chips and components. That is why it is a minimalist solution.

 

[DaveE]

I am certainly NOT suggesting fixed frequency PWM. I am saying a fixed width pulse will dump a standard charge through the inductor and LED. The average LED current will be proportional to the frequency of the software controlled pulse rate. That way you get efficient LED current conversion, and can adjust the brightness by changing the pulse rate. It eliminates current sense overhead inefficiency, and all those extra chips and components. That is why it is a minimalist solution.

Yes, I think that's a good approach. The user is the feedback mechanism.