Internal Temperature Sensor TMS320F28377SZWTS -40 to +125°C

Hi Fabien,

I think your understanding of the allowable junction temperatures are correct: T = up to 105C, S = up to 125C, and Q = up to 150C. 

Normally we wouldn't expect the device self-heating to make the device ~169C when the ambient temperature is 85C.  My guess is that there is some issue with the readings of the temperature sensor.  Is this a TMX device or TMS device?  

If this is actually indicative of the actual juntion temperature, this would probably indicate a short on the device is causing excessive current flow or there is something really hot on the board next to the device that is conducting huge amounts of heat through the device pins.  In either case, a thermal camera would be the ideal debug tool, but you can do pretty well with a thermocouple.  Try taping the thermocouple directly to the device package while the system is powered down.  You can then sanity check if the device is heating up well beyond the ambient temperature while the system is running (the temperature right on the device package should be close enough to the junction temperature for a sanity check).

Hello,

Thanks for your quick reply.

To reply to your question we have a TMS version.

We have measured at 25°C ambient with a thermal camera the temperature on the DSP; we measure about a bit less of 50°C (on the package) when we get the value from the DSP  we get about 117.

We have done also the measurement with a thermocouple on the package where we get 120°C (with 85°C ambient). It was in a previous version of our product where the dissipation where not perfect.  At that time we where getting a value of the temperature junction of about 200.

Is it normal to get a so high difference ? I suppose no, right ?

Do you have a table on what we should measure on the ADC before adjusting the gain and offset ?

Which delta could we add between the temperature we have on the package and the junction ?

Do we have to modify the offset of this measurement ? How can we be sure that this has been trimmed by TI ?

Note:
We are using the 2 cores at 200MHz and also HRPWM at those frequencies.

I don't want to have a precise value I just want to be sure that in max ambient temperature we don't go above critical junction temperature on the DSP !!

Thanks.

Hi Fabien,

It is possible to calculate the expected junction to package temperature delta based on the device power consumption, but what you are seeing is way beyond expectations. I am pretty sure the issue is with reading the temperature sensor.

Have you tried using the unmodified "adc_soc_epwm_tempsensor" example in ControlSUITE? The only thing you may have to change is the VREFHI voltage in this line:

//Initialize the temperature sensor
//Note: The argument needs to change if using a VREFHI voltage other than 3.0V
InitTempSensor(3.0);

Some other quick questions:

*What VREFHI voltage are you using?
*Is this a custom board or TI development kit? If a custom board, what does the VREFHI driving circuitry look like?
*Are you using the ADC for any other tasks, and does it seem to be working? (It is probably a good idea to get the ADC working first, since the temp. sensor depends on the ADC)

The VREFHI was not properly defined in the software VS the board.
It seems that we have found our issue.
Thanks for your quick feedback.