Sot-89 buck IC, possible minimag buck driver

Plenty of room at D1/Z1 for an LDO instead. Given the high input voltage for the pt4115 it could probably provide that 1A to an XHP 35.

You should try downloading diptrace, it is really not that hard to figure out if you watch the videos on youtube.

The more I think about this, the more I think it could run an external FET and as such drive as much current as we want. With a 17mm driver it would open a lot of possibility. Not sure it would give any better performance vs the mtnmax we have now but should hopefully fit a soic MCU allowing for bistro or narsil to be used.

I’d love to see this realized. I have a Minimag I’d love to put one of these in. Maybe I’ll fiddle with Eagle a bit myself and see if I can get it to work into a pcb. Not that I could build a two-sided driver. :weary:

That’s a good idea well worth exploring for a larger board. Maybe it could eventually be made smaller but higher current means a larger inductor, higher wattage sense resistor(s), more capacitance, and a higher rated shottky along with the external MOSFET and any other parts needed for it. Opening the hole in the tube would allow enough depth for a two board design but 14.5mm is a hard limit on the diameter and not modding the tube puts a hard limit on the stack height a well. I don’t have anything against designs for other lights or purposes but I’m going to stick to these constraints for this one.

Here’s a link to the instruct able I saw on adding the MOSFET.

Building two sided drivers isn’t much harder. The mmu is the tricky part, I’d like to see if an untented via can be placed avoiding Cout to allow hand soldering of the hidden center pad.

Interesting link, I wounder why they use the external mosfet driver, seems like the PT4115 would be able to drive the mosfet directly seeing as it can handle up to a 1.5A load (far more then an FET needs). It is far simpler and more compact then any buck IC’s I have seen but if you have to use an external mosfet driver it looses that benefit.

Just get yourself a reflow (or even hot air) gun and reflow the driver properly. It is 10X easier and gives a far better result.

The hardest part with that center pad is not putting too much solder paste on it and it causing the MCU to lift up and not make full contact with the other pins.

Hello Rufusbduck and Texas Ace,

Thanks for making the effort in the very nice cut-out diagram (pretty nifty work there! :) ), and the study for the buck driver. The PT4115 is quite nice but it is actually fairly large and only works as a buck, i.e. it wouldn't work if we decide to say use 2 NiMHs. So I did a little bit of thinking over lunch today and did a very quick schematic and layout study… pretty busy this weekend (will be out of town) but once I get back I think this will make for a fun project! So here’s what I came up with in about half hour or so..

Specs:

  • 4-Switch (Internal) Buck and Boost topology. This allows:
    • Input from two Alkaline/NiMH, 1 lithium or 2 Lithium Cells (~2V to 9V input)
    • Constant Current output 10W max (e.g. 2A at 3V+ or ~1.5A at 6V+)
      i.e you can use 3V or 6V LEDs
  • Switching Frequency 1.2 to 2.4MHz
  • 256 Levels of true constant current output, no PWM! Might be able to do 1024 if i can find a suitable part!
  • ATtiny84A MCU with Off-Memory and OTC
  • Battery cut-off and Temperature monitoring
  • 6-pin ultra-small Molex header for easy programming
  • 5mm Diameter, 5mm dia. Spring
  • Finally, hopefully some nice to have solder jumpers for easy configuration without need for re-programming
    • I have 3 GPIO left on the MCU so... 8 possible different mode configs!

The downside to this is that it may be a little difficult to hand solder. However it will pack in a bunch of functionality into a 14.5mm board. (By the way, why 14.5mm? Why not 15mm? Does 15mm not fit a Maglite?).

Above is a quick render of the placement study I did. As you can see everything generally fits (pretty tight though!) – I need to do a little wiggle around and careful placement but I ran out of time. Also note that the biggest thing here is the inductor but this as a 12A inductor – I just need something closer to 3A so it will be a LOT smaller IRL! Likewise the resistors and other passives will need to be optimized for size after more study in the power circuit.

Due to the small space ideally I’d route this as a 4 layer board for better performance and EMI control, but I’m pretty sure I can get it to work just fine on 2 layers. Finally - thanks much on the stack height - the inductor and caps are the biggest things here but again they will be resized smaller to fit the power requirements :). Also.. do you know how big the ground ring needs to be around the edge of the board (on both sides?). Above shows a 1mm ring but it would be nice if they could be narrower.

More to come soon!

Wow, that’s super! 15mm won’t quite fit, 14.3 slides in easily and 14.5 barely fits. Even at that the driver would have sit on the edge of the pill rather than inside a shelf and be soldered to that side(inductor side not spring side) at two or three points so the parts might need to fit within a 13-13.5 mm circle. I just picked the pt4115 to get the party started. It’s smaller than the ax20002 and bigger than qx9920 but main reason was it seemed perfectly compatible with mcu’s and firmware already on hand. If I can’t hand solder it then as long as Rich can and is interested in stocking a built version then that’s ok too. He programs the mmu before placing it on a driver so the header isn’t as much a priority as minimizing the total footprint on the inductor size(smaller means thicker pill wall). If not then it might be a problem. Did my paper driver get any laughs? Eagle is a lot of trail and error for me but this let me see how things might fit prior to that bit of laptop pain. I was willing to settle for 1A given the modest amount of heat sinking available but I guess whatever can fit is the best size motor. I’ve done some preliminary pill work and should be able to have a mock up soon. At that point everyone can describe it’s shortcomings in detail. Hope fill there will be some viable suggestions for improvement as well as how a machined version would have to differ.

For starters try for a 1mm ground ring on the top, .75mm as an alternative, not sure if .5mm would work though I suppose a notch could be made in the pill for the inductor to hang right to the board edge. I will keep trying to come up with the PT based design simply to provide one that is more readily a DIY Oshpark project. I guarantee that if yours can be made I’d buy it first though.

Orsm work RBD and loneoceans. Love it. Good luck not that you will need it. :+1:

Very nice work!

What IC are you using for the buck/boost?

I had never even thought of using the tiny84 with the DFN footprint, I love that! Does it use Tiny85 compatible firmware? It is so nice that it has extra pins as well.

The only downside is the lack of being able to easily re-program it. If we could come up with some firmware that would work with both clicky and e-switch this is the way to go in the future bar none.

With the latest OTSM this should be very possible as well and free a lot of space on the PCB’s.

One thing to keep in mind you will at the very least want to add in a resistor before the MCU power input to help reduce voltage spikes. Also, C2 being that close to the spring would worry me if the render is accurate on height. When a battery is inserted the spring could bend and short to the cap.

Overall I am super impressed and really liking the direction this is going.

It’s likely there is only room for a 5 x 2 mm brass button or one of the short 5mm springs anyway.

Subscribing! For me, Mini Mags carry a lot of nostalgic value. I’d love to have one with internals such as this.

There are so many of them, if only a tiny fraction use this it would still be a large number. The niteize drop ins are pretty lame relative to even .5A. Getting up to 1-2A with all our favorite mode options would rock. Operating from alky’s to liion’s would make for a HUGE variability in cell options.

I have a few of the 10W mr11 drivers on the way. When they arrive I’ll bypass the diode bridge and see if it’s as simple as hooking up a linear driver pwm signal to the dim pin to control output. I’m excited about what loneoceans is doing as it obviously represents the ideal outcome designed by someone who knows what they’re doing. Being a hack I can’t help trying to pursue a different approach even if it has more limited success. Maybe both will be of use.

I quoted this info on the copper pill I have merely as a reference. I’m pretty sure OD, ID, can be increased slightly(.1 -.2 mm) and inside depth can be increased up to another 1mm. The final form of the top portion above the end of the tube is undertermined as a machinist would need to chime in on practical realities and help to come up with a workable design. A pill made from thin wall brass tubing (K&S #8141) has an OD of 14.29mm and an ID of 13.58mm but I doub the reality of machining something that thin. Keeping the components within a 13mm(.75 mm ground ring) circle and total stack thickness of 7mm should allow 1mm thick ceiling and more reasonable wall thickness of maybe .65mm(same as original pill). The simple expedient of opening the tube to a continuous ID (14.5mm) and sanding the ano would increase available pill depth by probably 2-3mm but crosses the “no modding of the host” line. If at all possible I’d prefer not to go there but it all depends on what can be done with the driver.

A two cell boost/buck driver with the capabilities outlined by loneoceans could find uses in lights that already use a 15mm driver. in th next few days I’ll try to work on eagle but my laptop is 13 years old not very reliable.

tiny84 is not compatible with tiny85. The tiny84 has more pins because it has more, uhh, features! :smiley: IIRC, MikeC is familiar with tiny84. Maybe you could talk with him about it. But, if you just compare the datasheets, a lot of the differences will be easily apparent.

The tinyAVR series are all based on the same architecture and are all 8-bit Atmel RISC, so they're all compatible :). However what DavidEF probably meant was - no you can't just directly re-program the ATtiny84A with the same .hex - it will not work. That said, I glanced through some of the popular firmware (e.g. Narsil, Bistro etc), and it seems like everything that they do on the tiny85, will work just fine on the tiny84A. And I suspect it will be just... doing some pin definition changes, and at most making sure the timer prescalers and compares are set right.

Best of all though is that the tiny84A not only comes in a smaller package, but is also a little more powerful and have more GPIO so overall it's more flexible. Personally I'd have preferring just going with a small STM32 or PSoC42x series SoCs, but they do seem a little overkill for this application, and are a little less ubiquitous to use than the Arduino-supported AT-series, so that's why I chose the tiny84A. :)

Very nice, I will have to look at the tiny84. With the extra pins it would open up a LOT of options with the only downside being not able to reflash it easily. Although if the firmware was good enough then you would never have to reflash it anyways.

Did you list what boost/buck IC you used in your design?