Where did OP go?
I think the first step is to make a large (prototype?) driver for something like a srk. Make it 3A, that shouldn’t be too much of a problem, right?
If that is working properly try to make it smaller and smaller. If you have a working boost driver in de 24/22mm size (or 17mm) then look at increasing the current.
Also for the smaller sizes, to maintain enough room maybe a stacked PCB driver is a option? (Two pcb’s on top of each other.)
I would really want to make a single 26650 mtg2 light. That would be my first mtg2 light…
If you need extra room, does anyone ever stack the drivers in the tailcap, and turn the battery the other way ? Would also be less heat on the driver, and maybe more heatsink in the head.
stacking circular drivers is better than the bad idea i originally had
I think some work has been done with that boost IC in the past. It’s been recommended to me, but I haven’t spent any real time looking at it yet. I’m currently working on the AA boost stuff and tidbits.
I’ve said this before, but it’s been a long time. This “huge prototype” concept just doesn’t make sense. What we built into 17mm depends on what can fit. The same applies to 20/22mm & etc. Starting at >40mm doesn’t really give us any insight into what fits in 17mm. The layout isn’t similar. Most of the concepts we are applying here are understood on some level or another - and then bodged to get everything to fit into a flashlight. We know that a product like TPS61088 can drive an LED, so making a big driver just to test it out doesn’t achieve anything.
I suppose that this could be compared with someone suggesting that we build a “really big car” to test out the ideas before we build a normal car. 
This comes up a lot and seems like a good idea until you look at it from the correct perspective. In a typical driver we’ll say that:
- The LED connects to 2 things (LED+ and LED~~, some of which may also be the same as BAT+ or BAT~~ depending on the type of driver)
- The clicky switch interrupts either BAT+ or BAT-, we don’t care which.
- The driver connects to several things: BAT+, BAT-, and the LED. The LED may constitute either 1 connection or 2, but no less.
On either a 7135 based driver or an FET driver we see that the LED connects to BAT+ in addition to the driver, so that’s:
- LED connects to BAT+ and the driver
- Switch interrupts BAT+ or BAT-
- Driver connects to BAT+, BAT~~, LED~~
If we move the driver to the tail how would we facilitate those (required) connections? (A battery tray is the only somewhat sane solution. It’s generally not a great solution for us.)
What about this……
DESCRIPTION :
1.Work with 1 x Li-on battery or 2 x Li-on batteries in Series Connection
2.Upper Diameter: 18.5mm，Bottom Diameter:20.5mm, Thickness <10mm。
3.When the volt is below 3V on 1 x battery or below 5.8V on 2 x battery, the light will alarm.
4.Anti-reverse protection
5.LD-29 modes——2 groups
1、Hi 3000mA- Mid 700mA— Lo 100mA
2、3000-700-100-Strobe-SOS
6.Runing 5 seconds on Low mode, the light will flash, then you can change the group by turning off and on.
7.With Mode Memory
That is (ostensibly) the venerable LD-29 from LDCH. HKJ has reviewed it twice (1 / 2) and it’s gone through at least one generation since then. It is a buck driver, not a boost driver. If you aren’t familiar with the terms, take a look here: lagman - Understanding the difference between Linear, Buck, Boost and Direct Drive drivers
EDIT: I went and looked at the ones sold on FT. The one you linked to looks the same… at first I (incorrectly) thought it looked rather rough in comparison with my memories of the current generation LD-29.
There are laser drivers on rectangular boards which fit lengthwise into small tubes. Then there are stacked boards as previously mentioned. And if you use an optic which is shorter than a reflector, it leaves room in the head of the host for these longer taller drivers.
The Acebeam HC10 headlamp running an MT-G2 on a single 18650 looks really small in diameter. The board is no doubt bigger than 17mm but the host sure looks small.
And we should keep in mind that we sent someone to the moon when? 1969?
I’ll concede that all of that is true (I don’t know if it is or not). However, I was specifically referring to the size that will no doubt be required for a very high current boost driver for single-cell Li-Ion flashlights running XHP series emitters well above their ratings. I would expect it to at least require a huge inductor, a very large boost chip, like the one I linked earlier, and of course, a MCU or PIC to control it all. All of that takes space, ya know.
AFAIK, that driver has never been attempted. Many here think it can’t be done. I’m not smart enough to doubt it can be accomplished with current technology, so I’ll just foolishly say I’m sure it CAN and hope somebody soon proves me right! O:-)
I happened to look over the current-version LD-29 again. The last time I looked I thought that the floating/daughter PCB looked very different from previous versions. Turns out that they’ve just flipped the floating PCB upside down. I’m sure that this gives a tighter stack, so that’s a good thing. The reason the current LD-29 listings specifically mention 6v LED compatibility is that even though this was always a 1s/2s driver it didn’t previously work correctly with 6v LEDs. Output voltage had an upper limit of around 5v IIRC. I forgot about that when I wrote my previous post or I would have mentioned it.
shorter version: LD-29 will (now) run a 6V LED but only from two cells. Input V must always be higher than output V no matter what the cell/LED arrangement.
to drive an xhp series to good output with a boost driver will be VERY demanding on a single cell.
forget laptop pulls.think tool cells like vtc5,25r,ect.
what chip are you using?
i doubt it needs 3 47uf parallel.a single ultra low esr mlcc or poly can save enough board space for your attiny.
As a matter of fact, theomajigga used TPS61088 for the design in the OP. PCBs with two or more different layouts are pictured in the gallery linked from the OP. The assembled demonstration board shown in the video is not the layout shown in post #21, but does match one of the ones piled on the desk in the other pictures. In any case, Post #21 shows a voltage-controlled driver (rather than current controlled). Theomajigga’s Eagle markings indicate a 6v set voltage (but the board could have been built for another voltage during assembly just by changing resistors).
It’s possible to implement TPS61088 as current-controlled for our purposes, but not as simply as some other controllers.
The example of the DQG driver should be enough to show why a single cell high output boost driver is a Bad Idea. I don’t understand how that’s not obvious.
That driver stock only does 1.5A output on turbo. If you want to make one that’s worth doing (3A at least) that means you have to more than double the input current. Much more than double. Because with any cell, the more amps you suck out of it, the lower the voltage falls. The lower the voltage falls, the more amps the driver tries to pull. Rinse and repeat until you smell smoke. You can estimate an input current with the best cell available (of whatever size and capacity), with the cell fully charged, of around 7 amps, possibly more, for just a 3A output. If you can make the driver survive at the beginning of the cycle with a fully topped up cell, it will self-immolate when the cell starts running out of juice. That’s just not an acceptable tradeoff in safety and reliability just to get a relatively piddly 3 amps output.
The DQG driver isn’t a flawed example, there’s nothing wrong with the design. It’s not a case of not being able to handle higher output because it wasn’t done by the right smart person or that they didn’t try hard enough or they didn’t try for more because they didn’t think there was a market for it or because it would be too expensive. This is just how boost drivers work.
And this is not anything even remotely related to the current levels a ultra low resistance direct drive driver can handle without issue. A boost driver by nature has to switch the input on and off in order to create that voltage boost, and that switching creates enormous heat loads and that’s why these types of drivers tend to catch fire when pushed beyond safe limits. As temperature of those switching components increases so does resistance and that means the driver tries to pull even more current just to keep up which makes even more heat. It just gets exponentially worse.
If you could power it with a bench power supply and hold the Vin at a rock solid 4.2V at all times it would work just fine. But give it a constantly falling input voltage and eventually something breaks.
Thanks Comfy, that makes enough sense for even ME to understand.
Meh. I suppose it depends on your goals. The DQG driver doesn’t self-immolate when operated within spec, at least AFAIK. The components on the driver are relatively low current IIRC. The AO4468 which is used on the driver is only rated for a small fraction of what modern high power packages are rated for - and the modern packages are of equivalent or smaller size. SO-8 is poor at dissipating much energy and wasn’t intended for stuff like the AO4468
The idea that a driver can’t possibly stand up to high switching currents or be safe switching high currents doesn’t hold any water. TaskLED has high efficiency boost converters. The Meteor M43 uses a boost converter - I don’t know how efficient it is but it seems to get the job done, right?
I’m not arguing that boost drivers don’t work or are unsafe or that big output current isn’t possible. Where it falls apart is when it’s a single-cell-specific boost driver with big output current.
3A boost driver with…
- Single cell, single ‘6V’ XHPwhatever = no.
- 3 series cells, 3 series ‘6V’ XHPwhatevers = no.
- 3 series cells, TWO series ‘6V’ XHPwhatevers = YES!
- 4 series cells, 3 series ‘6V’ XHPwhatevers = YES ALSO TOO!
The Taskled Maxflex was originally rated at 1300mA but later versions were reduced to 1200mA. Given a big enough space I’m sure it can be done but the point I take from CC is that in the sizes we generally use its not practical and maybe not possible to push an mtg/xhp with a boost driver to the same level as can be easily reached with 2x18350 cells and a Zener modded DD driver. A single 18650 light has a limited amount of driver space and the Maxflex and DQG are already too big for it. An HBflex is 35mm and still only goes to 3A output where his same size buck driver does 6.6A. How much space would it take to get 8A to a 6V led? I’m thinking 4p in an SRK or bigger soup can might be enough after adding more length for a deeper reflector and vastly more heat sinking than is typical. It just seems all wrong to me, like pushing on the wrong end of a lever. Yes, it’s already possible to vastly under drive a 6V led by maxing out a boost driver in a single cell light but to what purpose when a triple will be more powerful and have more run time? It’s fun to stick a 6V led in a light with 2x18350’s, turn it on and giggle but a relatively compact and useful 6V light should have both the cell capacity and the size to handle all the extra heat. Do it if you want but don’t expect to come even close to what’s possible with lower voltage LEDs and a single cell.
EOR
You can have…
a. Output of 3A or more
b. Vout double Vin
Pick only one.
Dale, something I forgot to mention. Your measurements of your light with the DQG driver… the stock light’s 3S XPG2s have a higher vF at 1.5A than the single XHP70. Input current in the bone stock light is much higher than in your light because of that difference in voltage, even though the output current is the same.
This is all pretty academic, but according to the technical info page at TaskLED the HBFlex can do 7v-in 15v-out w/ 3A… at 80% efficiency. Reducing the ratio slightly increases efficiency a lot. 18v boosted to 36v at 3A also achieves >92% efficiency. Just barely enough to break your pick-one comfychair. 
FWIW the HBFlex is 35mm in diameter, so to simply fold it in half would require…. [insert circle math] a driver of ~25mm+. Realistically it wouldn’t work that way because the inductor is large enough to spoil a 25mm layout - plus the bottom side needs space for a spring pad. Not to mention the layout considerations required to get the heat out to the edges of the driver where it can be sinked by the flashight body. I haven’t looked over the specs of the large Cooper inductor, maybe a Coilcraft could be slightly smaller. Meh.
Granted, ~3v input is a whole different ballgame. I’m not recommending that any doubters go try and build a high-current boost driver. All I am saying is that unilaterally poo-pooing boost drivers won’t do anyone any good. Newbies are still going to have tons of false impressions of what a boost driver could do for them until they they read up on why it’s not going to work the way they think. Anyone who already has the know-how to develop one doesn’t need to be told the limitations. For my own part I’m in no hurry to try and develop a small, high current boost driver which can operate on 1s.
