Thermal simulation query

[Techman_7]

Hello,
In cauer thermal simulation as below -

Source1 is the PWL pulses acting as power dissipation in MOSFET. The simulation works but if the Rthca (R5) is changed, then still the junction temperature remains the same. (Logically its correct, it should not change as per circuit configuration).

But practically Tj should change if Rthca (R5) is changed.

What modifications are needed in above simulation?

 

[KlausST]

Hi,

In my understanding source1 needs to act like a constant_current not like a constant_voltage.

Klaus

 

[danadakk]

Thermal Simulation and Analysis Software in the Cloud | SimScale

Use thermal simulation to analyze and optimize your product early in the design phase with the world's leading browser-based CAE software.

www.simscale.com

https://www.infineon.com/dgdl/Infineon-Thermal_Modeling_of_Power_Electronic_Systems_3.0.pdf-ApplicationNotes-v01_00-EN.pdf?fileId=8ac78c8c92416ca501926358d80e4c65

 

[Techman_7]

Hi,

In my understanding source1 needs to act like a constant_current not like a constant_voltage.

Klaus

I have created behavioural voltage source whose values are equal to power dissipation in the MOSFET. The source1 waveform looks fine w.r.t. power dissipation in the FET.

 

[KlausST]

Hi,

this means: You don´t want to correct it?
If so: why do you ask for "what needs to be modified"?

Klaus

 

[FvM]

I have created behavioural voltage source whose values are equal to power dissipation in the MOSFET. The source1 waveform looks fine w.r.t. power dissipation in the FET.

The waveform may look correct, but the model uses no physically plausible mapping of power/temperature, thus can't give correct temperature values.
Please review the linked Infineon application note. If you map thermal resistance to electrical resistance, temperature corresponds to voltage and power to current, not voltage. Therefore power dissipation must be modelled by a current source.

 

[Techman_7]

The waveform may look correct, but the model uses no physically plausible mapping of power/temperature, thus can't give correct temperature values.
Please review the linked Infineon application note. If you map thermal resistance to electrical resistance, temperature corresponds to voltage and power to current, not voltage. Therefore power dissipation must be modelled by a current source.

Hello,
I tried with current source but Tj variation is inversely proportional to Rthca value. I think Tj and Rthca should be directly related to each other....

--- Updated Tuesday at 1:41 PM ---

Hi,

this means: You don´t want to correct it?
If so: why do you ask for "what needs to be modified"?

Klaus

Because, the junction temperature not changing even if Rthca is changed. This should not happen, thats why asking what modifications needed.
Rthca and Tj should be directly proportional...

 

[KlausST]

Hi,

Because, the junction temperature not changing even if Rthca is changed. This should not happen, thats why asking what modifications needed.
Rthca and Tj should be directly proportional...

if you have a voltage source .. an a resitor connected to it... Why should the voltage change when you change the resistor?

I can only repeat: use a current source.

I tried with current source but Tj variation is inversely proportional to Rthca value. I think Tj and Rthca should be directly related to each other....

makes no sense at all.

--> Current through a resistor:
* voltage is directly proportional to resistor
* voltage is directly proportional to current.

Nothing "inversly"..

Klaus

 

[FvM]

I tried with current source but Tj variation is inversely proportional to Rthca value.

Not sure what Tj is in your model. I'm my simulation it's potential of upper left node. V(tj) = I(p,avg) * sum(Rth) + V(tamb), thus temperature proportional to Rth.