The inductor is primarily responsible for storing energy / smoothing output in this circuit or buck converter. Capacitor is helpful for smoothing, but cannot work alone.
Think about what happens when you put a capacitor in parallel with LED right after MOSFET.When MOSFET opens, capacitor presents itself as an electric short circuit - there is nothing limiting the current except parasitic resistances and reactances (battery, MOSFET, wires). So in an ideal case, capacitor will achieve smoothing, but the result will be equivalent to 100% duty cycle - direct drive form LED’s point of view. There is hardly a point to that though, one can achieve the same result without large capacitor and oversized MOSFET by simply driving regular MOSFET with 100% duty cycle. In practice, current will be limited by those parasitics and the capacitor’s smoothed voltage will be reduced relative to direct drive. But that too will be pointless because it will be lossy similar to linear driver, except the heat will be dissipated not by a linear drive chip, but by battery/wires and mosfet.
The beauty of inductor is that it limits current, and does so without loss (well, almost). When MOSFET opens, inductor limits current rise rate (V=L*dI/dt) and stores energy in form of magnet field. That means neither MOSFET nor battery are overworked or need to be sized for excessive currents. When MOSFET closes, all that energy stored by the inductor is converted into current that flows through LED (and flyback diode).
For completeness, it should be noted that capacitor can also limit current almost without losses. For that it needs to be in series with the load. But that only works with AC and it does not provide any smoothing.
No it will not.
Take a look at how a capacitor smooths a rectified sinusoid.
It is NOT the same as having a 100% duty cycle.
A capacitor in parallel does what an inductor in series would do, but with an electric field instead of a magnetic field.
Excuse me but it looks like dsk3 intends to add PWM smoothing to a direct drive MOSFET driver, doesn't it?
I commented this somewhere in some thread around here months ago, I said something about placing an inductor in series with the emitter. Henrik said I was just missing a diode for a buck converter (or something like that), but to me the point was just to add PWM smoothing. Regulation is present in some MOSFET linears like LD-25 or led4power's offerings, but at the cost of burning out all excess voltage and power (not the same thing and better done with regulated buck, boost or buck/boost engines if space, budget and willpower are present).
A picture from selfbuilt on a review of a RRT01. He comments there is no pwm, but he measured this “noise”. It looks a lot like an inductor charging and discharging to me. Perhaps there are some drivers doing what has been suggested above, putting an inductor is series with the LED to smooth PWM. Some capacitance across the LED would certainly help smooth it out as well.
Led4power made(makes?) drivers using a mosfet to regulate current but it’s not as cheap and the software is proprietary. The Attiny 13A mcu has been around for a long time and has tons of support and documentation. It’s less a question of what is better but more a case of being able to personalize a UI with an adequate level of efficiency. Add to that the “I dare you to stack more 7135’s than this” or slaving extra boards for more current or different levels then the 7135 driver becomes an easy to understand and exceptionally flexible platform for those of us with a yearning to mess with drivers but lack the knowledge, equipment, and skills to fine tune a boost or buck regulator, most of which don’t use the mcu so many have acquired the necessary bits to flash for themselves.
There’s a lot of history behind this which makes for some interesting reading. There IS a lot of demand for other drivers which is also part of that history but mostly it boils down to board size flexibility, programming flexibility, cost, and ease of use. They’re simply the most convenient driver going.
Aaah yes, the output capacitor sbslider came to recall it. In essence, an LC smoothing circuit tuned to the corresponding switching PWM frequency. All of this, of course, adds bulk and cost to the driver.
Seems to me that reducing the power wasted as heat should be a primary goal because the problem with most lights today is that they’re constrained by the thermal mass and not the power output capability of the battery. If this improvement in efficiency cost a few dollars extra, maybe it’s worth it to have a slightly higher regulated output in return.
All drivers waste energy as heat. Until better drivers are available that are both cheap and open sourceed firmware the 7135 driver will be well represented. At any current worth worrying about the leds swamp the heat produced by these chips.
The main problem is not the heat, it is the fact that energy is being wasted resulting in shorter battery life.
It is only 7135 drivers that regulate current by converting excess voltage to heat.
Buck or PWM drivers can obtain much higher efficiencies throughout the operating range simply because they operate in a different way.
It’s possible to get 95% or more with a good design.
Obviously more expensive and larger, but still a better option than 7135 for efficiency.
I’m aware of how buck or boost drivers work but most still seem willing to trade that efficiency for something they can flash with a custom ui. Then again, find a 17 mm buck driver you can reflash, or 15mm, or 10mm. Your qualifications are exactly why it’s not better. A lot more expensive and doesn’t fit. No amount of “better” will fix that.
There are threads going back several years with the same discussions over and over. Nobody’s argueing that buck drivers can’t be more efficient, they just aren’t being made in the sizes needed or with mcu’s that can be tinkered with. There have been some attempts at producing drivers and loneoceans is the most current member I know of. There are literally countless 105C clone designs run through oshpark that use the same package of driver chips with however many 7135’s are called for. All of whose pieces are visble to the naked eye and can be built up with even modestly capable soldering skills.
There’s a whole lot more to this than just declaring one circuit better than another. It also has to be available.
With a flashlight that is not always the case, due to the low voltage difference between battery and led and the fairly high current.
The efficiency of a linear driver is not that bad in flashlights with LiIon batteries and it can be difficult to match it with a switcher over the full operating range. With a fresh battery the switcher will probably be best, but when the battery depletes the linear is best.
As LEDs keep getting lower and lower vf buck drivers will become more usable with single cell flashlights.
Buck is already the best option for anything that uses 3 or 4s.
linear is definitely not best when the battery depletes because then the LED brightness is no longer regulated, it just drops as the battery voltage drops.
With lower vf LEDs, liner driver efficiency will also drop even more, since there is a larger voltage difference between the battery and led.
They are better than buck, because they can be made with lower on resistance (About one inductor lower), i.e. they will maintain regulation longer than buck and maintain higher brightness when they goes out of regulation.
If the vf is low enough and battery voltage is high enough then buck can keep the LED in regulation literally until 0% battery.
For example a 6v led and 3s battery.
Even when the battery is near dead at 8.1v the LED will still be driven at 100%.
This is impossible to do with a linear driver because the voltage difference is too high it would need to burn off dozens of watts, depending on what current you’re driving the LED at.
And if you reduce the voltage difference, then you can’t keep it regulated.
Lithium batteries do not hold a constant voltage so everything from 4.2 down to 2.7 is fair game, so you either deal with very high efficiency losses or you deal with an LED that starts dropping in brightness when the battery is only half discharged.
A microcontroller unit is quite simple, like a pedal car. Porting existing software to PIC or other microcontrollers could be done. Pretty sure it is quite an accessible thing. Is there motivation? I see lots of good drivers with PIC microcontrollers which would thank a little reprogramming, just a little. For example, removing the <6.7V high to mid stepdown in the LD-29 driver, or making adjustments to low voltage warnings, etc.
They’re popular because they’re available or can be made for any single cell light from 10440 up as well as larger hosts, cheap, easier by far to design and solder up one or more from Oshpark since the components are small but still large enough to hand solder, mcu’s can be purchased with a number of different firmware options or one can flash their own. I keep hearing how great buck drivers are but where are they? Where is the firmware? I was ok using the old KD P7 3-mode buck driver for bike lights but it was limited in both input voltage and output current and reputedly had terrible efficiency(worse than any linear driver) along with being 19mm so was quite difficult to shoehorn into lamp heads and wouldnt fit at all into any p60 pill or 17mm driver pocket. As far back as I became a member and beyond the 17mm highly efficient buck driver has been a holy grail and while I expect it to arrive at some point I don’t hold much hope of it being less than 5x as costly, very little that it would fit a 14500, and none at all that it would fit a 10440. Remember the B in BLF and consider how many lights would get built if the driver cost twice as much as the host and 5x as much as a linear driver.
I also question the part in the thread title about limited current.
Mosfet based linear drivers became popular mainly due to modders wanting to max out current draw and getting tired of stacking 7135’s. Dtp copper boards complemented the max output goals as did searches for the cells with the highest current/least sag. Not much concern about efficiency or run time, just raw power. Even the best buck or boost driver at any cost can’t match the current to the led of a dirt cheap chip stack or mosfet linear driver which can fit in virtually any host, single cell or other.
That brings me back As a kid, I’ve build a few of big and heavy power supplies to power the music amps. Hunting for biiiig electrolytic capacitors to complement a heavy transformer was memorable.
Anyhoo, there is no question that a capacitor will smooth the rectified sine wave to its PEAK voltage under low or no load, which IS exactly equivalent to 100% duty cycle for PWM wave. That’s again only if parasitic losses (diode bridge, etc.) don’t limit current. But there is more:
The source of the sine wave prior to rectification is usually an AC transformer, which adds reactance (inductor) to the circuit. That is what makes the classic PSU to be able to provide smoothed wave even when load current reduces voltage below its no-load/peak value. The presence of inductor in some form is essential.
True for buck/boost drivers. That’s because they implement excessive functionality in a restrictive form factor. But the DD driver will easily match and exceed the current. And they are readily available (for example, FET17-NUV $7.95) And adding just three passive components between the driver and led should make it’s output comparable or even smoother than linear driver output with less power loss. The cherry on top is that it can be implemented even without modifying the driver board, just splicing those components into wiring between driver and led.
Looking at the responses in this thread I feel somewhat assured that I have not missed some obvious flaw. I’ll wait for couple more days and unless someone points out a fatal error, I’ll proceed with calculating and procuring required elements to modify a FET driver that I already have in hands.
As a kid, I’ve build a few of big and heavy power supplies to power the music amps. Hunting for biiiig electrolytic capacitors to complement a heavy transformer was memorable.