Build Log: The 1.5V Project

[calipsoii]

Hi CPF! :wave:

The intent of this thread is to serve as an ongoing build log for a project I've been working on for a little while now. I always enjoy reading the build logs of others and I firmly believe we can never have too many people documenting their projects (so if you're working on something, post about it so I can follow along!). Hopefully this thread will serve as an interesting read or a cautionary tale, depending on how things go :devil:.

I should mention that this build log will be a little different format than I would normally do. I very much prefer to complete a project, document the whole thing with pictures, and then do a write-up afterwards. It makes for a nice clean post with lots of visuals. In this case though, the scope of the project is much bigger and I'm simply not done! So I'll be adding to it as I go along.

Enough of that though. Without further ado, let's get down to business!


The 1.5V Project

Not much a name, I know - it was the first thing that popped into my head when I sat down with this grand idea. It kind of stuck and now all my documentation is labeled that way, so I'm just going with it. The project started as a wild idea to create a 1xAA light. It's always been my favorite format and I've yet to find the perfect AA light (probably because my definition of perfect is constantly changing). Which is actually part of the reason why I set out to try and make my own; one can never be satisfied until they control all the variables.

The criteria at the outset of the project was pretty simple:

• Runs on 1xAA
• Doesn't need to be very bright, just usable
• Couple hour runtime
• MUST use a reflector (I really dislike optics)
• One fixed output level.... BUT the output level must be programmable through an interface and not by swapping resistors or using trimpots or opening the light
• Must must MUST have an emitter I like. I don't like the XM-L2 nor am I particularly fond of the Nichia 219. In today's marketplace a light using something other than those is a rare beast
• A clicky body for clipping to my jeans pocket and a twisty body for riding in the 5th watch pocket
• NO KNURLING ON THE BODY, just the ends. My lights spend most of their life squeezed between my lips and I hate it when the knurling shreds them all up (especially as we head into winter)

There ya go! There are no shortage of 1xAA lights on the market and I'm sure if I looked hard enough I could find something to tick all those boxes. That'd make for a pretty boring build log though, so let's pretend this light doesn't exist and we'll make one instead.

[----- I'm going to update this post as the work progresses. I will add a link to each section below for ease of navigation -----]
​

	Prototyping the Circuit
	Building the Test Platform
	Pogo Pin Programming
	Solder Stencils (and high blood pressure)
	Writing the Firmware
	Building a Better Socket
	Rolling the Die(s)
	Machining the Driver Pill
	Machining the Head & Body
	A Clip Named Misty
	Punchin' The Numbers
	The Sum of its Parts
	Beamshots and Afterthoughts
	Runtime Graphs
	Mini Update: Needs More Threads! Mini Update: Cutting Ahead Mini Update: To Trim a Reflector
	I See You've Constructed A New Light...
	Mini Update: Thoughts on v1.2 Mini Update: Punching a Square Hole
	Quality vs. Quantity
	A Host Does Not a Light Make
	LED's and 003's
	Practice Makes... Diamonds?
	Instruction Manual (or: How Do I Work This Thing?!)
	A Dazzling Display of Color
	Instructional Videos
	SOIC Programming
	PCB's Fit for Royalty
	Back to the (Bread)Board
	Now I Know My ADC's
	Machining the Gen. 2 Driver Pill
	Machining the Gen. 2 Head
	Machining the Gen. 2 Tailcap
	Machining the Gen. 2 Body
	Putting It All Together
	15VP v2 Measurements
	3rd 4th Time's the Charm

​

 

[calipsoii]

Prototyping the Circuit

With a general idea of what I wanted the circuit to do, I set to work reading datasheets and ordering parts. When they arrived a few days later (DigiKey shipping to Canada is amazing!) I was excited to hook them all up. Previous projects have taught me that a little planning can save a lot of headaches though, so I started with that.

The squares in an engineering draftbook are excellent representations of the solder pads on a chunk of protoboard:

The wires connecting the microcontroller in the middle are TINY enamel speaker wires. They're about as wide as one of the lines in your thumbprint. So if the soldering looks sloppy, you'll forgive me

The million dollar question of course is: does it work? Well let's grab a couple batteries (one new L91 and one very dead alkaline)...

... and hook them up and see what voltage we get out of it!

Looks good! It's very gratifying to find out none of your enamel wires are shorting against each other. Added a green LED (Vf = 2.3V) to indicate whether we're cooking with gas when a cell is connected

In my excitement I couldn't resist hooking up a power LED to see if it worked (probably not a smart idea without some current control but everything survived)

It's easy to lose sense of scale when looking through a bunch of macro shots, so here's one to give you an idea how bloody small the step-up chip is

With the boost converter prototyped, it was time to look at the aforementioned

• current control
• programmability

I started with the programmability bit since this is more up my alley. This simple protoboard just has an Atmel programming header and an input and output header. The only way I could get the microcontroller to route its wires nicely was upside down, so it got soldered dead-bug style.

The uC board was hooked up to my AVR programmer to confirm everything worked (it did - discrete components with large legs are so lovely to solder) and it was time to move on to the current control board. As you can see, the boards are quickly multiplying... :sweat:

The prototyping ended shortly after that when I realized that my enamel wires simply would not fit into the 0.4mm pitch of the LED driver. No way, no how was that getting soldered. The end result looked a little bit like the board had been through an autopsy. :sick2:

No matter, we've come to far to quit now!

>> Return to root post
>> Continue to next post in the series

 

[calipsoii]

Building the Test Platform

When last we left off, I was discovering how amazingly, impossibly small 0.4mm pitch is on a DFN chip. With a box full of Digikey parts mocking me though, I had to know whether this circuit was going to work so it was time to head into Eagle and make a couple real PCB's.

(I should mention that I asked an electronics engineering grad whether my circuit diagram was going to work. He wrinkled his nose, called the design 'perverted' and said he didn't know. Ah well, only one way to find out.)​

The circuit design in Eagle took 5 weeks. I was 3-for-3 on ordering PCB's and discovering I'd overlooked stupid errors in my design so I spent extra time on this one. When the boards arrived there were no obvious defects.

Well, almost no defects. I failed to take into account the amount that the router bit would remove as it made the boards circular and it trimmed an excessive amount of my heatsink/ground pad. This ended up being a serious pain in the neck later on, so I figured I'd mention it now.

Soldering on professional PCB's is a breath of fresh air after working with enamel wire and protoboard. The 2 boards (boost + driver) turned out very nicely and all the little 0.4mm legs reflowed wonderfully in my $55 Black & Decker toaster oven.

I've found over the years that any time/effort you expend up front on something usually pays dividends later on. With that in mind, I wanted to make the test bed functional, accessible and easy on the eyes. This was the final design; it's about half complete in this image. All the major components (electronic guts, LED, switch, battery) are swappable and easily probed with a multimeter. The wood block is a stained chunk of oak.

To finish it off, a warm-white Osram GDP gets strapped to the aluminum heatsink and a piece of clear plexiglass pins the driver boards down.

I whipped up some very alpha firmware and loaded it via the programming header (6 pins in the foreground). The software just displays that the primary/secondary LED's are working and that the LED driver is setting levels properly. I initially set the output to max and the light started flashing rapidly - it took me a minute to realize this was the thermal protection of the LED driver kicking in! The PCB in the foreground seared a mark into my finger when I touched it.


With the test bed complete it was time to think about how the firmware was going to work!

>> Return to root post
>> Continue to next post in the series

 

[STiFTW]

Very cool, saving this thread to watch the build!

 

[blah9]

Yes, I am doing the same. Thank you for posting this as well as future updates! I'm very interested.

 

[AnAppleSnail]

Amazing to read your methodical progress!

 

[Steve K]

Looks like fun!

Having looked around for boost converters that can run off of one AA cell, I have ask.. what boost converter are you using??

 

[calipsoii]

Having looked around for boost converters that can run off of one AA cell, I have ask.. what boost converter are you using??

Linear's LTC3528. Only comes in DFN package so it's a little fiddly to solder but the chip seems to work nicely. Only thing I'm worried about is that its efficiency at high drive currents doesn't seem amazing. Need to test it more before I pass judgement though.

 

[Steve K]

hmmm.... I didn't have that part in my list. The datasheet does hint that 200mA is typical for output current with a single cell input. Since it's a voltage regulated boost converter, it's harder to predict what the output current will be when driving a LED. Are you using the voltage limit as a way to limit LED current (i.e. setting the voltage fairly low, such as 2.9v), or are you relying on the mosfet current limit as a way of limiting output current? Or maybe just using a single cell battery is sufficient... it's not easy to get a lot of current to flow through an inductor without much voltage. The inductor resistance might be enough to limit current.

and while I'm asking questions... I can't help but notice that you have two boards, and both have inductors on them. One board is the boost converter circuit, and the other is for the Atmel uC. Is the Atmel running as a buck converter? That inductor probably isn't on the board just for looks, so it appears that you are running two converters in series. That's not going to do much for efficiency. Or is the boost converter just to provide power for the Atmel, and it's running as a boost converter to drive the LED??

If you could post a schematic, that would answer a lot of questions!

oh, for what it's worth, I found a nifty boost converter designed to run a high power LED from a single AA cell. It's the PAM2805 from Diodes Inc. It appears to be designed specifically for flashlight use. The only problem is that it doesn't appear to be available in small quantities right now.

 

[calipsoii]

Hah, no way I'm posting a schematic Steve K. Then you'd know exactly how little electronics education I actually have. Sort of like finding code I wrote 10 years ago - it did the job for 10 straight years, but man, it sure ain't pretty. I have a little mucking around to do but I have a feeling I'm going to lose the 2nd inductor. It was an experiment that hasn't worked all that well so far. Assuming I can coax the thing into doing what I want I'll post more details later!

Right now my big problem is heat - wood is an excellent insulator! A nice-looking choice but probably not the ideal one when ratcheting up the current flow.

 

[darkknightlight]

Very cool build idea! I can't wait for the next update.

 

[gunga]

While I don't have the same design criteria as you, I think this is fine work. Very cool to design and build your own light! Kudos, a fellow Canadian too eh?!

 

[nickelflipper]

Right now my big problem is heat - wood is an excellent insulator! A nice-looking choice but probably not the ideal one when ratcheting up the current flow.

Subscribed! I love seeing real world driver build ups.

I don't see any vias for the LTC 3528 power pad? so, no associated bottom side copper? Linear fails to mention how much copper area is required to meet the data sheet results. Looks like a 17mm board, so have you tried it in a P60 (or ?) pill for extra heat sinking?

 

[calipsoii]

Sorry for the lack of replies guys, work's been pretty busy this week. Gotta finance this hobby somehow!

While I don't have the same design criteria as you, I think this is fine work. Very cool to design and build your own light! Kudos, a fellow Canadian too eh?!

Yessir, from smack-dab in the middle of Alberta! Well, even if the design doesn't interest you, I hope I can entertain you in the process.

Looks like a 17mm board, so have you tried it in a P60 (or ?) pill for extra heat sinking?

Oh man, if it was 17mm diameter I'd be able to have a huge grounding/heatsink ring around the driver board. I wish! Everything barely fit at the size it is now.

You're absolutely correct about the lack of heatsinking nickelflipper. Aside from the bit of copper on the board, there is no place for it to go right now. I hope to rectify that this weekend, assuming all goes according to plan. Wish me luck!

 

[Mike S]

Wow, nice work. I can't believe you successfully hand soldered that DFN package.

Have you tried prototyping with homemade PCB's? I've found that it works well enough that you typically don't have to order professional boards.

 

[calipsoii]

Spent a few hours today turning this:

Into this:

A little rough around the edges, but pretty damn good for a first attempt at a prototype I think! There are pictures of the machining process, of course, but I do need to save something for my "Building the Pill" chapter. :devil:

I think that this should help immensely with dissipating the heat from my test boards so I can finish writing the firmware. They fit snug as a glove. There is a threaded slot on the side for a tiny hex head set screw but nobody in town carries them so I'm waiting for an internet order. It'll tighten the pill up so things don't bounce around.

Going to solder components on those 2 boards tomorrow and see if that helps with the heat!

Have you tried prototyping with homemade PCB's? I've found that it works well enough that you typically don't have to order professional boards.

Hi Mike S!

I have actually done a little etching at home:

There's no way on Earth I could etch a 0.4mm pitch DFN footprint though. I basically consider having the PCB's professionally done the cost of my sanity.

 

[John_Galt]

Very interesting build!

Looking forward to some updates but I have a few questions about the overall circuit design.

-you built this to use a 1.5v powersource. What changes would be necessary to have the same functionality from a 3v source? I have several older lights that ive been looking to upgrade (arc ls, longbow micra, etc)
-what is the maximum and minimum drive levels currently? And what sort of efficiency are you looking at at each?
-im sure this will be detailed, but what sort of ui are you looking at?

 

[John_Galt]

Very interesting build!

Looking forward to some updates but I have a few questions about the overall circuit design.

-you built this to use a 1.5v powersource. What changes would be necessary to have the same functionality from a 3v source? I have several older lights that ive been looking to upgrade (arc ls, longbow micra, etc)
-what is the maximum and minimum drive levels currently? And what sort of efficiency are you looking at at each?
-im sure this will be detailed, but what sort of ui are you looking at?

 

[Hallis]

amazing work. Can't wait to see the pill build. Might want one of these little things for myself as i've got a 1aa mod that I used to have a LuxIII/14500 setup in.

 

[TexasLumens]

Danged nice work!! I like the thought put into the project.... even the test platform is almost surgical!!! Nice!! Anxious to see more. :goodjob: Dan.