Transistor temperature question

[rcbuck]

I have long wondered about transistor dissipation numbers given in data sheets.

I have a feedback RF amplifier stage that uses a 2N3904. The data sheet says total dissipation is 625 mW which is to be derated at 5 mW for each degree above 25. After running for 30 minutes the transistor case measures 40 degrees C. That means 625 - 75 = 550 mW maximum dissipation at 40 degrees C.

The transistor is drawing 20.8 mA with a Vce drop of 9.49 volts. That equals 197.4 mW dissipation. The transistor is hot to the touch but not too hot to prevent continuous holding on to it. Do I need to be concerned about it being hot? The circuit is in open air and will run a little warmer when it is put in the enclosure (which is vented).

Maybe I am being overly cautious but I just don't like parts that run hot.  Is the 2N3904 actually designed to run for years at this temperature with no negative effect?

[TimFox]

The 625 mW at 25^o C, derated by 5 mW/K, is for ambient temperature (air outside the case).
The other rating, 1.5 W at 25^o C, derated by 12 mW/K, is for case temperature.
Your 40^o C case temperature implies a power rating of 1.2 W, so your 200 mW dissipation is conservative.

[Cerebus]

What you need to worry about is the junction temperature.

You know your power dissipation (197.4 mW) and you know your case temperature at steady state (40ºC).

The Fairchild data sheet for the 2N3904 in a plastic TO-92 case gives a junction to case thermal resistance of 83.3 ºC/W. So, your 197.4 mW will lead to a temperature difference from junction to case of 16.4 ºC, adding that to the case temperature gives you a junction temperature of 56ºC in round figures. According to the datasheet the rated "Operating and Storage Junction Temperature Range" is -55 to 150ºC, which leaves your junction temperature a very comfortable 94ºC below its maximum.

[TimFox]

The ONSEMI datasheet for the 2N3904 I quoted is a bit easier to understand for this non-heatsinkable transistor.
It only quotes power dissipation as a function of case temperature or ambient temperature.
https://www.onsemi.com/pdf/datasheet/2n3903-d.pdf

[rcbuck]

What you need to worry about is the junction temperature.

You know your power dissipation (197.4 mW) and you know your case temperature at steady state (40ºC).

The Fairchild data sheet for the 2N3904 in a plastic TO-92 case gives a junction to case thermal resistance of 83.3 ºC/W. So, your 197.4 mW will lead to a temperature difference from junction to case of 16.4 ºC, adding that to the case temperature gives you a junction temperature of 56ºC in round figures. According to the datasheet the rated "Operating and Storage Junction Temperature Range" is -55 to 150ºC, which leaves your junction temperature a very comfortable 94ºC below its maximum.

Thanks for the detailed answer. I was somewhat confused about junction temperature vs case temperature. I did not look at the Fairchild data sheet.

The ONSEMI datasheet for the 2N3904 I quoted is a bit easier to understand for this non-heatsinkable transistor.
It only quotes power dissipation as a function of case temperature or ambient temperature.
https://www.onsemi.com/pdf/datasheet/2n3903-d.pdf

Tim, the numbers I quoted were from the Onsemi data sheet. They seemed to be simple enough but your reply gives more detail about the actual device rating. Actually you can put a heat sink on the TO-92 case. I have a dozen or so of these on hand that I bought about 25 years ago. You could probably epoxy a small copper sheet deformed with a drill bit to fit the back side of the TO-92 and accomplish the same thing.