Learning Forum

[dirk.linnenbruegger]

Dear Forum,

actually I’m wondering whether there is a reasonable mapping between heatsink parameters to Thermal (Static) Solution parameters like Convection in Air and/or Heat Flow.

From the heatsink datasheet I get the following information:

Thermal Resitance:  3.7 – 1.25 K/W

Dissipation Loss:  20.2 W

Surface: AL-natural

So, is there any information that can be used to parametrize Convection in Air and/or Heat Flow?

 

In addition, is there also a relation between FPGA power consumption, i.e. 30W, to Convection in Air and/or Heat Flow?

For example, can I set the Heat Flow of the FPGA to 30W or is this value too high or even too low?

 

Best Regards

Dirk

[Gaurav Sharma]

Hi dirk. linnenbruegger

I am sorry I could not understand your query completely. For a better understanding can you please provide me following information:

1. What program are you using: Discovery Live or Discovery AIM?
2. What heatsink datasheet are you mentioning?
3. Can you elaborate on the calculations for Thermal resistance and Dissipation loss
4. Are you referring to your application when you say FPGA. Do you mean field programmable gate array device, or are you working with the standard heat sink example of Discovery Live?

Please provide more details and also share your model, if possible.

Thanks & Regards,

Gaurav

[dirk.linnenbruegger]

Dear Gaurav.

Many thanks for your prompt response. I'll prepare a test case for you and I will provide more details on my questions later on.

 

However, this are additional questiones arised recently to my former post into the forum, as you can see here: 

https://discoveryforum.ansys.com/t/h4ahpl

 

Since I'm currently preparing such a test case for a customer presentation in the next weeks I expect that such questions may come within that session.

 

Regards

Dirk

[Gaurav Sharma]

Hi dirk. linnenbruegger 

I will be waiting for your test case.

Also, the other post you have referred does also require your model for further investigation, as requested by my colleague Naresh. I hope you understand that its important to replicate user's observations at our end for further debugging. Since the sample case Naresh tried and shared with you, did not lead to any ambiguous observation, it becomes important to look at your model where the observation is actually being produced.

Regards,

Gaurav

[dirk.linnenbruegger]

Hello Guarev,

please find attached the IC with heatsink as well as the datasheet of the heatsink.

 

So, it doesn’t matter what kind of IC you are using. In this example we have a step model of a Xilinx FPG. It also could be a simple rectangular body.

From the Xilinx datasheet you can find a power consumption of 30W. So I took this value as a starting parameter for heat flow simulation and assigned it to the whole IC body.

In addition I’ve started with the default parameter of 10 W/m2C for Convection in Air and I was wondering that I got a surface temperature >600°C. Changing this parameter to 30~40 W/m2C the results seem to be more reasonable.

 

And finally, I found the datasheet of that heatsink I’m using for this simulation. I don’t know if there any given parameters that could be used for the thermal simulation setup. Maybe you have an idea or an answer to this issue.

 

Please let me know whether you have any additional questions.

Regards

Dirk.

🛈 This post originally contained file attachments which have been removed in compliance with the updated Ansys Learning Forum Terms & Conditions

[Gaurav Sharma]

Hi dirk. linnenbruegger 

I actually wanted to have a look at your simulation file but anyways with the given information, following is what I can suggest:

1. It is possible that though you wanted to add a volumetric heat flow, but by mistake you ended up applying heat flow to surfaces only which may ultimately lead to unexpectedly high temperature values at the surface. Please confirm if the application is correct.
2. The convection value to be used has to come from the environment in which the application is used. Often designer's know what would be the best value to be used, either from experience or by the data collected from experiments conducted. 
3. The additional parameters listed in the datasheet shared, do not seem to hold any relevance for a thermal study in Discovery Live, except thermal resistance. 
4. Discovery Live uses thermal conductivity as material input that ultimately governs the thermal resistance as conductivity is reciprocal of resistivity.

I hope that answers most of your questions. please let me know if I have missed something or if there is any other further question.

Thanks & Regards,

Gaurav

[dirk.linnenbruegger]

Thanks Gaurav_ANSYS 

I think that we must keep this situation here as it is.

From my point of view I expect that most our customers (electronic designers) will not have any knowledge or experience with convection in air parameters. They just want to get a fast result for their simulation setup and they don't want to care about any parameters. 

So, my understanding is to go ahead with the default value for Convection in Air.

And maybe there are some more additional information available from IC vendors that would match to the Heat Flow / Heat Flux parameters. In this case it's not a good idea to equal power consumption of the IC with the Heat Flow parameter, meaning 30W power consumption equals to 30W Heat Flow of the same step file simulation body. So, I'm really not sure what kind of value makes sense in this case.

In further investigations I found out that a value of 2~3W for Heat Flow parameter leads to reasonable simulation results, also for internal and external flow when I put the DUT into a housing.

Do you think this is a way to go?

Thanks

Dirk

[Gaurav Sharma]

dirk. linnenbruegger 

Thermal results does depend significantly on the value of convection used. The default convection value in Discovery Live is just to start and for accurate results, it is expected the user will provide a more appropriate value.

Coefficient of convection depends on how your application is mounted and its environment. For instance, if there is fan that works to remove heat from the region by forced convection, higher convection coefficient is to be used. Whereas, if the convection is natural, a lower value is required. We DO NOT recommend any value, and its upto user to decide what is the best value for his/her simulation. Once a convection value is provided by user, Discovery Live calculates temperature distribution based on the inputs provided. I hope you understand that we do not expertise in customer applications and customer is responsible for using best inputs to the simulation and setting up the simulation right. Again, its beyond my understanding to comment on what heat flow should be used.

In case user is now sure about certain input parameter such as convection value, a google search to see what value other users have used might be helpful.

I might be able to provide some suggestions but only after reviewing your setup thoroughly. Again, its difficult to extend any suggestions unless I look into the working simulation file. Thanks for understanding.

Regards,

Gaurav

[dirk.linnenbruegger]

Hi Guarav,

Many thanks for your reply. This gives me good background information for discussions with customers!

 

Have a nice weekend

Dirk