Face palm.
Question: Can the function of the inductor be separated over two smaller inductors on two separate boards?
Yes. I don’t see why not.
I wonder, what with all the things that have changed comparatively lately, has anyone taken a look at Mattaus’ Knucklehead?
In what way? For my part the answer is no, I haven’t.
RMM has his MTN-MAXlp working well AFAIK. If I decided to work on a buck driver again I’d probably want to focus on either higher current or lower cost than RMM’s driver. I don’t think that the LED2001 IC can do either of those things for us. I might be tempted to look at the QX5241 again, or the more common QX9920.
I was trying to look back at what the issues were with the Knucklehead, I know I built a couple but I’ll be danged if I can remember what the issues were… still got a box full of components here for those as well.
IIRC they tended to *poof* the $2 LED2001 ICs. IIRC you may have gotten them up to around 3A or so without failure when running the MT-G2. At the end of the day I think we decided that they should certainly be safe to around 2A.
I was thinking I got it to around 3A with the MT-G2, but when I started to say that I didn’t know for sure and couldn’t remember if it was one cell or two. With my memory being so fallible, even a “clear” memory has to be questioned. Sucks, but it is what it is.
The Knucklehead is a buck driver, so that MT-G2 result was certainly with 2 cells.
Thanks. I’m like that rotating raffle ticket shuffler, all the information is in the cage but pulling it out in any kind of order is next to impossible!
Was following this driver that theomajigga was building, but he has been MIA for over a month now.
That is unfortunate. Lucky for me I only found this thread today.
However theomajigga should you pick this up again, vesture will take a pile of these boards 
I can’t apologize enough guys and gals I was building this for the girlfriend at the time, well along with that relationship went the motivation to finish the project. I’ll start working on it again here soon. I will send some new boards to be printed. I do use the light every once in a while, but until the heat thing is figured out, it’s nothing more than a “I did it.”
I haven’t had a chance to read through all of the comments, but I’ll pick this project back up here.
SORRY!!
No need to apologize! We all understand that life happens. Glad to have you back!
Sorry to hear about your troubles. Please feel free to bury yourself in this project as consolation, we’ll understand. 
and probably appreciate it a good deal more anyway.
Wait, so you got de-motivated by a breakup and re-motivated by a new relationship, or, well, none of my business I guess.
We’re just glad your back!
What he said!! :+1: Welcome back!!! 
Now get crackin!!!
Have quickly read upper 178 post and have seen lots of foolish post and much less useless one.
What can I say:
- All boost drivers have similar circuit engineering with difference in inductance value and input and output capacitor values.
- Сalculating values of this components shows that we need to use one that big and expencive to have low input ripples and good efficincy.
- There is defenetly no simple way to make high power AA boost driver (it will need one step of boost for mcu and components work and second one for led).
- There is defenetly no simple way to make high efficincy 1xLi-ion->1x3vLED boost driver (it will need one boost step and second buck one).
- There is no need in copying driver circuits from lights - all secrets are in software.
- There is defenetly no problem if Vout is several (three, four) times more than Vin. Efficincy don`t goes down when difference increase.
comfychair again had wrotten lots of nonsenses.
Boost driver is driver, it will ever have regulated output current. If output power is 20W, with fully charged INR18650 cell current from cell will be 6A-ripples (4V on load, 80% eff.). With fully discharged INR18650 cell current from cell will be 8A-ripples (3V on load, 80% eff.). Is it to much for this cells? -No. Will you ever make more powerfull boost driver on 17mm board? -No. What is the problem?
Lots of you are using FET, DDs that are not drivers at all, they do not manage current in turbo and this situation calls short-circuit. It may be good if you want to be similar to chinise manufactures with “2000 lumens T6”, but not good if are arguing that boost drivers are bad at empty cells (boost ones will be able to gave hole power till the end, what is about your FET?).
There is defenetly no simple way to make high efficincy 1xLi-ion->1×3vLED boost driver (it will need one boost step and second buck one).
There may not be a simple way, but there’s definitely a way, as evidenced by the long and perfectly flat runtimes (outside of high modes that hit the thermal limit) of most Zebralight products.
There is no need in copying driver circuits from lights – all secrets are in software.
Examining the circuits of lights with boost drivers known to give good performance tells a driver designer what hardware is capable of delivering that performance. It’s easier to iterate through multiple software designs than multiple hardware designs.
Lots of you are using FET, DDs that are not drivers at all, they do not manage current in turbo and this situation calls short-circuit...
boost ones will be able to gave hole power till the end, what is about your FET?
True, FET drivers don’t have stable output, but the buck drivers used in most 1x18650 lights don’t have stable output on the highest mode either. A Nitecore MH20 (note: emitter swapped to Nichia 219C and fan-cooled to prevent thermal throttling) isn’t any more stable in its output than a FET-driven 219C triple at 22% PWM (note: 22% PWM because higher modes result in overheating sufficient to damage components if run continuously). The most significant difference is that the FET driver has a higher low-voltage cutoff than the MH20.
