TPS7A47: Thermal performance

Hi Nandini,

There aren't alternative devices that wouldn't require a significant design change. The automotive version may have slightly better thermals, but what is the ambient temperature in your application? Do you already have boards made? 

Regards,

Nick

Hi Nick,

We are a few days away from Gerber release and that's why don't want to change much in design.

1. Ambient application is 30 degC. Based on Rthja of 32.5degC, it looks like we have margin (125degC-(1.668W*32.5degC) = 70degC.

2. In thermal simulation, Tcase is obtained from CFD analysis. Power dissipation is calculated as given in my query. Tjdatasheet = 125C

[Tcase+(Rthjc*Powerdissipation)] = Tj-sim

We find the Tj-sim to be about 26deg more than Tjdatasheet. This causes the thermals to fail.

Is this approach correct?

3. a) Regarding the Automotive part, is there any difference apart from better thermals that we need to look out for in terms of performance/characteristics in case we want to change to it from the existing part?
b) Can you pls confirm if it's a drop-in replacement for the existing part?

Thanks

Hi Nandini,

In the case that you already have the top-case temperature (from a real board measurement or simulation), you should use the ψJT parameter instead of the RθJC(top) parameter to calculate the junction temperature. Using the RθJA parameter, I estimate the junction temperature to be 30C + 1.668W * 32.5C/W = 84.2C. I can't do the calculation using the ψJT since I don't know what the simulated top-case temperature was, but from your calculation does that align? The board layout will affect the effective RθJA, so it's likely that it's not exactly 32.5C/W, but with that said the board layout would need to be very poor for the ~40C (125C - 84C) margin to be used up. Can you share a screenshot of the local layout near the LDO? The ψJT is defined in a way that makes it highly board layout-independent, which is why it is useful for estimating the junction temperature from a top-case temperature measurement on a real board. 

Nandini Muralidharan said:

3. a) Regarding the Automotive part, is there any difference apart from better thermals that we need to look out for in terms of performance/characteristics in case we want to change to it from the existing part?

There won't be any significant differences between the devices. The automotive device goes through more qualifications, and the guard bands during final test for some of the EC specs may be a bit tighter, but overall, it's the same device. 

Nandini Muralidharan said:

b) Can you pls confirm if it's a drop-in replacement for the existing part?

Yes it's a drop-in replacement for the catalog version. However, from my calculations I doubt that you need the automotive version since the difference in thermal performance is so marginal and you have so much junction temperature margin. 

Regards,

Nick

Hi Nick,

I can share the layout screenshot tomorrow & talk to the simulation team about your suggestions. Meanwhile, thermally simulated top case temperature is 116.6 degC. Can you shed any lights based on this? 

Hi Nandini,

With 116.6C top case temperature, the junction temperature would be estimated at Tj = 116.6C + 0.3C/W * 1.668W = 117.1C. That still seems quite high, so I would suspect that either the layout is very poor in regards to thermal heatsinking, or that the thermal simulation wasn't appropriately representative of the device. I would expect that it's the latter because the JEDEC High-k board layout from which the thermal metrics are derived does not have a particularly thermally-optimized layout, so it's not difficult to exceed the thermal performance based on the datasheet thermal metrics. Do you have any details of how the thermal simulation was performed? 

Please also note that the U.S. holiday is tomorrow and Friday, and so I will not get back to you until Monday in a response to a comment made tomorrow. 

Regards,

Nick

Hi Nick,

Thanks for the details - our thermal team is checking on your suggestion. Meanwhile, they asked if we can have the detailed CAD model (with inner details) or DELPHI-model for this device that you can share with us?

thanks

Hi Nandini,

I've reached out to our modeling team to see what we can do about getting you a more accurate CAD model for the lead frame. I'll update you when I hear back. 

Regards,

Nick

Hi Nandini,

Our thermal modeling engineer said we can create a Delphi model and send it to you. Which software tool is your team using? 

Regards,

Nick