pretty much what Georges80 told me - the reverse polarity is provided by a FET and the total overhead (he didn’t break it down) was ~1.1V, so my plans won’t really work and losing ~2W to heat losses in a 12W light by using a linear driver is a bit galling.
Now, using a higher power higher voltage version of this buck-boost driver would be just the ticket 
Understood, I was just thinking with all the tiny13A’s floating around, people could just float one off a spare or duff driver? Plus you shouldn’t have to calibrate the MCP9700, although it is no sweetheart on accuracy either. Routing PB5 or PB1 from the programming pad over to the thermistor under the inductor would be easy. The pads would be there, and it would be up to the individuals discretion on whether to populate it or not. Cost for the thermistor, piece of mind?, is $0.25-0.40.
For future Buck Boost board revision.
Moved the PWM pin to PB1. Reasoning was to take advantage of what appears to be a much more flexible Timer1 OC1A period register of the ATtiny25,45,85. PB1 for the ATtiny13A still lands on OC0B, so should be OK.
Placed the MCP9700 thermistor (sot23 pkg.) on bottom copper. The trace from PB5 or RST line is used for the a-d conversion, any problem with that? AFAIK PIC’s don’t give that option on the MCLR line, just digital input only.
The drawback of using PB5 for the A/D conversion is that will disable the ISP. Everytime you want to re-program the chip, you must enable the reset by high voltage fuse re-store. Simple programmer does not have the high voltage fuse re-store function.
Well thanks for that bit of info Microa, I did not know that. I’m expecting everyone to keep the AVR (and GCC) newbie in line :). If not a-d, then you can use RST pin for a digital input (like for an electronic switch) without a problem, correct?
Will have to do some more pin swapping. OK, PB0/MOSI is on the job now for a-d/ and the thermistor option.
I think that using the RESET pin as an input may also mess up ISP programming. ISP programming uses RESET to get the chip into programming mode.
Thanks for the heads up. O.K. then, took a read of the data sheet. It looks like to gain any of the alternate pin functions of PB5 you have to program the RSTDISBL fuse like you say, and HV programming to get it back. Decidedly unfriendly behavior 
. +1 for the PICS….hehe. That leaves RB0 as the only flexible or optional pin.
A new Sourceforge project, FlopDriver has been created to support this project. stupid name, I know , FlopDriver, or Flashlight open project Driver. Just starting to eek out the details. No software or anything else other than the Beta gerber files. People should wait for an update before actually manufacturing the boards.
The online gerber viewer is pretty good at viewing gerber files if people have an interest in that sort of thing.
Bummer, no solder mask (other than a few wisps here or there) on the Beta boards. The silkscreen came thru though :), for an unusual effect. This makes the build especially challenging.
The error might be due to calling out surface mount GND annular rings, and fitting the components inside?, messed up gerber plots?,I don’t know. Will definitely go with import DXF copper pours next time with a copy to solder mask. Going to have to try some Sharpie ultra fine point action, in lieu of the soldermask, and see how it goes.
The board came out 5-6 mil oversize, so a little extra file action to fit the pill. Next time I need to use less pixels for silkscreen text.
Mattthemuppet and relic38, sent out boards today. Included resistors and caps to build one board. Parts list were enclosed to complete the build if you want to try your luck.
OOPS!
Soldermask looked OK on the OSH Park render screen.
Send me the gerbers and I’ll tell you if it was a mistake in the gerbers or a mistake on OSHPARK’s side.
support@brightlightmodules.com
PPtk
PilotPTK,
Gerbers are ready for download at FlopDriver, the sourceforge project. Link here:
http://sourceforge.net/projects/flopdriver/
Gerbers are fine. OSHPARK Screwed The Pooch on this one. Send them an E-Mail - They’ll re-make these for you for free.
PPtk
PilotPTK, thanks for confirming, thought I was going a little crazy.
This is second in a row for flubs from OSH Park. Last time was for missed board outline holes, which turned out to be human interpretation of what I intended. Got a full refund on that one also. I hope Laen doesn’t think I’m a broken record.
I appreciate the extra eyes on the build. I can send a couple of extra boards, and 1X resistor/cap kit if you want, just PM me.
bit of a bugger, but the problem is with the labels, right? (I think, I get a bit lost with the technical talk). If that’s the case, I can just refer to a picture when I’m putting stuff in place. Thanks for the board and bits though, I’ll need to get a list together of the other parts and where to get them from. I also have a USBasp and SOIC8 clip to order too. How is the Atiny (13? 85?) used - to turn the inductor on and off by PWM to produce different LED drive currents? Will something like the BLF-VLD code be applicable to this driver?
I’m getting very interested in the whole programming side, even if I know diddly squat about it right now, so this would be an awesome if steep learning curve. I even have an application for it too, in a Nichia 219 headlamp that’s next on my build list.
No, The problem is the solder mask. The purple stuff. It’s missing.
oh, right. I’m guessing that’ll make things a touch difficult then?
mattthemuppet,
Not having the solder mask, makes it a bear to solder paste and hot plate, as it brings the possibility of bridges between tracks, pads, etc.
The ATtiny does the PWM for low current modes, checks the battery for low voltage, and toggles a couple of pins on the LTC3454 for different current settings (say 25-75-100%). The LTC3454 also handles driving the inductor, diode, and LED.
Heh, I need a kick in the pants, as I’ve been procrastinating on the programming part.
Thanks for the board, not sure if I can reflow the main two bits without the mask. I’ll give it a try. The other parts I’ll hand solder so that should be OK. Just got to watch the solder getting across traces.
I have a programmer from Fasttech, not sure it’s the right one. I’ll figure that out when I get there.
If you get the properly masked boards, feel free to send a couple my way again.
ah, I understand now. As I said, steep learning curve Nothing like making something hard even more challenging! You mentioned using a sharpie - will that prevent solder flowing where it shouldn’t? With the circuit diagram I should at least be able to check for bridges by continuity.
Still reading up on programming. Something like an externally programmable set up like lupodrv would be awesome, but I’ve no idea how to do that yet.
Well it’s hard to tell whether the sharpie helped as a resist, maybe marginally, it kind of made a mess. On the small LTC3454 DFN package I used my sharp pointy curved tweezers to pick out the solder paste between all the pads under a magnifying loupe, same for the other components.
I am conflicted about clicky programming. I appreciate a simple no mode memory interface myself.