Thanks for the great simplified explanation. The Radiation heat transfer rate is insignificant for small temperature differences. This becomes obvious with a simple calculation which boils down to the low constant of proportionality, Stephen Boltzmann constant, of 5.67E-8.
@nbsr1 I was probing the tab surface just underneath the screw, using the screw as a bit of a wedge. It seemed reasonably stable, and gave much higher temps than the heatsink which is what I wanted to show.
Thanks for the heat sink design info. Components like DO-35 diodes, where a heat sink is not an option, dissipate heat into the PCB through the leads. But this goes both ways. I know of a circuit design where all the components are within spec, but the heat from 1/2 watt resistor killed a diode next to it. The circuit went back for repair, and the same diode died again and again. A fan was added to the panel after the 5th time, and it hasn't died since.
@asus3571 Sorry, I get too many requests for personal design questions, so have to now say no to all of them. The EEVblog forum is available for asking questions like this.
Great video - thanks. It would have been very interesting to see how much difference adding the thermal paste between the transistor and the heatsink made.
Hi Dave, Good basic introduction. In the US this is called steady state heat transfer analysis. However, because we are dealing with heat transfer, strictly the power analogy should be in heat per unit time units. English = Btu/h, Metric Calorie/sec or Joule/sec. So we can work in Watts but the analogy is Heat per unit time. The US symbol for heat is Q. As interesting and much more important to power semiconductors is dynamic heat transfer. Sinking is sized for hot spot flow.
Great video! Seems in the forced air-flow example you didn't take into account the "spreading resistance", so if you add 1/3, you get 1.6 C/W which is much closer to the 1.8C/W measurement.
The color of a solid surface being black had little impact on the emissivity (e), since most of the thermal radiation happens at wavelengths above visible range ( Infrared) . Anodizing is more to do with surface finish and not black color.
Very informaive and formative. How about a tutorial on how to work with thermal impedance for pulsed operation and the associated thermal heatsink calculation? Best regards!
@asus3571 Can't post links here, google it, first hit. The link is also in my channel header graphic and also in the credits at the end of every video.
I read somwhere that the radiation of black heatsink compared to usual metal surface is up to about 10% (depends on temperature). When you have forced air cooling that advantage goes down fast as convection part of the heat transef is much higher. That's why you don't see many black CPU heatsinks. As Dave said, orientation of the heatsink matters, direction of airflow as well. Heatsink designed for forced air cooling have many thin fins to reach high surface, doesn't work well for still air.
You made a mistake at 16:30. You took the wrong curve / graph. It is 10W, and 26 deg C, on the left scale. So 2.6 deg C/W. Very close to the nominal 2.70 deg C/W. The slope of this curve (starting at 0 W, 0 deg C) is the thermal resistance.
I have a rather unusual case in which I need to calculate how hot a simple plate of copper will get--that is the objective! the heating sources are just a few power components with given thermal resistances, etc in the data sheet (it doesn't matter what they are--just so they are big enough for heating the plate. I simply need to know how hot the Cu plate will get after things are stable. So, I think it comes down to this: All I need to know about the plate is the thermal resistance from plate to ambient (no fins, no flowing air, and air ambient) I figure that all I need is a way to find the plate thermal resistance, given: The plate material, Cu here, the dimensions of the (rather small) plate i.e. L x W x H, and some sort of chart or formula for thermal resistance of this plate-to-air. Is the logic sound? THANKS FOR ALL YOU HAVE PROVIDED WORLDWIDE!
Dave, Many modern power applications operate in pulsed and impulse modes of power switching. Can you please do a thermal design sequel addressing thermal time constants and explain how the Capacitors shown on your thermal circuit model are handled ?
How should I find thermal resistance(Junction to ambient) of a resistor? Like it was not given in a datasheet that I read. Also, is that we can find the thermal resistance if I know the power rating?
I have seen some horrible heatsinks. I have learned that they also collect dust too. I have also seen tons of gold, silver and other color heatsinks in PC/computer type components.
What's should be the size of heatsink (length and width) for Tda2030A Ic if 4 numbers of such Ic is to be mounted on same heatsink and each Ic is to deliver upto 10 watts power which are all powered by 12-0-12 v 3 Ampere transformer ?
how can we calculate copper area on pcb for heat transfer.. or this video shows heat sink for heat transfer if we can use copper pad (open masking) then how to calculate PAD size ..
Hi Dave. Thank you very much for your informative video! I have a question about the forced air flow. For the radial fin heatsink you used for TO220 package, does it matter the direction of air flow? I was always imagining that we put a small fan on top of heatsink so that the air flows through the channel of fins instead of blowing directly at heatsink surface.
Hello, I’m struggling to calculate a problem. I have an internal heat source but I also want to account for an ambient temperature of 50degC. Does anyone know how to do this?
Nice video! Regarding heat-to-ambient thermal resistance for still air vs forced air conditions, isn't it logical to assume that the resistances from the two curves from the thermal curve are in parallel? For very low speed of air (close to still), the thermal resistance because of air cooling is near infinite, and hence the resistance of still air curve dominates. For high air speed, the forced cooling graph dominates. Might make sense?
hello so the thermal resistance is measured in ºC/W, good. in your example you use 10V@1A=10W, very nice... how about 5V@2A??? in a nutshell, how does the current through the transistor affect the thermal increase? thank you! =)
Starting at about 5:09 You say "for every ONE Watt the power increases, that the temp increases "X" degrees". If it's degrees PER Watt (C/W), Shouldn't an increase of ONE Watt cause an increase of ONE degree?? ... and an increase of x WATTS cause an increase of x DEGREES??? ... and same for --decrease--. I thought I had it but seem to be a bit confused now.
Hi Dave, your video was very informative. If a CPU heatsink is rated for say 95W Maximum TDP, does that mean it can dissipate 95W of heat from any source?, or is that some form of proprietary measurement that only matches up with the TDP rating of CPU chips?
I suggest changing the title - not really heatsink "design", more like heatsink "basics". Was looking for how one can design the shape and look of the heatsink. Fantastic video though, great job.
this was the most amazing vid ive ever saw being addicted to cooling you would be the man to ask as far as cooling motherboard vrm's and mosfets as well as gpu' parts that generally dont get much attention with cooling im in the process of doing some cooling mods to my mobp gpu and considering a new cpu cooler are yo open for ?'s
Nice vid realy, Well, anyway your heatsink is so big compared to the transistor. I would use a small "clip-on" style heatsink to bring the junction temperature down, like a thermalloy P/N 6073B , some like that...
I have used mica transistor insulators for years and lately I see some rubberized versions, anyone know where i can buy this rubberized stuff from in large sheets like A4