*BLF LT1 Lantern Project) (updated Nov,17,2020)

Can someone explain to me how burning the extra voltage above Vf in a FET instead of a 7135 is going to gain efficiency? What are the switching losses in PWM’ing a 7135? The proposed circuit basically uses a FET as a varistor. It’s 2018 and we’re back to using current limiting resistors?

Why isn’t there a true constant current boost circuit? A 1S4P or 2S2P battery config and a two 4 LEDs strings would operate in boost regardless of the battery voltage. This would give constant brightness, ensure each LED gets the same current, and should have greater overall efficiency than just throwing away power in a FET turning it into heat.

I agree, no additional moonlight mode is needed. Still, it’s interesting to see how low it will go.

I have to agree.

To play devils advocate:

Part of the reason may be the costs and complexity, though. Getting the boost driver right would require multiple revisions, especially to integrate it into a firmware like Andruil. Sure, there are dedicated smps ICs out there, but that increases the BOM and manufacturing effort.

For those who really care about such things, the BLF designs are all made to be easily modded, which many will do.

I don’t know if it would make the driver cost more, but I put that into the so what category. It wouldn’t be a lot. It is supposed to be the ultimate lantern right?

I don’t think it would require very many revisions. A prove out of the design should be all that’s necessary. I hope the design gets a prove out before they make the lights no matter what the design is.

The FET based linear driver is more efficient in all modes except the max mode because the LEDs are actually being driven at a lower current. LEDs become much more efficient with less current. A 7135 based driver always drives LEDs at the set current, so they have identical effeiciency in all modes.

A boost driver would be better though, I agree, because the LEDs in the lantern are not driven with high currents. This their VFS are rather low. FET vased linear drivers are good for medium to high currents.

Amazing work on this! ( i just got home form ym few days away trip)
This looks like the driver that will be the one to run the BLF LT1 lantern. :slight_smile: Just curious, for those who wish to mod the maximum lantern outputs, is “R1” the resistor that will regulate the maximum output? ( meaning simply swapping out that resistor with a different ohm value will change the overall output/amp draw?

yep, 1S4P as the same configuration for the Q8, which in this case with the lower maximum amp draws it can run on even one 18650 or four.

It’s not constant current regulated. It’s regulated (by throwing away excess voltage/power) until the battery voltage drops below Vf + the Rdson * current. At which point it is no longer regulated.

This lantern is crying out for a boost drivers and series strings of 4 LEDs.

Theoretically yes. Maybe… You’re only saving switching losses. Nearly all the losses or gains in efficiency at varying current levels are due to heating of the die. The die heating at 175mA vs. 350mA with a 50% duty and a reasonably fast PWM will be virtually identical meaning there is no efficiency loss/gain in the LED itself. It’s only the switching losses of the PWM.

I’ve yet to see anyone quantify the switching losses, so for all I know someone is chasing a sub 1% gain.

Actually, I prefer the linear driver for this project. That way, the lantern is less likely to suddenly die on me when the cells run low, as often happens with boost drivers.

The linear driver can just drop out of regulation and gradually lose output, whilst still producing enough light for me to find my spare cells and another light to use while I change them.

That has nothing to do with ho boost drivers work. It just depends on how smart the U.I. of the light is. It should be no problem slowly reduce the brightness at the end of the runtime.

This. And given the tendency of Chinese manufacturers to play the part swap game, there’s something to be said about a proven reliable design where few things can go wrong. I like a nice driver as much as anyone but I like a working light even more.

Can you prove this? I don’t believe it. LEDs have a lower forward voltage at lower current levels. This means that the total power consumed is less (instead the power is burned off in the FET, which is not a problem when cooled properly). Also, the reason that LEDs get dimmer with higher current density is not actually heat. You can cool an LED to negative temps and it will still become less efficient at higher current levels. This effect is called droop. Of course heat is an additional problem that can further reduce brightness. The LED will definitely be cooler with the FET because it runs at lower power. In addition to this it’s also brighter…

Fair enough, but someone has to specifically implement that in the boost driver, whereas a linear driver just works that way with no extra effort.

I was going to answer similarly but The_Driver beat me to it! If you want to know about the rated difference in efficiency at the LED itself, simply look at the LED datasheet and find the current over Vf curve. For CREE LEDs it’s even easier. Just check out the CREE PCT here and input any LED from their list, and they show you the voltage, wattage, lumens output, and overall efficiency (lm/w) of the LED at various current levels. Every part of the lighting system will have its own efficiency calculation. It isn’t just at the FET, nor is it only from heat.

Why do we want it to dim? How about some sort of flashing indicator or a brief flash of the lantern itself (like once a minute)?

Again, can you quantify the gain? Most of the power gain you’re claiming is irrelevant because it’s not a system level gain. A small amount of the power is just burned off in the FET instead of the LED. The system efficiency is not improved from a battery run time perspective. You’re just moving the heat around. With a boost driver you could actually save on input power and gain system level efficiency with the lower Vf.

Frankly, I don’t care much about a linear FET vs. 7135s. Both are bad solutions. The right solution is a true constant current boost driver.

I’m interested in thus. Are we still able to sign up?

Added Twinnie to the interest list at number 964. Welcome to BLF!

One thing I would like to make sure ends up in the final design is a solid method of hanging and carrying the lantern. Keep in mind the added force of momentum when walking with the weight of 4x 18650s and the battery compartment swinging and bouncing around.

I understand both sides of the argument. A setting in the U.I. would make everybody happy. In my eyes a lighted switch with multiple colors is the better solution to warn the user. If they want to they can then reduce the brightness.

Sure:
Lets say you are driving a Cree XPL HD 3000K 90CRI (very similar to Samsung LH351D) in bin U3 at 2A and 50°C with a 7135-based driver. You will get 588lm at 91.2lm/W (efficiency for just the LED, together with the driver it’s lower). Now you want to dim to 50% of this current. The LED will thus still be running at 2A, but with PWM. So you will basically double the battery runtime with half the brightness (294lm) at the same LED efficiency of 91.lm/W.

With FET-based linear driver the LED would have the same brightness and efficiency at 2A, but at 1A it would be brighter and thus more efficient (see here). The above XP-L produces 330.5lm at 1A (109.9lm/W). Thats 12% more brightness for free. A fet-based linear driver is cheap just like a 7135-based one.

Both drivers are running at the same total power (depends on the current pulled from the battery and is independent of the LED). The difference becomes more prononouned as the maximum current the driver needs to be capapable of is increased and especially of you like to use low modes.

At 350mA and 30°C the above XP-L LED has an efficiency of 130.6/lm/W. Thats already 45% higher compared to the the same LED when PWMed at 2A and you can of course lower the current even more.

Of course a boost driver is the better solution from a technical standpoint, but if we are set on a linear driver we might as well use the better kind.