Maximum safe temperature of housing- Mtbr.com

# Thread: Maximum safe temperature of housing

1. ## Maximum safe temperature of housing

How hot can the "typical" led bike light housing get before the led(s) are in danger of being damaged?

I pose this question because, in practice, most of us make some kind of judgement about whether the housing is "too hot", above which it is felt the led(s) are in danger.

But how hot is "too hot". A housing (or anything else) becomes too hot to comfortably touch at around 60 DegC, and there is probably an intuitive feeling that anything much above that is "too hot", but there is no scientific justification as such for that belief.

I am not asking what is the maximum desirable operating temperature for the housing and leds. That is a different question, and it is usually found that once the bike is moving, the housing stays not much more than warm, which we all agree is OK. However, if the bike is stopped for any length of time with the leds at full power the case temperature usually rises alarmingly, and it would be useful to have a scientifically based estimate of how hot the housing can get before the leds are in danger of being damaged. The maximum safe housing temperature can be estimated as follows.

The maximum semiconductor led junction temperature (Tj) is 150 DegC. For an MC-E or XR-E, previous threads showed that at maximum current, Tj will be around 28 DegC higher than the led heatsink point on the back of the led. In other words, at maximium safe junction temperature, the back of the led will be at around 122 degC. Previous threads also showed that the temperature difference between the back of the led and the mating surface of the heatsink/case will be around 8 degC, or perhaps even less if the led is epoxied directly to the case with a thin (<0.1mm) bond thickness, but let's be conservative and say 8 degC. However, we don't usually get to feel the case at that (hottest) point, so we need an estimate of the temperature difference from that point on the housing, to the exterior housing surface where we would normally touch it. This depends on the thickness and method of construction of the housing. However, some back-of-the-envelope calculations for a "typical" housing of reasonable design show that the outer surface of the housing will be somewhere around 12 DegC cooler that the point at which the led is mounted.

Adding up all these temperature differentials leads to the conclusion that the maximum safe housing temperature will typically be around 150-28-8-12 = 102 DegC.

For a really top-class housing, with top class mounting of the leds, this number could be improved to maybe 112 DegC, and of course, even with a perfect housing the number can't get better that (150-28) = 122 degC.

So, as a rule-of-thumb, with average housings of fair quality, you shouldn't need to worry about damaging the leds until the outer surface of the housing reaches around 100 DegC, which is pretty damn hot.

I am in the process of machining the housing for my dual MCE light, and will include small diameter thermocouple access holes so I can precisely measure the temperature at various points along the thermal path from led solder point, out to the outer housing surfaces, and I'll report on my findings. Until then, the above estimates should be something-like right, and better that sheer guesswork ...

2. My own non scientific rule is if it is too hot to touch then it is too hot inside.

There are some posts of mine where I have a temp probe drilled down just at the back of the Leds , with some basic data but I am rushing out now will try and find the posts later.

I try to put the Flex driver on the led heatsink as close as the design allows so it does the temp monitoring and go with the lowest setting that works well in the UK 50 degrees c works well but with higher ambients a higher setting would be better . or better cooling and more mass to give a buffer for the stops

I have never had any leds turn blue and die yet and it would be interesting to see what the casing temp gets to when they do give up but it is a costly experiment.
and then the driver might die first .

3. To me it seems to be a human factor issue. I don't want my light to burn me. That limits me to 70C or so.

4. My three MC-E's in 2p2s came in at 18.2V @ 500mA. I glued the LED's to the heatsink using Wakefield H-1518, I also used lots of pressure. I'm going to use a 5400mAH li-ion battery (8 cells in s/p). Driving the Maxflex with my power supply I get this current draw @ 16.8V:
level 1 = 150mA
level 2 = 250mA
level 3 = 500mA
level 4 = 900mA
level 5 = 1.45A
Reducing the voltage to 14V raised current 10-20%. So I'm gonna set the med-voltage alarm to 14.0 volts. The battery will be flat soon at this point.
In still air, starting with a case temp of 35C, overtemperature setting set to 70C (Maxflex is glued to case for HS'ing), the light ran for 3 minutes on high before the OT kicked in. At that time, case temp was 55C. After 10 minutes on reduced power (level 3) case temp had dropped to 54C. I'll tell you what though, you don't want to hold the case at 55C! I think it will be fine riding
Ok CC now, how's your analog driver coming along? I like your posts, don't fall asleep. lol James

5. Originally Posted by vroom9
To me it seems to be a human factor issue. I don't want my light to burn me. That limits me to 70C or so.
Seems like a reasonable approach to me!

But it is worth knowing that the leds are safe even for hand-frizzling case temperature up to around 100 DegC.

If you spit on the case and it sizzles, then time to turn the current down .....

6. Originally Posted by OldMTBfreak
My three MC-E's in 2p2s came in at 18.2V @ 500mA.

Ok CC now, how's your analog driver coming along? I like your posts, don't fall asleep. lol James
Hi James,

It's not clear to me how your 3 MC-Es are wired in 2p2s, but as the total voltage is 18.2V, it seems more like 2p6s. Doesn't the product of "P" times 'S" have to equal the total number of leds, which is 12. Is that 500mA per die within the MCE?

My MCEs have arrived, and measure Vf=3.15 at full rated current of 700mA per die, or 2800mA with all 4 dies in parallel. That is significantly less than the spec of 3.4V, which means they are better than spec. As the rated lumens is for a fixed current (350mA), a lower than spec forward voltage means the leds are more efficient than spec - I love it!

This also means that any fears I might have had about a linear regulator dropping out of regulation, with Ncells=Nleds, have completely evaporated. I will run 4 18650 cells in series (14.8V), with the MCEs in series parallel, 2p4s, so led voltage will be 3.15x4 = 12.6V. But here is the ultimate irony. With a forward voltage that low, I now feel inclined to go for a switching (buck) regulator, because the Vin-Vout differential is now too high for what I regard as acceptable efficiency from a linear regulator. All I can tell you for certain is this. Whatever home-brew regulator I eventually end up with, it will achieve >95% average efficiency over the discharge cycle. I used to design 5kW switching buck regulators for electric bikes (another hobby), and achieved 97% efficiency. I also design and build (1 to 5kW) ultra-high-efficiency brushless, ironless rare-earth DC motors (for bikes) and again, anything less than 95% efficient is not acceptable.

Just for now though, my attention is on building the dual MCE light. I spent the weekend machining the housing, and it is virtually finished - should mount the leds in the next few days just as soon as I machine the jig for locating and clamping the led while the epoxy cures.

I'm glad OldMTBfreak likes my postings, because he will be subjected to a few more once the light is completed and I take some measurements ...

7. A man after my own heart. No guts, no glory. lol Yep, 2p6s wiring. Each star wired 2p2s. I sure was happy that the Vf is low on the MCE's. That Vf measurement I gave you was 500mA per string, 1A total. Lower than the XRE's in another light (1A=13.6V). This lets the boost reg I'm using be more efficient. I am going to up the overtemp to 90C. I'll let you know what sort of case temp that'll be.

8. Originally Posted by OldMTBfreak
A man after my own heart. No guts, no glory. lol Yep, 2p6s wiring. Each star wired 2p2s. I sure was happy that the Vf is low on the MCE's. That Vf measurement I gave you was 500mA per string, 1A total. Lower than the XRE's in another light (1A=13.6V). This lets the boost reg I'm using be more efficient. I am going to up the overtemp to 90C. I'll let you know what sort of case temp that'll be.
I won't be responsible for any damage!!! Also, Troutie made the good point that the driver may die from oveheating before the leds, so keep an eye (or temperature tuned finger) on the driver.

9. Don't forget that when the LED temperature rises, even if it isn't cooking hot, its efficiency is dropping very fast.

10. Originally Posted by Calina
Don't forget that when the LED temperature rises, even if it isn't cooking hot, its efficiency is dropping very fast.
Absolutely true, and not a small effect. For an MCE, efficiency falls by 0.28% per DegC.

When the bike is moving and you need the light, then the lower the junction temperature the better, and hopefully way, way less than 150 degC. That is a different discussion, but equally important if we care at all about efficiency. As previously discussed, at full power the junction temperature will be roughly 50 DegC hotter than the exterior casing. Many report that the case temperature is warm to slightly hot when riding, which equates to maybe 40 DegC, which corresponds to a junction temperature of 90 DegC.

Assume in the worst case that the casing temperature rises by 50 Deg when stopped, corresponding to Tcase=90 and Tj=140 DegC. The loss in efficiency as a result of this rise is 0.28x50 = 14%, which is unlikey to matter when stopped, but we would not wish to suffer such a loss in effciency when the bike is moving.

An obvious question arises, as to how large and heavy we should make the light housing, so as to run cooler and increase efficiency and run time when riding? For example, by adding 50gms of metal to the casing and increasing the surface area, the light runs cooler, effciency is increased, and run time extended. However, that extra 50gms could also have provided one additional 18650 battery cell. If you started out with 4 cells (typical), then adding one more increases run time by 25%, but there is not a hope in hell that adding 50gms to the housing can provide a 25% increase in efficiency. You will soon figure out that OVERALL system performance is optimised by designing the housing to be pretty much as light as possible. Concentrate on designing the housing to be thermally smarter, not grossly heavier.

11. Awesome discussion on temps! I have built a couple of 4xpe lights running at about 500ma in a pretty standard solid machined round housing, and the housings get hot! I can hold my hand on them, but it is a little uncomfortable. But, I am in AZ and the outside temp when I started the ride tonight was 114f.... so it was really a torture test for the lights to see how hot they would actually get. Good news, nothing burned up so I should be good to go (and from this thread, I am thinking that I am waaaaay under the 100c casing temp....

CDCDCD good threads and thoughts on the lights the past couple of weeks... good DIY info and stuff to think about... w

12. Originally Posted by tamen00
Awesome discussion on temps! I have built a couple of 4xpe lights running at about 500ma in a pretty standard solid machined round housing, and the housings get hot! I can hold my hand on them, but it is a little uncomfortable. But, I am in AZ and the outside temp when I started the ride tonight was 114f.... so it was really a torture test for the lights to see how hot they would actually get. Good news, nothing burned up so I should be good to go (and from this thread, I am thinking that I am waaaaay under the 100c casing temp....

CDCDCD good threads and thoughts on the lights the past couple of weeks... good DIY info and stuff to think about... w
While it is true that up to 100 DegC casing temperature is unlikely to damage your leds, I just say again that my "100 DegC rule" is intended to apply to abnormal, infrequent, extreme events such as accidentally leaving the light on full power for an extended period when the bike is stopped, and is NOT recommended as an appropriate casing temperature during normal operation.

Personally I would aim to keep the JUNCTION temperature below about 100 DegC during normal operation, which means putting enough fins and surface area on the light to keep casing temperature in the range 45 to 55 DegC when riding.

13. Concentrate on designing the housing to be thermally smarter, not grossly heavier
.

Quite true but within reason .
again on CPF there was a very interesting post about removing metal from a maglite head and how it caused the led to run hotter but looking at it with the fins it looked like it would run cooler , so what looked like a good idea was in fact not
and there were thermal pics to back it up .
but there are no results for forced air cooling as in a bike light just handheld

I do try to make my lights as lightweight as possible and have had some crazy lights that if stationary for a few minutes would cook eggs on but at a slow riding pace are stone cold so it is the airflow that is the key and an acceptable mass to buffer the stop time incase you forget to dim it .

Lets see some pics of the work in progress

14. Originally Posted by cdcdcd
Personally I would aim to keep the JUNCTION temperature below about 100 DegC during normal operation, which means putting enough fins and surface area on the light to keep casing temperature in the range 45 to 55 DegC when riding.
Yeah, that would be nice to keep the casing temp on the lights in the 45 - 55 degc when riding, however the ambient air temp to start with was 45 DegC... I will post pics of the lights, however they seem to be performing fine and they do get pretty hot... much more so than I would like, but I see no way to avoid it.

FWIW... a light and motion Seca 700 gets much hotter than my lights and only runs for about 10 minutes while riding in this heat before it kicks into medium or low mode. I have a 12 hour night race I am testing them in this weekend, so it should be much cooler, I will see how hot they get on the long climbs without much air going over them...

I also agree with troutie... post some pics of the lights you have built and the designs... it would be very interesting to see them!

15. I did a little test yesterday and made a pair of very simple LED lights to light my drill press, micro lathe and also to help me see when soldering. They won't get air flow so wanted to see how hot it is safe to let them get.

I put a temp probe on the suface that the led was direct bonded to, no optic so was easy to place the probe.

The conclusion is don't sail to close to the wind as LED's will vary from batch to batch just like most things electronic these days. You could have recieved a good batch that dosn't mind being red hot or a batch that can't quite take it so hot and how will you know?

Anyway one died at 79 c the other 106 c One was from Lumitronix the other from DX (DX one took the most heat) both SSC P4 U bin.

16. Originally Posted by yetibetty
I did a little test yesterday and made a pair of very simple LED lights to light my drill press, micro lathe and also to help me see when soldering. They won't get air flow so wanted to see how hot it is safe to let them get.

I put a temp probe on the suface that the led was direct bonded to, no optic so was easy to place the probe.

The conclusion is don't sail to close to the wind as LED's will vary from batch to batch just like most things electronic these days. You could have recieved a good batch that dosn't mind being red hot or a batch that can't quite take it so hot and how will you know?

Anyway one died at 79 c the other 106 c One was from Lumitronix the other from DX (DX one took the most heat) both SSC P4 U bin.
That's rather disturbing in the light of my estimate that around100 DegC should be safe. I note theP4 has a lower max junction temperatureof 145 DegC, but that'snot enough to explain your observations. What kind of probe did you use,and how large was it? Are you sure the probe was at the same temperature as the surface you were measuring? The best method is to drill a hole into the surface being measured, and insert a small thermocouple or thermistor bead into the hole. How many watts were you inputting to the led? What kind of thermal epoxy did you use, and how thin was the bond thickness? I guess I'm hoping there was something odd about what you did or how you measured it, otherwise I can't explain your observations.

17. Originally Posted by cdcdcd
That's rather disturbing in the light of my estimate that around100 DegC should be safe. I note theP4 has a lower max junction temperatureof 145 DegC, but that'snot enough to explain your observations. What kind of probe did you use,and how large was it? Are you sure the probe was at the same temperature as the surface you were measuring? The best method is to drill a hole into the surface being measured, and insert a small thermocouple or thermistor bead into the hole. How many watts were you inputting to the led? What kind of thermal epoxy did you use, and how thin was the bond thickness? I guess I'm hoping there was something odd about what you did or how you measured it, otherwise I can't explain your observations.
Arctic Alumina was applied to the LED slug then held down with an old optic and light pteasure in a small vice, the probe was a tight fit in the hole that you see on second pic.and the probe has a small smear of Arctic silver 3 on it.

I think a lot depends on how the LED's are stored before you recieve them and maybe what they go through in the post ect. and perhaps their age.

Some data sheets mention baking LED's before use if they have had stoage in humidity.

Lets face it though 80 c is stupid hot for a housing (as I found yesterday) and you won't want anything that hot on a bike.

None of my bike lights go above 52 c as the bFlex sees to that.

I did the test for reasons of safety as I don't want to burn down my work area.

EDIT: Both LED's were run in series from a 1A (1000ma) buck puck and an 11.1v 2600mah Li-ion battery pack.

18. The probe

19. I live in NW Florida, normal day temps are 33-36C. The moving air has always cooled my lights ok. The light cases are cool to touch in the winter. I did get a kick (more like a burn) out of touching the light housing when it was 56C. I only have one ride (road ride) on my new light, But it doesn't overtemperture trip during the ride. 33C start temp at 9:00. The OT setpoint is nice to save the led's from damage. I reset the OT setpoint to 80C. I'll get some more measurments soon. One last thing though, adding mass mearly adds to the time for equilibrium. More fins would be cooler, just more mass, longer time to stablize. As I posted before, GOOD THREAD. James

20. Originally Posted by yetibetty
Arctic Alumina as aplied to the LED slug then held down with an old optic and light pteasure in a small vice, the probe was a tight fit in the hole that you see on second pic.and the probe has a small smear of Arctic silver 3 on it.

I think a lot depends on how the LED's are stored before you recieve them and maybe what they go through in the post ect. and perhaps their age.

Some data sheets mention baking LED's before use if they have had stoage in humidity.

Lets face it though 80 c is stupid hot for a housing (as I found yesterday) and you won't want anything that hot on a bike.

None of my bike lights go above 52 c as the bFlex sees to that.

I did the test for reasons of safety as I don't want to burn down my work area.

EDIT: Both LED's were run in series from a 1A (1000ma) buck puck and an 11.1v 2600mah Li-ion battery pack.
Hmmm. Can't see anything wrong with your measurement, except that it would have been nice to actually measure the 1 amp, but I'm grasping at straws in saying that. The only conclusion I can draw is that your SSC emitters do not meet specs re max junction temperature, at least for the one that failed at 79DegC case temperature. Maybe so, but the inevitable conclusion appears to be that it's best to be conservative and try to stay well below rated junction temperature at all times.

I have no experience with power leds, but plenty of experience with power semiconductors, and know from experience that silicon semiconductors are incredibly rugged and usually fail at junction temperatures well above published maximum of 150 or 175 DegC. A striking example is the common experience of observing power semis "unsolder themselves" without damaging the device ..... Not that anything I would design would do that, of course.

As far as I know the storage and baking of leds prior to intalling has to do with reliability of soldering, and the same advice is given for all surface mount components, so should have no relevance when epoxy bonding is used. If SSC have supplied product that does not meet spec they need a good kick up the arse, for there is little point in publishing specs unless the product conforms. Thanks for performing the test (at your expense!) by the way - very interesting.

21. If I remember correctly the recommendation for baking the LEDs is because companies are using soldering bathS and the high temperature is likely to cause water vapor bubbles in the silicone if the humidity level is too high.

22. Originally Posted by cdcdcd
Hmmm. Can't see anything wrong with your measurement, except that it would have been nice to actually measure the 1 amp, but I'm grasping at straws in saying that. The only conclusion I can draw is that your SSC emitters do not meet specs re max junction temperature, at least for the one that failed at 79DegC case temperature. Maybe so, but the inevitable conclusion appears to be that it's best to be conservative and try to stay well below rated junction temperature at all times.

I have no experience with power leds, but plenty of experience with power semiconductors, and know from experience that silicon semiconductors are incredibly rugged and usually fail at junction temperatures well above published maximum, which is usually 150 or 175 DegC. A striking example is the common experience of observing power semis "unsolder themselves" without damaging the device ..... Not that anything I would design would do that, of course.

As far as I know the storage and baking of leds prior to intalling has to do with reliability of soldering, and the same advice is given for all surface mount components, so should have no relevance when epoxy bonding is used. If SSC have supplied product that does not meet spec they need a good kick up the arse, for there is little point in publishing specs unless the product conforms. Thanks for performing the test (at your expense!) by the way - very interesting.
I enjoyed doing the test, it kept my brain cells active for half an hour and makes me feel safe knowing that I can leave my lights on in my very messy workshop (@ 500ma )without risk of fire, the LED's are only \$4 so quite a cheep price for peace of mind.

For me it has sadly just backed up what I have always thought and that is to take the spec sheet with a pinch of salt.

Strange how the cheep DX one took the most, those Lumitronix ones cost a fortune.

23. Originally Posted by tamen00
I also agree with troutie... post some pics of the lights you have built and the designs... it would be very interesting to see them!
I have taken pics of the machined housing for my new dual MCE light, and will post them in a thread ASAP, and when I work out how to do it. Some posts provide links to be clicked on to bring up a picture, others show the picture immediately which is preferable, but at this stage I don't know how to include either in a posting. I'd better try posting a picture on this thread first so I know how to do it. Have finished the led mounting jig - hope to mount leds today.

I'm also looking ahead to providing beamshots. Trouties beamshots are really good, and I'll try to find a similar kind of "real world" track with trees. Question for Troutie - how far away is the white-trunked tree in the distance, just to the left of the track? I'll try to find a similar tree at a similar distance.

What are the standard camera settings used by Troutie and others?

Need shutter speed, aperture and ISO.

Also need to know the focal length, expressed as an equivalent for a 35mm film camera.

24. Originally Posted by cdcdcd
I have taken pics of the machined housing for my new dual MCE light, and will post them in a thread ASAP, and when I work out how to do it. Some posts provide links to be clicked on to bring up a picture, others show the picture immediately which is preferable, but at this stage I don't know how to include either in a posting. I'd better try posting a picture on this thread first so I know how to do it. Have finished the led mounting jig - hope to mount leds today.

I'm also looking ahead to providing beamshots. Trouties beamshots are really good, and I'll try to find a similar kind of "real world" track with trees. Question for Troutie - how far away is the white-trunked tree in the distance, just to the left of the track? I'll try to find a similar tree at a similar distance.

What are the standard camera settings used by Troutie and others?

Need shutter speed, aperture and ISO.

Also need to know the focal length, expressed as an equivalent for a 35mm film camera.
Yes it would be nice to see pics of your new housing, simply click on go advanced, you will then find a paperclip for attatching images straight from your PC.

Camera settings are in the beamshot thread at top of page.

25. cdcdcd

there ar a few ways to post pics
1 as yetty says
2 the forum has a gallery which you can upload to and link to
3 and online photo host like Photobucket . which is my preference as casual browsers can see the pics , with option 1/2 you need to log in to see the pics

maximum size allowed on boards is 800 x 600

distances

camera settings

6 seconds @ f4
white balance set to daylight

26. I got some testing done today. Start temp 31C, OT trip in 5' 45", case temp 65C, OT set to 80C. Cools quickly at level 3 and will go back to high in 2 minutes. This is in still air. Ok, here's what's hapnin. This is at 16.8 volts input, Maxflex controller, stars wired 2p6s, levels 1 through 5.

Input Current Watts drawn Actual watts Vf Eff
.15 2.52 2.49 16.57 98.8
.25 4.2 3.89 16.9 92.6
.5 8.4 7.71 17.53 91.6
.9 15.12 14. 18.19 92.6
1.45 24.36 22.25 18.7 91.3

So, cd; here's a real world sample. Input current will rise at lower input voltages; about 20% greater at 14 volts input. I believe my 1st measurement on Vf was wrong. This should be accurate. Man, that case is HOT at 65C (150F). Since the case is red anno, I guess it's red hot! lol You can't touch it for long. James

27. Originally Posted by troutie-mtb
cdcdcd

there ar a few ways to post pics
1 as yetty says
2 the forum has a gallery which you can upload to and link to
3 and online photo host like Photobucket . which is my preference as casual browsers can see the pics , with option 1/2 you need to log in to see the pics

maximum size allowed on boards is 800 x 600

distances

camera settings

6 seconds @ f4
white balance set to daylight
Thanks. I'll try photobucket. I have registered with photobucket and uploaded a test image to them, but then what do I do ???

Re standardised beamshots, all shots really need to be taken at the same zoom setting or, to be more technical, at the same lens focal length, normalised to the equivalent focal length for a 35mm film camera. What camera do you use, what is the zoom range, and where within this zoom range do you take your standard beamshots?

Maybe I should start a new thread to discuss this and get it sorted. If photos are taken at different zoom levels, the result will affect the apparent width of the beam, as well as beam intensity recorded on the picture. Basically, if the camera is "zoomed out", the beam will appear narrower and more intense. Clearly then, it is a waste of time comparing beamshots unless they are all taken at the same zoom setting. My feeling is that Troutie's standard beam shots are about right, which is why I would like to know at what focal length (zoom) they were taken, so I can use the same.

28. Testing, trying to display an image ...

That's my long range offroad explorer, and a decent light would be useful. Full suspension, CVT transmission, 200-800mpg, 33cc 4-stroke, but of course I only use the engine for emergencies.

If this works, I'll post some pics of my new light.

29. Originally Posted by cdcdcd
That's my long range offroad explorer, and a decent light would be useful. Full suspension, CVT transmission, 200-800mpg, but of course I only use the engine for emergencies.
That thing deserves its own thread! Love it.

30. Just a quick update, ordered a few more SSC P4's from DX.
The run quite happy at almost 90c. Far too hot for me to ever touch the light to point it in the right direction so will run at lower current but I was interested to find out why I had odd results in the last test.

Must have recieved some poor quality ones before as these don't appear to mind one little bit, not even the slightest change in colour tint.

31. Good info on the temp of your workstation light, and good to know they are doing better than the last batch.

I found out this weekend the effects of ambient air temp on the lights. Since I built my last couple of lights, running either 3XPE or 4XPE at 500 mA, I have only had a chance to ride them on my road bike in AZ... ambient air temp for those rides has been between 42 and 46 degrees C... the lights get pretty hot even when moving... hot enough that you can hold your hands on the housing but it is a little uncomfortable.

Last night I did a 12 hour race that was at night, and in the mountains... ambient air temp was 18-24 degrees C, and the lights barely got warm to the touch, even after riding extended climbs very slowly... what a huge difference! Even feeding between laps when the bike was just sitting, the temp increase in the house was not noticeable. Very cool...

I was a little worried that the heatsinking that I have was not enough... but after these results I can see that the outside air temp really plays a huge role in the running temp, much more so than I would have guessed. If the outside temp was a little lower I could increase the current to the leds safely, but as it stands the lights work excellent and provide plenty of light for night racing.

32. Ambient temperature does have a huge impact on things! When I was back in Florida and the air temps would be in the high 80's or low 90's (deg F), cooling was a much bigger issue than it is here in Colorado with cooler night temps.

33. Originally Posted by znomit
That thing deserves its own thread! Love it.
Yeah! nice machine!

34. Originally Posted by yetibetty
I did a little test yesterday and made a pair of very simple LED lights to light my drill press, micro lathe and also to help me see when soldering. They won't get air flow so wanted to see how hot it is safe to let them get.
That seems like a really good idea for old leds. What drivers are you using?

35. Originally Posted by unterhausen
That seems like a really good idea for old leds. What drivers are you using?
That was the idea, to use up my oldies but as they are going cheep I've bought a few more.
I have a couple of 1000ma Buckpucks laying around that I could use but I will just run them from the mains with an old mobile phone charger or similar DC power supply and a resistor to keep the current correct.

On a slightly different theme, I converted our bathroom to all LED lighting a year ago as I got fed up changing blown light bulbs. I used these Elara strps and they are great http://www.ledsupply.com/elarastrip.php

36. I just added some plots of housing temperature versus time, at various currents, in the thread "FLATPACK - Dual MC-E design". I thought it best to keep all the data for this light in the same place in that thread, but mention it here cos it is relevant to the discussions on this thread.

37. Originally Posted by cdcdcd
I just added some plots of housing temperature versus time, at various currents, in the thread "FLATPACK - Dual MC-E design". I thought it best to keep all the data for this light in the same place in that thread, but mention it here cos it is relevant to the discussions on this thread.
cd, I hope you don't mind but I will stick the final results of my bike light test here, I can only do a very simple test but it could be interesting to see if you get similar(though more scientific) results.

I made a quad XRE a while back and although I like it a lot, it would get too hot for me, fine on the bike but I also use it for night walks and it was either too hot to hold or the bFlex would turn the power down.

So I reworked the housing and found that lots of very thin fins are a lot better than fewer thicker ones

With my lux meter and the light ponting at the same sheet of white card, I found that light output starts to drop when the housing reached 49c and at the same time vf would drop quite a bit also.(themal protection turned off for test) So although the LED's could take a lot higher temp there was no point running them higher than 49c.

After messing with the housing I can now leave the light on full (1A per die) in 23c ouside air temp and a very slight breeze and the bFlex doesn't trip (set at 50c). Obviously runs very cool on the bike.

38. Motorcycle manufactures also found in testing, that lots of thin fins were better than fewer coarse ones. Nice light Yeti, nice test. Where did you get your lux meter? I need one. James

39. Thanks, I have an "environmental"meter and there are a few around these days and are designed for the health and saftey bods to measure the work enviroment and they measure sound(db) as well as light(lux), temp, and humidity and it is all built into the multimeter. My brother works in health and saftey and tested it againts his \$\$\$\$\$£££££££ work provided meter and all is well.

I tested the electronics side of things againts my normal "electronics only" meter and they match up. So I think they are good enough for what we require. Just don't go dx style cheap.

If I didn't have the light meter my eyes would never have noticed the drop in output, these meters are very sensitive.

Also found that I play my music too loud.

My meter is from here but the site is bloody nuts! http://www.applegate.co.uk/company/12/05/761.htm

EDIT: this limk is better http://www.metrel.co.uk/

40. cd, I hope you don't mind but I will stick the final results of my bike light test here, I can only do a very simple test but it could be interesting to see if you get similar(though more scientific) results.

I made a quad XRE a while back and although I like it a lot, it would get too hot for me, fine on the bike but I also use it for night walks and it was either too hot to hold or the bFlex would turn the power down.

So I reworked the housing and found that lots of very thin fins are a lot better than fewer thicker ones

With my lux meter and the light ponting at the same sheet of white card, I found that light output starts to drop when the housing reached 49c and at the same time vf would drop quite a bit also.(themal protection turned off for test) So although the LED's could take a lot higher temp there was no point running them higher than 49c.

After messing with the housing I can now leave the light on full (1A per die) in 23c ouside air temp and a very slight breeze and the bFlex doesn't trip (set at 50c). Obviously runs very cool on the bike.
That's a very sweet light. Nicely made, and the final version with the many thin fins is pretty close to optimally designed for heat dissipation, I suspect.

I find your information interesting. What are the dimensions and weight of the final light? It's pretty impressive that it runs full current (1A/die) and still stays cool in still air.

If we assume Vf=3.2 per die at 1A, then at 1 amp you are dissipating 12.8W, for a temperature rise of about 25 DegC. The thermal resistance of your housing to still air is therefore 25/12.8 = 1.95 DegC/W.

My "flatpack" housing has a still-air thermal resistance of 50DegC/11.7W = 4.27DegC/W, which is twice as bad as yours! Your "slight breeze" would help, and your housing may (or may not) be slightly larger, but evidently your many long, thin fins are providing excellent cooling.

As per my thoughts on the flatpack thread, the flatpack design could be improved by having thinner and longer fins, but machining many long, thin, closely spaced fins on a mill is more work than with a circular design such as yours machined on a lathe.

As you realize, I'm interested in optimum design and quantitative comparison, so I'm interested to know the weight and dimensions of your quad XR-E light.

I'm also interested in using my light for more than just riding, and envisage that it will fit into a cradle/base for many fixed-use applications such as camping. In this role, cooling will be greatly enhanced by thermal transfer from the pivoting mounting points on each side of the housing, to the cradle/base assembly. In this case the battery will fit into the cradle, and the cradle will also have a handle so the light can be conveniently carried in one hand for night walks etc. Or, the cradle can be placed on any flat surface, and the "light-head" pivoted to throw light where it is needed. I have tried the light in still air in a roughly built cradle, and it runs cool. So it is interesting that we have similar applications, yet different designs, and I'm always keen to steal the best parts of other peoples designs!

41. cd, sorry for the very short reply but I have to get some sleep.

Hope these pics help and I will comment when I am awake enough to see!
I have included the size of the back plate as it is aluminium and is thermaly connected so must help a little.

Hope these help

42. Originally Posted by Calina
Don't forget that when the LED temperature rises, even if it isn't cooking hot, its efficiency is dropping very fast.
As you can see a nice 600lm light soon turns into a 450lm light if it get too hot....

K

43. Originally Posted by ktronik
As you can see a nice 600lm light soon turns into a 450lm light if it get too hot....

K
Ha ha, lucky me! But unfortunately I'm not brave enough to ride my bicycle when the temperature drops to minus 15 or 20 degrees C and the nights are long.

It is not that polar in the summer though, yesterday the thermometer hit 32 degrees.

#### Posting Permissions

• You may not post new threads
• You may not post replies
• You may not post attachments
• You may not edit your posts
•