I agree that constant current dimming is ideal, but using LED PWM to get lower modes doesn’t necessarily defeat the purpose of the buck driver. The buck driver allows a large input voltage range to be used efficiently.
Well are there any that actually interperet/ internalize a PWM input though? And if they do are they not just a resistor and cap in the IC? The Ti and the one NF showed have analog setpoint inputs. I'm not sure inputting PWM with a cap and resistor is so shoe-horned. It seemed a little that way until finding it in the standard application diagram. Are you sure the ones you used didn't call for a cap and resitor on the control input. Maybe you just didn't take much notice?
That's half of what it does obviously. But it seems strange to not use the other half.
Can you sent link to him as blf member? I hope he would make test if Ill sent him one driver.
Like I said, I only skimmed the thread, didn’t really catch what the proposed solution was, only that it seemed to fix an issue that is already not an issue in a lot of cases.
I am sure there are both ones that sense PWM and that use resistors and caps be it internal or external.
Really it is not that hard if that is all you are talking about. The issue is driver space. Except for something like a SRK driver space is already silly tight for a buck driver that can handle high currents. Adding anything at all to that would be interesting.
At this point, I would not be sure that the ones using constant current aren't doing some secondary linear regulation. I mean they shouldn't be, but I don't know anything until I know.
This is wrong if we’re talking about switching regulators designed specifically as LED drivers. Switching regulators with a topology for driving LEDs typically (again, it depends on the specific driver…) have PWM input dimming which sends a corresponding PWM signal to the LED string, analog dimming which sends a constant current to the LED, or both. General purpose switching regulators typically don’t have these features, which often make them much better for our purposes since analog dimming is still very possible.
The whole point of PWM dimming in LEDs is that it preserves color reproduction and avoids tint shift. Linear, Maxim, TI, etc., aren’t designing most LED drivers for us or even for flashlight manufacturers. They’re designing them for the automotive industry, for interior lighting, for industrial applications, and for a number of other sectors. In most of these cases, color reproduction is fairly important and efficiency less so — PWM is very appealing if you’re designing interior lighting for a shopping mall. Efficiency doesn’t matter as much, since you’re running the LED string off an AC main, but you do want all the lighting in your newly-designed shopping mall to be both pleasing to the eye and reproduce colors accurately so that shoppers buy things.
Here’s an old Digikey article which talks about exactly this.
The way we use LEDs has very very little in common with other applications (no engineer in their right minds would put into production a design which drives LEDs 5-6x past their rated currents, which is what we do on a regular basis). PWM dimming isn’t usually ideal for us since we’re trying to squeeze every last mAh from our batteries…but it’s great most other places.
Bingo!
Sure I can see saving two components would make a big difference in a flashlight, though convoy S2's won't need a buck until forward voltages come down more. You didn't miss any solution, just what we've been discussing now. It's not earth shattering. Right now the only decent buck I know how to buy though is the MTN-MAX( maybe I just don't know), so it would be a shift I think in what's available for modding. I don't want to criticize the MTN-MAX either. Getting any buck into such tiny spaces doesn't look so easy to me. It's more like a challenge that's out there, to get a better one.
near the end of the thread I made on a SRK buck driver I think I posted a new TI buck IC that RMM pointed me towards. I have not had time to read the (very long and detailed) datasheet but it is what I plan to use when I finally get around to messing with buck drivers again.
Yes, I linked to that data sheet in the OP.
http://www.ti.com/lit/ds/symlink/lm3409-q1.pdf
It's the one I based this plan/example on. FIgure 15, page 9, in that document clearly shows that their idea of PWM dimming, and what the PWM dimming pin input does, is just turn the driver entirely on and off. The figure shows a square wave LED current. However, that driver also has an anolog dimming input. So you can use that the same as NF's example, send pwm to it, with a low pass filter or some other form of PWM to DC converstion, but then you need to add the two filter components to the board at least.
I think a programmable buck flashlight driver with true analog dimming would really be a new thing. It's not clear to me that it exists.
Interesting, I should have taken the time to really grasp what you were saying before posting. That is my bad, sorry.
This is an idea, although sending an analog signal to the dimming pin will cause it to lower the current? Or is it an on/off switch?
No worries.
It has two dimming pins. One is documented for PWM. That's an on or off, and it pulses the LED. The other labeled IADJ, takes a voltage level. That one regulates the current output in continuous DC fashion as desired, but, it requires an analog input.
As said before, replace the word buck with boost anywhere you want I think (and double the work).
I think I will be using this idea with the buck driver I am (re)starting work on. You can see it here: The Texas Buck driver series, Q8 / Skyray King 2S/4S buck driver RELEASED!
I will need help with part selection among other things if anyone cares to help make it happen!
Here is no idea, just some members dont undestand what it R-L-C circuit.
:person_facepalming:
Definitely finding some details in the calculations that are causing some challenges for the buck over a range of circumstances but none of them have to do with "understanding LRC circuits". More like IV loss across the diode, the need for sense resistors, and gate drive losses. The primary losses are simply caused by the need to have a diode, a fet operating at a high frequency, and sense resistors, with high current running through them all. Resistance obviously plays a role in all parts. The diode can be removed by more complex synchrounous bucks, and the resistors improved with amplifiers, but probably not for a small flashlight driver.
The only thing so far that seems to strongly impact transition to moderately low power though is the gate drive, and that might turn out to be an issue. Indeed running at lower frequency helps that but does bite back in an earlier transition to discontinuous mode for inductors that fit in flashlights.
Anyway, for the record the simple resistive dac discussed above has already been invented and is called an R-2R ladder network, reuquiring no transistors outside the MCU. Of course you need a spare pin for every bit of precision and yet another pin still to PWM down to moonlight, so it's pretty limiting.
The PWM to cap thing will work pretty well though. It works best with the cap fed from a voltage divider not just a resistor. I've worked out the details of how the voltage divider interacts with duty factor (the cap bleeds to ground across both resistors during off time, but the result ends up guessably simple), ripple, RC time, and even corrections from the bias current on the IC (in the case of the LM3409). It's not any big deal, but I'll try to summarize some it here soon. It looks like for the LM3409 we'll be able to get about 1% control resolution/ripple in worst case with no ripple at max output, so significantly better than a 3 bit DAC I'd say.
I've posted the details of how to set this up here: