[CLOSED] Electronics Cooling Tech Preview:Feedback and Discussion

[heath.houghton]

Hello Feedback Hub,

A public Tech Preview is now in Fusion 360 called Electronics Cooling. Electronics Cooling is a tool that allows Fusion360 subscribers to visualize the temperatures and air movement on their electronics assemblies. The target Fusion360 user for this tool is someone that would like to predict the temperature of their electronics components and see the air movement in their designs.  The Fusion360 user doesn’t need to have prior simulation experience to make use of this tool. It is technically a simulation and thus is integrated into the simulation workspace as a type of simulation in Fusion360, but the user experience and technology has been designed such that we feel the user doesn’t need to have prior simulation experience to gain value from the electronics cooling.

Electronics Cooling Key Features

• Use geometry as-is – no geometry simplification or modification of your actual design needed to run a thermal simulation
• Electronics Design integration – it is not a requirement to use electronics design, but if you do there is enhanced automation to materials assignment right now with other attributes coming straight from the board and its components in the future
• No simulation experience required – the user experience is aimed at creating an absolute minimum number of user inputs to get from manufacturing model to providing temperature and airflow visualizations and with no simulation experience required
• New results visualization experience – completely new visualization engine with an experience targeted specifically for the needs of electronics cooling

Share your feedback in this thread

Share what you like, dislike, any surprises you encountered, or areas of improvement related to Electronics Cooling workflows, example projects, or online help materials.

We encourage you to upload a video of your screen, take screenshots at key moments, or other documents to share feedback.  

Topics may include:

• Understanding the terminology
• Entering Electronics Cooling
• Assigning materials, heat loads, or assigning fans
• Visualizing results

Learn more about the Electronics Cooling Preview on the Fusion 360 blog.

Thanks,

Heath Houghton

Fusion 360 Product Manager

[rbameMA3EK]

This is an awesome new simulation feature.  Two features despareratly needed, anisotropic thermal conductivity for PCB material and the ability to simulate at higher altitudes. Our company simulates electronics for avionics and the ability to simulate at altitude is key. 

[heath.houghton]

@rbameMA3EK Thank you for the feedback regarding your needs for avionics design.  We are actively pursuing how to get more thermal accuracy and detail for the PCB board itself. The main thing is we want to do everything as automated as possible.  There is always the fallback of creating an anisotropic material manually and assigning, or going a step further and having a wizard create the material based on specified copper layers, but we want to get to a point of automation where users don't have to type in the values for the material for boards that have the traces laid out. Phase 1 might be the user creating a material that has anisotropic properties for the PCB, but we have a vision for making it easier than that.
I have a question. Do your 3D models have the copper on the board? The alternative is the PCB with the components positioned but no copper traces.  Also, if you don't mind me asking, what software are you using for designing the circuits and laying out the PCB? This type of information is valuable to understand the overall workflow. 
The different air properties is something we definitely need to consider to suit avionics.

Do your designs use fans, natural convection, or are they using some other means of moving air? If you could post an image of some of what you've run in e-cooling, that would be a treat for us.  As an example, here is a screenshot of a model I was messing around with.

[wvillers]

This has the potential to be a very cool application 🙂  A few things I'd like to understand, since everything happens under the hood:

1. What type of CFD technology is employed: finite volume? Boundary layer? Turbulence Models?
2. What kind of meshing is done? Is there a way to show the mesh?
3. How are the components modeled: 2-resistor? block of resin? 

At first look, it seems like a complete black box, which is understood to be the goal, no argument and ertainly no complaint.  But, for the users a bit more elaborate, would you consider fan curves, orthotropic PCB material, junction temperature output? 

[remi.rayet]

This is a feature I've been waiting for sooooo long! Thanks! 😁

Really exited that you bring it to Fusion 360.

I've just tried a simulation. The simulated assembly is a box with a fan on one side and a device to be cooled attached to the opposite face. The device has a circular radiator around it in the same plane as the opposite face of the fan. There are some electrical connectors on the box too which are 3D rendered in the model, except one where there is only a small hole in the box. 

The result of the simulation is not giving what I expected: see below pict: the air flow gets out the box through the small hole (well, ok, that's right 🙂 ) but not through the circular radiator...😶

I assume that the model simplification done automatically before running the simulation is considering the circular radiator as a plein object instead of a partially open one.

Any suggestion to turn around this and get more accurate flow simulation at this stage of development of the tool?

Thanks

Remi

PS. a feature to simulate with respect to altitude (atmospheric pressure) would be great too.

[rbameMA3EK]

We work with a lot of outside vendors.  Typically we do not have any control over the PCB layout or software.  Most of our customers use Cadence or Altium.   

[heath.houghton]

@wvillers , Thank you for the questions.  Questions validate our direction or give us new perspective on the direction we should be taking the software and we greatly appreciate them.

1. We are using a solver we have developed over the last several years.  At a high level, it uses voxels for the mesh, it solves as a full transient, and uses LES turbulence.  We have seen very good results from it and the usage of voxels is what gives us the ability to utilize pretty much any geometry.
2. We show the results on the physical geometry, not on the mesh.
3. The components are currently modeled as each body getting a singular physical material.  We have been discussing the possibility of adding an optional junction temperature Theta Jc capability.

All of the capabilities you mentioned, anisotropic PCB, junction temperature, fan curves... all are either under current development or in the backlog to get worked on as we move forward.

[wvillers]

@heath.houghton thanks for your answers, it's even cooler than I thought 😁  Another one on my list is heatpipes, vapor chambers and liquid cooling (done the real way, not the fake high K way).  Very exciting product!

[heath.houghton]

@remi.rayet thank you for the image and detailed information.  This looks like a model that e-cooling should be able to handle.  Maybe the fin spacing on the radiator is smaller than the voxel resolution and thus the radiator is seen as solid, like you stated.  It is hard to tell from the single image.  I'd like to dig in and fully understand why you are seeing an unexpected result.  Ideally we would have an online meeting where you could show your model and we could have a discussion.  I am in the USA, so if you propose a time that fits your schedule and is during waking hours for USA, I'll send an invite for a zoom meeting.

[heath.houghton]

@rbameMA3EK , Does this mean you receive them as the native Cadence or Altium files, or as a neutral 3D format like .step?  Are the copper layers (traces and vias) physically represented in the files you receive?

[rbameMA3EK]

We receive the PCB files as step or idf files.  We never get the native layout files.

[heath.houghton]

@rbameMA3EK , Thank you for the engagement as this is very helpful.  When you receive the step files, do they include the copper traces as solids, as a canvas, or not represented at all?  Knowing this will help us understand how much our planned automations for anisotropic PCB materials will be utilized for workflows like yours vs. electronics designs created in Fusion 360.

[eugenepentland]

How do these simulations differ from doing a normal thermal simulation? I work on designing military RF equipment. I was trying to compare a simulation with natural convection of 5W/m^2*K at 25C on all components vs an electronics cooling simulation at 25C with no fans. The temperature on my hottest components were at 420C in the electronics simulation vs 220C on the thermal simulation. I can provide a sample file if needed.

Also will you be adding the ability to add a cold plate to the simulation and time to steady state? Both would be very useful for us.

For your reference, most of our PCB design is done in Altium as of right now. I have been taking the PCAD file, importing it into Eagle, then taking the .brd file and putting that into fusion. That lets me have the vias/traces in the CAD. It doesn't always work perfectly but its enough to get a better simulation.

[heath.houghton]

@eugenepentland Wow, that is hot! In the thermal simulation, it is an FEA and you are assuming the heat removal rate with the film coefficient.  Electronics cooling is actually calculating a CFD analysis in the background and giving an actual simulated heat removal, local air temperature and velocity, etc.  Without seeing the model, I am making a bit of conjecture on the reason why the results would be so different.  The results being warmer for the electronics cooling simulation could indicate to me typically one or a combination of  things.  My first three guesses based on your description would be:

1. that the assumed film coefficient for the FEA thermal is a little high
2. there is "parasitic heating" (hot air rising off a component hits others) and this is not accounted for in FEA thermal
3. the heat spread due to traces and vias is accounted for better in the FEA thermal vs. the e-cooling

The third situation might be the case today, because the FEA is directly meshing the copper traces, while e-cooling is using a voxel mesh.  In electronics cooling, if the traces are quite smaller than the voxel size, the thermal effect of the copper traces is quite reduced.  We are in the works to bring in anisotropic properties for the board and automations that would make the spreading effect of copper traces and vias accurately taken into account regardless of modeling practice.  That work has not made it into the technical preview yet.  If you want me to take a look, you can send me a private message with the .f3d file. 
p.s. Thank you for the workflow information.  It is really helpful to understand how you are getting the model created, what information might be inherent in the model based on it's origination, etc. So much of what we are doing for the software is about getting as accurate a representation of the physical setup as possible, all while automating anything and everything we can.

[GeekFieldGuide]

Thanks for adding this preview, it's been very useful for doing some simulation for a 3D printer hot end design I've been working on.

I have a project for a charity that I'm also working on that could use another feature for this tool. The ability to simulate peltier cooling devices would be extremely helpful to me for this project. I can obviously use it to sort out the hot side of the module, but being able to simulate the full system would be a huge time saver for the initial testing phase.

Thanks!

[Anonymous]

Hi,  is there any plan to add contacts?   Ignoring the contact resistance is often a major source of error that I have seen in electronics cooling

[brianP9KPV]

Hi Heath,

I was able to run some simple test cases and found the tool to be intuitive for pre and post processing.  Is including radiation as a heat transfer mechanism on the roadmap?  This does add complexity, but would help with accuracy in natural convection situations where the radiation contribution is significant.

[pd700]

Hi Heath, Great preview with lots of potential and value. One item that I think would that would add to an accurate a representation of the physical model would be an option to add an air filter on the inlet or exhaust of the enclosure to the simulation.  This would account for the pressure drop and flow distribution.  Typically these are type porous media type filters such as foam, mesh, EMI filter depending on the product.

[pd700]

Hi Heath,

Great preview with lots of potential and value.

One item that I think would that would add to an accurate a representation of the physical model would be an option to add an air filter on the inlet or exhaust of the enclosure to the simulation.  This would account for the pressure drop and flow distribution.  Typically these are type porous media type filters such as foam, mesh, EMI filter depending on the product.

[stephen_mcmillan]

I'm also looking to simulate a 3D printer hot end. Can you share any details of your setup?

And for the original feedback request: include the effects of water cooling.