AnsweredAssumed Answered

Heat sink with TO-247

Question asked by Travis Rowell on Dec 18, 2013
Latest reply on Oct 24, 2016 by Daniela Coblas

Like • Show 1 Like1
Comment • 8

Hello,

I am having trouble accuratly modeling TO-247 packages in simulation. I was hoping to get some help on how best to set this up. My TO-247 package consists of three components, a copper case, a die(the transistor), and a plastic case. I modeled the die as silicon, the copper case as copper, and the plastic case as epoxy resin. For the insulator I used the manufactuer spec sheet to build a material (the spec sheet wasnt very good I had to leave a lot blank). I am still not getting accurate results. My prototype has far different values then simulation. I took a cross section of the temperature distribution and I found that the temp of my junction was the same temperature as my heat sink. I know this is not possible. Can anyone who has done this type of work before give me an accurate method for modeling this situtaion? I would much rather use the junction to case resistance, and insulator to heatsink resistance provided by the manufactuer then rely on material selection in simulation. I have upladed my heat sink model for extra clarification.

Thank you,

Travis

Attachments

Heat Sink Question.rar.zip647.0 KB

No one else has this question

Outcomes

Jared Conway Dec 18, 2013 11:33 PM

To do what you want for setup, get the electronics cooling module. He 2r input method does exactly what you want. Or convert the values and make your own 2r.

To understands why you sim may be different, you will have to elaborate on your test setup and conditions. The assumptions you think your analysis makes and some numbers from your test. Blindly looking at simulations is really difficult.

Actions

Travis Rowell @ Jared Conway on Dec 19, 2013 1:38 PM

I do have the electronic cooling module. I tried to use it for this project. However I get an error message when trying to apply it to my plastic case. I get the message cannot find face between case and junction or something to that effect. I was told that the two parts I make the 2R must be the same height, width, and thickness. These TO-247 packages do not follow that model. I may be being a bit picky here I dont know, but I didnt want to change the shape of my package to fit the solidworks model of a 2R. As to my real world setup. It is exactly the same as the setup you see in the picture below. I have a fan in front of the heat sink, each TO-247 package should have 17.833 watts running across it. I use a thermal couple to measure the temp of my heat sink. The thermal couple measures 74C which is very close to what solidworks told me it would be. However my junction temp should be way hotter. Instead solidworks is telling me my junction temps match my heat sink. This is not correct. I applied the 17.883 watts to the small sliver of chip on my model. How does solidworks calculate the heat from that? Would I be better off putting a more realistic temp of 100 degrees on that chip instead of putting a wattage?

Actions

Jared Conway @ Travis Rowell on Dec 19, 2013 5:10 PM

"my package to fit the solidworks model of a 2R." > the method used by solidworks is the standard method for 2r

"I have a fan in front of the heat sink, each TO-247 package should have 17.833 watts running across it." > what you've shown is 17.833 shared among all the units. not per. (default in flow)

"It is exactly the same as the setup you see in the picture below." > please describe the details of both the physical test setup and analysis setup. there may be a small detail in here that is causing a difference. for excample you haven't said if you are using internal or external analysis, what your fan is, what the environmental conditions are, is there radiation, do you have gravity enabled...etc. if you have a heat problem already and are using17.xxx for all the components, it will probably be much hotter now which indicates more likely a setup problem to start with. IE you weren't just getting lucky that the heat sink temp was 100 and the rest was off.

Actions

Mark Keown Dec 19, 2013 1:52 AM

You will need the manufactures Rj-c (resistance junction to case) and Rj-b (resistance junction to board) for the device. Adjust the thermal conductivity of the layers as done in the attached picture. You can use two or three layers. With two layers apply the power to the interface surface and with three layers apply power to the center solid. Good that you have a model to check your simulation.

OR adjust the thermal conductivity until it matches your bench test results.

Actions

Mark Keown Dec 19, 2013 8:00 PM

For the issue of applying power to the interface - for some reason you have to use a surface filter if the feature is not an assembly.

Actions

Jared Conway @ Mark Keown on Dec 19, 2013 11:57 PM

shouldn't need to

Actions

Mark Keown @ Jared Conway on Dec 20, 2013 3:17 AM

As for the power issur you need to apply 17.833 w x 6 parts = 107W this should heat you up a bit...

Actions

Daniela Coblas Oct 24, 2016 7:59 AM

Hello, I have to do same simulation. Did you managed to get good results?

Can you explain to me better how you built your 3D model, how you defined your materials? It is very important for me

Actions

8 Replies

Jared Conway Dec 18, 2013 11:33 PM
Mark Correct
Correct Answer
To do what you want for setup, get the electronics cooling module. He 2r input method does exactly what you want. Or convert the values and make your own 2r.

To understands why you sim may be different, you will have to elaborate on your test setup and conditions. The assumptions you think your analysis makes and some numbers from your test. Blindly looking at simulations is really difficult.
Like • Show 0 Likes0
Actions
- Travis Rowell @ Jared Conway on Dec 19, 2013 1:38 PM
  Mark Correct
  Correct Answer
  I do have the electronic cooling module. I tried to use it for this project. However I get an error message when trying to apply it to my plastic case. I get the message cannot find face between case and junction or something to that effect. I was told that the two parts I make the 2R must be the same height, width, and thickness. These TO-247 packages do not follow that model. I may be being a bit picky here I dont know, but I didnt want to change the shape of my package to fit the solidworks model of a 2R. As to my real world setup. It is exactly the same as the setup you see in the picture below. I have a fan in front of the heat sink, each TO-247 package should have 17.833 watts running across it. I use a thermal couple to measure the temp of my heat sink. The thermal couple measures 74C which is very close to what solidworks told me it would be. However my junction temp should be way hotter. Instead solidworks is telling me my junction temps match my heat sink. This is not correct. I applied the 17.883 watts to the small sliver of chip on my model. How does solidworks calculate the heat from that? Would I be better off putting a more realistic temp of 100 degrees on that chip instead of putting a wattage?
  Like • Show 0 Likes0
  Actions
  - Jared Conway @ Travis Rowell on Dec 19, 2013 5:10 PM
    Mark Correct
    Correct Answer
    "my package to fit the solidworks model of a 2R." > the method used by solidworks is the standard method for 2r
    
    "I have a fan in front of the heat sink, each TO-247 package should have 17.833 watts running across it." > what you've shown is 17.833 shared among all the units. not per. (default in flow)
    
    "It is exactly the same as the setup you see in the picture below." > please describe the details of both the physical test setup and analysis setup. there may be a small detail in here that is causing a difference. for excample you haven't said if you are using internal or external analysis, what your fan is, what the environmental conditions are, is there radiation, do you have gravity enabled...etc. if you have a heat problem already and are using17.xxx for all the components, it will probably be much hotter now which indicates more likely a setup problem to start with. IE you weren't just getting lucky that the heat sink temp was 100 and the rest was off.
    Like • Show 0 Likes0
    Actions
Mark Keown Dec 19, 2013 1:52 AM
Mark Correct
Correct Answer
You will need the manufactures Rj-c (resistance junction to case) and Rj-b (resistance junction to board) for the device. Adjust the thermal conductivity of the layers as done in the attached picture. You can use two or three layers. With two layers apply the power to the interface surface and with three layers apply power to the center solid. Good that you have a model to check your simulation.
OR adjust the thermal conductivity until it matches your bench test results.
Like • Show 1 Like1
Actions
Mark Keown Dec 19, 2013 8:00 PM
Mark Correct
Correct Answer
For the issue of applying power to the interface - for some reason you have to use a surface filter if the feature is not an assembly.
Like • Show 0 Likes0
Actions
- Jared Conway @ Mark Keown on Dec 19, 2013 11:57 PM
  Mark Correct
  Correct Answer
  shouldn't need to
  Like • Show 0 Likes0
  Actions
  - Mark Keown @ Jared Conway on Dec 20, 2013 3:17 AM
    Mark Correct
    Correct Answer
    As for the power issur you need to apply 17.833 w x 6 parts = 107W this should heat you up a bit...
    Like • Show 1 Like1
    Actions
Daniela Coblas Oct 24, 2016 7:59 AM
Mark Correct
Correct Answer
Hello, I have to do same simulation. Did you managed to get good results?
Can you explain to me better how you built your 3D model, how you defined your materials? It is very important for me
Like • Show 0 Likes0
Actions

Heat sink with TO-247

Attachments

Outcomes

Related Content

Recommended Content