EDIT: I forgot to preface this by saying that this is exciting work so far. Thanks for showing it off!
Wait, you showed a tailcap draw of 4.2A earlier. I don’t think we’re looking at 97 efficiency here… more like 65% efficiency, right? Here are my assumptions: Output = 1.5A @ 6.4v = 9.6W Input = 4.2A @ 3.5v = 14.7W~ Efficiency = 9.6W / 14.7W = 0.65 = 65%
I take it that this design is CV then?
IIRC that’s the direction Lamdba went with the boost driver for VaraPower lights. It was a fairly different situation, but my understanding was that the driver worked well and provided a lot of power. If you are interested I think that the best information on that driver is available over here, starting with post #20.
It helps me! Thanks for posting the link, but don’t forget to include the post # since BLF’s forum software doesn’t behave nicely with direct links depending on your “view” settings. FWIW that’s Post #1616 over there.
I have to admit after seeing claimed 97% efficiency every other posts just looked like a blur. I’ll have to play skeptics too, that’s just way beyond what I can believe.
But if the OP is reading and really working on this, don’t let our skepticism get to you, keep up the good work.
KD one is interesting, but it's big, specs are vague, and thinking there's some gotcha's with battery and LED configurations. It may be 3V-9V input and require at least 2 cells in series in order to support the 6V LED's, for example. I doubt it's doing boost and buck in that size and price.
It’s huge compared to the driver theomajigga is shooting for. 22mm isn’t in the same ballpark as 17mm.
As Tom E pointed out - it’s certainly just a buck driver. I’d be willing to wager quite a bit of money that the driver was based on either QX5241 or QX9920. Drivers of that size and type are common.
The new and very small Zebralight SC600 MkIII is running a single cell XHP35 - one owner measured about 4A at the tail. I assume their XHP35 is a 6V LED.
I wonder what size board they use? It has to be tiny and short (no stacked boards).
We already have single cell boost drivers in the Fenix PD40 (MTG2), Shadow JM35 (MTG2 and XHP50), Acebeam H10 (MTG2), Zebralight SC600 MkIII (XHP35) and SC63 (XHP35), so why hasn’t anyone cloned any of them yet?
Furthermore, Zebralights are known for maximum efficiency, not just small size.
Clone a boost driver? Uhhh - several have tried, several have died. Ok, not died, but gave up - same thing . Easy to say, not so easy to do, plus the costs could be insane for small qty runs. There's read locks on MCU's, part #'s filed off, mystery unknown parts with no identifying #'s, extremely small parts, multi-layer PCB's (layers you cannot see but are internal).
Quicker/cheaper to design from scratch, by the right engineer of course. But it's not quick or cheap.
In addition to what Tom E mentioned it’s hard to clone a driver we haven’t seen. Certainly we’ve seen the driver of the DQG tiny 26650 [comfychair] (don’t really like this setup currently) and the driver of the JM35 [M4D M4X] (new to me - probably similar to the DQG). I can’t easily find images of the PD40 driver or the others. I certainly haven’t seen detailed pictures of those drivers.
Often pictures aren’t enough. Take a look at that first thread I linked to - it took us a long time to figure out the basics of that driver, even with a willing comfychair prodding and taking voltage / resistance measurements and posting lots of large, clear pictures. If everyone involved had had one in hand the reverse engineering could have gone much faster. Tom E pointed out that it can be easier to design a fresh boost driver from scratch. This is accurate, but as Tom E mentioned one must understand the concepts involved in order to do it. While a circuit like the DQG 26650 for example can be reverse engineered without an understanding of those principles/concepts… it takes a large amount of work to determine values and part numbers for all the components. Sometimes it parts just can’t be identified without both measurements and a large knowledge of the industry… which most folks don’t get without already understanding the those principles/concepts mentioned earlier… bringing us back around to doing a fresh design rather than copying others.
I have an interest, but like others I have limited time. I don’t have the expertise to just lay out a problem free high-current boost driver in a day or two. RMM put in some work on a fresh boost driver over the summer but was unhappy with the results. Besides RMM, myself, and theomajigga I’m actually not aware of anyone else on BLF who’s (a) demonstrated that they have the knowledge to implement a boost driver like the one in this thread and (b) expressed an interest in doing so.
The driver in the JM35 might not be the best, but it’d be a very handy driver to have available on the market. Although the size of it would limits it’s use I guess. But I love high output lights of that size. My FourSevens MMU (SST-90) must use something similar, and frankly I much prefer these to multi 18650 lights.
Your FourSevens MMU w/ SST90 would not use a driver of the type we are discussing (boost). The SST90 is a regular “3V” LED rather than a “6V” or “12V” LED. Without seeing a picture (I’m always happy to be shown a picture of course!) I’d say that the driver is most likely linear. I don’t remember the Vf curve of the SST90, but I feel safe assuming that linear is the best fit for the SST90 + 26650 combo.
Thanks cncyana. The board is “uncrowded” partly because it’s huge. I estimate that the dimensions are approximately 18x20mm! Offhand I’m going to take the weird angles as a “bad sign” that they had trouble getting everything to fit with the high-power traces.
Unfortunately no-one is going to clone anything with low resolution pictures like those. If a person was very motivated they could ask vinhnguyen54 for some very clear, detailed photos showing the IC markings.
. Easy to say, not so easy to do, plus the costs could be insane for small qty runs. There's read locks on MCU's, part #'s filed off, mystery unknown parts with no identifying #'s, extremely small parts, multi-layer PCB's (layers you cannot see but are internal).