Will heat sink temperature rise to Tj

Thread Starter

marcf

Joined Dec 29, 2014
256
I suspect that I know the answer to this, but will the Heat Sink rise to the temperature of Tj.

If one were crazy enough to cause Tj (junction temperature) to go to 125C by allowing a high power dissipation in say a LM317 to heat the junction temp to 125C, would the heatsink reach this temperature as well?
 

GopherT

Joined Nov 23, 2012
7,983
I suspect that I know the answer to this, but will the Heat Sink rise to the temperature of Tj.

If one were crazy enough to cause Tj (junction temperature) to go to 125C by allowing a high power dissipation in say a LM317 to heat the junction temp to 125C, would the heatsink reach this temperature as well?
No, there is always a temperature gradient. The junction is the hottest temp. There is no way the entire heat sink will be at the same temperature if you are heating one point on the heat sink to 125C.
 

Thread Starter

marcf

Joined Dec 29, 2014
256
Is there a way to calculate what the heat sink temp will stabilize to assuming it is enclosed with no air flow?
 

GopherT

Joined Nov 23, 2012
7,983
Is there a way to calculate what the heat sink temp will stabilize to assuming it is enclosed with no air flow?
Completely sealed? No air flow? Heatsink INSIDE the box. Then I retract my statement above. Where do you expect the heat to go? Is the box insulated, too?
 

Thread Starter

marcf

Joined Dec 29, 2014
256
No, I am sorry if you misunderstood me, assume the heatsink at 25.0C ambiant in a room in a well ventilated project box There are no fans, the air is still.
 

crutschow

Joined Mar 14, 2008
23,349
There are several thermal resistance values that you need to consider, including, junction-to-case, case-to-heatsink, and heatsink-to-air.
These all add together to make the total junction-to-air thermal resistance value, which is what you use to calculate the junction temperature from the junction power dissipation.
 

Thread Starter

marcf

Joined Dec 29, 2014
256
Thank you very much for the information, it has caused me to do some research.
I would appreciate if you would do a 'sanity test' on my conclusions.
Using this site's formulas:
http://www.petervis.com/electronics/Voltage_Regulator_Heatsink/Heatsink_for_TO-220_Voltage_Regulator.html
It indicates that for a LM317 with the TO220 case style with a Vin = 25v, a Vout of 15v and a load of 0.7A, a heatsink with a thermal resistance (Theta_SA) of 9 C/W should be used. This would maintain a junction temp of no more that 125C in a 25C ambient temperature environment with no forced air cooling.
From http://www.irf.com/technical-info/appnotes/an-1057.pdf ,
It indicates that a heatsink with a thermal resistance of 8C/W would result in a thermal rise of 56C in the heat sink when 7W is disipated into it. I assume that the 25C ambient temperature should be added. This would raise the heatsink surface temp to 81C (177F).
Theta_SA = (Tj-Ta)/Pd – (Theta_Jc of the LM317 (5C/W)) + Thermal Resistance of heatsink compound (1.3C/W) all over 7W = 7.9C/W
125C-25C/7W -(5C/W + 1.3C/W) = 8C/W * 7W = 56C
I see that the temperature of the heatsink will be a gradient as you pointed out, but would not most of the heatsink rise to its surface temperature?
 

crutschow

Joined Mar 14, 2008
23,349
Due to the high heat conductivity of aluminum, the heat sink will likely have only a small heat gradient over its surface and volume.

Although 125C is the maximum junction temperature, I would keep it no more than about 100C for good reliability.
Note that the ambient air can also be higher than 25C, especially in a box where power is being dissipated.
 

Thread Starter

marcf

Joined Dec 29, 2014
256
In my research, I found that the military spec for Tj is 110C.

I know coming in contact with voltages of <50V will result in no hazards, but coming in contact with 100C could result in severe burns.

I would go with the finger test.
If it is hot enough for an 'ouch' it is too hot.
What do you think a good range for Tj and Ta is?

I think 140F and 80F would be reasonable.

Also, isn't there a heat limit for the PWB to prevent discoloration?
I assume that the temp rise extends to the leads soldered into the PWB as well.
 

mcgyvr

Joined Oct 15, 2009
5,394
touch temperatures differ depending on the standards your product will be tested to and some allow "hot surface" stickers if you exceed that rating.

Tj max is specified in the datasheet for each and ever electronic component and should be followed. There is no one-size-fits-all Tj temp
Max PWB temperature (and for any connector,etc.. or really its plastic raw material) is typically specified as the UL RTI value (usually 105 to 130 deg C for standard FR4) which accounts for aging over time.

The closer to Tj/RTI,etc..... the shorter the components life. (and it can be quite a drastic curve)
All in all.. you want your components/enclosures,etc.. to be as cool as possible and never above the touch temperatures (for exposed components) or Tj temps (for IC's) or UL RTI (for PWB/plastics) ratings for each and every component.

Then throw in flammabilty rating/oxygen index, mechanical issues, vibration, blah,blah,blah and you become an Engineer :)
 

crutschow

Joined Mar 14, 2008
23,349
140°F (60°C) for Tj would likely mean a much larger heat sink than necessary.
I think 100°C or so would be okay (which would make the transistor case temperature about 65°C).

You might also consider an LM317 with a TO-3 case, which has a lower theta-Tjc (2°C/W) and thus will reduce your heat sink requirements.
 

Thread Starter

marcf

Joined Dec 29, 2014
256
Thanks for all of the responses. I have learned a lot. I was just using the LM317 as an example.
This is so much better than going on google!
I look forward to finding more things out, and hopefully contributing from time to time.
 

dl324

Joined Mar 30, 2015
8,918
In my research, I found that the military spec for Tj is 110C.
According to this from a TI datasheet, LM117 can operate up to 150C, LM317 is 125C:
lm117Tj.jpg
Military grade parts are grossly more expensive than commercial grade. The Military has the benefit of not caring how our tax dollars are spent.
I know coming in contact with voltages of <50V will result in no hazards, but coming in contact with 100C could result in severe burns.

I would go with the finger test.
If it is hot enough for an 'ouch' it is too hot.
What do you think a good range for Tj and Ta is?

I think 140F and 80F would be reasonable.
For the life of me, I can't fathom why people think too hot to the touch is too hot for pieces of silicon. Manufacturers build in safety margins when they spec parts, so a junction temperature of 125C is safe for LM317; if derate specs are followed. Besides, the part includes over temp and over current protection.
Also, isn't there a heat limit for the PWB to prevent discoloration?
I assume that the temp rise extends to the leads soldered into the PWB as well.
Some heat will be dissipated through the leads and copper on the circuit board, but for serious heatsinking, a TO-220 package should be standing up to maximize air circulation. A TO-3 should be mounted so the heatsink fins are on the exterior of the enclosure; unless it has a fan and ventilation.
 

crutschow

Joined Mar 14, 2008
23,349
......................
Military grade parts are grossly more expensive than commercial grade. The Military has the benefit of not caring how our tax dollars are spent.
......................
It's not that they don't care, it's that, if you're a soldier in combat, I don't think you'd want your equipment to fail because of some cheap commercial part which has not been tested to give a high confidence that it will operate under harsh military conditions (heat, cold, moisture, vibration, etc).
It's the added testing of the part to help insure its reliability including at the high and low temperature extremes (for which commercial devices are seldom if ever tested in production), the requirement for hermiticity of IC packages, and the documentation of this testing along with the part serial numbers, that causes the large price increase compared to a jelly-bean commercial component.
The actual part fabrication costs are generally trivial compared to the costs to insure reliability.
 
Top