# Case temperature of mosfet during pulsed load

Discussion in 'General Electronics Chat' started by nepdeep, Nov 18, 2012.

1. ### nepdeep Thread Starter Active Member

Sep 14, 2011
140
0
What is the case temperature of the MOSFET during pulsed load...the maximum junction temperature is metioned as 175..should i take value less than 175 or less to calculate the maximum power dissipation...?

2. ### crutschow Expert

Mar 14, 2008
21,338
6,117
The case temperature depends upon the transistor current, the pulse duration and frequency, and the thermal resistance of the junction-to-case and the heat sink its attached to.

175°C is very high for a maximum allowable junction temperature. I would use no more than 125°C for good reliability.

3. ### JMac3108 Active Member

Aug 16, 2010
349
67
Reliability goes down as temperature increases. I think its a squared relationship?

When I got my first design job my boss told me to keep everything under 100C worst case (maximum ambient temperature + temperature rise) for two reasons.

(1) reliability of the semiconductor

(2) He said that they observed that FR4 PCB would blacken and get weak if exposed to greater than 100C for long periods of time (many years), even though the FR4 was rated for this temperature.

I've been designing to these rules ever since and have had good success. As in all designs, there are tradeoffs involved and you have to decide what is best for your product.

nepdeep likes this.
4. ### nepdeep Thread Starter Active Member

Sep 14, 2011
140
0
Hi,

Thanks for the advice. I have used the thermal transient characteristic curve of the mosfet to calculate allowable peak power during the transient response...

the x-axis represents the ton....or pulse time...y-axis represents the power...
now i can see that at some point the power dissipation of even 100[W] is possible at duty cycle of 20%...that means 20[W] average power...but
from my calculation ...my average power dissipation
Power * dutycycle* Thermal junction to amb +temp amb = Max junc temp
Power * duty cycle*30[K/W]+25=125
Power average=power * duty cycle=100/30=3.33[W] which is much lower than 20[W] dissipation...

what i am confused at is....can I not dissipate peak power of 20[W] average if for very shor time...or ...should i limit my duty cycle to ...say below 5%...if i have to use higher value of peak power say...100[W]...when 10[A]at 10[V]...then the maximum dutycycle allowed is
pmax * duty = 3.3
100* duty =3.3
duty=3.3/100=3.3%

5. ### bountyhunter Well-Known Member

Sep 7, 2009
2,498
510
I read your post three times and don't understand what you are trying to do. Maybe if you could explain the operating conditions. I think you are making this more complicated than it is. Just calculate the average power (peak power times duty cycle) and calculate the junction temp based on average power times total thermal resistance (package thermal resistance plus heatsink thermal).

I also advise keeping max junction temp under 125C for good reliability.

Last edited: Nov 20, 2012
nepdeep likes this.
6. ### ifixit AAC Fanatic!

Nov 20, 2008
650
118
Hi nepdeep,

You need to consider the absolute pulse width, or on-time, as much as the duty cycle because of the device transient thermal response time.

A PWM signal with a duty cycle of 20% and a Ton time of 1 second will allow the junction to heat up to full temperature for that peticular power level. However, if a PWM signal with a duty cycle of 20% and a Ton time of 10 microseconds is used, then the junction will only heat up to approximately 10% of the maximum peak temperature for that peticular power level.

In each situation the average power dissipation is the same, and the average junction temperature is the same, but peak junction temperature is different. Therefore, by using a high enough PWM frequency, you can ensure the maximum junction temp is not exceeded during the on time.

Your device spec should have a transient thermal response time graph in it somewhere. This graph allows you to get transient thermal impedance (Zθjc) from dutycycle and Ton time.

Tj = Power * Zθjc(t) + Tc.

Regards,
Ifixit

nepdeep likes this.
7. ### bountyhunter Well-Known Member

Sep 7, 2009
2,498
510
Which is why I asked for some details on the actual usage. It's true, "transient thermal resistance" only has a meaningful definition in cases where the pulse is fast enough that the power pulse does not cause the die to saturate with heat and reach it's final temperature. If that happens, however long it takes, that is not a transient case it is steady state.

nepdeep likes this.
8. ### nepdeep Thread Starter Active Member

Sep 14, 2011
140
0

this is what i want to do...and...as you all suggested ...i calculated the above graph using the transient thermal impedance graph....

OK i calculated the max junc. temp using thermal impedance and graph
I just wanted to ask one question?

i.e. do i have to care about the ambient temperature and case to ambient temperature during the transient response...

okey simply...can i just use the formula
Tjmax= Powermax*JC thermal impedance + Tcase
I took TJmax=125 and Tcase =95

Sep 14, 2011
140
0
10. ### nepdeep Thread Starter Active Member

Sep 14, 2011
140
0
GOT an answer....i added the transient behavior and the average power dissipation and draw the new graph with both combined....thanks for every suggestoin from the experts....thanks you very very much