[WIP] °12mm° DD+single-7135 driver: double sided 10440 fun for Dual-PWM

Just to be clear, although you probably realize this: There is not a super-duper amount of copper connecting R1 and the big copper-covered area labeled “BLF”. That area is there primarily to allow a person to solder the strap in place - that’s it!

Thanks wight, maybe add the estimated height and possible shaving option to 11mm to the OP? these smaller builds is often as much limited by hight as by width so i think it would help getting an idea where i could fit.
I actually don’t understand what “strap” means in this context, are you referring to the bat +, one would have to place over the mcu?

Ok nice so it is possible to fit in 11mm that opens up some more possible hosts something like this could fit widthwise anyway, height is still a question thought.

The 10mm layout i was wondering about was a single sided 10-11mm with only a small DD FET, mcu and as small as necessary smd’s without the 7135 like your A17DD-SO8, for a tiny and slimmer option, if something like that could be fitted on one side with your new understanding on how tight you can place these tiny FET’s, if you only use the tiniest footprint needed like these 2 new drivers.
If some shaving of the attiny or fet is required that would be worth it, if thats would make it possible i think.

  • I will update the OP soon with the shaving option. There are a lot of variables affecting total height. (Mostly how much work the builder puts in IMO, plus PCB thickness.)
  • You are understanding correctly. Like a strap used to hold down a load on a truck or trailer. It (generally) must be affixed at both ends at stretch over top of something.
  • 10mm is very small, the standard footprint for the MCU is approximately 8.5mm diagonally. Someone (not me) needs to go out and determine what clearances / keepouts a driver like this would require. I do not think single sided & low profile is practical. Note that this FET is actually bigger than the one used on the BLF Tiny10! (And much higher performance.) It makes much more sense to eliminate the ATiny13A and move to a smaller (and more difficult to program) MCU.

One possible way of building a 10mm single-sided driver with an ATtiny13A and LFPAK33 MOSFET is to stand the MOSFET on end. You can easily trim 0.5mm of length from the MOSFET, so I’d expect the full standing height of the driver to be 4.4mm including the PCB. You’d need to move down to something small, probably 0402, and I think there would be an issue with C1 (it would probably need to remain 0603).

v005:

https://oshpark.com/projects/YUYEfb8i

  • Pulled divider via fully underneath the MCU. This way working on your strap will be easier & you’re less likely to cause a short while scraping & soldering the strap.
  • Fixed several small DRC issues (clearances) I missed in v003. Many clearances on this driver take full advantage of the design guarantees, it’s extremely tight.
  • Changed via pattern in center from 3/4 diamond to skewed parallelogram.
  • Moved R1 down slightly. (Gives more room for soldering the strap. Should still be compatible with an unmodded SOIC clip.)
  • Moved D1 down very slightly.
  • Changed via pattern for LED+ to a straight line of 4 vias.
  • Added copper pour under 7135 to better accommodate shaved 7135 units where the tab has been removed.
  • LED solder pads are now precisely across from each other.

excuse my noobness….

but I’ve been wondering this for awhile. What is the purpose of the vias everywhere?

A via is a plated hole. It electrically connects things on both sides of the PCB.

If you like, download and install gerbv (it’s a free program) and then go to OSH Park and download the zip of one of my boards. Extract the zip and load the files as layers into gerbv. This may help you see what’s going on.

so there are actually 2 layers of traces that need connecting at certain points through the pcb layer? That makes much more sense now.

These are double sided PCBs, so there are traces on both the top and the bottom. That’s why the OSH Park images show both sides. Single sided PCBs are only used in a small number of specialty situations these days. If you take a look at the PCB (s) you recently got from RMM you’ll see that it has traces on both sides.

Thanks for thinking this over and filling in my blanks of understanding in this.
Ok, so one possible way could be to stand the LFPAK33 FET up :slight_smile: if you wouldn’t have told me the difference between a linear fet driver & dd det driver just yesterday i would have thought that it would be overheated when standing up with less heatsinking.

some different ideas……i hope some of these could make it a little bit more possible.

What if one used the PowerPAK SC-70-6 2 x 2 or the 1,9mm x 3 mm PowerPAK-8 size FET + the attiny13a-ssu, would it be possible to squeeze everything together so i could fit then with small smd’s?

Or would it be possible to place the smaller 3x3, 2x2 FET under the attiny13a-ssu if you bend down the legs so it could fit?

Or is it possible to use one of the other packages that the attiny13a come in something like the ATTINY13A-MMU 10M1 MLF that one is 3x3 and should take much less space on the pcb without those long legs, programming it i guess is only possible before the reflow or maybe there is tiny bits of legs that is accessible after.

Can you explain to me what you mean by clearances / keepouts?

EDIT
i haven’t succeeded in quoting only parts of someones post, thats why it gets a little long.
EDIT2
not anymore thanks to Rufusbduck :slight_smile:

Select the text you want to quote then copy/paste using the quote button in the text editor.

Thanks Rufusbduck, now i finally got how to do it :slight_smile:

I got a new nice 10440 host today the Jexree mini 5$, a very nice little aaa light but it goes single mode when i use a 10440.

Every part is easily accessible, lens,led star,driver,even the bottom spring. Driver compartment is 6,6mm deep, driver is 12,6mm wide, led star sits on a shelf thats is 13,8mm wide.

So i wonder, is it possible to make this 12mm driver a little wider for maybe 13mm sanded down to 12,6mm for a nice i hope press fit, or the even better fit for this because the driver sits on a little shelf, the single sided 15mm DD FET 7135 driver to 13mm sanded to 12,6mm if you use smaller smd component and if needed stand the FET up.
Or only the MCU+LFPAK33 FET, would that make a single sided 13mm driver sanded down to 12,6mm possible?

I am sorry wight if i ask for to much, i really don’t know how much work anything like this would be to figure out, but i kind of got it in my head that i really want to build some really tiny and really powerful lights, but i don’t yet have access to or currently enough fabricating skills to rebuild a light for whats available.

That’s the same light I had in mind for this driver,but I haven’t received mine yet. How deep is the pill area in the flashlight? Deep enough to push this one up inside and use a contact board?

Good choice, i bought a bunch of nice little aaa lights in hope of finding my perfect host, and this one really hits the spot :slight_smile: it is small but surprisingly roomy where it needs to be. and it is as of yet the easiest and fastest flashlight to disassemble i have got so far.
And it headsinks really well for something this tiny, the whole flashlight gets warm quick when using a 10440 not just the head like some with worse heatsinking.

When you say pill area do you mean how deep is the part where the led star sits? that is 13,42mm deep without the bezel.
Or do you mean the part under that, because from the other side the one i call “driver compartment” is 6,6mm deep so because wight said this driver could be assembled in just under 5mm, we would have 1,6mm to spare for a possible contact board, but that would not leave much room for wires so it would be really fiddly to assemble if even possible.

Edit
This is wights answerer about the hight of the assembled driver
“Without your strap? 4.9mm w/ a mediocre build. 4.8mm w/ a good build (optimal amount of solder, 7135 and MCU both right against the PCB). Should be able to sand components down somewhat. Best case scenario I think you could reduce the top side by 0.5mm (limited by C1) and maybe the same on the bottom side (I don’t know how far you can sand the ATTiny13A). After all that you’ll have to add your strap’s thickness.”

So 1,7 + possible if one would sand down some components 0,5+0,5mm more so 2,7mm for an extra contact board and wires, still tight though but better :slight_smile:

Yeah I was talking about the driver area, thanks for the great info. The contact board I have in mind is from an ultrafire protection circuit, so it’s 18mm now, but very easy to sand down and very thin. The wires will be the tough part, I’ll probably have to use pvc coated because the silicone is much thicker.

Edit: it sounds like a lot of work though. If wight can expand the board enough for a press-fit, that would be a simpler solution

Yeah this 12mm board expanded to 13mm to be sanded to fit w would be a much better option, but when you get yours you will see why i asked if it was possible to make a single sided 13mm sandable to 12,6mm if he used smaller smd components and the
standing FET trick :slight_smile:
Because how this lights driver kind of almost need to rest on the tiny shelf in the driver area because it is a twisty, and the bottom of the driver connects its minus on a non anodized shelf on the top part of the battery parts shelf and the twisty action would probably be much sturdier if one could use an almost drop in mechanical configured driver, and i hope to be able to build these good enough to be able to gift to people and with this drivers possible low volt protection on the 10440 it could make it a giftable lithium aaa size light to less technical minded friends :santa: :slight_smile: with a short turbo timer of course :wink:

Thanks for the pvc wire tips, any tips on where to get some appropriate ones? and are you sure pvc can stand the crazy heat something this small and hot can produce, i thought it was silicon or teflon that one should use on this kind of builds.

What do you plan to use for led? i have a Noctigon XP16 with a XP-L V6 3D that i will sand down to 13,8mm for a press fit in, so i can get some extra heatsinking on the sides also.

I’m not planning that hot of a build, just around 1.25 amps on high with a 30 sec timer down to 0.35a. Moon, Low, and Medium using the 7135, and High using the FET. So I’ll use a neutral XPG2, and planning to just use an aluminum star. This will be an edc light, so practicality and run time on the 10440 are priorities. I’m not planning any mode memory, so hopefully 4-modes isn’t too annoying with a twisty

Interesting set up, yeah something like that would probably be well balanced for edc gift lights and lower cost also i will keep that in mind, for me i am still new at this so i want MAX lumens also :), i understand the fascination dims 8) a bit with time but for now it is still exiting.

So when you say high you mean turbo down to 0,350a as in medium low and moon and non turbo on the 7135? what lumens range can you get on the 7135 and a xp-g2? EDIT i checked djozz graph and the answer is further down.

Can you tell me how the selection of the 7135 is handled? is it automatic from the pwm levels or something else, and how much lumens could you get at max from only the 350mhA 7135? sounds 120 lumens from a xp-g2 and 160-170 lumens from a xp-l v6 about right? do you have any guesstimated runtimes one could hope for?

I thought a copper dtp star would improve the efficiency and extend the runtime also, but i have been wrong about some similar assumptions lately so maybe it is not.?

I could be wrong on the details. I’m still working it out myself, I’m basing verything on other driver threads right now. These drivers utilize the dual PWM outputs on the STAR firmware, so I believe you can tell the MCU what output and PWM level to use for each mode.
On the XPG2 I’m guesstimating 120 lumens on medium and 400 on high at the emitter.
.
I want a mode range of Moon (~1lumen) > Low (30 lumens) > Medium (120) > High (400)
.
My efest 10440 is 350mAh, so assuming 85% efficiency, on medium that’s about 50 minutes of runtime.
You are right that copper helps, but I thinking that at my current levels it won’t get hot enough to make much of a difference. I very well could be wrong though.

That mode range sound perfect for an edc in this size, i guess i will build at least two for me then, one crazy look at how much light your hand is throwing J)
and a normal practical one, the problem is the twisty, if i could find as good host as this but with a button, i can really see some more and higher modes be useful but as you said more than 4 probably gets annoying, and the thought of twisting something scorching hot doesn’t sound all that pleasant |( as an edc