Internal VREG temperature sensor offset (S12ZVML31)

[igorpepelyaev]

Dear Sir or Madam,

I need your help for internal VREG temperature sensor offset at S12ZVML31.

According to the family reference manual v.2.11:

VHT(temp) = VHT(150) - (150 - temp) * dVHT

VHT(150) = 2,4V

dVHT = 5,25mV/C

Voltage regulator uses external bipolar transistor.

Sensor (untrimmed) output signal always has offset 38C (At 30C shows -2C)

Sensor (trimmed with 0x88) output signal always has Offset 18C (At 30C shows 12C)

Sensor (trimmed with 0x8F) output signal is nearly perfect.

According to the reference manual at untrimmed sensor VHT(150) should be 2,4V which is not the case.

Why does VHT(150) deviates from reference manual?

Could VHT(150) deviate from one S12ZVML31 to another?

What is possible to do with it?

Could bandgap reference voltage be somehow used for temperature sensor calibration?

Best Regards

Igor

[danielmartynek]

Hello Igor,

The VHT(150) value is different on every device. Typical value is 2.4V but min and max values are not specified. Therefore, it could be 2.3V for instance. The temperature sensor needs to be calibrated at known temperature. Because the slope of the characteristic also varies in range 5.05 – 5.45 mV/C it is better to perform calibration at minimum and maximum temperature.  

CPMUHTTR register is rather for threshold temperatures THTIA and THTID setting.

You can refer to AN3624 S12X Temperature Sensor, the sensor on S12Z is the same.

Also, very good description is in the AN3031 Temperature Sensor for the HCS08.

Regards,

Daniel

[igorpepelyaev]

Hello Daniel,

Ok, let's look at high temperature interrupt. This interrupt is completely specified in family reference manual with CPMUHTTR = 0x88

                                                  Minimum typical and maximum values

High Temperature Interrupt Assert(9) 120 132 144 C

High Temperature Interrupt Disassert 110 122 134 C

Our measurements (With CPMUNTTR = 0x88) show that interrupt assert takes place at 102C and disassert at 93C. This is definitely outside of the specification. How could it be?

Will it be the case for all S12ZVML31 devices? Or it could be different from device to device?

Best Regards

Igor

[danielmartynek]

Hi,

If my understanding is correct, these values 102C and 93C are calculated using this equation:

VHT(temp) = VHT(150) - (150 - temp) * dVHT,       

where VHT(150) = 2.4V

But if the sensor is not calibrated, the VHT(150) is unknown.

Also, please check your reference voltage.

Regards,

Daniel

[igorpepelyaev]

Hi Daniel,

Values 102°C and 93°C are measured with thermocouple. CPMUHTTR was set to 0x88.

We have placed the thermocouple on MCU and adjusted SW to report CPMUHTCTL_HTDS and temperature measured with internal temperature sensor. Then the temperature was very slow changed from -40 to 150°C. As a result we have now the difference between true temperature and temperature measured by MCU internal sensor (Always 18°C) and we have seen that CPMUHTCTL_HTDS  was set at 102° and was cleared at 93°C (Measured with thermocouple).

Test was done on three PCBs and result is the same (Difference is negligible).

Supply voltage was set to 7.5V and GDU was switched off.

There are now two important questions.

1) Why do we have a Deviation from Family Reference Manual?

2) Is this temperature Offset the same on all S12ZVML31 or it could deviate from device to device?

Regards,

Igor

[danielmartynek]

Hi,

When you say internal sensor, you mean ADC temperature sensor and it was always higher (18C) than the surface temperature, right? 102C and 93C are the values of the thermocouple (surface temperature)?

The Vreg sensor measures internal junction temperature of the MCU. Since the voltage regulator dissipates most of the power it will be in form of a hot spot. The Vreg sensor is close to that hot spot, so it will always read a higher temperature. ADC sensor is not as close to the regulator as Vreg sensor. ADC sensor should be also calibrated.

So Vreg temperature is higher than temperature of ADC which is higher than the surface temperature measured with the thermocouple. To get an understanding, you can change power dissipation of the MCU by changing supply voltage, which should also change the temperature at which you measure the thresholds.

There could be difference on each device but it should be in the range given in RM.

Regards,

Daniel 

[igorpepelyaev]

Hi Daniel,

Thank you for support and patience.

So one more time: in our measurements we have compared the following three values: temperature measured with thermocouple (mounted on MCU serface and evaluated with external, high precision device), temperature measured by VREG temperature sensor (calculated with default values from family reference manual) and high temperature interrupt

Our measurements (With CPMUHTTR = 0x88) show that interrupt assert happens at 102C and interrupt disassert happens at 93C. These values are measured by thermocouple, mounted on MCU surface and evaluated with external high precision device.

Could it be that MCU VREG internal temperature is 18C higher than on MCU surface? 18C seems for me too much. MCU supply voltage was 7.5V and GDU was deactivated.

Generally, how could it be that in family reference manual, high temperature interrupt assert and disassert tolerance is defined and tolerance for VREG temperature sensor output voltage at 150C is not defined? According to AN3624, ihigh temperature interrupt assert and disassert is done as a result of comparison between VBG (Constant value) and VREG temperature sensor output voltage. See figure below. Therefore, high temperature interrupt asser/disassert and VREG temperature sensor (uncalibrated) should actually have the same tolerance...

Best Regards

Igor Pepelyaev

[danielmartynek]

Hi,

The reference manual (Chapter A.1.8) describes relation between junction temperature, ambient temperature, total chip power dissipation and package thermal resistance. As per AN3624 (Chapter 2.2), the calibration of this sensor should be done based on that equation. Surface temperature is not ambient/junction temperature.

The device must be allowed time to stabilize at ambient temperature using an oven or a thermostream. Then based on the equation the junction temperature can be calculated.

I don’t know the power dissipation so I cannot tell if the difference is too high.

Please measure the bandgap output voltage, it is not exactly a constant, it should be in range 1.14; 1.20; 1.28V (Table B-1).

Regards,

Daniel

[igorpepelyaev]

Hello Daniel,

We will use HTDS (High temperature detect status bit) to detect over temperature, as for HTDS we have min and max values in Family Reference Manual.

Thank you for your support.

The topic could be closed.

Best Regards

Igor Pepelyaev