Doug S
Flashlight Enthusiast
The Li14430: Lot\'s O\'Lumens; Under 1 inch cubed!
It all started with this post Holy Sh*t! 6 months ago today and ended with the light pictured above. The linked post is one of my technical ramblings, in this case about the optimal cell choice for building a luxeon based light as small as possible while being able to run regulated at least 350 mA for 1 hour minimum. I had the good fortune of attracting a couple of the forum "build it from scratch" heavyweights, tvodrd [Larry] and Chief Wiggum [Mike] into a collaboration designing and constructing of a light meeting this criteria set. We even had McGizmo [Don] on the hook for awhile until he saw that we were truly out of control and withdrew himself into an advisory role.
Working with these guys, none of whom I have ever met, has been a blast. There was an excellent synergism in exchange of ideas in developing this design. It was nice to be able to do the board and mechanical design as a coordinated effort allowing us to get the features and mechanical interfaces just right.
The design is based on using a size 14430 Li-ion rechargeable cell which at 6.6 cm^3 is the smallest available cylindrical Li-ion cell. In contrast, a 123 cell is 7.8 cm^3. The choice of the power source is the basis of the name of this light, the "Li14430". I must say, however, after seeing Larry's extreme attention to detail, that Anal-lux was a close runner up for a name. This guy's work has to be seen to be believed. Due to the close match been Li-ion cell voltage and Luxeon emitter voltage, a linear constant current regulator was designed and used rather than a switching regulator. This provides excellent efficiency. While it will vary with Vf bin and drive current chosen, efficiency of energy delivery from the cell to the emitter averaged over total discharge is on the order of 90%+ and is higher at high drive currents unlike switchers which become less efficient at higher drive currents. This is because at higher currents the emitter Vf rises and the cell voltage sags thus reducing the difference between Vf and cell voltage. With a linear current regulator, it is only this voltage difference that is dissipated external to the emitter.
Mechanically, the light consists of a body tube, a center section and a bezel. The body and bezel are 7075T6 (a high strength Al alloy) with anodize type III, class 1 finish (About .002" build.) They were chemical film treated for the inside and subsequently stripped on the outside prior to anodizing. The centers are alloy 145 TeCu (Tellurium Copper- a machinable Cu alloy) which were electroless nickel plated (professionally) and gold plated by Larry. The front opening surrounding the optic/reflector was vacuum metallized (aluminum). A gold plated 2-56 screw in the bottom of the body tube forms the negative cell contact. A module consisting of the circuit board, heatsink, and emitter is fastened into the center section with a pair of 0-80 screws. This arrangement permits the emitter and board to be assembled into the light without the use of any epoxies or other hardening compounds. It is thus possible to remove the board and emitter for change out or repair. The bezel end of the center section can accommodate either a modified 2XAAA MM reflector or a turned down NX-05 optic.
Assembled weight is 23g without cell and 40g [1.4oz] with cell.
This link shows the module and center section: module and center section
This link shows a complete exploded view of the light: Exploded view
Shown in the link above is a reflector version of the light.
This link shows the board only: board
The circuit board is configured to provide a LOW and HIGH level. In the photo linked above of the board only, two gold contacts may be seen. The springy looking one actuates the LOW level and the solid cylindrical one the HIGH level. On/Off operation is via twisting the head. The LOW contact makes contact with the positive cell terminal first. Further twisting of the head causes the HIGH contact to make with the positive cell terminal.
The cells used have capacities in the range of 550 to 650 mAhr. Provided that a relatively low Vf bin emitter is chosen it is possible to exhaust almost all of the cell capacity before falling out of regulation. I have configured one of the lights produced for low and high levels of 110 and 518 mA respectively driving a "J" Vf bin Lux III in an effort to meet the 1 hour minimum regulated criterion. Testing reveals a runtime of 74 minutes on high before loss of regulation. A quirk of how the regulator works results in the highest output occurring immediately before loss of regulation. CPF's own Runtime Roy agreed to do tests on this 110/518 mA version of this light and his results can be seen here:
Runtime low
Runtime high
Fairly early in the testing of the safety aspects of using these Li-ion cells in a light of this size, it became apparent that it would be possible to run at currents of 700 mA or more without safety problems or exceeding maximum rated die temperature for a LUX III. While I am paraphrasing the actual conversations a bit, it went essentially like this:
The guys : so you are saying that your safety testing shows that we could run at 700 mA or more and these things aren't going to blow up on us?
Doug : Right.
The guys : Will your proposed circuit design coupled with these cells permit operation at 700ma or more?
Doug : With selection of low Vf bin emitters, yes.
The guys : F*ck your 1 hour runtime criterion, give us the current!!
Consequently, the majority of the boards were fabricated with the LOW current set to 60 mA and the HIGH currents in the range of 700-900 mA.
So how does this light measure up to the small and bright objective?
Dimensions when OFF are .667"X2.67" for a volume of 0.93 cubic inch or 15cm^3 for our SI friends. Other lights in the sub 1 inch^3 range include:
Larry's original CR2 light .70"X2.125" for .82 inch^3
Dorcy 1AAA at .92 inch^3
ARC AAA at .53 inch^3
Larry's original NLS at .40 inch^3
If we use that most basic measure of performance, total lumens actually exiting the light divided by volume, the single Nichia 5mm lights measure poorly due to the limited output of a single 5mm LED. ARC reports 3 lm output for the ARC AAA. Dividing by the volume of .53 inch^3 gives a measure of about 6 lm/in^3.
Looking at the NLS, this light uses a luxeon powered by an ArcMania microconverter. While I have only checked one sample and it is quite possible that there is considerable variation from sample to sample, this converter appears to be able to drive a luxeon at about 100mA off of either a fresh alkaline "N" cell or a NiMh. I have measurements that suggest that a "T" bin Lux III can produce about 15 lm at 100 mA. How much actually gets out of the light? I have other measurements that suggest that in lights using a mineral glass lens and an optic, approximately 75% of produced light actually makes it out the end of the flashlight. For a light using a reflector the figure is around 80%. Since the NLS uses an optic, an estimated output at 100 mA is then 15X0.75 = 11.25 lm. Dividing by the volume of .40 in^3 gives 11.25lm/.40in^3=28 lm/in^3.
Now lets consider an Li14430 running a "T" bin Lux III at 700 mA. Per the Luxeon binning document, AB21, the minimum output of a "T" bin at 700 mA is 67.2 lm. Remember that this is at a junction temperature of 25C. At these power levels we cannot ignore the output reduction due to heating. We can calculate it though. With the "J" Vf bin Lux IIIs I have used, I measure a Vf of 3.4V at 700mA giving a power of 3.4X.70A = 2.38 W. The thermal design of the Li14430 is very good and the die to housing thermal resistance is no greater than 15 C/W. At 2.38W this gives a temperature rise for die of 2.38WX15 C/W = 36C over the temperature of the housing. Now here it gets a bit arbitrary. The case continues to heat up for a long while after being turned on. Let's arbitrarily take the measurement at 1 minute after turn on. It turns out that the center section temperature rises by about 10C in the first minute. The total temperature rise of the die over ambient in the first minute is thus 10C + 36C = 46C. The Luxeon datasheets show that a white Luxeon loses about .34% of output per degree C rise. If we assume an ambient of 25C the loss from the initial 25C value is thus 46C X .34%/C = .16. Our actual emitter output is now a minimum of 67.2 lm X 0.84 = 56 lm. If using an optic version of the Li14430 the actual lumens out is thus a minimum of 56 lm X 0.75 = 42lm. Dividing by the volume of .93 in^3 gives 45 lm/in^3 or 2.8 lm/cm^3. The equivalent calculation for the reflector version yields 48 lm/in^3. BTW, if I remember correctly, Larry reports that a CR2 driving a BB700 can stay in regulation for about 8 minutes. Assuming equivalent emitter and thermal performance, the lesser volume of the CR2 would yield 51 lm/in^3.
I'm getting tired of jabbering about this. Hopefully the other guys will chime in with some more details. In the meantime, here are a few more pictures that may be of interest:
This photo shows top-bottom, Arc AA, Li14430, CMG infinity
This link shows Li14430 with Surefire E2e
comparison
This link shows a close up of the tail keyring attachment. Would you believe that Larry pressed a stainless steel bushing in the hole the ring goes through??
keyring
This link shows the bezel end of the center section. This is fabricated from a copper tellurium alloy but Larry had it vacuum aluminized to catch the few photons that leak out of an optic
center section
This is the module with heatsink for the reflector version. The gold and silver header pins are used to precisely locate the emitter on the heatsink pedestal.
module-reflector
And last, check out this really cool 3 section drawing by Larry. Not much wasted volume in this light!
Section drawing
BTW, in case anyone was thinking about asking, a total of 11 of these were made. There are no plans for future production. It is likely that at least 6 of these will eventually find their way into the CPF marketplace. We are still putting them through their paces to see if there are any problems not yet identified.
EDIT: For those interested in a history/timeline of this project, check this post Li14430 project timeline
EDIT: For those interested in the safety testing associated with use of these Li-ion cells, check this post Li14430 project: li-ion cell safety testing
EDIT: For those interested in the electrical design aspects of this light, check this post: Electrogeek stuff
It all started with this post Holy Sh*t! 6 months ago today and ended with the light pictured above. The linked post is one of my technical ramblings, in this case about the optimal cell choice for building a luxeon based light as small as possible while being able to run regulated at least 350 mA for 1 hour minimum. I had the good fortune of attracting a couple of the forum "build it from scratch" heavyweights, tvodrd [Larry] and Chief Wiggum [Mike] into a collaboration designing and constructing of a light meeting this criteria set. We even had McGizmo [Don] on the hook for awhile until he saw that we were truly out of control and withdrew himself into an advisory role.
Working with these guys, none of whom I have ever met, has been a blast. There was an excellent synergism in exchange of ideas in developing this design. It was nice to be able to do the board and mechanical design as a coordinated effort allowing us to get the features and mechanical interfaces just right.
The design is based on using a size 14430 Li-ion rechargeable cell which at 6.6 cm^3 is the smallest available cylindrical Li-ion cell. In contrast, a 123 cell is 7.8 cm^3. The choice of the power source is the basis of the name of this light, the "Li14430". I must say, however, after seeing Larry's extreme attention to detail, that Anal-lux was a close runner up for a name. This guy's work has to be seen to be believed. Due to the close match been Li-ion cell voltage and Luxeon emitter voltage, a linear constant current regulator was designed and used rather than a switching regulator. This provides excellent efficiency. While it will vary with Vf bin and drive current chosen, efficiency of energy delivery from the cell to the emitter averaged over total discharge is on the order of 90%+ and is higher at high drive currents unlike switchers which become less efficient at higher drive currents. This is because at higher currents the emitter Vf rises and the cell voltage sags thus reducing the difference between Vf and cell voltage. With a linear current regulator, it is only this voltage difference that is dissipated external to the emitter.
Mechanically, the light consists of a body tube, a center section and a bezel. The body and bezel are 7075T6 (a high strength Al alloy) with anodize type III, class 1 finish (About .002" build.) They were chemical film treated for the inside and subsequently stripped on the outside prior to anodizing. The centers are alloy 145 TeCu (Tellurium Copper- a machinable Cu alloy) which were electroless nickel plated (professionally) and gold plated by Larry. The front opening surrounding the optic/reflector was vacuum metallized (aluminum). A gold plated 2-56 screw in the bottom of the body tube forms the negative cell contact. A module consisting of the circuit board, heatsink, and emitter is fastened into the center section with a pair of 0-80 screws. This arrangement permits the emitter and board to be assembled into the light without the use of any epoxies or other hardening compounds. It is thus possible to remove the board and emitter for change out or repair. The bezel end of the center section can accommodate either a modified 2XAAA MM reflector or a turned down NX-05 optic.
Assembled weight is 23g without cell and 40g [1.4oz] with cell.
This link shows the module and center section: module and center section
This link shows a complete exploded view of the light: Exploded view
Shown in the link above is a reflector version of the light.
This link shows the board only: board
The circuit board is configured to provide a LOW and HIGH level. In the photo linked above of the board only, two gold contacts may be seen. The springy looking one actuates the LOW level and the solid cylindrical one the HIGH level. On/Off operation is via twisting the head. The LOW contact makes contact with the positive cell terminal first. Further twisting of the head causes the HIGH contact to make with the positive cell terminal.
The cells used have capacities in the range of 550 to 650 mAhr. Provided that a relatively low Vf bin emitter is chosen it is possible to exhaust almost all of the cell capacity before falling out of regulation. I have configured one of the lights produced for low and high levels of 110 and 518 mA respectively driving a "J" Vf bin Lux III in an effort to meet the 1 hour minimum regulated criterion. Testing reveals a runtime of 74 minutes on high before loss of regulation. A quirk of how the regulator works results in the highest output occurring immediately before loss of regulation. CPF's own Runtime Roy agreed to do tests on this 110/518 mA version of this light and his results can be seen here:
Runtime low
Runtime high
Fairly early in the testing of the safety aspects of using these Li-ion cells in a light of this size, it became apparent that it would be possible to run at currents of 700 mA or more without safety problems or exceeding maximum rated die temperature for a LUX III. While I am paraphrasing the actual conversations a bit, it went essentially like this:
The guys : so you are saying that your safety testing shows that we could run at 700 mA or more and these things aren't going to blow up on us?
Doug : Right.
The guys : Will your proposed circuit design coupled with these cells permit operation at 700ma or more?
Doug : With selection of low Vf bin emitters, yes.
The guys : F*ck your 1 hour runtime criterion, give us the current!!
Consequently, the majority of the boards were fabricated with the LOW current set to 60 mA and the HIGH currents in the range of 700-900 mA.
So how does this light measure up to the small and bright objective?
Dimensions when OFF are .667"X2.67" for a volume of 0.93 cubic inch or 15cm^3 for our SI friends. Other lights in the sub 1 inch^3 range include:
Larry's original CR2 light .70"X2.125" for .82 inch^3
Dorcy 1AAA at .92 inch^3
ARC AAA at .53 inch^3
Larry's original NLS at .40 inch^3
If we use that most basic measure of performance, total lumens actually exiting the light divided by volume, the single Nichia 5mm lights measure poorly due to the limited output of a single 5mm LED. ARC reports 3 lm output for the ARC AAA. Dividing by the volume of .53 inch^3 gives a measure of about 6 lm/in^3.
Looking at the NLS, this light uses a luxeon powered by an ArcMania microconverter. While I have only checked one sample and it is quite possible that there is considerable variation from sample to sample, this converter appears to be able to drive a luxeon at about 100mA off of either a fresh alkaline "N" cell or a NiMh. I have measurements that suggest that a "T" bin Lux III can produce about 15 lm at 100 mA. How much actually gets out of the light? I have other measurements that suggest that in lights using a mineral glass lens and an optic, approximately 75% of produced light actually makes it out the end of the flashlight. For a light using a reflector the figure is around 80%. Since the NLS uses an optic, an estimated output at 100 mA is then 15X0.75 = 11.25 lm. Dividing by the volume of .40 in^3 gives 11.25lm/.40in^3=28 lm/in^3.
Now lets consider an Li14430 running a "T" bin Lux III at 700 mA. Per the Luxeon binning document, AB21, the minimum output of a "T" bin at 700 mA is 67.2 lm. Remember that this is at a junction temperature of 25C. At these power levels we cannot ignore the output reduction due to heating. We can calculate it though. With the "J" Vf bin Lux IIIs I have used, I measure a Vf of 3.4V at 700mA giving a power of 3.4X.70A = 2.38 W. The thermal design of the Li14430 is very good and the die to housing thermal resistance is no greater than 15 C/W. At 2.38W this gives a temperature rise for die of 2.38WX15 C/W = 36C over the temperature of the housing. Now here it gets a bit arbitrary. The case continues to heat up for a long while after being turned on. Let's arbitrarily take the measurement at 1 minute after turn on. It turns out that the center section temperature rises by about 10C in the first minute. The total temperature rise of the die over ambient in the first minute is thus 10C + 36C = 46C. The Luxeon datasheets show that a white Luxeon loses about .34% of output per degree C rise. If we assume an ambient of 25C the loss from the initial 25C value is thus 46C X .34%/C = .16. Our actual emitter output is now a minimum of 67.2 lm X 0.84 = 56 lm. If using an optic version of the Li14430 the actual lumens out is thus a minimum of 56 lm X 0.75 = 42lm. Dividing by the volume of .93 in^3 gives 45 lm/in^3 or 2.8 lm/cm^3. The equivalent calculation for the reflector version yields 48 lm/in^3. BTW, if I remember correctly, Larry reports that a CR2 driving a BB700 can stay in regulation for about 8 minutes. Assuming equivalent emitter and thermal performance, the lesser volume of the CR2 would yield 51 lm/in^3.
I'm getting tired of jabbering about this. Hopefully the other guys will chime in with some more details. In the meantime, here are a few more pictures that may be of interest:
This photo shows top-bottom, Arc AA, Li14430, CMG infinity
This link shows Li14430 with Surefire E2e
comparison
This link shows a close up of the tail keyring attachment. Would you believe that Larry pressed a stainless steel bushing in the hole the ring goes through??
keyring
This link shows the bezel end of the center section. This is fabricated from a copper tellurium alloy but Larry had it vacuum aluminized to catch the few photons that leak out of an optic
center section
This is the module with heatsink for the reflector version. The gold and silver header pins are used to precisely locate the emitter on the heatsink pedestal.
module-reflector
And last, check out this really cool 3 section drawing by Larry. Not much wasted volume in this light!
Section drawing
BTW, in case anyone was thinking about asking, a total of 11 of these were made. There are no plans for future production. It is likely that at least 6 of these will eventually find their way into the CPF marketplace. We are still putting them through their paces to see if there are any problems not yet identified.
EDIT: For those interested in a history/timeline of this project, check this post Li14430 project timeline
EDIT: For those interested in the safety testing associated with use of these Li-ion cells, check this post Li14430 project: li-ion cell safety testing
EDIT: For those interested in the electrical design aspects of this light, check this post: Electrogeek stuff
Last edited: