Hello everyone,

I was wondering if my transistor respects the spec or not. I have already posted my question to onsemi.
Here the link, if anyone of you has some hint, feel free to comment. thanks.
https://community.onsemi.com/s/ques...-the-transistor-mjd340-is-within-thermal-spec
if you cant see my post is because probably they still did not approve it, here below a copy and paste of the question:

How to verify if the transistor MJD340 is within thermal spec?
my MJD340 consumes 1.4W as average power consumption. In the datasheet, there are two PDmax values one refer to the case when Tx=25C the other refer to the case of Tcase=25C. Rthjc=8C/W so in my case : delta_Temp_case= 1.4W*8=10.2C => Tcase= 10.2+20=30.2C assuming Tcase_nominal =20C. if I look at the plot in Fig.7 which refer to Tc it seems to me that the maximum power conumption would be still over 12.5W and I would be within spec. However, if I use Rthj-a =80C/W then I would get : 1.4*80=112C=delta Temp => Ta=112+20=132C. Looking at Fig.7 referring to curve Ta, the max power consumption allowed would be something around 0.25W and I would be out of spec. Shouldnt these two calculation give the same conclusion?

 

Rthjc is the thermal resistance between junction and case. That tells you how much hotter the junction is than the case. It doesn’t tell you anything about how hot the case will get under any given circumstances.

Rthja is the thermal resistance between junction and ambient when attached to the specified copper area. That one will tell you how hot the junction will get when you know the air temperature.

so Δcasetemp will be 10.2°C, but that just means that the junction will be 10.2°C warmer than the case. It doesn’t tell you how warm the case will get. If you were to keep the case at 20°C by some huge heatsink, then the junction would be 30.2°C.

On the standard footprint the junction will reach 132°C, and Rthjc tells you that the case would be 10.2°C lower than that: 121.8°C which is rather too warm.

Rule of thumb: that size of transistor case will dissipate 1W with no heatsink, no more.

 

got it, thanks I am a bit rusty on this concepts. so I guess, I will need to review them. If you have any link or book as reference to suggest it would be great otherwise I will search some random links in internet.

 

Don’t forget that the datasheet will tell you directly all the good things (like it will dissipate 15W under some ideal circumstances) but give you just about enough to work out the less favourable things for yourself (i.e. It gets damn hot at 1W soldered to the board)

 

No they won't, and can't. The fallacy is in your argument that, in the first case, the case temp is the same as ambient which it won't be. The radiating surface of the case is so small that the only way it can be held at 25C is with an infinite heat-sink. For normal power transistors a case temp of 75C is generally accepted (that is the effect of the 'derate at 0.12C/W above 25C' clause below the 15W dissipation figure at a junction temp of 150C), though few manufacturers actually quote performance at that case temp. IXYS is one of the few that do for their linear MOSFETs. So the junction would be acceptable at, say, 30C below Tj(max), (typically 150C), or 120C and so max dissipation is (120-75)/8.33 = 5.4W and you'd need a heat-sink + thermal compound with a combined thermal resistance of better than (75-25)/'5.4 = 9.2C/W to achieve that at 25C ambient. That's hard to do for the SMD case types hence the recommended pad layout and 1.56 W max dissipation rating.

Real power transistors are in non-SMD cases, the best being TO-247plus, though the venerable TO-220 is good for 1.6W standing up with no heat-sink at 25C ambient.