HID flashlight design

Hi guys,

This is my first post here, and I wanted to run a flashlight idea by the folks with experience doing this sort of thing.

I have played with HID headlight retrofits for cars for a few years, and enjoyed the performance. Finally the prices on these units are getting down low enough that its in a reasonable price range to be useing for tinkering.

I would like to build a d-cell maglight sized flashlight featureing an automobile type projector lens/reflector assembly complete with the automobile HID bulb socket/placement.

I have a pretty good deal of 3.6v 4400mAh lithium cells sourced from laptop battery packs.

Due to the legnth of the projector lens assembly and ballast size, I was thinking of perhaps grouping the cells into 2 parallel clusters of 4cells in series to yeild a pack of 8800mAh capacity at 14.4v.

From experimentation, I know that the powersuppys in the ballasts draw slightly less power when switching 14.4v rather than 10.8v (3 cell groups). Output to the lamp being current controled of course stays the same reguardless of input voltage to the ballast (within ballast parameters).


So, this is the part where I am out of my domain. First, I have found chargeing Li-ion cells to be a finicky task, and I need ideas towards what design it will take to charge such a pack.

Secondly, has useing a projector type headllight assembly allready been tried and rulled out for some reason that I haven't thought of?

Any feed back would be great.

-Luke

BTW, Great forum guys!

There are some astonishingly awesome builders on this form - you will get your answer, even though I can't supply it.

A guy did just build a "flashlight" by hacking a Chrysler 300C HID light into a flashlight-like body, so it would seem like projector-style systems (if you can get them in a tube like that) can and will work. The beam pattern out of the 300C system was a little odd - very horizontal and not so round, like you might expect out of a car headlight.

I have to say, though - if you build these, I will buy one. I just got bit by the HID bug and I have to say a 3 or 4-D sized HID projector would just tickle me pink.

Hey Luke, are you the same Luke whose posts I frequently loved to read on The Old One's forums??

Anyway, there are several HID iterations that have been floating around this forum. Most of the home-brews that use a Maglite as a host use the HID lamp and ballast made by Welch-Allyn. The W-A ballast unit is a cylindrical form factor and is almost the exact correct size to use in a Maglite (the Maglite needs a little lathe-boring). The optics on the HID handheld lights, or at least all of the ones I've seen here, have been a conventional parabolic reflector with a glass window. I have not seen an automotive style handheld light that uses a focusing lens... perhaps that makes too tight of a spot and not enough "spill," which many people consider to be advantageous for a handheld light.

BTW, the W-A HID ballast comes in different voltage versions... most popular here is the 12-volt version. Your Li-ion cells are probably nominally rated at 3.6v, which means they're probably closer to 4.20v hot off the charger, so three of them will make the 12-volts necessary to run a usual HID ballast.

I have a Microfire Warrior K2000R handheld HID, and it uses three 18650 lithium ion cells in its pack.

Anyhoo, spend about ten hours reading all the posts on the site and you'll have an idea of what people here have done... and trust me, they've done just about everything.

(BTW, I had a Civic EG with a 307 shp B20 under the hood, ITBs and BIG Comp Cams camshafts haha... built in no small part due to TOO's help, and probably yours.)

Yep!!! Same Luke! I really miss that forum. Its always a pleasure to run into an old Endyn board member. 307hp B20 EG? That must be a delightful handful. After getting becomeing an engineer for Microsquash, I've developed a respect for the k-series. Right now I'm working on a datsun 510 powered by an f20c, and I'm hunting for an Ariel Atom in a reasonable price range to use as a weekend fun car. Great to hear from you!

Anyhow, back to the flashlight design...

I would like to achieve a handheld compact torch with at least 5000+ lumens output while getting at least 90minutes run time.
So, I need some help from the experience guys.

ASSumeing an automotive headlamp HID efficiency doesn't drop below 65L/w while getting overdriven from 50w to 77w, then I will be looking at needing ~80w (assumeing 96% ballast efficiency) to hold for 90mins... It would be great for someone with greater insight on efficiency drop as MH is overdriven who would like to say if my 65L/w is reasonable to expect from a 1.5 powerfactor.

This will mean a 12v battery will need to be ~10Ah, and by useing 2 groups of 3 Li-ion cells at 4.4mAh each, I will be short on reaching my goals. If I bump voltage up to 2 packs of 4cells, I will be close, but still not quite reaching my goals. I don't like the way that groups of 5 or 6 cells pack together, which kinda places me at 7cells.

7cells with a 0.125" wall would yeild a handle ~2.85" (72mm) This seems acceptable, as I have very large hands. 2 groups of 7cell clusters would also yeild a pack with the energy storage I require to meet my goals. The additional surface area of the larger diameter handle should also increase heat dissapation area, which I assume will become quite an issue in a compact 5000L handheld device.

The downside is that 14cells doesnt form common voltages easily. It makes 3.6v, 7.2v, 25.2v, or 50.4v. Pardon my ignorance, but I dont belive that any standard HID ballasts use any of these voltages? Fortunately, I belive useing the 50.4v pack, I could make my own ballast fairly easily, assumeing I could canibalize some other lights starting circut. The starting circut part seems like a PITA to build, the regulated current supply part seems like a snap with a Pic micro, a few FETs and a cap. Once again, please pardon my ignorance if there is an additional complexity involved in HID ballast design that I am overlooking.


Sable- Could you perhaps find the link to the fellow who used an automotive type projector reflector assembly in a flashlight? The non-round beam is actually interesting me in a pretty big way. With a twist of my hand I could optimize where my lumens are going, rather an just lighting up a large field that I am generally only interested in seeing a perticular area of. It would seem a flatter beam would be ideal for walking on a trail or something, perhaps this is why headlight manufactures arived at this beam pattern?


Thank you again for the warm greeting on your very nice board!

Best wishes,
-Luke

Welcome Luke!
Unfortunately the popular Welch Allyn cylindrical ballast/Solarc lamp is only 10w. Worse, it's just rated at 530L so efficiency is 53L/w.

The trouble with a Mag sized 35-50w HID is most cheap and available ballasts will be box shaped, which would be no problem for a lantern shaped light like a Maxabeam. Or even imagine a cheap boxy plastic 6v lantern with a killer 50w HID ballast hidden inside!

Some previous projects have included H3 based HID conversion kits in a Thor, which is hardly D-cell sized. But then 80w for 90 minutes is a tall order for a D-cell sized light, unless it's 6D+

Yes, I agree that the biggest problem is getting a ballast small enough (at that wattage) to fit a MagLite. Just look on ebay to get a feel for the size/shape of the boxes... they are too big. If you look at something like "The Beast", you can see this large pancake right behind the reflector that is holding the ballast... not very sexy. Like others have said, the Welch Allyn ballasts are small enough, but then you are very limited in output power.

Second is the heat issue. 96% efficiency sounds really optimistic to me, and with 77W going into a 50W ballast, you'd better make sure you don't cook it. I am curious where you saw a 96% efficient ballast.

Charging Li-Ions is easy, especially if you are familiar with PICs or AVRs. Google it and you will find that you just need a current limited supply for initial charging. Then as the cells get close to 4.2 V, you need to reduce the current and just stop the charge once the cells hit 4.2 V. No peak detection or anything like that is necessary. Atmel has a reference design with code for a charger using their ATtiny15 MCU and a simple boost converter. Microchip probably has a similar reference.

As far as charging goes, you can only safely charge 3-4 cells in series. Doing more than that will give you trouble with cell matching and possible overvoltage, which can be catastrophic with Li-Ions. However, with the proper PIC, you could make a simple resistive cell balancer that bleeds off charge from the faster charging cells. With a balancer setup, you could do all 14 cells in series, but at reduced efficiency. I assume you'll keep the charger separate from the flashlight, so extra heat should not be a problem.

Btw, if you are comfortable with all of this, you *could* just hack an automotive ballast to fit whatever cell combo you want. They all use the same topology: a flyback converter to boost the voltage to the lamp voltage followed by a full-bridge. You could add/remove windings from the flyback and tweak the input passives to match your battery pack.

And now for making your own ballast. First of all, it isn't as easy as it first appears. DO NOT make the HID ballast from that resonant flyback schematic that is floating around on the internet somewhere. It will almost certainly fry your bulbs (via acoustic resonance). The reason that the automotive ballasts all use the same topology is because it is still pretty much the only reasonable one for 12V operation. It would be much easier to build the resonant designs (they don't need an ignitor for cold start-up), but metal halide bulbs cannot operate from about 1 kHz to 1 MHz because of the acoustic resonance problem. Thus, the ballasts use a square wave in the 100-400 Hz range, and since it has to be free of harmonics, it must be operated at 100% duty cycle. This means that regulation has to be done in the flyback inverter.

For a good starting point, get the spec sheet for the TI UCC2305/3305. You can also get free samples :rock: At any rate, I think you'll find that using a PIC is not even worth it when there is this nifty controller chip lying around.

As for the ignitors, you'll probably need to buy one. They usually come as a separate box, but now some of them are part of the base or even in the bulb itself.

And thus I end my epistle. I think you'll find, as I have (and no doubt many others have too) that this is not as easy as it first looked. But that is why it is so fun.

Btw, Luke, the ignition circuit is only complicated if you want hot-restrikes. If you only need to start cold lamps, you can design for 4 kV peak, which is much, MUCH easier. The ballast puts out around 600 VDC for starting, so you only need a 10:1 ratio for the cold start. You could do that with a spark gap, a cap, and a small transformer.

Hot-restrikes require much more... 25+ kV, and many of the commercial ignitors are capable of supplying close to 100 kV at high current. You're looking at something similar to a coil-on-plug setup for that, although as you can see, they manage to make them small.

wulfgang said:

Btw, Luke, the ignition circuit is only complicated if you want hot-restrikes. If you only need to start cold lamps, you can design for 4 kV peak, which is much, MUCH easier. The ballast puts out around 600 VDC for starting, so you only need a 10:1 ratio for the cold start. You could do that with a spark gap, a cap, and a small transformer.

Hot-restrikes require much more... 25+ kV, and many of the commercial ignitors are capable of supplying close to 100 kV at high current. You're looking at something similar to a coil-on-plug setup for that, although as you can see, they manage to make them small.

Click to expand...

Wulfgang, do you work in the HID industry or a related one. You seem to know a lot for someone who's interest is only casual?

I am at University of Illinois, so I have free access to all of the latest and greatest research papers in this area. Also, some of the people in my MS thesis research group were modeling ignition processes in MH lamps a few years ago.

But yes, my interest is only casual at this point since I need a good bike light for riding to/from work. I just joined CPF a few weeks ago after finding so many good ideas here.

Hiya Luke!

Here's the link to the feller that created the "Headlight On A Stick" - it's kind of ugly, but sure gets the idea across!

Headlight On A Stick


- Sable

Thank you for all the great info guys!!!

I realize that this is a very tall order to build, which happens to be exactly why I am interested in doing it.

I ordered 4 different types (I'm at least hopeing they are different inside) of Ebay el-cheapo HID retrofit kits for cars. I'm going to open and hopefully be able to de-pot the ballasts in each one, and find out which setup will be best suited for input voltage jumping, overdriving and component relocation to be possible to fit inside 2.5" tube. The units which are poorly suited towards this task will be re-potted and used to improve my headlights in my cars.

Sable- Thank you for that link. I think that beam pattern he showed looks like it will suit my purposes much better than a round beam pattern that I see from most flashlights. I will go ahead and hack/cut/smash apart the projector style headlight assembly I have laying around and see how compact I can get the unit sized down to without comprimiseing the efficiency of the design.

For anyone who asked about buying one, I'm afraid I am only planning on building 1 of these for personal use. I will however make explict directions to DIY your own unit after my testing and re-testing and re-testing results in something that I find to be a useful end product.

Case construction will likely be custom machined and tig welded aluminum, or perhaps even a epoxy and carbon fiber hybrid design. Being practical, it will likely will require the increased heat dissapation capabilities of a full aluminum body rather than a hybrid of sorts.

The battery chargeing is still something that I am worried about... Since I must have the light dive capable, I'm thinking perhaps a series of flush copper studs epoxy mounted into a strip of delrin plastic and flush mounted into the end or side of the flashlight body. Neodymium magnets epoxy mounted into drilled pockets in the back of the plastic strip would be aranged to match opposite polarity neodymium magnets mounted in the chargeing clip which would contain a series of spring loaded contacts to firmly connect to each of the required charging leads inside the battery pack. From Wulfgang's information, It would seem I will require at least ~10 charging contacts to keep the overvoltage cell damage risk to a minimum. Seems like it will be an awful PITA to get this charger setup perfect... I would love to just canibalize or reverse engineer someone else's design of this sort of battery pack. If anyone knows of a compact sized pack (and adaptable charger) meeting these sorts of power storage needs, I would love to source one.

Thanks again for your help!
-Luke

Yeah, I would just ditch the series Li-Ion setup and just try to make a 12V pack in series/parallel. You can get 18650's at about 8 W-h each, so for 80W you need 10. Round it up to 12 or down to 9 so that you have 3 or 4 sets of 3 in series (called a 3S pack). Then you just parallel the series packs (3S3P or 3S4P). This is good because a set of 3 18650s will just barely not fit as a triangle inside a MagLite D body. So it is close to the correct size. Fortunately for you, McMaster sells a CF tube that is just a bit larger ID and will fit the 18650s.

If you end up with any pack that has parallel sets of 3S, charging is a sinch. Many people will not even monitor the intercell voltages in a 3S setup, but to be safe, it is probably a good idea. For this setup, you only need a 4-wire charging connector, which is what ThunderPower, Kokam, etc. use on their Li-Poly R/C packs.

But the question still remains... can you drive your 50W lamp safely, reliably, and efficiently at 77W, and will the ballast maintain that high efficiency?

liveforphysics said:

The battery chargeing is still something that I am worried about... Since I must have the light dive capable

Click to expand...

Have you considered doing a non-contact charging method? I think the principle is called mutual inductance... imagine the two coils of a transformer. Put one coil in your light, the other coil in a charging cradle. Drop the light into the middle of a toroidal coil winding. It will have to be an AC field in the toroid to get this to work, but it's probably do-able and not that difficult.

The HID kits I ordered from ebay came. I purchased 4 "unique" kits from 4 different sellers. Well, it turns out that I purchased 4 of the same kit with unique packageing and unique stickers on the front of the ballasts.

I installed one of the kits into my daily driver, and one into my girlfriend's daily driver. We are both VERY pleased at the difference in lighting, and I calculated that in roughly 9,000miles of night driving the kit will have paid for itself through fuel savings.

I was about 80% expecting them to be all the same, which I supose is OK, it simply means that I must find out how to adapt this perticular cheap/common style of kit towards compact flashlight usage.


I cut/broke apart the outer houseing of the projector style headlight retrofits that I had laying around. I am NOT impressed at all. I was expecting some sort of narrow glass tube with internal reflectors that efficiently would focus the light. What I found was a fairly standard bulky headlight reflector with a little glass disk held by 3 stamped metal legs directly infront of the bulb. The disk appears to have some fresnel looking ridges in the back side. If I had to guess, I would say its just an alternative from the status-quo blocking plate method of protecting on comming traffic's eyes from directly seeing the intensity of the plasma chamber (or W wire).

If anyone has some deeper grasp of projector headlight lens bennifits, I would love to get some input.

So far, I am dissapointed with the original plans, and will likely considder useing a conventional type reflector. However, for some reason, it seems ALL high-end cars are switching to this style of reflector, and I would like to think it has some merit beyond asthetics.

I have lots of aluminum tubeing laying about, and I am excited to get started on the body of the light, but I am at a standstill until I find an answer as to how small I can shrink a pumped up HID ballast package, and what the diameter of the reflector assembly I choose will be.

If anybody has some suggestions out there, I'm all ears.

Thank you and Happy New Year,
-Luke

here is my first HID lamp


http://www.mittgallery.net/albums/album38/DSCF0001.sized.jpg

was a H3 xenon ..

Use 2x 6volt battery..

liveforphysics said:

... We are both VERY pleased at the difference in lighting, and I calculated that in roughly 9,000miles of night driving the kit will have paid for itself through fuel savings.

Click to expand...

That is an interesting way to justify an HID upgrade, but I think 9k miles is far too optimistic. I just did a rough estimate using regular gasoline (114kBtu/gal at $2.50/gal) with a 38% efficient engine, 90% ballast, and 80% belt/alternator and got 25,000 hours to make $100 (vs. a 55W halogen). Even if you drive all of the time at night at 10 mph, you're looking at 250k miles to break even!

But maybe you bought your HID kits for a really good price and/or have a better justification scheme. I'd like to see what you did, so I can show my wife

wulfgang- I would be pleased to go over my rough calculations.

First, the bulk of my driving is commuting on the freeway. My daily driver requires 7% throttle position at 3800rpms to maintain 68mph. This throttle position number is arbitrary here, as its just a measurement of a pot afixed to the throttle body, but it becomes useful for this calculation.

Ok, so, knowing that the car requires 7% throttle at 3800rpm to maintain 68mph, I made special note of this to check out the power output at the wheels on the next time the car was on the dyno. So, useing a mustang inductive load type dyno, I measured the power at the wheels at 7% throttle opening and 3800rpms to be ~12whp (16.6ft-lbs @ 3800rpm). Now, I give this at least a +-15% range due to barro pressure and temp changes as well as dyno's reading a +- a couple % at very low power measurements.

So, now that we know roughly the energy rate delivered during most of my driving, and I know my average fuel economy is 38mpg, it becomes a fairly simple calculation.

The energy savings amounts to (55w+55w) - (35w+35w) = 40w. Sure doesn't look like much However, 40w is .054hp.

12whp-0.054hp = 11.946, which is 0.45% less energy required to travel at night.
This means 38mpg turns into 38.171mpg
So, it's a simple equation at that point.

My car's all feature very high compression engines that run more efficiently with 92 octane fuel. If pick $2.60/gal as an average price for premium, then we can find the amount of miles before it pays for itself.

38mpg/2.60 = 14.6154 Miles/Dollar
38.171mpg/2.60 = 14.6812Miles/Dollar

So, make an equation which sums the price of the HID kit, which in this case was $108 shipped. (I got a hell of a deal )

X = number of miles for breaking even point.

(14.6154m/$)X - (14.6812m/$)X = 108$

0.0658X = 108$

X= 1641miles to break even point.

Now, being an engineer, I multiply that figure by a saftey factor to compensate for all the goofy stuff like ballast losses, alternator losses etc. I used a saftey factor of 5, mainly because I couldn't belive it could pay for itself so quickly, and because I belive the factor of 5 should very very conservatively compensate for all the unfactored things that could boost or buck the calculation.

wulfgang- It should be said that the engines I use are very high compression Honda VTEC 4cylinder engines in a performance state of tune. My tuning always has emphasis towards increaseing power through lowering BSFC.

These calculations for a stock American or German car would be quite a bit off, perhaps by an additional factor of 2-3 due to poor engine design and minimal attention to detail.

liveforphysics said:

X = number of miles for breaking even point.

(14.6154m/$)X - (14.6812m/$)X = 108$

0.0658X = 108$

X= 1641miles to break even point.

Click to expand...

Nice one. My wife might buy that

But I don't. You need to invert those numbers as in:

X/14.6154 - X/14.6812 = 108

Then you'll get 352 kmiles, which is in the same ballpark I calculated. Nice to know that we get roughly the same number with two completely different approaches. However, you'll notice that I only did one headlight, so you'll only need 12,500 hours of driving to break even, not the 25k I said.

Still, it only makes sense if your car is brand-new. Otherwise you need some other justification, such as safety or coolness.

Wow! How embarrassing!

I had thought my calculation seemed a bit on the 'hopeful' side of things... Thank you for catching that wulfgang. I am also suprized at how similar the calculations turned out after you performed the last step correctly on my own method.

I bought an off the shelf 35w HID light from costco yesterday. I discovered that it seems to be a fairly silly amount of light to use for my applications, which are generally all close quarters type working. However it was amuseing to illuminate distant trees and things. After playing with it for a bit I've decided that I dont require my compact sized custom flashlight build to have more than perhaps a 50watt HID setup, which still seems like overkill, but diving with it at night might be nice to have that much light. Perhaps I will design/mod a ballast to have a high and low setting, enableing me to switch from perhaps 15watt (or whatever the bulb low threshold is) to high power of 50watts (or whatever experimentation shows I can boost to safely).

I really appreciate all your help here, and if you would like, I could send you one of the kits I dont end up useing. That way no justification needed for the wife .

Best wishes, and thanks for the help!
-Luke