Dear team,
When my customer use RθJC(top) listed in our datasheet to calculate Tj, Tj will be larger than 150C, so they hope we can help evaluate the actual RθJC(top) value based on their dissipation model as below. They add one 1.8*2.0mm thermal pad at the bottom side of chip, and they add heat-conducting glue with thermal conductivity 2.0W/m.k at the top side of chip. Could you please help evaluate the value?
Thanks & Best Regards,
Sherry
Hi Sherry,
I don't think you should use RθJC(top) to calculate TJ directly. See this section of the semiconductor thermal metrics page:
This assumes the majority of the heat is dissipated through the top of the driver, which is incorrect and leads to very high TJ estimates. ΨJT controls for this and would be better to use.
I can't help evaluate the thermal metrics of the device with this heatsinking; it will involve too many variables such as the PCB dimensions and airflow that I can't know or control. Regardless, this scheme should help, although I don't think adding heatsinking to the top of the chip will make a huge difference. The reason is the same as what I mentioned earlier; most of the heat will flow through the thermal bad and bottom of the chip, and only some will flow through the top. Because of this, adding heatsinking to the top will have a limited effect. The best way to know will be to evaluate it on the board while operating. Your customer may be able to get a pretty accurate Tj measurement using the pullup/pulldown resistance of the input pins.
I know it is often suggested against, but I have found RθJA method of calculating TJ to be accurate enough for gate drivers.
Thanks,
Alex M.
Hello Sherry,
There is no direct way to measure the Tj of the device, so I have seen people come up with ways to determine the Tj based on the change of certain device characteristics.
For example, you could determine Tj of a device using this chart:
By measuring the startup current (hard to do while the device is running). Or you could look at prop delay:
So if the turn on delay is 15ns, you know Tj is around 100 C. I have seen this done in a few different ways, so these are just examples. This is one way to calculate Tj, rather than using thermal modeling. It has drawbacks also, so this is just one possibility, especially if the customer thinks the Tj calculation is off.
Thanks,
Alex M.
