So errr, don’t order SiS414DN to use on this driver. I must have missed the fact that it’s actually PowerPAK 1212-8, a much smaller package than LFPAK56. That’s why the “package limited” values seemed so low!
Is this kind of FET smaller than the one used in the BLFTiny10DD or 15? If it is maybe it could be updated to a A10DD-(some nice acronym for the new smaller FET) with room for a battery positive on one side so you don’t have to build one over the mcu for example.
It’s slightly larger than SOT23 (and significantly higher performance IIRC). It appears that PowerPAK 1212-8 is also known as: LFPAK33, DFN3.3x3.3, PQFN08B, EDFN 3x3, TSDSON-8, Micro8 Leadless, and MLP08S according to this document from NXP.
There are even smaller modern power-MOSFET packages which might be good choices for the task you mentioned. It appears that there is a SOT-363 compatible footprint being used now which is referred to as things like “PowerPAK SC-70-6” and “UDFN2020-6” and probably other things. It’s more like 2x2 rather than 3.3x3.3. Maybe something in that size could be found the for the smallest DD drivers. That said, if the larger 3.3x3.3 pacakage fits it’s what I’d probably go with.
Thanks for the answer wight, i realized that it was larger myself after my post so i deleted it, i am glad you saw it before i did
Of course if it would even be possible to make an even tighter 10mm FET driver i would be very interested, i have a bunch of small 10440 possible hosts coming that i just have to somehow put a XP-L V6 and a tiny driver in, but i think the slimmer the driver could get the less modification i will probably have to do to make it work and leave more room for heatsinking.
What do you think of the SI5415AEDU-T1-GE3, would something like that work?
it is a 1,9mm x 3 mm PowerPAK-8 ChipFET Single
When I first built the Tiny10 FET the intention was for an Efest IMR10440 to run it at around 3A. But I was urged to test it with larger cells. As I recall it went up to almost 6A with bigger 18650 cells before I quit, wanting to use it in my favorite light and not kill it.
I have a new one in that light now that is pushing an XP-L V6 2C at 966 lumens from the Efest IMR10440 at 3.2A current draw. My cells are old, not fully charging anymore, so I have a couple of new ones on the way. Looking to get 1000 lumens from this 3” neck light. (the 2502 FET is the one it uses, I think)
Looks pretty awesome. This looks a whole lot easier to fit into a Cypreus EDC than the BLF17DD I used. I especially like that it’s compatible with retaining rings, since I hate soldering in something I might want to reflash later. And it’s ready for zener and off-time cap and everything!
Any chance of adding a single 7135 chip somewhere, attached to a PWM-capable pin? Giving it “moonlight special” hardware would be icing on the cake. No more need to use tricky fast PWM=0 hacks to make a high-powered light work at low levels, not to mention longer runtime on modes under ~350mA.
Edit: For my purposes, additional 7135 chips wouldn’t matter. I just want the lowest modes to work better without sacrificing the ZOMG WTF BBQ blinding output offered by a FET. One 7135 could handle modes from 0.1 lm up to about 50 lm, and the FET could handle the rest.
What would be cool would be if an optional 15mm daughter board with 7135 to be attached to this great piece of work which would in effect keep the bottom flat for the retaining ring. That way maybe there is the possibility of adding up 4 or 5 7135's to suit one's needs.
What would be cool would be if an optional 15mm daughter board with 7135 to be attached to this great piece of work which would in effect keep the bottom flat for the retaining ring. That way maybe there is the possibility of adding up 4 or 5 7135’s to suit one’s needs.
Mattaus already made a 16mm daughter board which carries up to 6*7135. It’s in the OSH Park projects thread. Depending on the goals that is a good option.
And how would the daughter board mate to this driver for control...and is there firmware yet that can handle it? Sort of like that 7135+FET Turbo that you have sitting around the million other ideas ( :) ) except this keeps the retaining ring side clear and you do not have to struggle to find 1/10th of millimeter. LOL
Eh, the PCB Mattaus shared is intended to be used with 1 PWM signal for the whole PCB. So you’d have 1 PWM signal for that board and 1 PWM signal for the FET. [EDIT: This should be workable with the current STAR_off_time firmware.] You would not be able to achieve as low a LL/Moonlight as you the single 7135 setup or the 6*7135+FET Turbo setup (which is setup for 1*7135 w/ PWM + X*7135 w/out PWM + FET w/ PWM).
Mattaus’s PCB uses 4 wires: GND, LED-, LED+ (just passthrough), and PWM. You airwire the whole thing together. It would definitely be possible for me (or Mattaus or whoever) to layout a pair of PCBs specifically intended for stacking, but I don’t see me doing it anytime soon.
I don’t know of any specific instances where the hybrid 7135+FET driver won’t fit where people want it to. I’m sure they exist and I wouldn’t mind hearing about them (in that thread…).
It looks ready to go to me. What are we using for moonlight with a single 7135 these days? I think you posted some good info in RMM’s store thread, but I’ve lost track.
Awesome! Best of both worlds.
I don’t have a single-7135 driver at the moment, but the moon level depends on both that and on the specific LED used. Like, for an XM-L2 it can probably run at about 6 or 7, but an original XM-L may need 9 or 10.
This driver looks like it should do almost everything for almost everyone. The only thing missing is a zener, and there are other drivers available for that. I’m not really into multi-serial-cell lights anyway; my only MT-G2 light was pretty disappointing so I’m not eager to make more.
It seems that we are having some problems with Zener builds. I thought I’d tested one out on a little PSU at least, but now I’m not so sure: both DBCstm and RMM have reported [by PM] being unable to build a Zener configuration with v024 and have it work!
RMM reports that placing the cap which normally goes at C1 on the Zener pads and then stacking the Zener on top of that works. That’s pretty doable and a great way to salvage existing boards if needed.
We know the schematic to be good, the same schematic works for other designs. RMM reports that his new 17mm & 20mm drivers of similar design are functioning, so it seems clear that this is a layout issue on mine. I’ll re-test with the Zener pretty soon.
In the meantime, here’s an UNTESTED revision intended to address RMM’s suggestion of moving the C1 closer to the MCU. C1 is now fed through a single thin trace, R1 has been moved so that it is no longer between C1 and D1/R3, and C1 and D1/R3 are connected through a massive trace, and all of this is happening right next to the MCU’s Vcc pin. I also increased the size of LED+ slightly and doubled down on the GND vias. EDIT: and no, LED+ and MCU Vcc do not actually touch. It’s just OSH Park’s rending system that displays it that way.
Since I’d been playing with the layout again for this driver it struck me that what I’d really always wanted was LED+/- pads that were across from each other. IMO that’s a key aspect when it comes to very,very, tight builds. I set about shuffling things around, but they wouldn’t play nicely. Then I recalled that the LFPAK56 footprint I was using sucked. Switching to a normal Power-SO8 footprint helped, but it took me a long time to achieve the level of symmetry I was looking for. Without the OTC it’s actually really easy to do all kinds of great symmetrical patterns (ask me how I know, oops!). In the end this was the best I could do. Now that the signature look of the original driver is entirely gone, I’m not sure whether to call it something else or what.
Also note that I didn’t spruce up the Power-SO8 footprint any for LFPAK56 use. While it should function fine I’ll probably do that at some point.
As far as functioning with the Zener, this should be very similar to v030. (so in theory they both should work fine :p)
Oops, don’t use this one. I made a minor error with the vias & GND ring on top. While this one should work, I recommend v024, v030, or wait for v042.
Now moron’s like me can stack the MCU on the MOSFET footprint.
That looks nice Wight, dont’ think you should change the name as the Revision number suffices. That’s just me though. I’ll have to get some of these, and solder paste stencils, and give em a go. (starting to get more than I can remember what goes where) I need to use up the ones I have and switch over to these… my memory doesn’t multi-task well.