LMG5200: Device can not be selected in TI PCB Thermal Calculator

Greetings Beat,

We are looking into this and will get back to you shortly.

Thank you!

Masoud

Hello Masoud

Since I'm having the LMG5200, built into a DC-DC-converter, on my table now, December does not help me a lot.

Is there a similar device I could select to get approximate results?

Regards

Beat

Hello Masoud

Thanks for the offer - I happily accept it.

The print I have has two LMG5200 mounted on top side (H-Bridge configuration).

Both modules are thermally well connected (lots of vias)  to a cooling area on the bottom layer (40 um thick)  of the print. The whole area is ca. 22 x 46 mm.

The print is vertically mounted, no forced cooling.

What I would be interested is a thermal resistance equivalent circuit, and the corresponding values. I tried the circuit as in the attached .png. It considers basically two paths of thermal flow: One via top side through the case of the module, and the second path through the vias to the bottom side via the PCB cooling area.

The left thermal resistors in the equivalent circuit are the ones given in the datasheet, the right two are unknown to me.

Is the thermal equivalent circuit ok?

If yes, what values should I take (or what total Rth junction - ambient)?

If you propose an other thermal equivalent circuit: Could you send me a drawing and the corresponding values?

Thanks a lot

Beat

Hello Masoud

Yes, I had already read the tech note.

I agree that almost all the heat will go trough PCB. The unknown is just what my case to ambient with the given copper area would be. Since you offered me to " answer any specific thermal designs questions ", I was hoping you could give me numbers, ca. in the same way the tool would do.

Regards
Beat

Hi Beat,

 

Some suggestions for your consideration. The circuit thermal path looks okay with us; however, we don’t have the Rth_CA number. Indeed, the top path can be considered negligible since most heat will be dissipated through bottom. Rth_BA depends on your specific layout (copper plane size/thickness, via dimension/number). We don’t typically measure the Rth_BA. One possible way to leverage TI’s thermal calculator would be picking up a similar QFN part (thermal pads size) which is already in the library and refine/import your PCB design for calculation. You can then get the thermal resistance accordingly. This can give you a ballpark number for design analysis. As LMG5200 uses a unique package, we don’t have a good recommendation so far, but we are working on adding it into the library. Definitely the most accurate way would be to measure it directly.

 

Regards,

 

Rui

Dear Rui

The dimensions of the PCB cooling area (thickness, area) are given in an earlier message in this chat. If something else is missing, please let me know exactly what you need.

Regards
Beat

Dear Rui

Why can the copper plane to ambient Rth,Cu_A not be calculated? How does then the TI PCB Thermal Calculator do?
The tool gives a curve for Tjunction=f(PCB Copper Area). This is exactly what I am looking for.

Since the tool does not yet have the LMG5200, it would be very helpful to get all Information I Need to calculate the same Thing manually.

Regards

Beat

P.S. Although I estimate a lot your explanations and the time you take for me - I have now been waiting quite a bit of time for an Information which should be in the TI-Tools according to the LMG datasheet. I do really not understand if you try to tell me that something the tool obviously does should not be possible!

Dear Beat,

Unfortunately we do not have access to the algorithm or equations used by the tool as one does not exist.

We would be happy to answer questions directly related to the LMG5200 specifications. However system level thermal design questions such as the one in your last post is beyond the scope of this forum.

Thanks!

Masoud

Dear Masoud

It seems you misunderstood my question. In one of your first answers, you offered "In the meantime, We would be happy to answer any specific thermal designs questions you may have on this forum."

What would help me a lot, and what I have been asking for in my latest answer, was exactly the same type of Information as the TI PCB Thermal Calculator tool does provide for other TI devices, but unfortunately not for the LMG5200.

In your latest answer, you write you can not give me this Information, "as one does not exist". I'm not sure what you say does not exist, the tool or the equations.

So I take one more attempt to explain: What would help me is to get Information about junction temperature, or alternatively the Rth from PCB cooling area to ambient, depending on PCB cooling area - exactly in the way the TI PCB Thermal calculator gives for the other devices.

Since the tool is there, and it is a TI-tool, the Information does for sure exist in TI. And it would really help me.

Hi Beat,

The TI PCB Thermal Calculator mainly applies to the components with exposed thermal pads (ESP) to PCB. However, LMG5200 has a unique package and it does not fall into this track (no ESP).

The calculator gets the T_junction v.s. PCB Copper Area curve based on over a thousand CFD simulations with package and PCB copper area variations. The simulations are then correlated with experimental measurements. Please see the application note (link attached) for better explanation. This requires a significant amount of effort. Our modeling team is currently working on it, but it needs time to happen. We appreciate your patience and support with TI GaN. Please let me know if you have further questions.

www.ti.com/.../slua566.pdf

Regards,

Rui