You should probably get some coffee, this story is long...
Updates can be found at the bottom of the post.
- The Reflector
- The LED
- Electronics and Batteries
- Planning Phase
- Building the light
A few years ago I was lucky to be able to cheaply purchase an old, used Maxabeam Head. The Maxabeam is the farthest throwing, commercially available, portable spotlight on the market. More Infos concerning this can be found here in sma's nicely detailed thread. Two details are special:
- The head contains a relatively large, electroformed, parabolic, precision reflector
- The head itsself is quite large and having a modder with a lathe build one would require a lot of time and effort
Next to a D-Maglite:
Since then it has been my dream to combine this head with an LED to create, in terms of throw, probably the most extreme, classic LED flashlight (only one LED with reflector).
Why LED? Because in many ways LEDs are much more practical compared to the specialized HID bulbs which normally get used in such specialty spotlights:
- cheaper by an order of magnitude (taking the electronics and the batteries into consideration)
- dimmable to any preferred brightness
- higher efficiency
- much lower total power consumption leading to smaller batteries and/or longer runtimes and less wheight
- no dangerous UV / IR radiation
- can be turned on and off randomly
- can be left running for any length of time in any orientartion
- no risk of bulb explosion
- no high voltage
- layman compatible
Michael (tlf member RC-Drehteile, now here on BLF as RC-Lights, an experienced light builder who owns a lathe) kindly took on the project and really outdid himself! It was much more complicated and lengthy than anticipated, but now after around four months the light is finally finished! He went to a lot of effort to incorporate my numerous wishes and to withstand my growing impatience.
I will go into a bit more detail on it here. Electroformed means that a precisely formed, negatively shaped mandrel, typically made of stainless steel, is coated with Nickel while sitting in a suitable electrolyte. The nickel electroform made this way then has all the shape- and surface features of the mandrel down to the smallest detail. This process enables a comparably cheap production with a repeatable, high precision.
The especially precise (parabolic) shape allows for a large difference between the size of the light source and the size of the reflector. Normal flashlights don't usually require this to be considered because comparably speaking most LEDs are rather large. At some point it does become important though. The Maxabeam has a very small lightsource (the bright spot in the arc of the Xenon shortarc bulb has a diameter of around 0.125mm - 0.005in) and only produces its extreme luminous intensity due to the precise reflector used because only this way the entire surface of the reflector can be lit up by the small point (see this thread for an explanation).
After the Nickel has been electroformed it is coated via electroplating or vacuum disposition. There are different materials for different purposes. The most common is probably aluminium because it's cheap and reflects 90% of visible light. The Maxabeams reflector is coated with the noble metal Rhodium. Rhodium coatings are considered the most robust because they are corossion resistant, very hard und are unsusectable to UV-radiation. Since the Maxabeam is meant for professional use in the military, where maritime use is very likely, the corrosion resistance to salt water was probably the primary reason for this coating. The main disadvantage of Rhodium compared to Aluminium is the lower reflectance of visible light (70-80%, see here).
Dimensions of the Maxabeam reflector:
- inside diameter large opening: 118mm
- inside diameter small opening: 11mm
- minimum focal length (the smallest distance between the focal point and reflectors surface when using an LED): 20mm
- for all parabolic reflectors this is horizontal from the focal point to side of the reflector
- when using a bulb which also emits light behind itself this value is half of the horizontal distance
- this causes the unused area in the middle (when using an LED) to have a diameter of 40mm
- maximum focal length: 94mm
Part of my motivation to finally start the project this year was that there is finally an LED which is better for throwers than the old (de-domed) XP-G2 from 2012 - the Osram Black Flat Gen2 (LUW HWQP).
- higher luminance - approx. 250cd/mm2 (see Photons Test and and also this thread about luminance )
- for comparison: XP-G2 max. 220cd/mm2, but only with very green tint
- no de-doming necessary
- cool-white (6500K "ultra white")
- basically no bad tint (especially not greenish/yellowish)
- center solder pad not electrically neutral
- Size and positioning of the solder pads does not precisely match those of XP-LEDs
- DIE size: 1mm2 (even smaller than XP-G2, here you can find a size comparison of the hotspots using the same reflector, but please ignore their brightness since they are direct driven)
- even smaller hotspot
- even more difficult to focus
- You never know which BIN you have since Osram doesn't specify this
- low lumen output (approx. 790 LED Lumen according to djozz's test)
((790 / 4) + (790 * 3/4 * 0.75)) * 0.97 * 0.95 = ~592 "theoretical ANSI" lumens
I factored in how much light hits the reflector, the reflectors reflectivity, the UCLp lens and a small loss for heat with very good heatsinking. This is all based on the 790 LED lumens which djozz measured.
Luminous Intensity (throw):
Now to the likely most interesting value, which is actually rather easy to calculate (see here).
Luminous_intensity[cd] = luminance_LED[cd/mm^2] * area_of _reflector_as_seen_from_hotspot[mm^2] * reflectivity_of_reflector[%] * transmission_of_lens[%]
- Luminance of LED: 250cd/mm^2 (of course there is some uncertainty here)
- Area of reflector (lit up area of circle as seen from the hotspot) = 10936mm2 - 1256mm2 = 9680mm2
- Reflectivity of reflector with Rhodium coating: 75%
- Transmission UCLp lens: 97%
Maximum possible luminous intensity when turning on the flashlight: 1,760,550cd (2,654m ANSI throw).
To make it more realistic one could now subtract an additional 5-10% for the heating up of the LED.
Minimum distance for accurate measurement of luminous intensity:
Many of you know that one cannot measure the true luminous intensity (throw) of a thrower in 1m distance if the value is supposed to be accurate. The reason for this is that in this distance in some lights the reflector is not completely lit up by the LED as seen from the sensor of the lux meter. One has to go further away and adjust the value for the longer distance. TLF member sma once studied this in more detail here. We both have the theory that it depends on the difference in size between the reflector and the LED. I have tried to calculate this for my light based on his results with an Olight SR-95 UT.
Olight SR-95 UT:
LED-diameter (SBT-90): 3mm (if it were a circle without the corners)
Reflector diameter: 75mm
Minimum measuring distance: 2.5m (approx 1/10 of the proportion)
LED-diameter (if it were a circle without the corners) = 1mm
Reflector diameter: 118mm
Minimum measuring distance: 11.8m
BTW here also for the Maxabeam:
Light point diameter: 0.125mm
Reflector diameter: 118mm
Minimum measuring distance: 94.4m
I can't say how accurate these values really are. I don't know in what way the corners of the LED need to be taken into account. Theoretically they reduce the distance a bit (think about the hotspot size of lights with bulbs, the filament is basically just a slim glowing bar).
To be able to have an understanding of the focus of this light beforehand I tried to calculate the spot size. For this one needs the following relation, which sma explained to me. It can be found in this wikipedia article on lenses, unfortunately there is no English version. One also needs the diameter of the light source and the maximum focal length of the reflector.
A = B / G = b / g
- A = projection scale
- G = Diameter of projected ("real") image (the hotspot)
- B = Diamater of light source ("object") = 1mm = 0.1cm (here I also ignore the corners of the LED DIE)
- g = Distance between reflector opening and hotspot
- b = maximum distance between the light-source and the reflector opening = max. focal length = 94mm = 9.4cm
So if we want to calculate G for a specific distance g the equation looks like this:
G = (B * g) / b
Of course, for distances smaller than 11.8m this calculation is invalid.
Example values (distance - spot diameter):
- 12m - 12.8cm
- 30m - 31.9cm
- 100m - 1.1m
- 300m - 3.2m
- 1000m - 10.6m
- 2000m - 21.3m
So the spot is very, very small for an LED light with reflector. These values need to be understood though! I did not account for the corners of the LED DIE and there is also a large corona around the hotspot because the reflector is so deep. I have only calculated the size of the part of the spot with maximum luminous intensity.
Electronics & Batteries:
I wanted the light to use a buck driver which is able to fully exploit the capabilities of the LED regardless of the battery voltage and which can deliver much higher LED currents to enable the light to be upgraded in the future with more powerful LEDs. Additionally I wanted an over discharge protection since the light would have multiple cells in series.
Two things led us to use the Ampere! driver from the German company pcb components:
- I have always wanted to use this driver in a light because of it's supposed high quality even though there are so many disadvantages (expensive, large, can only be dimmed via external dimming module, only pwm dimming, multiple defects reported in the German forum [but all were replaced, very good customer service] etc.)
- It is the only buck driver known to me that has actually been tested to >=10A (theoretically there is another one available now, but it has never been actually tested)
The inductor is rather large:
Because the Ampere! driver itself can't really do anything except drive the LED, we combined it with the Stripe v4 Dimmer from the same company. It has the features I wanted, i.e. overdischarge protection, dimming with 1-2 electronic switches, connections for 1-2 status LEDs and multiple configuration options. There is even a new beta-firmware (v2.5), which offers the possibility of ramping (step-less dimming) with only one electronic switch. I wanted this. After a few starting difficulties we managed to flash it onto the dimmer.
Part of the dimmers features can be conveniently configured via small DIP-switches, which we have set to the following configuration: 00011000 (1 = switch up). The "Akkuwächter" (overdischarge protection) was also activated, the light does not dim prematurely, the remaining charge level of the batteries is indicated by two status LEDs and the PWM frequency was upped from 200Hz to 2kHz. The new single-switch-ramping was also activated and works nicely with a dimming time of 3s (from 0 to 100%).
During testing we noticed that unfortunately the driver produces a high-pitched whine whenever it is not at 100% or <10%. This makes the whole ramping thing a bit less fun, but just today I found out that there might be a fix. It also seems to ramp with linear current, since the brightness does not change linearly (making adjustment of low brightnesses more difficult). The voltage values of the "Akkuwächter" are also a bit outdated, they seem to be for older Li-Co cells which have high voltages when empty. The status LEDs say that the battery is empty when there is still 40% left in it. Luckily custom values can be set with help of the PC Software and and the separately available USB programmer. So yes, the driver and the dimmer do have some downsides, but they still have a ton of configurable features!
The light was to be run with three 18650s in series for a good compromise between available energy and weight. The average of 11V would ensure a high compatibility with many different buck drivers in case a different one was ever needed. The battery runtime with 4,5A and a Vf of 3,7V with 90% driver efficiency and protected NCR18650GA cells should be around 2h without any form of dimming/step-down!
I have had this idea for years. Lets start with my first (ugly) concept drawing, which I made a few years ago (in paint ):
The following things were important to me:
- Everything should feel like a production quality light. There were to be no small annoying things which I would always have to keep in mind when using the light.
- Many pronounced cooling fins directly where the rod with the LED on it sits (and oversized in case a more powerful LED is later put into the light)
- Electronic switches for operating it where ones fingers are when holding the light
- Status LED(s), which show the remaining battery charge
After the start of the project we really focused on the planning. Most of the light was planned in CAD. It was decided that the body would be made of multiple parts because otherwise it would be very difficult to connect the body to the head and mount the electronics inside. It was also decided that the batteries would need to be inside a specialized battery cage, which has both contacts on the same side since we couldn't have the current flow through the body. The reason for this is that the center solder pad of the Osram Black Flat is not electrically neutral and we wanted optimal heat transfer via a copper PCB which electrically connects both.
The setup from left to right:
Head -> connection part of body with LED on copper rod -> part of body with the electronics inside a "sled" and the electronic switches on the outside -> battery tube with battery carrier -> tailcap
The electronics were to be put into an insert or "sled" which gets put inside the middle part of the body. This insert would have screwholes for fastening it to the body and holes for the wires on the front side and connection plates for the springs of the battery carrier on the backside.
A first concept:
As luck would have it I had the spare battery tube and battery carrier of my Magicfire Scorpion laying around. The battery carrier even has a mechanical switch in the back which is nice. Our alternative was the battery carrier of the Aceabeam K40, which would have been very expsenive as a spare part from China.
After the general planning we started with the finer details like the outside design.
Building the Light:
After finishing most of the planning, Michael started making the body.
He began with the copper rod for the LED. It has threads on the outside for heat transfer and focusing. It is wider at the top so that we could use a normal 16mm pcb and screw it down properly. The 11mm narrow reflector opening was not widened because of the danger of warping the reflector. This meant that the LED rod could only be inserted from the front, a delicate operation.
The LED was then mounted. The leads go through two openings under the rim and then down through the hollow rod to the driver.
Next up was probably the most important part - the connection piece between the head and the remaining body.
Four sides were leveled to have a transition to the design of the head.
Three M6 socket head screws were to connect it to the remaining body.
The four screw holes for the connection to the head were then added. They go through the cooling fins.
Thus this part of the body was finished and Michael continued with the electronics section.
It was supposed to have knurling everywhere except for where the switch and status LEDs are. This didn't quite work out because this piece of the body was a bit too long for Michael's tool.
Three flat sections were integrated to accomodate the electornic switches and status LEDs.
The last large, outside part to be made was the endcap. It was supposed to be especially massy to make the light less top heavy. An opening with rubber boot for the mechanical switch of the battery carrier was also needed. The switch needed to be electrically isolated from the body.
Now it was time to anodize the large metal parts since they were finished. The body was to be anodized black to match the head. For this the parts were first vibratorily finished to improve the surface and champfer the edges. Luckily Michael has built (and has experience with) an elaborate facility for this in his workshop.
Subsequently the switch inserts were constructed. Michael made them in his own style, having done this previously. The two status LEDs got their own insert. To give the light some color the inserts were anodized in a light blue.
Seems to fit:
The last part to be made was the insert for the electronics. First an aluminium disc was constructed which can be screwed into the front of the light, has openings for the LEDs leads and onto which the driver would be glued.
A partly open plastic cylinder makes up the rest of the outer shape. The (way too large) dimmer was mounted into it.
Massive copper contact plates for the springs of the battery carrier were put on the rear side.
For higher corrosion resistance the contact plates were gold plated. Michael also does this himself!
"Fun" fact on the side: 33 torx screws were used to build this light. This does not include the already present screws of the batterier carrier.
This marked the end of the building phase.
Now it was time for Michaels probably most annoying but at the same time most important work during this project - the optimal positioning of the LED in the focal point of the reflector in all three spacial dimensions. The calculation of the luminous intensity above already shows why this is so essential if one wants to fully make use of this reflector. Only when the LED is positions optimally the entire surface of the reflector will be lit up by the LED which leads to the maximum possible candela value (throw).
Our goal was to focus the light only by turning the copper rod (so only adjusting the height of the LED). During this the LED was to always be in perfect X- and Y-axis alignment. For three reasons this was more difficult with this light compared to most others:
- The enormous size difference between the LED and the reflector
- The general way of constructing LED flashlights compared to specialized HID spotlights - we don't have fine pitched screws for precise focussing
- The existence of at least five different sources of error, each of which could cause the LED to not be in optimal X- and Y-alignment at differents heights:
- The position of the reflector relative to the screw holes of the head (it's possible that the manufacturer didn't center it perfectly)
- The postion of the screw holes in the body relative to the head
- The position of the threads for the copper bar in the body relative to the screw holes in the body
- The position of the LEDs pcb on the copper rod relative to its rotational axis
- The position of the soldered LED relative to the pcb (unfortunately the solder profile of the Osram Black Flat does not perfectly match that of XP-LEDs, meaning that it might not center itself perfectly during soldering)
Of course some of these sources of error were of theoretical nature since Michael fabricates metal pieces to a high degree of precision. At his first focussing attempts (at night after 11 PM since it gets dark rather late in the summer here and it needed to be done outside) he quickly started to grasp what degree of precision was actually needed. He positioned the lux meter in a distance of 12m and started turning the copper rod. He had already noticed during earlier tests that only in a very small range of the focus a useable spot was possbile. 0.3mm of rotation caused a change in luminous intensity from 914kcd to 1,371kcd! At this point he wasn't able to reach an even higher value causing some disillusionment.
As it turned out though the body was not perfectly centered in relation to the reflector. Luckily Michael had thought of this beforehand and implemented some play in the screw holes to be able to adjust this a bit. In addition to this the LED pcb was not centered perfectly on the copper rod. Both problems were solved.
During the final focussing session a luminous intensity of 1.543kcd was measured. This value made us happy, also because we improved on standing performance "records" of similar, optimized lights from wiestom89 and gaston01 by 50%. The copper rod was then fixated with a bit of threadlocker.
Of course it is easier to somehow "beat" these values with a much larger reflector, but building an LED light that reaches the highest possible values for it's size in this type of light is time-consuming and difficult.
A note on the luminous intensity values:
As many of you know, optical measurements generally have relatively large tolerances since many different factors can influence them. Using a non-modded light measured according to ANSI-standards and made by a well known manufacturer, an Olight SR-52 UT, which has an XP-L HI LED with similar tint to the Osram Black Flat, Michael tried to reduce this uncertainty a bit. In addition to the similar tint this light has a perfectly regulated mid mode, in which the LED heats up much less compared to the highest mode. As it turns out, Michael's lux meter measured only 7/8 of the "real" value. The above values have been adjusted by +14.3% accordingly. Assuming that Olight only uses LEDs of a single Bin in this light a tolerance of +/- 6% would remain in addition to that of the lux meter itself.
Numerous problems arose during the planning and building of this light. Here I want to go into a bit more detail on them and our "solutions".
- Maxabeam-head doesn't have threads, but instead screw holes & a rectangular shape und the light was still supposed to have a fitting design
- Screws going through the cooling fins
- Body was supposed to have ample cooling fins even though it was to be screwed to the head
- Splitted up the body into multiple parts
- 10mm LED PCBs, which would fit through the tight (bulb) hole of the reflector, are difficult to screw down because there is not enough room on them next to the LED and the solder connections
- Copper rod with wide top and LED mounted onto 16mm pcb
- Small opening of Maxabeam reflector only has 11mm diameter and rod with LED and leads must fit through it
- Drilled openings into the rod and routed the leads through it
- Osram Black Flat has a large variance of maximal brightness
- Here our solution was to test six LEDs and use the best of them. It should be noted that we only tested until what current each LED got brighter and how bright it got relative to the others. We forgot to write down values for each at the same current. All were mounted on the same type of pcb and on the same heatsink.
- Copper rod too long in the beginning for correct positioning of LED in reflector (the Maxabeam reflector has a rather short minimum focal length)
- Shortenend it a bit
- Screws of the LEDs pcb and the solder joints blocked part of the high-angle light of the LED
- Replaced the screws with flatter ones and replaced the leads with strips of copper sheet which was encased in heat-shrink tube
- Light probably rather top heavy
- Added mass to the tailcap
- Light was supposed to be upgradeable with other LEDs in case better ones come to the market and should be able to drive LEDs with different Vfs and currents without having to replace the driver (especially because the new Osram Q8WP might actually be the new "king of throw")
- Used a driver which has a large range of possible currents (has been tested up to 11A) and where the sense resistor can be easily replaced
- More than two cells in series
- 2-3 electronics pcbs needed to be integrated into the body in a practical way and wired beforehand
- Put electronics into an insert (or "sled") which is inserted and screwed into the light after all the wires are soldered
- Two electronic switches needed to be placed in ergonomic positions
- Multiple flat sections near the front of the body
- Battery carrier needed which has both contacts on the same side and is electrically isolated from the body (because of Osram Black Flat)
- Used accessory of Magicfire Scorpion Transformers light
- A black aluminium battery tube, into which the battery carrier fits, needed to be found (to save Michael a lot of time and effort)
- Used accessory of Magicfire Scorpion Transformers light
- Problem with knurling of long (electronics) part of body
- Opted for shallow fins instead which match the spacing of the cooling fins in the front
- Reflector not perfecty clean (“hazy” look)
- The reflector needed to be cleaned under extreme caution. Luckily for us member sma anticipated this when cleaning his own Maxabeams and showed a seemingly safe method using soap foam which he even demonstrated in this Video. After sourcing the somewhat rare soap foam Michael was successfull in cleaning the reflector without damaging it.
- Picture taken right after cleaning (without the lens), the cleanings marks are form the original owner:
- Hole for the screw which fixates the electronics insert in the body was forgotten during planning
- There was still some room left next to the driver
- UCLp lens cracked with just slight (accidental) pressure from the bezel / head
- We haven't found a solution yet which would prevent this in the future
- Best LED died spontaneously during first test
- The second best one was just a tiny bit less good
- LED not perfectly centered on copper rod when turning it even though everything was done with a high degree of precision
- The centering of the body in relation to the head was not optimal
- UCLp lens has numerous small cracks on the edges and does not sit firmly anymore between the bezel and the head
- Here we also don't have a solution yet (when we do, I will order another lens)
- Tint of the LED turns very blue at 100% setting of driver which is a well known sign that it is getting too much current. When reducing the current just slightly the tint changes dramatically to a more normal cool white.
- Since this LED only gets brighter until 4.7A according to the above Test, we will reduce the current to 4.5A by switching out the sense resistor (the above candela values were measured at the maximum 4.7A with a lab power supply). Unfortunately the needed sense resistor (smd, 0805, 0.011Ohm, 0.5W) is currently hard to get here in Europe without paying horrendous shipping fees.
Since the light is now finished, it is time to show it off .
The classic size comparison incl. Mag 2D:
David and Goliath (especially amusing since the Zebralight actually produces more lumens):
I can't get enough of this reflector:
The light is very impressive outside. The beam has such a tight focus, extreme throw and an almost perfectly white tint. As a flashlight it is completely useless though, the spot is way too small to find anything. But then again that was never our intention. The only remaining application: cloud bounce!
I have already managed to do it.
A real beamshot comparison can be found here.
A few "fun pics" were also taken:
It really was an extremely extensive project! Even though we had to solve so many problems along the way, we still got an amazing, nicely working light as the end result. Because of the precision reflector and the extreme focus I think that in some ways it is a new category of LED thrower. I owe the whole thing to Michael who took on the project and saw it through the whole way!
Update January 15th. 2018: The light has been repaired & improved. It now has an even better LED which increased the luminous intensity to 1.7Mcd. Details can be found in post 114.
Additional pictures in posts: 124