BTW, has anyone considered XAR7030 inductors?
These are not very powerful for 8.5x8.5mm footprint, so I’m not sure if they make sense for us, but provide an unorthodox way to increase PCB real estate.
17mm and 50W driver is just for competition and hobby IMHO. More easy way is to find lathes and make wider diameter for 22mm driver or just proper host ;))
Whoa! I haven’t seen triple stacking before, some serious work there!
I’ve seen them mentioned here before somewhere. I haven’t taken a serious look at them though. For drivers with maybe not so much output I think they could be a good option.
Quite weak when used alone, but maybe with one on each side of the driver or 4 of them on a larger (but still quite flat) driver…
I agree with this. These days, I don’t know what I’d even need a 17mm driver for in my current collection or anything on my wishlist other than the Convoy M2 I mentioned earlier. 20mm is suitable for anything modern, esp 21700 lights.
I feel that most everything I know of with a 17mm driver is probably not worth receiving this awesome driver. I’m curious which hosts MikeC has in mind? I’m clearly missing something.
The exercise of fitting everything on a 17mm driver definitely makes upsizing easy, but may actually force compromises in layout too.
Hmm, I never considered that. Something to keep in mind!
I have a whole bunch of older hosts that take 17mm drivers and I want to use my own driver designs in them, what ever that design is. I’d just rather use all of these old hosts than sell or discard them, and messing about with lathes is not something I have any interest in at all. The first lights that will be seeing them are these: Build/Mod: Zoomie with Brynite B158B host (wider flood mod) I have custom copper pills for them and I’ll be putting 12V XHP35 HI or similar LEDs in them, they can’t fit larger than 17mm drivers.
But ultimately, yes, guilty as charged! It’s a hobby, that’s all it’s ever been for me. It’s satisfying to have got it working with 17mm and I learnt a few things in the process 
Up-scaling is easy, in it’s easiest form all I have to do is increase ground ring diameter. If I put a little more effort into it than that, well, if I can make it work with 17mm then I think I can make it work with more space without compromising layout too much 
BTW, as to height restrictions…
normally you have:
- X mm driver spring
- 1-1.5 mm PCB
- Y mm inductor
There are 2 height limits:
- from the PCB mounting point to the end of the shelf
- from the end of the tailcap to the front of the driver stack (so a cell fits).
If there is some component (inductor) protruding from the PCB towards the cell, it shortens maxmum supported cell length. Which may be OK.
But mounting a large inductor on the battery side would probably be too much except for extreme cases.
How about stacking 2 PCBs, one lays flat against the shelf, the other is in a regular position? The one in the regular position has a hole in the middle through which the inductor sticks into the battery tube. There is a battery contact at the top of the inductor.
You may even make the PCB against the shelf on a flex substrate to save another fraction of a mm.
Though I don’t like the complexity of it and I think that the stack would need some adjustments to get reasonable drop resistance because cell pressing the inductor against the shelf doesn’t seem good to me….though I have no idea how mechanically robust are these inductors.
HA! I have my first B158B coming in the post right now. Ordered it a few weeks ago, hoping to see it soon. I only had plans for NM1 and SST20w. I didn’t buy an extra pill, would love to have gotten a few spares back when there was a GB eons ago. I tried messaging AE sellers to get some w/o driver or led but no luck.
17mm is a weird size for 18650, normally the tube is at least 21mm so there should be space for a driver of that size. Isn’t a 17mm host just a bad design ?
That said I’m very impressed with Mike’s layouting skills for fitting that much components on 17mm. I really need to improve myself on that.
The issue I see is that on multi emitters hosts with a centered hole the inductor can block the passage for the LED wires so it will have to be moved closer to the side and the regulator with it to keep the traces short. On my driver I thought about using a wider inductor for lower DCR but decided against it due to that.
On another note, I see a diode on your board, what is it for ?
In the case of an S2+, with its “push” style design, and the driver being mounted in the pill itself, and held via a retaining ring which threads into the pill, 17mm is the maximum size.
If the retaining ring were not integral to the pill, but instead threaded into the head, sandwiching the driver between the pill and the ret ring, the driver could then be as large as the ID of the head. The ret ring is not even necessary as the body of the light can serve to compress the driver and create a solid GND connection.
But, there are two problems that need to be addressed: retaining the driver in the pill, and installing/torquing the pill into the head, both of which were previously accomplished by the integral ret ring, and this needs to be done in a manner that is suitable for such a universal host.
Possibly bad design, but Convoy S2, S3, M1 and M2s (?) all have 17mm. These are quite popular hosts, and I have quite a few of them 
Yes, this is a valid concern for sure regarding hosts with center holes. My current design for those kind of hosts is buck boost for low regulated output, and FET for full blast. For those drivers the inductors are not so big. But for sure it’s something to think about if I go with multi emitters in series. I don’t see any hosts for 17mm drivers with multi emitters that can fit the XAL7070 inductor.
My drivers and firmware support clicky (off) switches, momentary (E) switches, and dual switch (one of each). The diode is for what’s called “OTSM”, a method of measuring off time for clicky switches. On the “inside” of the diode there is a 0805 100uF cap which can keep the MCU alive on power down mode and count exactly how long the light is off until cap runs out. I have an interrupt that detects power off on my VBAT voltage divider, and instantly shuts everything down and goes into very low power consumption mode. With 100uF cap I can run the MCU for about 30 seconds without power, measuring off time with 125 millisecond accuracy. This makes short and long off press measuring for clicky switches very reliable.
@Hoop. I’ve thought of carefully cutting the driver side of the pill off (a few threads worth) with a hacksaw and turning it into a retaining ring in the fashion you describe. Yes, assembly won’t be very smooth, but that’s modding right? 
Thanks for the explanation Hoop and Mike.
@JaredM I suppose with enough slack wire the driver can just be turned sideways while the pill is tightened, and then a ret ring with the same threads as the pill could be installed. In that case, a 21mm driver could fit in the s2+.
I have to change this answer a bit… those 30 seconds of off time is for “normal” drivers. This boost driver has a couple of fat input caps that have to drain a bit before the voltage is low enough for the off time interrupt to trigger. That means everything is happily continuing normally for a moment after power is turned off by the switch, and once off time is detected there is less juice left to keep the MCU awake, so it’s about 8 to 9 seconds with this driver. I had forgotten all about his, it was too long ago I was working with this stuff :person_facepalming:
Anyhow, that was the last functionality test for V4 version, everything is working as expected so I can go forward with new V5 and test with MUX to eliminate Rdson from current sense. If there is interest for this V4 version I can make PCBs available for purchase on OSH Park, otherwise I’ll just do it when V5 is tested. Note that for V5 I’ve made little markings on the driver so that components for unwanted features can be eliminated as long as component pads with markings those markings are shorted. More details on that if V5 works out.
Did you measure current accuracy of HDR with charge pump?
I’m definitely interested on your driver, so availability on OSH Park would be great.
But I’m also eager to know how well mux performs, so I don’t know which version I’ll end up building.
Decisions everywhere…
I didn’t measure the accuracy, but I did notice some differences. Anyhow, I’d like to try the mux as I have use for it in other projects if it works well.
I also have to scale down my firmware for 1616. My 3217 firmware has a whole bunch of “engineering” functions and options for testing because I never had to care about space. Some of that stuff has to go in order to fit in 16K.
Ok, keep us updated 
Boards ordered from OSH park, now it’s just a waiting game.
I’ve built and tested the new version with the mux. I was looking at a few different chips but thefreeman suggested the TS5A3159A so that’s what I used. The muxing works very well, output levels look rock solid.
This particular driver I built can’t deliver the same amount of power with the same source but I think that’s due to the soldering issues I’m having with the inductors: Those that have soldered large SMD inductors on your drivers, what was your method?
I’ve had this issue before. It’s not the mux in any case, I’ve bypassed it and power levels are the same. I’ll build another one and check but first I need to make a little SMD hot plate or similar, I need a more reliable method for soldering on these inductors.
I’m out the door for a week, If there is interest I can make driver board available on OSH park and the provide BOM when I get back. The firmware isn’t fully assembled but I have all the parts. In any case if someone actually builds this driver they need the flashing kit regardless, so re-flashing later versions of firmware won’t be an issue. I used the ATtiny416 Xplained Nano, read more about flashing these newer MCU versions here: Adventures in TinyAVR 1-Series