I'm confused about heatsink C/W values. I've used online heatsink size calculators and two of the required inputs are ambient temp and max temp. I've input 25C for ambient and 65C for max temp, thinking that equals a 40C rise above ambient. Heatsinkusa says their approximate °C/W/3" values are based on natural convection and a 70C temp rise above ambient. Another heatsink manufacturer says their approximate °C/W/3" values are based on natural convection and a 60C temp rise above ambient. Based on that it seems the 70C and 60C above ambient would allow the chip temp to get 20C to 30C hotter than my calculator inputs. What am I not understanding?
The real concern is the device junction temperature. But the heat sink should not get so hot either,
so look for about 60C for the sink if possible. The mfr listing higher sink temps is just spec wars.
An example:
https://www.smlease.com/entries/thermal-design/how-to-select-a-heat-sink/
so look for about 60C for the sink if possible. The mfr listing higher sink temps is just spec wars.
An example:
https://www.smlease.com/entries/thermal-design/how-to-select-a-heat-sink/
Last edited:
HeatsinkUSA has less than 10 workers, less than 10 million US in sales.
Check it on BBB or similar business information sites.
They are a chop saw place, getting sections extruded outside (their tools), and shipped to warehouse...Extrusion plants are expensive.
On order receipt, section is cut with chop or hack saw, and shipped from warehouse...
Maybe tapping and milling is done.
Conrad in Aussie also small setup. Less than 20 people.
Use engineering text books, these guys are bit players.
But held in high esteem by some here, who think they are great.
A lot of their guidelines are to impress people, without design knowledge.
Like a kid with racing stripes on a cheap car pretending it is a sports car.
Tell us amp design, and do the calculations to hold the device case temperature at 65 degrees, with ambient temperature at 45 degrees, only convection heat transfer.
The delta-T should be 20 degrees Celsius only, device case to ambient.
Fin tip to ambient, farthest from source, about 5 degrees.
Then add 25% to the calculated rating.
And also de rate for poor mounting, bad / dirty surface, and so on.
Also do not exceed 80% voltage on supply, and stay within 75% of output.
Then the unit will be durable.
Check it on BBB or similar business information sites.
They are a chop saw place, getting sections extruded outside (their tools), and shipped to warehouse...Extrusion plants are expensive.
On order receipt, section is cut with chop or hack saw, and shipped from warehouse...
Maybe tapping and milling is done.
Conrad in Aussie also small setup. Less than 20 people.
Use engineering text books, these guys are bit players.
But held in high esteem by some here, who think they are great.
A lot of their guidelines are to impress people, without design knowledge.
Like a kid with racing stripes on a cheap car pretending it is a sports car.
Tell us amp design, and do the calculations to hold the device case temperature at 65 degrees, with ambient temperature at 45 degrees, only convection heat transfer.
The delta-T should be 20 degrees Celsius only, device case to ambient.
Fin tip to ambient, farthest from source, about 5 degrees.
Then add 25% to the calculated rating.
And also de rate for poor mounting, bad / dirty surface, and so on.
Also do not exceed 80% voltage on supply, and stay within 75% of output.
Then the unit will be durable.
Last edited:
It is easier to measure at high heat-flow, and with natural convection you get a "better" number.
You do NOT have to run the sink at 60 or 70 degree rise!
Same as designing with wood. You test heavy loads until it breaks. Then you de-rate for safety. A typical hunk of wood beam will take a fiber stress of 1000psi, with a big sag, and it might break. In houses we call this 300psi assuming the floor is full of people (party house). Sag is small and safety is high.
Or your 5-passenger car with you and your mate. Or your half-ton truck to fetch a 6-pack of beer.
In most cases, at ambient + 70, it will force the device into thermal shutdown, or destroy it, or at least cause reduced life.
40 + 70 = 110, most devices start to fail at 130 or so JUNCTION temperature. That means case is at 110 or so. Finish...
Like I said, those tables are more hype than practical.
40 + 70 = 110, most devices start to fail at 130 or so JUNCTION temperature. That means case is at 110 or so. Finish...
Like I said, those tables are more hype than practical.
I am thinking your calculation may be based on full power? Eg, you are calculating based on the amp running at full power all the time. If so, I believe that is why you are getting high temps in your calculations.
That is the correct way, it is not, to design for full continuous power dissipation?
Anything less will be a constraint on the user.
Anything less will be a constraint on the user.
Maybe online calculators are too generic or do not apply to the exact one you re using.
Go back to basics: define/calculate actual power dissipated > define acceptable die temperature > expected/acceptable maximum ambient temperature > do the math for acceptable die-to-ambient Rth
That is your goal.
Then substract power device internal Rth and the interface (mica-grease) Rth.
What´s left is acceptable heat sink Rth.
Browse the catalog and pick the one which matches your needs.
Online calculators are fine and always will show "a number" but usually not too clear on how they reached it.
Personally I always prefer to keep track of every step and be able to tweak each one as needed.