Experiment time: the "holy buck" adjustable driver

Flashlight Modding and DIY Parts
1

I’ve read about people running several amps into XM-L’s with success. Contemplating something similar, I decided to order a couple of ryansoh3’s Convoy C8 pills, and I have a couple of XM-L2’s on copper MCPCB’s on the way.

So now I need a driver. I’ve got a couple of options…

- Direct drive. I’m not a fan of this, I’d rather have regulated current in the LED.

- Nanjg 105C with stacked AMCs. Unfortunately the ryansoh3 pill doesn’t give any room to stack AMCs inside (not that I’m complaining!) I can only stack 1 extra layer of AMCs on the battery side. Assuming I put twelve 380mA AMCs on a driver, I end up with 4.56A. Which is “sufficient” but I’m a bit more nuts than that.

  • Use two 18340 cells instead of an 18650, and a buck driver. Doesn’t look like anyone makes one in a 17mm form factor, or without a big toroid.

So I decided to bang out my own petite driver. Behold, the “holy buck” 17mm driver.

Schematic:

http://imgur.com/MHKURFm

Switching regulator is a Linear Tech LTC1773 synchronous buck controller. I’m using it in “non-datasheet” mode - I’m putting ~650mV into the VFB pin of the device, low enough so the controller thinks the voltage isn’t high enough, and high enough to keep it out of frequency foldback. The chip will source a current out of the ITH (compensation) pin - putting a pot on there lets you command the ITH pin voltage, which sets the current threshold of the controller. With the component values chosen, you get a range of 0 to 10A. This is tested in LTSpice and appears to work.

Input voltage is meant to be two lithium cells - it can’t go beyond 10V without blowing stuff up. One cell is possible by shorting the 3.32K resistor and changing the FETs to something with a lower Vgs threshold, but I don’t plan on doing this myself. Output voltage isn’t regulated, and should work with 3 volt LEDs (XML/XML2/SST90/etc) or 6V LEDs (MT-G2).

I’m using a cheap power-on-reset IC for low battery protection, with the threshold set at ~6V. When the battery voltage starts to dip, the light will start cutting out - the POR chip has a ~1/4 second delay.

Here’s the PCB layout:

Top (LED facing) side of the PCB:
http://imgur.com/XdihOzI

Bottom (battery facing) side of the PCB:
http://imgur.com/BJOWQgb

The inductor (2.2uH IHLP3232, 4mm high) is centered on the bottom - plan is to add a strap of adhesive copper tape from the top surface of the inductor (and the + terminal of the battery) to the pad on the right. To the left of the inductor is the exposed current setting potentiometer - this allows the current value to be adjusted without removing the driver, changing components, etc.

Anyway, parts are ordered and the PCB design will be cleaned up and finished after I get my TVBG stuff shipped. I’ll keep y’all posted as stuff happens.

2

Sounds like a wonderful driver, I can not commend on the electronics, but a 0-10A adjustable 17mm driver suitable for any led including the MT-G2 is a dream come true :-)

3

Awesome! This type of driver in a 17mm form has been long-sought by at least a few members here. Synchronous topology is the way to go when small dropout voltage is desired, which is especially desireable when driving “6V” emitters with 2xLi-Ion.

I’m guessing a separate controller for modes would take up too much space?

4

Or a host that had a control ring or something that could set the current.

5

A 2x18350 C8 with VaraPower-eque external pot would be a ‘next-level’ mod :crown:

6

I went with synchronous rectification because assuming 50% duty cycle (say 2 cells feeding an XM-L), 8 amps output and a 0.5V schottky drop, there’s 2W to get rid of. Which is hard in the limited space of a driver.

I originally fit a Tiny13A on there in a 10-DFN, but there was no room to break out programming pads for it, and preprogramming would be tough.

You can hook up remote power/brightness control to this design if you want - remove the pot and connect a remote one (33K is the ideal value) for the brightness control, and put a switch in series with the 3320 ohm resistor for on/off.

Anyway, I have no idea how well this circuit is going to work - I’m pushing a fair bit of current through a small design, and I might run into thermal issues yet. I’ve ordered the bits to build just one, and if it works out, I might consider making them to sell.

7

Update: PCB and parts are ordered. Also ordered a bunch of TO-220 diodes which I’m gonna series up and bolt on a heatsink to make a fake load for testing.

I made a couple minor changes to the layout, and added a 0.1uF cap to the ADM803’s VCC pin, but the design is pretty much exactly as posted. I chose FDMC6679 for the P-channel and FDMC7572 for the N channel FET.

I went with Seeed Studio for the PCB, and they’re doing a DHL shipping special right now so I should hopefully see the PCB in a couple of weeks.

8

Nice driver, if there’s no thermal issue’s, I’ll be in for a couple.

9

[quote=gmarsh]

Part time/wannabe circuit designer and PCB layout enthusiast here.

How would you incorporate an MCU into this circuit? Can it be dimmed via PWM? I had a quick look at the datasheet but nothing sticks out to me. That's not saying much though! Basically I'm just wondering if a a PWM 'in' pad could be incorporated for external PWM control perhaps? I have a control board I designed for exactly this sort of reason that I use in a lot of lights and drop-ins I make. I also have a 2-cell, 4A MTG2 driver in the works but 10>4 ;)

- Matt

EDIT: Looks like you'd have to put a variable voltage source on the ITH pin now that I've read your first post properly. So a PWM input is not the way. You'd need something like ADC out to feed a constant voltage between 0 and 1.2V. I think. I don't know what i'm talking about now lol.

10

Could this driver be modified to run 1 XM-L by 4AA batteries (NiMH)?

11

I’ve heard these people do a good job, and for a lot less than even Seeed… and their stainless steel stencils are dirt cheap.

12

Feed the PWM signal to an R-C filter and voila… analog voltage.

13

[quote=Mattaus]

With the pot on the Ith pin and the Run/SS pin directly driven, the LTC actually starts up/shuts down pretty much instantly - fast enough so you can PWM the enable pin for brightness control. Here’s it running at 1KHz, 25% duty cycle:

Using this method you won’t be able to do ‘moonlight mode’, and you’ll probably hear the inductor chirping/buzzing at 1KHz, but it’ll work.

But the main problem is I’ve got no space. The smallest microcontroller I can find with both ADC and EEPROM is an ATTiny13A in a 10-DFN. I actually had this laid out on the board in place of the ADM803, with no extra parts required - problem is I have to preprogram it before I install it on the board, and I don’t think anyone makes a 10-DFN ZIF socket that would facilitate that.

14

Nice find. Looks like they use the same factory as Seeed/Itead.

15

There are lots of QFN test sockets on Ebay, but no DFN sockets. DFN sockets are made, but can be a bit pricey.

One possibility is to try z-axis tape and a DFN break-out board. It is double sided sticky tape that conducts electricity in one direction. Sparkfun sells it. It may or may not work for DFN lead pitch. There may be finer pitch z-axis tape available from other sources.

16

I quoted a DFN-10 socket from Loranger, almost $300. That Z-axis tape looks awesome, people are using it for 0.65mm BGAs so I figure a 0.5mm DFN should work fine. I could conceivably make an alignment/pressure jig that sits on top of a PCB, and make it work.

I’ll probably just stay the course on the 1 mode design though. This driver’s not really intended for full-featured, reliable, everyday-carry flashlights - it’s intended to provide a ridiculous amount of current in a small package.

17

So what you're saying is we can leave the pot on (and use it to set the maximum current) and then drive the run pin with PWM for brightness control?

You know this is BLF, right? People want a ridiculous amount of current in a small package so they can have a ridiculous amount of light in a small package. Sanity has no say here lol!

18

I wonder why DFN sockets cost so much more than QFN sockets… I’ve seen those in the $20 range.

19

I have an SST-90 light (3s 18650) that PWMs it's buck driver at 20kHz, and I can get nice moonlight levels. I don't know what buck-controller it uses (not synchronous though). Buck drivers often start pretty quickly, within a couple of µs.

20

If this makes it to production, I’ll take one. I don’t care for modes either, something like this is for a “WOW light” not edc. This combined with tofty’s switch means I just need to find a decent spring and I’m set to build. :heart_eyes: