Current industry wide practices no longer have such issues with power ratings and derating. Back in the early 80's at least, manu's started to understand the problems in consistently diebonding without voids, and also how to test for them in the production line. In the higher level design, thermal control modelling started really taking off also.
Nowadays, dissipation densities can be significantly higher without compromising quality or reliability.
Even Berquist boards work very well, and they are only copper epoxy bonded to aluminum... They're in the LED bulbs.
jn
Of course they are. But you didn't ask any question about convection, fans, radiation, anodizing, nuttin. You made a statement about greased washers and the manu's refusal (so to speak), to give you numbers. I explained why they cannot.
Of course there is. Black heatsinks look really really cool..
jn
I didn't ask a question. I mentioned that heat considerations are simplified but not exact. The example was that while a manufacturer would give derating in watts per degree, no one would give a firm value for the mica washer part of the chain.
There are many reasons for this, from the thickness and fill material of the grease, the heatsink flatness etc.
If heat dissipation were a very accurate problem we would also start to see tolerances, I.E. 150 W at 200 C die temperature +/- 5%.
Now you can do a really exact design if you want to do some testing, but it is much more practical to be close and on the right side of the smoke.
I know, and explained why. And yes you are correct, you did not phrase it as a question..
A very good side to be on.
I also like to wake up on the right side of the dirt as well..
jn
I was surprised at how much smaller a difference the emissivity made on a heatsink than I thought (it's been a long time since I looked at this). With any kind of forced air the black body radiation is just too wimpy. I knew a guy who painted all his steam radiators flat black (Krylon) ugly as sin and no help was noted. Yes I know a very careful scientific experiment.
jn does the fact that most of the area is "looking at itself" matter? Just ruminating here.
The National Audio handbook has a nice section on heatsinks and as I recall has enough data so you can calculate the improvement. The hotter the better.
But there is another good reason to go black. The IR thermometer reads much more accurately.
If you paint the radiator black, you loose some from the paint thickness. But the actual test would be input temperature vs output. The warmer you make the room, the more energy leaks out.
Now there is a nice program around from a former Ericsson engineer that covers heatsink design quite nicely. Today I just get error messages when I try the website where I found it.
Any radiation to itself just keeps it in the system. It has to radiate out of the system to lose the energy. It's the classic two wall thingy. Kirchhoffs law of thermal radiation...of course.
(I didn't remember that law and don't even know if I ever knew it, just found it on wiki...but man, made me look really smart, no??
jn
Actually, thermal drop through the paint is insignificant. The energy density is too low, the surface too thin.
Here's some data. Note anodized doesn't seem to be color specific..weird.
Surface Material Emissivity Coefficient
- ε -
Alloy 24ST Polished 0.09
Alumina, Flame sprayed 0.8
Aluminum Commercial sheet 0.09
Aluminum Foil 0.04
Aluminum Commercial Sheet 0.09
Aluminum Heavily Oxidized 0.2 - 0.31
Aluminum Highly Polished 0.039 - 0.057
Aluminum Anodized 0.77
Aluminum Rough 0.07
Aluminum paint 0.27 - 0.67
Antimony, polished 0.28 - 0.31
Asbestos board 0.96
Asbestos paper 0.93 - 0.945
Asphalt 0.93
Basalt 0.72
Beryllium 0.18
Beryllium, Anodized 0.9
Bismuth, bright 0.34
Black Body Matt 1.00
Black lacquer on iron 0.875
Black Parson Optical 0.95
Black Silicone Paint 0.93
Black Epoxy Paint 0.89
Black Enamel Paint 0.80
Brass Dull Plate 0.22
Brass Rolled Plate Natural Surface 0.06
Brass Polished 0.03
Brass Oxidized 600oC 0.6
Brick, red rough 0.93
Brick, fireclay 0.75
Cadmium 0.02
Carbon, not oxidized 0.81
Carbon filament 0.77
Carbon pressed filled surface 0.98
Cast Iron, newly turned 0.44
Cast Iron, turned and heated 0.60 - 0.70
Chromium polished 0.058
Concrete 0.85
Concrete, rough 0.94
Concrete tiles 0.63
Cotton Cloth 0.77
Copper electroplated 0.03
Copper heated and covered with thick oxide layer 0.78
Copper Polished 0.023 - 0.052
Copper Nickel Alloy, polished 0.059
Glass smooth 0.92 - 0.94
Glass, pyrex 0.85 - 0.95
Gold not polished 0.47
Gold polished 0.025
Granite 0.45
Gypsum 0.85
Ice smooth 0.966
Ice rough 0.985
Inconel X Oxidized 0.71
Iron polished 0.14 - 0.38
Iron, plate rusted red 0.61
Iron, dark gray surface 0.31
Iron, rough ingot 0.87 - 0.95
Lampblack paint 0.96
Lead pure unoxidized 0.057 - 0.075
Lead Oxidized 0.43
Limestone 0.90 - 0.93
Lime wash 0.91
Magnesium Oxide 0.20 - 0.55
Magnesium Polished 0.07 - 0.13
Marble White 0.95
Masonry Plastered 0.93
Mercury liquid 0.1
Mild Steel 0.20 - 0.32
Molybdenum polished 0.05 - 0.18
Nickel, elctroplated 0.03
Nickel, polished 0.072
Nickel, oxidized 0.59 - 0.86
Nichrome wire, bright 0.65 - 0.79
Oak, planed 0.89
Oil paints, all colors 0.92 - 0.96
Paper offset 0.55
Plaster 0.98
Platinum, polished plate 0.054 - 0.104
Porcelain, glazed 0.92
Paint 0.96
Paper 0.93
Plaster, rough 0.91
Plastics 0.91
Porcelain glazed 0.93
Quartz glass 0.93
Roofing paper 0.91
Rubber, hard glossy plate 0.94
Rubber Nat Hard 0.91
Rubber Nat Soft 0.86
Sand 0.76
Sawdust 0.75
Silicon Carbide 0.83 - 0.96
Silver Polished 0.02 - 0.03
Steel Oxidized 0.79
Steel Polished 0.07
Stainless Steel, weathered 0.85
Stainless Steel, polished 0.075
Stainless Steel, type 301 0.54 - 0.63
Steel Galvanized Old 0.88
Steel Galvanized New 0.23
Tile 0.97
Tin unoxidized 0.04
Titanium polished 0.19
Tungsten polished 0.04
Tungsten aged filament 0.032 - 0.35
Water 0.95 - 0.963
Wood Beech, planned 0.935
Wood Oak, planned 0.885
Wood, Pine 0.95
Wrought Iron 0.94
Zink Tarnished 0.25
Zink polished 0.045
jn
Of course the fourth-power law for luminosity versus absolute temperature is the crucial variable here. And the amount of radiative cooling is small for the sorts of temperatures we can tolerate for silicon. As higher-temp semis are developed the efficacy of radiation will get progressively more important. Of course it's about all you have to work with in space except for once-only phase change, understandably a messy business.
In the aluminum extrusion plants there is a quickly-learned policy for dealing with an extruded piece sitting on a table or the floor. You assume it is hot, and you don't touch it. Unless you can detect no convective currents whatever, it's not safe to suppose it is not hot, given the very low emissivity of "bare" aluminum.
It's a bit like handling high voltage capacitors lacking a visible shorting wire across the terminals.
In the early 90s, an eastern neighbor had the idea to offer his amp models in different (uni) colors, heatsinks anodised and the rest of the case powder coated.
Before you knew it, a lot of the local competitors entered the rave, with a brochure that read any RAL color.
(of course, some still couldn't shake their R&B programming)
Made zero difference whether the heatsinks were red or black anodised. Better to brush them clean on a regular basis.
(even smarter when you're lying)
Attachments
Only after having tried to pick one off the floor.
(naturally, the one that just came out the mold. gives great reflexes)
JN
Enamel paint is .8 for a single coat. Ever look at how much paint ends up on a steam radiator?
One building renovation I am familiar with, was supposed to be cosmetic only. The radiators exposed areas were painted. The designer doing the renovation didn't like the look, so they were stripped and repainted. The heating bill went down 20%, so I have been told. So a .09 for bare aluminum to a .9 is significant particularly for space applications. However in a building a .6 to a .8 or so actually shows up.
Enamel paint is .8 for a single coat. Ever look at how much paint ends up on a steam radiator?
One building renovation I am familiar with, was supposed to be cosmetic only. The radiators exposed areas were painted. The designer doing the renovation didn't like the look, so they were stripped and repainted. The heating bill went down 20%, so I have been told. So a .09 for bare aluminum to a .9 is significant particularly for space applications. However in a building a .6 to a .8 or so actually shows up.
I had never thought about it but a physicist freind pointed out that no concentration of solar rays can make a temperature more than the surface of the sun on any object.
Zeroth Law. And to forestall Ed's inevitable question, they went at night.
Stock radiators are cast iron and not very shiney. The point about the heat sinks is that there is in general no attempt to maximize radiative area, almost all of it is parallel plates closely facing each other so the emissivity is of little importance. You need to compute the blackbody radiation compared to the convection, it just does not add up (there are studies involving radiator covers).
Last edited:
shhh...takes all the fun out of it...
I was a docent once for the museum...guy brings a group of people in, says "any questions, ask John, he'll know the answer....".
I looked at the crowd, said..""And when I don't know the answer, I'll make up something, you'll never know I'm lying..."" Oddly, he's never asked me to give tours since...go figure..
Doesn't matter. The power density doesn't produce a very large temperature gradient through the paint. What, a kilowatt through a square meter in a radiator? Power semi's do a kilowatt per cm squared, or 10 megawatts per meter squared.
I always knew that flat black was best...matte black has a coefficient of 1..
Hmmm..I halved my heating bill with a can of non expanding foam from home depot and outlet foam insulators. Sometimes it's difficult attributing a change to one specific thing, specially if you do more than one thing different.
Hey, on those breakers....weren't they both mag and heat? Seems to me that all breakers have two regimes, one thermal and the mag for bolted fault.
Oh, big time. I was just gonna paste up to black paint, but said to myself...yo, stupid...it's just ones and zeros, put it all up...
Honestly, some of the numbers surprised me..I thought bare alum was better than that.
But actually, I saved it to word first..I need some of that stuff..radiation in a vacuum cryostat can be significant when one watt of heat leak at 4.5 kelvin causes a kilowatt of fridge power.
Interesting, hadn't thought of that..I suspect it is incorrect (no actual knowledge backing that up of course...). If you concentrate it on some object and get to the temp, them have it radiating to another using a different surface with a different coefficient at that temp...who knows...
me, no...
jn
- Status
- Not open for further replies.