Power Semiconductors With 5% Duty Cycle 1

[Tobalcane]

Power Semiconductors With 5% Duty Cycle

I’m doing some heat transfer analysis on power semiconductors with a 5% duty cycle at continues pulse. I know that for preliminary the Average Power (heat) dissipated will be Peak Power x Duty Cycle. Because it is 5% duty cycle, the device will heat up and then cool down with each pulse so the temp rise will be the same through the time used.

I’m locking horns with my manager about Peak Power. He wants me to also do an analysis using Peak Power. I guess my question is “Should the Average Power be enough for the analysis?” and “Why look at Peak Power anyway, does it matter in a 5% duty cycle?".

Thanks in advance for your time and effort.


Go Mechanical Engineering
Tobalcane

 

[IRstuff]

I'd agree with doing a peak power analysis, since the duty cycle duration is sufficiently long enough that there might be thermal equilibrium in the junction during the ON time.

The junction temp might get hot enough to cause metal migration into the junction, resulting in a shorted junction.

TTFN

 

[Tobalcane]

Thanks IRstuff! I broke down and started looking at peak power. I have been reading some stuff from Onsemi and they have some calcs for Peak Power. I’m surprised that Steinberg does not have any say on this topic.

Go Mechanical Engineering
Tobalcane

 

[hacksaw]

it is not only peak power, but also your peak current densities

 

[Tobalcane]

Hacksaw,
Thanks for responding. Now why do I have to look at peak current densities?


Go Mechanical Engineering
Tobalcane

 

[IRstuff]

Electromigration of the metallization is a function of temperature and current density. Electromigration can cause voids in the metallization, resulting in open fails.

TTFN

 

[Tobalcane]

IRstuff,
Will this happen near or above the max Tj? Is this also a function of time? As in will this happen in seconds or days of continues use?

I just wan to see if I am thinking right. When I do a transient analysis using the Peak Power, would the temperature rise come to equilibrium close to a steady state analysis using the Average Power?


Go Mechanical Engineering
Tobalcane

 

[IRstuff]

Electromigration, while dependent on temperature, can, and does, occur at room temperature. It's mostly a momentum transfer phenomenon, but the atoms are more mobile at higher temperatures.

The semiconductor temperature-time behavior is dependent on its thermal mass and thermal isolation, so it may, or may not, reach anything close to equilibrium.

TTFN

 

[hacksaw]

electromigration is one of the issues. as with IRStuff your junction temperatures will be anything but steady state.

 

[sreid]

Pulsed power ratings for semiconductors has been traditionally based on "Transcient Thermal Impedance." Go to International Rectifier

http://www.irf.com/indexsw.html

and search for, say, IRF250. Look at figure 11 in the data sheet. You should also be able to find an app note on the subject.