TPS549D22: Thermal Simulation of IC-Die vs IC-TOP

user1874945

Intellectual 300 points

Part Number: TPS549D22

Hi TI expert,

My team is current using TPS549D22 for a new project design. The application is as below.

1> Vin=12V(+/-5%)

2> Vout=3.3V

3> Iout=23.5A

There are three questions.

1. We do the test with Fsw=600Khz and Fsw=400Khz. The efficiency is 93%@600Khz, 23.5A, and 93.3%@400Khz,23.5A. Seems the efficiency is just affected a little by the switching frequency change. Could you explain the reason?

2. Per the thermal image test under 25 ambient, no airflow(18layers PCB), the hottest top case temperature of IC is 75.6 deg.C(600Khz, max load). How to calculate the junction temperature?

3. Per using TI Webench, the simulation result shows the efficiency can be 95%+. The IC power dissipation is 3.15W. And IC- Top is 60.6 deg.C and IC- die is 57 deg.C. Does this mean IC junction temperature is even lower than the IC top case temperature?

For our motherboard design, we usually have 18 layers or 20 layers PCB. The PCB can provide very good thermal dissipation path. But in our main board, the ambient temperature is not possible to be 25 deg.C. It may be as high as 55 deg.C to 60 deg.C because the air is preheated by the CPU or the switch chipsets.

Look forward to your comments.

Thanks,

Dora

over 7 years ago

0 JohnTucker over 7 years ago

TI__Guru**** 191445 points

Dora,

1. For higher load currents, conduction losses dominate. So the efficiency gains due to lower switching losses with lower switching frequencies may be small as you observe.

2. The thermal images in the datasheet use much fewer layers, 6 layers total.

3. The junction temp should be higher than the top side case temp. The difference will be small however. Can you share the Webench design? I'll take a look at it.

In webench, you can stipulate a higher ambient temperature,

0 user1874945 over 7 years ago in reply to JohnTucker

Intellectual 300 points

Hi John,

How could we calculate the junction temperature more accurately by measuring the TOP case temperature?
Should we use Junction-to-top characterization parameter(0.96 deg.C/W) instead of Junction-to-case (top) thermal resistance(18.3 deg.C/W)?

Here is the link for the Webench link.
webench.ti.com/.../webench5.cgi

Thanks for your prompt reply.
Dora

0 JohnTucker over 7 years ago in reply to user1874945

TI__Guru**** 191445 points

Dora,

That is correct. If you have the top case temperature and teh IC power dissipation, you can use teh junction to top characterization parameter to calculate the junction temperature.

0 user1874945 over 7 years ago in reply to JohnTucker

Intellectual 300 points

Hi John,

Thanks for the confirmation. The last question, why is IC-die lower than IC-top per Webench simulation?
webench.ti.com/.../webench5.cgi

Regards,
Dora

0 JohnTucker over 7 years ago in reply to user1874945

TI__Guru**** 191445 points

Dora,

That seems unusual. I will have to defer to the webench team. I have asked them to look into it.

0 user1874945 over 7 years ago in reply to JohnTucker

Intellectual 300 points

Hi John,

Thanks. Look forward to their feedback. I will ask a new question related with this.

Dora

Power management

Power management forum

TPS549D22: Thermal Simulation of IC-Die vs IC-TOP