[PART 1] Official BLF GT Group Buy thread. Group buy officially closed! Lights shipping.

BLF Special Edition Products
3281

There are MCUs with built-in SMPS functionality, but not the typical ones we use.

One problem is the relatively low PWM switching frequency. The L1 part on the GT prototype driver would need to be physically ~10x larger to get the same performance. There are other issues as well, like having proper low-impedance driver(s) for the FET (s).

Firmware can be relatively simple. The LM3409 used in the prototype driver uses simple ‘hysteretic’ control: Inductor current gets to set-point, FET switches off. Current drops below set-point far enough and the FET switches back on. No PID loops or compensation issues. But without a dedicated SMPS chip, the SMPS-related logic needs to be robust. In a buck driver a software glitch could be instant death to the emitter.

3282

Indeed, I have thought about using the MCU to control a buck before and it is very possible but also a lot more complicated. If you have the space and budget for the buck IC it is a lot better to just use that. If space or budget is limited then using the MCU would be a good idea.

For example in a 17mm buck space is so tight you might have to go with an MCU controlled buck just to have enough room to fit everything on the driver.

3283

What do you think is the smallest size to get a good buck driver with comfortable parts spacing, TA?

3284

It all depends on what you want it to do. The amount of current that the driver can handle is going to be directly related to the size and cost of the components. These are things you can’t escape.

For a low current driver, say 3A, you can get a pretty small driver, 17mm for sure, maybe even 15mm or smaller. For a higher power driver, say 6A, 17mm would be a tight fit for sure as seen by the mtnmax with it’s 5.5A limit.

That could be vastly improved on in the feature and firmware but the actual current could not be improved all that much without going to an external coil.

If you did go to a stacked coil inductor setup, then figure you could get a fair bit more current out of the driver, not sure exactly what components options there are but I would guess that a tall 17mm driver would max out around ~8-10A without potting the driver, although even that would be hard as shown by past attempts.

This is why higher voltage is going to become more and more important. Regulating more then 8-10A gets exponentially harder in these small lights. On the other hand regulating a higher voltage setup with lower current is easy.

For example this GT driver would die a very swift death if used in the Q8 with 20+A of current at 1s. You would need something like the Texas buck to handle 20A at 1s and even the Texas buck was only designed to handle 15A. The cost would go from $10 to$25+ to handle that extra current.

On the other hand if we raise the voltage to 4S we only need 5A of current to give us the same output power, this means we can use the much cheaper GT setup and get a more reliable driver as well since it will not run as hot.

As a side bonus we can maintain regulation longer with higher voltage.

This is why I am a big fan of higher voltage, it simply works better for all mathematical, electrical and scientific aspects.

This is also why I am a big fan of the 21700 cell. With the larger cell comes larger drivers, that extra space gives us a lot of wiggle room for driver designs.

3285

The switching frequency you’re using is much higher than I expected. With all the electrical noise and heat created at speeds like that in such a small space its surprising the FET’s turn off properly.

How fast can you run a converter before the efficiency really drops off?

3286

Yeah I know you love the series setup TA, but for lots of single cell lights it doesn’t make sense right? I mean a single 18650 has more energy then two 18350/16340 cells no matter how good these are.
For a big light like GT with low amp needing XHP35HI I get it too and now we are doing a buck driver.
But it is very interesting to see the discussion here and I must say seeing the GT driver I would not have guessed a 17mm buck was possible.

3287

The faster we run, the less inductance we need for the same output ripple current. So physically smaller inductor, given the same current rating.

The LM3409 chip can do up to 5 MHz, in theory, but apparently it is difficult to achieve good efficiency above 1 MHz. This number is also not fixed. This chip is using ‘constant off time’. So the frequency is tied directly to the duty cycle and efficiency, which in turn is tied directly to the battery voltage. I am aiming for 500 - 600 kHz with full cells. The frequency will drop from there as the battery drains.

Of course it is mostly the switching speed of the FET (and recirculating diode) that gets you when you push the switching frequency. And fast FETs have high Rds-on (so high I^2R losses). Part selection is a delicate balancing act. (This chip needs a P-FET, but apart from that, the nice low Rds-on N-FETs we normally use are way too slow for a switcher.)

3288

I’m in.

3289

Put me on the list :slight_smile:
Thanks.

3290

Put me on the list please. This light is too interesting to pass up.

3291

I haven’t looked into this stuff too deeply yet but making that work inside a flashlight is borderline magic to me. Getting current to flow then stop continuously for 0.00000035 of a second with all those square waves and harmonics bothering the gate across a 100+ degree Celsius temperature span seems so mind boggling. And that’s only one piece of the puzzle. Best of luck finishing the rest of it.

3292

WTF (still lolling for your username :wink: ) that is one of the awesome things DEL can do, scope the currents. It helped a lot with the Q8 driver and DEL is going to build the GT driver proto as well, yay!

Welcome new members, will update list later.

3293

MARK
And core list updated
556 (the program of r.Scott is going to sort entries into the correct lists, I will place my second entry at the end of all backup lists eaning we have AT THIS vERy MOMENT reached 555 and this first backup list is gaining importance for up to now all seemed to be able to get the number of GTs they wanted, now this might change, and as I wrote, the goal is to get a 1 special groupbuy-priced GT in the hands of as many people as possible.

3294

It’s your group buy, but I still think that is a wrong goal. :innocent:

3295

Thanks :slight_smile: . Switch-mode stuff does have a little bit of magic to them. Take a look at that boost driver thread for the next level up.

Using a well-documented controller chip, supported by the manufacturer’s design tools, makes it a little like ‘paint-by-numbers’, fortunately. The key will be to have it perform within the constraints and targets we have.

3296

I’d like to get one of this, too! put me down please

3297

Can’t pass this up, PLEASE put me down for 1!!

Thank you!!

3298

I’m interested, put me on the list please.

3299

MARK

Welcome to BLF!

Gonna update list in a little bit

3300

Will you print serial numbers on first 555?