Another Marwi or Performance Viewpoint conversion

[BrianMc]

Here is my first light that wasn't an MR16 plug-in DIY. The second will be the high beam and awaits results from aspheric lens tests.







Thanks to all those who had input.

Sorry for the cell phone pics. When I sort the camera for the beamshots, I will have room on the roll for better. (Yes, film.)

The heat sink is the pill and liner one I posted in this thread.

The intention was to use the Ledil Lisa2 RS XPG lens which at the time was listed as 11 degrees FWHM. I got the Carclo 10417 lenses in case I wanted a broader beam for my low beam. The Cute triple and triple MCPCB would have been a better choice though possibly a tight fit, had I known the Lisa2 RS XP-G lens was really 26 degrees FWHM. Oh, the bleeding edge of technology!


The lights and mounts were from Hoffmanamps (EL34). The driver is a Bflex from Taskled. Cutter supplied the light engines and optics. Batteryspace: Trailtech cables. Battery is a 12 volt 4.5 A-hr now 4 A-hr NIMH water bottle for the Cygolite Z Force 14 W HID.

The bulb covers were removed for use, and the bulbs, sockets and on/off switches removed. The momentary switch from the Bflex driver is now under the modified push button cap. A lens from a PC status led was shortened and fit into a hole drilled at the back tip of the bullet shape. The Bflex was mounted on the backside of the pill, and the three 20 mm MCPCBs on the front using AA paste. The pill was coated in Arctic Aluminum thermal paste then inserted in the liner matching the keyed ridge/valley and the liner forced into the shell which had been polished to the metal, then treated with AA. A tight fit: gentle persuasion with wood and a hammer was needed. There is barely room to get the wires in with 8-10 mm allowed for the lenses and 10 mm between the LED mount and the Bflex.

The body warmed up to just above body temp when I soldered the LED leads, so I think that means I have a good thermal path.

It was tested at the default 350 mA on a day ride at 4C. Cold as the bike frame after an hour, as it should be. After an hour of playing with the setup at room temps, (and playing with the high setting) it was lukewarm. More impressions when the beam shots and test rides are done. But it is amazing compared to a 3 watt luxeon (100 lumens in a 10 degree beam versus nearly 800 on 16 to 20 degrees. Nice rush. Hooked now, I suppose. Wanted to ride, but it's very late or awfully early.

 

[BrianMc]

Snow here. So no beamshots yet.

Wind dropped. So I did a test ride with the light. Bad icy patches so I kept it short: 2 miles.

Bflex on Duo Mode high level 1 A, low 0.5 A. Ambient temp 23 F, -13C. After ride, I did a hand touch temp comparison to the steel frame. They felt the same to me. Being so cold, the Triple XP-G's might have put over 1100 lumens on high, and 600 on low, out the front tonight.

Beam pattern: I am getting decent spill from left berm to the right ditch/lawns on narrow suburban streets. Main beam down the road is similar. Maybe I'd like less to the side and more in front, but on high I don't feel short of light on these streets at my speeds.

Much of the near spill appears to be light reflected off the copper heat sink. It seems warmer. Need a camera to be sure. Mounted this close to me, the spill affects my night vision. Bouces of tire and fender, too. It needs a light bar to put it out front more.

Impressions in 'traffic': two cars:
I got a ton of respect and sharing the road and actully stopping at the stop they roll through! I can't tell if the light helped as my riding after a blizzard is super weird here, and I could have scared them. I also increased my visibility a lot beyond the new light: clothes, tape, other lights.

I set the bike up against the Basketball (State Bird of Indiana) hoop post at the top of my drive with the light in either mode. Then walked from the neighbors, 250 feet (75 m) away, and 6 feet (2 M) lower, against it, both straight on, and to the right. The low is no issue at all, the high, isn't pleasant, but not blinding. I will redo as if driving a low sports car to be sure. After the motorist glare test, the light felt like it might be a bit warmer than the bike, but not much if it was.

Thermal management test: brought it in from 26 F (-13 C) to 70 F (21C). Left it on high for over 30 minutes. It took 20 to not feel cold (up to room temp). In another 10 minutes it warmed to just under luke warm. The Bflex thermal protection is programed at 50 C to power down. No response. I don't have an IR thermometer, but since I can access the heat sink and the MCPCB's easily I'll try a repeat with a plain old thermometer and see if I can verify thermal management.

All that copper I used should help a lot but if it sounds too good to be true....

 

[BrianMc]

Revisited testing thermal management success. After eleven minutes at 1 A, the heatsink temp went from 70 F (21.1 C) to 108.2 F (42.2 C) The body read 105.7 (40.9 C). So after 11 minutes of warming the temperature differential was about 2.5 F (1.4 C) between the MCPCB/heatsink and the body. If there was significant resistance to heat flow the difference would be bigger. :twothumbs

In fact, it appears that the ceiling fans in the room where the previous test was done, though on low and the bike at the wall and not in the direct flow, were enough to keep the light just below body temperature (just under luke warm) after 30 minutes. I don't think there are many days when the air is THAT still. :shakehead

I'd call that good thermal management. Probably could handle more XP-G's. All that copper made the light heavy for its size, though. No free lunch.
:sigh:

 

[Uzzi]

How thick is your copper heat sink? I just made a couple from 1mm pipe that fit in the front section and they are quite light. I made one from thicker pipe for the back plate to check if there was a difference. I wrapped and soldered a 5 mm wide ribbon of copper pipe around the circumference of the 34mm round plate that fits snugly on the housing.

I think that at the heat wattage I'm working with (3-12W), that there is enough conductivity to shed the heat. I think of it in terms of voltage wattage; if I think that a certain amount of voltage will short out the wire I'll use a thicker wire. I think that with an aluminum heat sink it would be more crucial to use thicker material, but with copper you can go fairly thin since its a highly conductive material, thus making a lighter heat sink.

 

[BrianMc]

How thick is the heat sink? Depends on where. I did go a bit overboard with this. I am making a lighter liner-less version for possible helmet use. Probably could have CNC'ed several Al bodies with the time this took. Nothing excels like excess.:twothumbs

Originally, the LED's were to be 5 mm from the edge of the pill, and maybe one in the center. Because of a change in lenses, the LED's are mounted so they are 7 mm from the sides of the pill. Directly beneath them is the thickness of teh MCPCB and the pipe cap end minus polishing 1/16" (1.5 mm), or close enough. Under the 2mm of MCPCB nearest the outside of the pill, the copper and silver solder extends 9/16" (14 mm) to the rear, and does so for another 1/8" (3mm) then the outside of the pill which is 1/32" (<1 mm) longer than the bushing insert. The pill tightly contacts the liner that tightly contacts most of the inside of the aluminum housing with AA to help out.

In case that is as clear as mud, here is the heatsink setup:

URL=http://img209.imageshack.us/i/light1a.jpg/][/URL]

In case you want to know why and how:

Why:
The aluminum body of the Marwi/Viepoint Bullet light is fairly thin, and very thin where the inside cylinder shape meets the inside cone 1" (25.4 mm) from the front threads. So heat transfer laterally in the body will be reduced compared to the thick-walled CNC housings designed by others here. There is minimal body area for heat transfer to the air (about 7 square inches, or 45 cm squared) in these MR11 housings. I wanted to use all of it as much as possible. The low beam is intended to be at 1A all the time (about 8 watts of heat to dissipate) if a beam with good cutoff is developed. Seven could be stuffed in and at 0.5 A most of the time plus a Maxflex and there would be 10.5 watts of heat. A short burst on 1A would be almost 20 watts of heat! Air flow helps, but you need to get the heat to the air interface as fast as possible. So a copper liner to spread the heat and increase the thermal path into the body was devised for this all out thermal management light, but a simple pipe cap heat sink for the other.

How I made the liner:
The copper liner is a 1.5" (38 mm) Mueller pipe union with a 1/4" section removed lengthwise to allow constriction to reduce it's diameter. This makes it a spring and it clamps outward with enough force to make assembly a bit difficult even with the inside of the body polished to remove the anodizing. A 1/2" (12.5 mm) section was removed from the front end. Cut outs on the back acccomodate the mount/tail cap, and fingers were formed and bent in, ground and polished to fit the conical rear of the light body and about 1/4" was removed from this end, too. Both the inside and out were sanded and polished reducing thickness to about 1/32" from 1/16". I'd guess it at less than 30% of it's original mass.

How I made the pill:
The copper pill is a 1" (25.4 mm) Mueller pipe shortened to 3/4" (22 mm) reduced in wall thickness on the outside only to about 1/32, lapped on the face to 1500 grit . This surface was thinned no more than needed to flattenand polish. In fact, about half of the stamped-in Mueller 'M' remains. A groove was made to match the union's ridge and lock the pill in place in the liner. The cap part of the pill was reduced to about 60% of its original mass. There was a lot sanding then fitting of the liner and pill in the housing and repeat.

I envisioned the 10 mm round MCPCB with Lisa2 RS lenses and holders mounted as near the outer edge of the pill as possible, reducing the thermal path length to the body.

A 1" x 3/4" (25.4 mm x 22 mm) copper bushing, cut to half length was soldered inside the pill because the locking groove almost cuts the pill in two, it would reduce hot spots, and add a short-term thermal mass heat sink for pulses of higher output from more LED's, if I wanted to add more later. This bushing has a 1/8" (3mm) thick wall. It's one end would be directly behind the outer MCPCB's. With silver solder, it at least doubles the weight of the pill. The back end is the shoulder that the Bflex is mounted on and because the pipe cap part of the pill extends just a bit further, there is a nice lip. It also provides a 10 mm deep air pocket inside the pill to isolate the Bflex from the thermal path but bathe it is the temperature of the light's interior and so, infer that of the LEDs. If a MaxFlex is fitted later, the heat sink boss for the thermal pad will be soldered to this chunk of copper. If doing it again I would use a shortened 1" copper pipe clamp from Mueller bent to fit. They are 1/2" (13 mm) wide and only 1/16" (1.5 mm) thick, saving half the weight of the bushing and still reinforcing the pill wall and thermal path along the pill sides.

Not as heavy as I thought, but no lightwieght, either:
I have no scale, but used a makeshift see-saw balance to get the crude mass. The entire light head and it's part of the QR mount has the same mass as a Duracell D alkaline battery. According to a Harvard pdf file that should be about 125 gm, but even if the Duracell is a beefier 150 gm, this is much better than I thought by its heft (the small size fooled me).

Performance:
By touch, the circular area around the light corresponding to the pill sides was detectably warmer than the area just behind the bezel or at the tail in front of the mount, but not by much. These areas were still above skin temp: they felt quite warm, just not as warm as that band where the pill is. That is as good as could be hoped.

Future:
I can fit only 3 of the aspheric lenses I am playing with in this light head. So no compact 7-up at low power for same output or a bit more at higher max run times, with the chance for short bursts of 'Turbo mode' on fast descents, until we have smaller narrow optics or XR-E sized LEDs that efficient.

 

[BrianMc]

Ran it for 30 minutes in a drawer at room temp. Back end was 105.3 F, (41 C) and the front just ahead of the pill was 107.8 F (42 C). The middle was above 108 F max the thermometer will output, but not by much as it did not feel much hotter than the front: about 110 F. If the difference measured before holds, then the MCPCB was less than 115 F or 46 C at room temp after 30 minutes of static testing. Excellent.

There is no way I am standing at at traffic light for 30 minutes with the light on high. The chance of no breeze of any kind are almost as slim.

Troutie reported that his 7-Up 'Not Square-Not Round' got warm to touch on full at a dog sled race (where ambient temps would be a lot lower than room temp) and that light is about 150 grams. An equal weight of copper is not as good a static heat sink, but maybe I can cram in 7 XP-G's to run at half power or less most of the time (with the right optics) and have enough reserve thermal capacity for max output on fast descents. I'll worry about that if and when the optics arrive.

I wasn't expecting it to run this cool. Apparently getting a short fast thermal path to the air compensates well for the lack of radiating surface. My hunch confirmed.

 

[BrianMc]

This is a small MR11 housing with no cooling fins. So I took a time-line on temperatures at points on the light.

The top is 1-2 degrees C warmer than points on the sides halfway beteween the shortest thermal path from the LEDs. So I took the readings down the top in line with an LED.

My nice Excel chart jpg won't go into Image Shack. And posting makes a hash of the table.

Room Temp was 23 C. The MCPCB maxed at 115 F or 46 C. The Body's hottest spot was only 2 C cooler, and the coolest part of the body was only 8 C less than the MCPCB indicating good heat flow.

The Junction T at equilibrium was calculated to be about 66 C. So it looks like I can ride at night at 115 F or 46 C and not have to worry about burning up the LEDs! If the nights get that hot here, there'll be lots more important things to worry about. The Blfex is set at 50 C and it was not triggered. If I get a job in the Australian outback, I should be good to go.

Only about 1" /watt surface area, and no fins. I guess getting that heat to the body ASAP works.

 

[BrianMc]

Refreshing this thread.

I have two key fob cameras I am testing to use for light videos. (http://www.buyincoins.com/key-chain-spy-camera-camcorder-dv-dvr-tf-card-product-1311.html).

They appear to be Type 8 (http://chucklohr.com/808/) not quite as good a video quality as the Type 3. They were purchased to test rear blinky lights, but I plan a new ride and record of my lights because I have altered both beams. Although these cameras aren't as good under low light as the Sonycam, the difference is small, and their resolution through to a captured video is much better, and I don't have to hand-hold or tape them to the bars as I did with the Sony.

So I wanted a daytime run around the same route as for the night, to test through to a video playing here, and to compare to the night ride.

They are small enough to go unnoticed if you wanted to record a commute or trail ride of an hour or less unobtrusively.

The front camera was mounted on the tip of the front fender, so cantilevered out from the fork crown mount, and it has less road vibration since the wheel with 32 mm trire at 60 psi, fork, and fender absorb a lot. The rear camera is on the top of the rear fender and the stays move the fender with motion at the rear dropout, so the road vibration is more noticeable. These videos have only been cropped at each end, and compressed for a High Speed Internet connection. No smoothing or other processing was done. The raw files are 640 X480 expanded to 720 X 480 and the compressed files here is about 4% as large, so some quality was lost. I have not reset the Date/Time stamps on either camera, yet.

Front:

<embed src="http://img706.imageshack.us/flvplayer.swf?f=Pfrontview" width="320" height="260" allowFullScreen="true" type="application/x-shockwave-flash"/><br/>

Rear:

<embed src="http://img87.imageshack.us/flvplayer.swf?f=Prearview" width="320" height="260" allowFullScreen="true" type="application/x-shockwave-flash"/><br/>

I will edit to add the front and rear night run. If this works this would be an inexpensive way to show your bike lights in action. Yea, an HD camera at $300 would be excellent, but...good enough is good enough.

 

[tandem]

Coming Soon From Key Fob Pictures...

Rear View From A Bicycle
Staring Brian Mc and introducing A. Bicycle delivering its Premier Performance

A touching story about a man, his bicycle, and the roads they travel together. Director and leading actor Brian Mc take us on a whimsical journey through a rural land. Unplagued by unnecessary dialogue this powerful tour de force is destined to become a cult classic.

Later this fall look for the thrilling sequel, Rear View From A Bicycle, In The Dark!

Warning: Those sensitive to motion related nausea may find this trailer unsettling. Repetitive viewing may lead to hair loss, skin mottling, and groin itch. Mime approved.

(I can't wait for the sequel with lights!)

 

[BrianMc]

Thanks Tandem. If I need a marketing agent...
I debated music but this was really a test. It also showed what the microphones picked up for those interested in the cameras themselves. If I get bored and tweak them, more refined versions will replace these.

File size is a HUGE problem. When I filtered some of the noise and vibration the file got even bigger. Tens times the size of the one I uploaded. There is a filter add-in that allows the frame to float in a larger black frame, but this video editing when you have to learn everything beyond splicing, eats up so much time! Back to good enough and have some Gravol.

Got more experience with the cameras last night. (another way of saying I did not get what I wanted.) Apparently the unlit suburban route I planned and the light from the bike lights is WAY too little light for these cameras. They appear to pool frames until enough light is gathered to trigger a frame save (increase exposure), so lit street signs are a smudged bouncing line, and only the brightest part of the light patches show up on a black background. Even more boring (if you can imagine that) just a red fuzzy patch with occasional red fuzzy blurred signs, or an elongated double white patch with occasssional 'speed striped' signs and fuzzy ball-stripe lights on black. As if Modrian decided to do organic forms instead of rectangles of solid color. They work better than my other choices on the lit street. So they are still on deck for the blinky test. They were not good on a prior helmet test night ride, but I blamed my moving head for increasing the data differentials between frames. Partly true, but mostly low sensitivity.

Enter plans B & C. Plan B: The Handicam showed most of the beams in my last attempt hand holding it and guessing where to aim with the high beam on the helmet (resulted in me aiming both the camera and the helmet light a bit low most of the time):

<embed src="http://img532.imageshack.us/flvplayer.swf?f=Pcap0003" width="320" height="260" allowFullScreen="true" type="application/x-shockwave-flash"/><br/>

The pavement details show as does the green grass, something way beyond the key fob cameras' low light capabilitites. This is of the old beams, so should be here as a handy reference anyway.

So duct taped to the seat rails for the rear loop and to the handlebar bag cradle for the front pass, I may get 'good enough' video to show what I see (or close enough with a little imagination).

Plan C: see if a friend who shoots video professionally has an affordable (hopefully free) solution I can borrow for an hour.

A bit like DIY lights, if first you don't succeed... Too bad about the low light sensitivity as these would have made a great way to get riding beamshots affordably. The technology WILL catch up.

Episode 5: Return of the Handicam.

 

[BrianMc]

Learn to do by doing - the 4-H Motto - has its moments.:thinking: (Junior High Sex Education probably isn't one of them if your daughter's in the class, so there are exceptions.)

Learned that the Sony Handycam import software is poor. Can you spell 'POS'? So the output is a lot better now. I will retry it with the blinkies, too. The Handycam has some built in vibration removal (thank goodness, can't imagine it worse) and I did not opt to spend the time to smooth it more with processing.

So here is Night Ride -The Musical, rolling around my neighborhood. :thumbsdow

I brightened it to match what I see as best I could to make it a reasonable facsimile of the real ride. The far end of the beam is easily seen by eye as is a greater spill, but the camera doesn't pick either up. At the top of the hill with the narrow bridge markers near the end of each loop, I could see the signs at the next intersection about 1/4 mile away, they show a bit in the 1 A loop until I drop the beam, I think, but not so well in the 0.5 A loop in the video.

The oncoming car crouded me a bit with over 2' on the right side of the car, but the driver did not drastically reduce speed and head for the ditch from being blinded, which was the previous reaction by every motorist to the old low beam. He gets more spill the closer he gets. A nice kid, but not the best driver of his age I've ever seen.

I dropped the high when I saw the glow of his lights (before the camera picked it up even with brightening BTW, so I may still have it a bit dark) by tilting it down and forgot to move it back up. So you can see the stacked square beams (projected LED dies) that make it up starting about 30 feet in front of the bike. This time I may have had the camera a bit ihigh as I missed 75% of the low beam hot spot. No biggie. 'F-you've seen one oval hot spot, seen 'em all. The helmet mount makes this aiming second nature, but puts the beam out of line with the camera on turns and adds a lot of queasy motion. A post credit piece shows a bit more than 1/4 mile of tail lights. Sometime you get more than you want or can stand. :naughty:

<embed src="http://img824.imageshack.us/flvplayer.swf?f=Mheadlightridethemusi" width="640" height="380" allowFullScreen="true" wmode="transparent" type="application/x-shockwave-flash"/>
Uploaded with <a target='_blank'

Enjoy. :sick2: Some say it's better with eyes closed. :sigh:

 

[BrianMc]

When you narrow the beam of a headlight you can reduce the spill so much that it is not too effective as a "be seen" light. However if it is powerful enough, that narow beam widens with distance and should work.

It is a good idea to check it out, though. So I have a short video from two of the return trips doing the blinky test. Note I have not done anythig with the date time stamp. Note too thet the high beam is on the handlebar for this when it is meant for the helmet so it can be aimed at cross traffic drivers thinking of hooking in front of me. Also the low beam can be tilted up and aimed left to increase its brightness. The videos indicate they are readily seen in the day from 1/4 mile away, though small and obviously, not overly bright. Of course, now I have these sorted, I am thinking of a generator hub and light system for the errand bike...
lovecpf

Have to look ove the Phillips light, somrthing about a generator version...

Anyway, this completes the tweaking and reporting.

<embed src="http://img233.imageshack.us/flvplayer.swf?f=Mheadlightsday" width="640" height="380" allowFullScreen="true" wmode="transparent" type="application/x-shockwave-flash"/>
Uploaded with <a target='_blank' href='http://imageshack.us'>ImageShack.us</a>

 

[BrianMc]

Got some decent lit street video so made this to show the reduced glare using the fresnel lens and shade, enjoy:

<embed src="http://img820.imageshack.us/flvplayer.swf?f=Mbeam" width="640" height="380" allowFullScreen="true" wmode="transparent" type="application/x-shockwave-flash"/>
Uploaded with <a target='_blank' href='http://imageshack.us'>ImageShack.us</a>

PS My gapper forgot the dern tape so the camera prop to make it not level so the picture would be, slipped out, oh well, it would only be another hour riding maybe with the second snow this fall falling. Not wet enoug until it landed to be sleet, last night. Weird. Over 90 on Sunday and snow last night.