What transistor is this?

[moamps]

Thermal resistance junction to ambient is basically about heat radiation from the surface of the device package, to the surrounding air. A similar size metal and plastic package will have similar surface area thus similar thermal resistance to ambient.

Hm, 2N3440 and  PN3568 don't have similar size, and have almost equal RθJA.
Another example,  for similar size: PN2222 has RθJA=2000C, and 2n2222 has 300 or 3500C.

 

[JohnRoberts]

Hm, 2N3440 and  PN3568 don't have similar size, and have almost equal RθJA.
Another example,  for similar size: PN2222 has RθJA=2000C, and 2n2222 has 300 or 3500C.

OK, I just looked up some data sheets and according to ON semi your numbers are correct. I didn't find a 2222 but 2222A 

Contrary to my speculation the plastic part also quotes 83.3 degree per watt junction to case,  while metal can version specs 150 degree per watt junction to case.  If case gets hotter all else equal it can radiate more heat as the specs suggest.

Also from those same spec sheets the metal 2222 is OK up to 200 degree C, while plastic 2222 is only OK up to 150 degree C.

I recall using the metal 2222 back in the 60s. Touching the can to see how hot it was could burn a small red circle into your finger tip  :-[. I learned to wet my finger with some spittle before touching hot metal cans. 

I will have to make a new guess based on the empirical evidence (specs).  The die surrounded by plastic on all sides apparently conducts heat out measurably better than the die only touching the metal package on the bottom with air ( a poor thermal conductor) above. The inexpensive metal base was probably steel, a good but not great thermal conductor.

JR

 

[moamps]

I recall using the metal 2222 back in the 60s. Touching the can to see how hot it was could burn a small red circle into your finger tip  :-[. I learned to wet my finger with some spittle before touching hot metal cans. 

Same here. We used BC107/177 pairs for headphones amps, and they became sometimes very hot ( usually when headphones's 6.3 jack was not fully pushed into socked and one channel was shorted ).

I will have to make a new guess based on the empirical evidence (specs).  The die surrounded by plastic on all sides apparently conducts heat out measurably better than the die only touching the metal package on the bottom with air ( a poor thermal conductor) above. The inexpensive metal base was probably steel, a good but not great thermal conductor.

It sounds as good explanation. I remember old GE Japan transistors in metal package (like AC188) filled with thermal paste  for the similar reason.

 

[JohnRoberts]

Same here. We used BC107/177 pairs for headphones amps, and they became sometimes very hot ( usually when headphones's 6.3 jack was not fully pushed into socked and one channel was shorted ).

It sounds as good explanation. I remember old GE Japan transistors in metal package (like AC188) filled with thermal paste  for the similar reason.

I was just figuring it out on the fly, but I learned something new from the exercise.

The lower thermal junction to ambient resistance of the plastic package suggests they can replace the metal package devices for many applications despite the lower max junction temperature (150 vs 200).
---
I am still unsure about the heat  calculus supporting 1:1 substitution of plastic power devices for metal devices, but after decades I am not aware of wholesale failures from plastic power devices in audio amps.

JR