lordpil's blog


its for thermal transfer
usually it makes another cooling solution work as designed, because in the real world all flat surface solerances suck
maybe there was a heatsink you guys forgot to put back
go away!

Posted by renesis at 15:16 | permalink | 0 comments


synth: i dont think its because the solder is a mechanical bond, either
you could have the thermocouple ends seperated by a conductor, if everything on the hot end was same temp, reading would be the same
almost always yes
let me guess in the military you used white thermal paste
13K
usually for thermal reasons
where are you going with this
its usually silicone based

Posted by renesis at 15:11 | permalink | 0 comments


nope

Posted by renesis at 15:05 | permalink | 0 comments


lots of DMM will have thermocouple modes
and those agilent multichannel data loggers will usually do thermocouples
yup
probes are expensive, so i dont use them
theyre usually big too, so they respond slower than wire twists

Posted by renesis at 14:56 | permalink | 0 comments


i think i use lm60?
whichever is calibrated 10mV/C
but sometimes those work well
ive wedged them into heatsink fins, covered them with thermal paste, work really well
but for most stuff i'll use J or K thermocouples
i just use thermocouple extention, strip off a couple inches , twist the ends, tight, clip so like 2mm of twisted wire left, tiny bit of solder on that (sometimes i dont even solder them, theyre held together by a few twists pretty well)
really fast response, which i guess offsets the slow response of the CA glue we use to stick them to everything

Posted by renesis at 14:51 | permalink | 0 comments


EMC you do because you get in trouble if your shit doesnt work right in most situations
but get thermal wrong and your shit just dont work
anyway, everything i explained is how i would do it if asked to figure it out for work and write some report to prove my work
IRL, you can just ramp up load and measure/feel temps, noting ambient temperature versus maximum expect ambient temperature
you can graph that shit out and prob extrapolate what you need
im doin em

Posted by renesis at 14:45 | permalink | 0 comments


but you decide it, if you cant get inside and determine was is limiting
its really hot, but most electronics will survive it
problem is, thats a surface temp, not a component temp
i would use 60C because lots of gear is designed to not get hotter than 60C on external surfaces
because a user feels more than 60C and they *know* some shit is wrong, whether it is or not
ideally, you base max temps by running temp sensors on sensitive components, with the system assembled normally
so you shit is outdoor i think, so it can maybe get hotter than 60C
and it might be made for higher ambient too
but its hard to say without opening something up, figuring out whats running hot, and data logging temps in typical operation with the unit assembled as normal
consumer products, and high reliability products, thermal is a lot of development time
circuits are pretty easy, but part selection and mechanical design relating to thermal usually isnt trivial

Posted by renesis at 14:40 | permalink | 0 comments


30C ambient, 60C max, thats a 30C difference
so you can dissipate 10W inside the panel at 3C/W (because 30C/10W)
so lets say your internal impedance is 1 ohm
im pulling most these numbers out of my ass btw, to show example
cool
with 10V output, but i think you kinda get it
11 ohm load, on 12V unloaded panel, with 1 ohm internal impedance, would pull 1A
and the output voltage would drop out to 11V
1V dropped across the internal impedance, which is 1W, which at 3C/W, would raise the panel surface temperature to 33C, assuming 30C ambient
well max temp is a constant

Posted by renesis at 14:35 | permalink | 0 comments


and how much the voltage will drop out for a given load
look for the hot spot on the panel under load
when you find hottest spot, test with two diff loads again
calculate the wattage dissipated inside the panel
and measure the temp rise on the hot spot
do this twice, make sure the C/W numbers come out similar
an external surface should really get above 60C, in normal outdoor ambient, lets call it 30C
so you can rate the panel at 30C by how much power dissipated in the panel for it to rise to 60C, or 75C, whatever you decide for ambient and max surface temp
wait an hour for temperatures to stabilize
say it raises 3C per watt dissipated in the panel

Posted by renesis at 14:30 | permalink | 0 comments


one million
volthz
those seem pretty big
take them to sun, measure voltage unloaded, put a resistor on them, measure loaded voltage, calc output impedance, change resistance, measure output impedance
make sure output impedance calcs close to same with diff loads
well i would test with like, 100mA and 250mA, some shit like that
so R = 12V / 100mA, and 12V / 250mA
so like 47 ohms and like 120 ohms
when you know that, you can calculate how much power is dissipated in the panel

Posted by renesis at 14:25 | permalink | 0 comments


gotta have a standard to test against!

Posted by renesis at 14:05 | permalink | 0 comments


Top | Add to Technorati Favorites


© 2007 lordpil.   XHTML 1.0! CSS! Site design by GNAA  Blog Engine by pbx | MULTI2 | ian hanschen | lolwat