deathshadow
Newly Enlightened
- Joined
- May 23, 2008
- Messages
- 5
Now, I'm not a nube to electrical circuits, but definately a nube to dealing with this much thermal energy as a result of using just diodes.
I'm starting out small with parts out of my junk bin before I sink any money into the 'good stuff' - I've got a pair of the 'allegedly' 28,500mcd 10mm ultra high brightness white from radio shack, a pair of the 7000mcd 5mm high brightness whites, four of the 5000mcd 10mm reds - and a variety of assorted red, green, yellow and tri-color in a variety of packages. (This is all leftovers from three of the Shack's 'grab bags') - and I'm planning on recycling a 7.2v 1500mah pack from when I still cared about RC cars. (nice since I already own it and the charger) - Still playing with the ideas for casing and form factor, though I'm planning to troll Wally world and Home depot tomorrow. I'm actually thinking on making the basic shape in a really thin/easy to cut plastic or metal (since my only real cutting tools are a set of tin-snips), then reinforcing it in plasti-dip. (assuming I can find anyone who still carries it locally)
In any case - the reason for my posting: I'm looking at a lot of the technology people are throwing at doing this, and coming up with a few questions.
Driver circuit - I can understand tossing in a conditioning resistor... even an absurdly low-pass one - I'm experimenting now with four 10 ohm to parallel sets of 2 in serial - which is giving me 3.1-3.0v per bulb at 21ma - The resistor seems to be enough so that the initial power on does not overvolt during that fraction of a second before the diodes... diode - but wouldn't a capacitor and a zener diode up-circuit from the switch be more effective at regulating this? I'm seeing some really complex circuits in use - given that my existing voltage would be so close to my target (I'm thinking four sets in parallel of 2 LED in serial) should I explore those further?
Also - has anyone tried doing a Bike LED using a Joule Thief type circuit? I was thinking on wiring in one in parallel to the regular circuit, perhaps using a zener or transistor with a refernce as the cut-in circuit or using a three position switch. Power dips below where it can keep the LED's lit, you use the Joule Thief to keep 'em going.
Oh, and those are 1 watt resistors I'm using. I calculated about 3/16ths of a watt thermal - so like any good engineer I doubled it... and then some. (mostly it's what I had in my parts box)
Collimators and lenses - While I could see maybe adding a lens to turn the circular output into a oval, illuminating a wider arc more brightly, most of what I'm seeing seems to be trying to catch back-light (of which most LED's have maybe 5% their total output) or just bend the arc of output wider. In testing I can barely tell the 28,500mcd's are on from behind, any light that does reach back seeming to come more from refractive effects inside the housing (which is a lens itself). The 'built in' lenses on the units I have seem to be putting out about 30-35 degrees for the smaller whites, and 10-12 degrees for the larger ones - are all these extra lenses, reflectors, housings and such more needed for the surface mount ones, or are people just getting greedy to milk every last ounce of light in the direction they want? I mean, the number one complaint about LED's in the consumer market is they are too directional, but the inclusion of reflectors would indicate they are not... Though - being white LED's are just a blue with a phosphor, is the phosphor emitting a wider arc, hence the need for the extra optics?
Cooling - I understand semiconductor cooling from a transistor standpoint (lord knows I've built enough two phase tesla coils), but have never dealt with LED's in an evironment where they would make enough heat for failure to even be considered... I'm looking at the materials some people are using to cool LED's, and shaking my head in confusion. With the surface mounts aluminum seems to be the weapon of choice, but as anyone who deals with IC's will tell you, aluminum SUCKS for transferring heat. I've even seen some people trying to sink to their housings in materials ranging from the exotic (titanium) to the mundane (cast iron) - The titanium ones likely never feel hot because they don't conduct heat worth a damn, burning up the lamps - while the iron or other heavy metals can store heat, but don't give it up easily enough to make a good heat sink (there's a reason skillets are made from cast iron - they heat up, then they stay hot a LONG TIME) - remember, many insulators can store energy for long periods of time.
With the type of bulbs (5mm and 10mm round) I'm using, I'm seeing people go nuts wrapping their legs in heat-shrink wrap, then flooding the area with artic silver adhesive... Not only is arctic silver sufficiently electrically conductive to be a concern (negligable does not mean 'not a problem'), it could induce capacitance... Worse, since you aren't making contact between the thermal adhesive and the connectors, the only place for it to try to pull heat from is the housing - which is a low -thermal conducting plastic.
Topping it off, much like regular arctic silver, it has a fraction of a percent the thermal conductivity of actual surface to surface contact. Arctic silver, both regular and adhesive, like most heat sink compounds is designed to just plug the microscopic holes between the two surfaces - you are not supposed to use it as a 1mm or more gap filler. Remember, a half ounce tube of arctic silver is supposed to be enough to do twenty or more computers. It's supposed to be a thin film, not half a tube for a 1cm square.
The material I'm planning to work with is obvious - a soft metal easy to work with, it's the #2 thermal conductor (and way more affordable than the #1) of choice in electronics... COPPER.
The thing is, since copper is also electrically conductive, I'm thinking on every two LED's (which are going to be in parallel) will share TWO separate copper sinks, with their anodes and cathodes silver soldered directly TO the heat sinks. We know silver conducts heat well, so this should be a much better approach, right? I'm going to use plumbing half inch 'straps' for the mounts separated by insulating foam - Though I'm worried that the two parallel sheets with foam between may act as a capacitor... I may attach copper 'VGA RAMSINKS' to them as well for extra mass, and more importantly extra surface area - since mass holds and transfers heat, surface area releases it - which is why REAL heat sinks are nothing but fins.
Watersealing - I'm seeing people going nuts waterproofing their battery box and light housings... while I can understand the concern, it seems like they are doing so at the cost of cooling and going a bit 'overboard' in terms of where they put their waterproofing... while often neglecting the most important weak point - the switch.
Electrical switches, especially in a DC circuit are wrought with problems. Arcing at voltage can often make it not turn off, and when liquid is added to the equation you can often build up enough capacitance to blow the whole circuit... if nothing else, water + metal switch === zap. One of the projects I saw online the guy flooded every corner with RTV sealant, but left a great big bat-handle metal switch facing UP on the panel he handlebar mounted. Needless to say, I'm going to a marine supply place tomorrow to find a waterproof switch.
Shouldn't sealing be applied to JUST the exposed circuit areas and not the whole housing? Shrink wrap and a ounce of RTV, not a whole tube flooding the device?
Likewise, I'm planning on cutting ***SHOCK*** VENT HOLES in my housings. Resistors make heat, lighting LED's make heat - where there's heat, airflow is your friend. I'm seeing 700 lumen projects where people are sealing the entire assembly air-tight. I don't care how much heat sink you are putting in there - no airflow == overheat. (I'll probably make a scoop for the bottom and vents on the top)
------------------------------------------
In any case, I plan on buckling down and starting my build either tomorrow night or sunday afternoon - when I do I'll photodocument as I go so you guys can laugh at me I can solder, I can make my own PCB's - but my skills at things like housings SUCK! (I might downgrade to using balsa as a form to do a resin casting - that I can do)
I'm starting out small with parts out of my junk bin before I sink any money into the 'good stuff' - I've got a pair of the 'allegedly' 28,500mcd 10mm ultra high brightness white from radio shack, a pair of the 7000mcd 5mm high brightness whites, four of the 5000mcd 10mm reds - and a variety of assorted red, green, yellow and tri-color in a variety of packages. (This is all leftovers from three of the Shack's 'grab bags') - and I'm planning on recycling a 7.2v 1500mah pack from when I still cared about RC cars. (nice since I already own it and the charger) - Still playing with the ideas for casing and form factor, though I'm planning to troll Wally world and Home depot tomorrow. I'm actually thinking on making the basic shape in a really thin/easy to cut plastic or metal (since my only real cutting tools are a set of tin-snips), then reinforcing it in plasti-dip. (assuming I can find anyone who still carries it locally)
In any case - the reason for my posting: I'm looking at a lot of the technology people are throwing at doing this, and coming up with a few questions.
Driver circuit - I can understand tossing in a conditioning resistor... even an absurdly low-pass one - I'm experimenting now with four 10 ohm to parallel sets of 2 in serial - which is giving me 3.1-3.0v per bulb at 21ma - The resistor seems to be enough so that the initial power on does not overvolt during that fraction of a second before the diodes... diode - but wouldn't a capacitor and a zener diode up-circuit from the switch be more effective at regulating this? I'm seeing some really complex circuits in use - given that my existing voltage would be so close to my target (I'm thinking four sets in parallel of 2 LED in serial) should I explore those further?
Also - has anyone tried doing a Bike LED using a Joule Thief type circuit? I was thinking on wiring in one in parallel to the regular circuit, perhaps using a zener or transistor with a refernce as the cut-in circuit or using a three position switch. Power dips below where it can keep the LED's lit, you use the Joule Thief to keep 'em going.
Oh, and those are 1 watt resistors I'm using. I calculated about 3/16ths of a watt thermal - so like any good engineer I doubled it... and then some. (mostly it's what I had in my parts box)
Collimators and lenses - While I could see maybe adding a lens to turn the circular output into a oval, illuminating a wider arc more brightly, most of what I'm seeing seems to be trying to catch back-light (of which most LED's have maybe 5% their total output) or just bend the arc of output wider. In testing I can barely tell the 28,500mcd's are on from behind, any light that does reach back seeming to come more from refractive effects inside the housing (which is a lens itself). The 'built in' lenses on the units I have seem to be putting out about 30-35 degrees for the smaller whites, and 10-12 degrees for the larger ones - are all these extra lenses, reflectors, housings and such more needed for the surface mount ones, or are people just getting greedy to milk every last ounce of light in the direction they want? I mean, the number one complaint about LED's in the consumer market is they are too directional, but the inclusion of reflectors would indicate they are not... Though - being white LED's are just a blue with a phosphor, is the phosphor emitting a wider arc, hence the need for the extra optics?
Cooling - I understand semiconductor cooling from a transistor standpoint (lord knows I've built enough two phase tesla coils), but have never dealt with LED's in an evironment where they would make enough heat for failure to even be considered... I'm looking at the materials some people are using to cool LED's, and shaking my head in confusion. With the surface mounts aluminum seems to be the weapon of choice, but as anyone who deals with IC's will tell you, aluminum SUCKS for transferring heat. I've even seen some people trying to sink to their housings in materials ranging from the exotic (titanium) to the mundane (cast iron) - The titanium ones likely never feel hot because they don't conduct heat worth a damn, burning up the lamps - while the iron or other heavy metals can store heat, but don't give it up easily enough to make a good heat sink (there's a reason skillets are made from cast iron - they heat up, then they stay hot a LONG TIME) - remember, many insulators can store energy for long periods of time.
With the type of bulbs (5mm and 10mm round) I'm using, I'm seeing people go nuts wrapping their legs in heat-shrink wrap, then flooding the area with artic silver adhesive... Not only is arctic silver sufficiently electrically conductive to be a concern (negligable does not mean 'not a problem'), it could induce capacitance... Worse, since you aren't making contact between the thermal adhesive and the connectors, the only place for it to try to pull heat from is the housing - which is a low -thermal conducting plastic.
Topping it off, much like regular arctic silver, it has a fraction of a percent the thermal conductivity of actual surface to surface contact. Arctic silver, both regular and adhesive, like most heat sink compounds is designed to just plug the microscopic holes between the two surfaces - you are not supposed to use it as a 1mm or more gap filler. Remember, a half ounce tube of arctic silver is supposed to be enough to do twenty or more computers. It's supposed to be a thin film, not half a tube for a 1cm square.
The material I'm planning to work with is obvious - a soft metal easy to work with, it's the #2 thermal conductor (and way more affordable than the #1) of choice in electronics... COPPER.
The thing is, since copper is also electrically conductive, I'm thinking on every two LED's (which are going to be in parallel) will share TWO separate copper sinks, with their anodes and cathodes silver soldered directly TO the heat sinks. We know silver conducts heat well, so this should be a much better approach, right? I'm going to use plumbing half inch 'straps' for the mounts separated by insulating foam - Though I'm worried that the two parallel sheets with foam between may act as a capacitor... I may attach copper 'VGA RAMSINKS' to them as well for extra mass, and more importantly extra surface area - since mass holds and transfers heat, surface area releases it - which is why REAL heat sinks are nothing but fins.
Watersealing - I'm seeing people going nuts waterproofing their battery box and light housings... while I can understand the concern, it seems like they are doing so at the cost of cooling and going a bit 'overboard' in terms of where they put their waterproofing... while often neglecting the most important weak point - the switch.
Electrical switches, especially in a DC circuit are wrought with problems. Arcing at voltage can often make it not turn off, and when liquid is added to the equation you can often build up enough capacitance to blow the whole circuit... if nothing else, water + metal switch === zap. One of the projects I saw online the guy flooded every corner with RTV sealant, but left a great big bat-handle metal switch facing UP on the panel he handlebar mounted.
Needless to say, I'm going to a marine supply place tomorrow to find a waterproof switch.Shouldn't sealing be applied to JUST the exposed circuit areas and not the whole housing? Shrink wrap and a ounce of RTV, not a whole tube flooding the device?
Likewise, I'm planning on cutting ***SHOCK*** VENT HOLES in my housings. Resistors make heat, lighting LED's make heat - where there's heat, airflow is your friend. I'm seeing 700 lumen projects where people are sealing the entire assembly air-tight. I don't care how much heat sink you are putting in there - no airflow == overheat. (I'll probably make a scoop for the bottom and vents on the top)
------------------------------------------
In any case, I plan on buckling down and starting my build either tomorrow night or sunday afternoon - when I do I'll photodocument as I go so you guys can laugh at me
I can solder, I can make my own PCB's - but my skills at things like housings SUCK! (I might downgrade to using balsa as a form to do a resin casting - that I can do)
Last edited:
