Citivolus
Enlightened
Something that has been nagging at me for a while is runtime charts. Some reviewers do them with a fan, and some without. This can have a very dramatic impact on the graphs and even the run times, as some drivers use thermal regulation to keep the head temperature down. This regulation can reduce their current draw, increasing their overall run time on a charge while at the same time decreasing net light output.
Justification that I have seen given by some for not cooling during testing include speculation that the human body is not a good heat sink, while others say that it is. I figured I would try to find out for myself, to see if I should really be testing with a cooling fan running. I tried to set this up in a way so as to not influence the results.
I took a Dereelight CL1H, my favourite test subject of late due to the developing mythology around it, and attached a thermocouple to the bezel. The thermocouple was attached on the outside where the pill makes contact, and held in place using a thick elastic band which also prevented the thermocouple from directly reading air temperature. The thermocouple was located about 1cm up from where my index finger contacts the barrel in a hammer grip.
Meter accuracy was about 0.3C. Ambient temperature in the room was 22.8C. My hand temperature at the start of the test was 33.1C. I had a small fan handy for the second stage of the test, with an uncalibrated air flow of between 2 and 3 metres per second.
To try to reproduce real world results, I carried the CL1H around in my pocket for about 20 minutes before connecting for testing. This raised its temperature to 26.7C, which I set as my starting temperature.
After connecting the thermocouple as described above, I gripped the CL1H in a hammer grip making certain to keep my fingers clear of the thermocouple as that could also influence results. I then started logging the data and turned on the light. I continued to log the data for 20 minutes, doing a bit of flexing every once in a while to ensure proper blood flow. Had I been active outdoors, blood flow would have been higher. The final temperature after 20 minutes of running in my hand was 39.9C, a delta of 13.2 degree Celsius.
For step two, I allowed the CL1H to fan cool to the previous starting temperature. When it reached 26.7C, I placed it on a pelican case raised off the floor, and put it again in the path of the fan. I restarted data logging, and turned on the light. Again I logged for 20 minutes. This is an identical cooling configuration to that which I use for my own previously published runtime tests. The final measured temperature was 39.4C, 0.5 degrees below the hand held test and a delta of 12.7 degrees Celsius. This is a difference in temperature rise of 3.9%.
The graph which resulted was interesting to say the least.
As you can see, the air cooling had a very, very slight advantage over cooling by hand. However, I would say that the equilibrium temperatures under these conditions would have been within 1C of each other, and it is a fairly trivial difference. Had the air flow been only slightly lower, the curves would likely have been identical.
Had I been outdoors and active at the time, there would have been additional cooling air flow over the bezel, as well as my blood circulation would have been higher. These would have impacted the graph and potentially represents a source of inaccuracy.
From the graph I would conclude that at reasonable "summer" temperatures, the human body is basically as good as a fan at cooling a 5 watt power source. As mentioned previously, had my air flow rate been only slightly lower, the two curves would have ended up identical to each other.
I will continue to use a fan while testing, confident that it is not unduly impacting my test results due to thermal effects.
For those who requested an uncooled temperature graph, here it is:
Ambient was 4 degrees Celsius cooler by the time this test was done, which will impact the curve.
Notice how the temperature never actually plateaued, despite reaching over 48C here. When I pulled the battery out and measured it, it was still 38.4C at the positive end and 33.4C at the negative. Judging by that, at this point we were dealing with the battery partially cooling the LED and the temperature could have gone much, much higher if left to its own devices.
Regards,
Eric
Justification that I have seen given by some for not cooling during testing include speculation that the human body is not a good heat sink, while others say that it is. I figured I would try to find out for myself, to see if I should really be testing with a cooling fan running. I tried to set this up in a way so as to not influence the results.
I took a Dereelight CL1H, my favourite test subject of late due to the developing mythology around it, and attached a thermocouple to the bezel. The thermocouple was attached on the outside where the pill makes contact, and held in place using a thick elastic band which also prevented the thermocouple from directly reading air temperature. The thermocouple was located about 1cm up from where my index finger contacts the barrel in a hammer grip.
Meter accuracy was about 0.3C. Ambient temperature in the room was 22.8C. My hand temperature at the start of the test was 33.1C. I had a small fan handy for the second stage of the test, with an uncalibrated air flow of between 2 and 3 metres per second.
To try to reproduce real world results, I carried the CL1H around in my pocket for about 20 minutes before connecting for testing. This raised its temperature to 26.7C, which I set as my starting temperature.
After connecting the thermocouple as described above, I gripped the CL1H in a hammer grip making certain to keep my fingers clear of the thermocouple as that could also influence results. I then started logging the data and turned on the light. I continued to log the data for 20 minutes, doing a bit of flexing every once in a while to ensure proper blood flow. Had I been active outdoors, blood flow would have been higher. The final temperature after 20 minutes of running in my hand was 39.9C, a delta of 13.2 degree Celsius.
For step two, I allowed the CL1H to fan cool to the previous starting temperature. When it reached 26.7C, I placed it on a pelican case raised off the floor, and put it again in the path of the fan. I restarted data logging, and turned on the light. Again I logged for 20 minutes. This is an identical cooling configuration to that which I use for my own previously published runtime tests. The final measured temperature was 39.4C, 0.5 degrees below the hand held test and a delta of 12.7 degrees Celsius. This is a difference in temperature rise of 3.9%.
The graph which resulted was interesting to say the least.
As you can see, the air cooling had a very, very slight advantage over cooling by hand. However, I would say that the equilibrium temperatures under these conditions would have been within 1C of each other, and it is a fairly trivial difference. Had the air flow been only slightly lower, the curves would likely have been identical.
Had I been outdoors and active at the time, there would have been additional cooling air flow over the bezel, as well as my blood circulation would have been higher. These would have impacted the graph and potentially represents a source of inaccuracy.
From the graph I would conclude that at reasonable "summer" temperatures, the human body is basically as good as a fan at cooling a 5 watt power source. As mentioned previously, had my air flow rate been only slightly lower, the two curves would have ended up identical to each other.
I will continue to use a fan while testing, confident that it is not unduly impacting my test results due to thermal effects.
For those who requested an uncooled temperature graph, here it is:
Ambient was 4 degrees Celsius cooler by the time this test was done, which will impact the curve.
Notice how the temperature never actually plateaued, despite reaching over 48C here. When I pulled the battery out and measured it, it was still 38.4C at the positive end and 33.4C at the negative. Judging by that, at this point we were dealing with the battery partially cooling the LED and the temperature could have gone much, much higher if left to its own devices.
Regards,
Eric
Last edited:
