Suggestions for a custom LED vehicle flashlight using a mostly/all plastic body

Hey everybody, I'm new to CPF and was hoping to get some advice from anyone willing to share their knowledge.

I'm looking to create a custom flashlight to replace one that was included by the factory in a mid-80's sports car. The original flashlight was powered by a pathetic 0.5w incandescent bulb and two AA batteries. These factory flashlights slid into a slot in an interior console and are hard to find these days as many of them have been lost/broken over the years. I bought my car used and the flashlight was missing. I am looking to create a new more modern replacement for the original lights using LED's and modern components. There is not a huge market for these but I have had some others express interest in a replacement and I am considering eventually offering the parts as a kit to them in the future.

My goal is to create a much brighter light with much better battery life than the practically useless original light. I also want to include a multi-mode driver to give the normal brightness and flash pattern choices so the light might also function as a warning light, etc. I have a few questions/concerns due to the unique size/shape of the light.

• The slot the light fits into is a rectangle about 17mm x 55mm. This limits the size of many of the internal components I can use. There is just enough space to use 2 AA batteries or 4-5 AAA batteries. 9v batteries are too large to fit into the slot. CR123s or 18650s are not even close to fitting inside the space.
• The factory light is an all plastic design. I believe it was ABS plastic or something similar. It was not weather tight in the slightest and did not even have a lens covering the bulb.

My biggest concern is heat management. I plan on making my replacement by 3D printing an ABS housing similar to the original. I know I can probably use a few 5mm LEDs and make a light moderately brighter than the original, but I'd prefer to use a more modern high power emitter. Without having a metal body I am worried about heat buildup from whatever emitter I eventually go with. The size of the slot being 17mm tall limits the vertical height of the emitter to about 15mm (assuming a 1mm thick plastic shell) Most of the emitters I've seen are mounted on MCPCBs that are 17mm or larger but I've seen some that might fit. I have a CNC router that can cut aluminum so creating a custom heatsink, or a partial/full aluminum body for the light is an option but I would prefer to keep the majority body plastic to match the look of the original.

1. Is there any way to calculate what the maximum wattage one can power an LED at if the only real thermal management is a MCPCB that does not have any significant thermal bridge to a metal flashlight body? ABS has a glass transition temperature around 105°C so I'd like to keep the MCPCB safety below that if possible.
2. Would one get a better thermal management powering multiple higher power LEDS at a lower voltage than powering one LED at a higher voltage to get the same lumen output?
3. The factory light used two AA alkaline batteries. For a light designed to be stored daily in potentially hot/cold car interiors what would be the safest modern battery solution?
4. Does anyone know of a widely available driver/emitter combo that would be suited to this application? I suspect I may need to look into making a custom PCB for this project but if there is a suitable drop in solution I would prefer to go that route especially for my first prototype.

Any pointers you all can give, or resources you can point me towards would be appreciated. Thanks in advance!

The efficacy of an LED typically does decrease with increasing current. Thus multiple emitters at lower current will have higher efficacy than a single identical emitter combining the currents. However this effect is most pronounced when driving emitters in the vicinity of their ratings. Given that you will probably be running the emitter well below specs, the effect would probably be too small to warrant the trouble it takes.

To attach numbers to this discussion would require selecting an emitter and choosing a drive level.

It's true that optimal thermal management would have a direct metal path from emitter to the outside of the light body, this isn't necessary at low enough power levels. Running the emitter at 1/2W, you should be able to get 60-80 lm from it, which is probably around 10 times what the original was. At this level only around 0.3W is heat (the rest coming out as light), and just the MCPCB inside a plastic housing should be sufficient, if you got a 17 mm one and trimmed the edges to fit in the 15 mm allowed.

If you were to make a custom heatsink with large surface area within the plastic shell, you could probably get into the 1-2W range safely.

Making calculations including a plastic case is possible, but there are so many variables it would be hard to make such calculations with any accuracy. I you wanted to experiment with what a particular design can actually handle, a thermocouple meter could be used to measure the temperature of the MCPCB. You can get a decent one remarkably cheap - around $30.

Thanks for the information. That was exactly what I was hoping for, a rough estimate of power levels to start out with. I suspect the rest will be trial and error on my part once I start piecing together a prototype or two. I'll start by looking to see what emitter options are out there with the highest efficacy near my target range and go from there. If I can eventually get into the 1-2w range and get 150-300 lumens output that would be nice, as the estimates I've seen for the light output of most cell phone LEDs is around 40-50 lumens, and everyone carries around one of those in their pockets these days.


I suspect I probably will eventually end up going with a custom heatsink that is at least vented to the outside air if not completely external. Since most of the emitter designs I've seen have electrically isolated thermal paths, I am also wondering if I attach the LED to a standard f4 type PCB with a large enough thermal plane connected to separate heatsinks might be sufficient as it would probably be easier to fabricate and fit my space requirements better.

BLgecko said:

cell phone LEDs is around 40-50 lumens, and everyone carries around one of those in their pockets these days.

Click to expand...

Well, almost everyone does. I don't. Say what you will.

BLgecko said:

I am also wondering if I attach the LED to a standard f4 type PCB with a large enough thermal plane connected to separate heatsinks might be sufficient as it would probably be easier to fabricate and fit my space requirements better.

Click to expand...

At power levels of a watt or two, this is quite a reasonable solution. My first custom light used 6 Luxeon Rebel LEDs on a standard FR4 circuit board, running about 10W total. The back side of the PCB was mounted to the heatsink. It works fine. There is some significant science here, and it would be wise to study it to make sure you at least approach an optimal solution.

Another option that wasn't available to me at the time, is to make a custom aluminum-backed PCB. Believe it or not, if you can make your PCB single-sided, you can get 10 small boards delivered for about $60 total. There's also free and easy software available to design it.

Thanks for all the advice. I think I've decided on a possible emitter/driver combo. If I am reading the spec sheets correctly and interpreting the graphs right, I should be able to get about 105lm at around 0.5w and 168lm around 1w with a Cree XP-L2 at 85°C. I looked at the XP-G3 too, which is would give slightly less lumens and is a bit cheaper, but the lower thermal resistance of the XP-L2 (2.2°C/W) I think is better for a plastic housing.

I am thinking that a driver based on the A17DD-L FET+1 featured here ( https://kandepet.com/deconstructing-a-flashlight/ ) might work for my needs. With the opensource and customizable firmware I should be able to just disable/delete the higher amp modes to keep it within the wattage/temperature limit I determine is safe for the housing. It will also allow me to set multiple lower light levels and/or flash patterns. I could probably just delete the FET from the schematic as I wont see power levels that high but I might leave it there for possible future upgrades. That driver looks like it was designed for use with an 18650 in mind though so I am still looking to see if it will work with a 2AA (Alkaline or lithium)combination.

The back side of the PCB was mounted to the heatsink. It works fine.

Click to expand...

I was actually thinking of mounting the heat sinks to the copper side of the board (single sided) and keeping them electrically isolated but connected to the thermal ground for the emitter. The unique shape of this housing makes it feasible to mount them on the emitter side, unlike a normal tubular flashlight. If your light was cooled sufficiently from heatsinks on the back side of a fiberglass board I would think that connecting them to the emitter side has a real good chance of working at the power levels I am looking at. Though I will definitely keep studying up!

I've made a few circuit boards in the past, and played around with Eagle and a few other software programs. I have what I need to etch a board, and my CNC router should be able to cut a PCB too if I don't go too thin on the traces. I have a few friends who are much more versed in electronics who can help me in that aspect too so for prototyping I should be set. This project seems more and more doable each day.

The way I calculate it, you get 0.5W at about 2.6V, 192 mA. At 192 mA, it looks to me like you get about 20% of the flux you would get at 1050 mA, or about 100 lm if you are using the very highest bin in the data sheet.

At that current, you get about 2-2.5 hours runtime before the battery voltage drops enough that the light comes out of regulation. That's only about 20% of the nominal capacity of the battery. I'm using data from the Duracell coppertop datasheet.

I get 1W at 376 mA, 2.66V, and not quite 40% of rated output. So almost 200 lm from that same emitter.

The coppertop chart is hard to read at that point, but it looks like an hour or less of regulated runtime, or only about 14% of nominal battery capacity.

A quick check shows 4 AAA cells isn't an improvement over 2 AAs. If you could manage 6 AAAs in a 3S2P configuration, that's a big improvement - around 10 hours at 0.5W, over 4 hours at 1W.

Of course, all this assumes that driver will run properly down to 2.6V. You'd want to verify that too.

Thanks for the reply. I am glad my figures weren't that far off from what you estimated, but I definitely would want more battery life than that. What would your opinion be on adding a DC-DC boost converter to feed the Driver such as this one http://www.ti.com/lit/ds/symlink/tps61020.pdf

While I would lose some efficiency, I wonder if I would still get more total run time out of using two AA connected in parallel feeding into a boost converter, then feeding that 3V output into the FET+3175 driver. Thoughts?

That would certainly work. But if you are going to build a circuit yourself, it would be much more efficient to boost directly to the LED. This would require a regulated current boost converter instead of a regulated voltage one, but they are available, specifically designed to drive LEDs.

I'd take a look at Microchip MCP1643, MPS MP3412, or Linear Technology LTC3490. I would definitely run two cells in series with any of these. You'd get way more regulated runtime from this configuration than what we've been talking about before.

Thanks for the continued help and suggestions! The chips you recommended look very promising. i'm putting together an order now for some components so I can build a few circuits and test them out. I think I might order one of each. The max amp output is a little lower than I'd like in case I can push the temperature/wattage of the assembly higher, but it looks like they were still well within my initial current range for testing. It looks like Analog Devices even offers free samples for the LTC3490 so that is a bonus too!

I was thinking of running the batteries in Parallel as a safe guard against over discharging a battery in case one forgets down the line and were to use Lithium AA batteries in the light. With this light being basically a replacement for emergency or occasional use and otherwise left in a car for a long periods of time I figure it may often have less than ideal voltage. With the shutoff voltage of the boost converters I was looking at being around 0.6-0.9v I figured that is pretty unlikely to over-discharge lithiums wired in parallel. In series though I think it might be even more likely to allow one battery to over discharge. The good thing is all the options I was looking at and you suggested have an input voltage range wide enough for me to try wiring powering the circuit either way. I also kind of like the idea that the light would still function in parallel if one only had 1 AA available, albeit for a much shorter run time.

You suffer a big hit on the efficiency running cells in parallel. Note how the efficiency graphs drop off rapidly below 1.5-2.0V.

That's why I recommended series. I think a single driver could work with 1-2 cells, either parallel or series, so maybe a test would be in order to decide.