RM57L843: OTP Temperature Sensor Calibration Values

Hello Demitri,

The link below demonstrates how to use the on-chip temperature sensors:
www.ti.com/.../spna216.pdf

We will look at your measurement.

Regards,
QJ

The document has been reviewed and my code lines up with what is described.

The OTP Calibration values seem to be wrong.

Sure:

float s2_sum_temp = otpREG->S2TEMP1 + otpREG->S2TEMP2 + otpREG->S2TEMP3;
float s2_sum_val = otpREG->S2TEMP1VAL + otpREG->S2TEMP2VAL + otpREG->S2TEMP3VAL;
float s2_temp_x_temp = ((otpREG->S2TEMP1 * otpREG->S2TEMP1) + (otpREG->S2TEMP2 * otpREG->S2TEMP2) + (otpREG->S2TEMP3 * otpREG->S2TEMP3));
float s2_temp_x_val = ((otpREG->S2TEMP1 * otpREG->S2TEMP1VAL) + (otpREG->S2TEMP2 * otpREG->S2TEMP2VAL) + (otpREG->S2TEMP3 * otpREG->S2TEMP3VAL));

s2_slope = (((s2_temp_x_temp * 3) - (s2_sum_temp * s2_sum_temp))) / (((s2_temp_x_val * 3) - (s2_sum_temp * s2_sum_val)));
s2_offset = (s2_sum_val - (s2_sum_temp / s2_slope)) / 3;

float s3_sum_temp = otpREG->S3TEMP1 + otpREG->S3TEMP2 + otpREG->S3TEMP3;
float s3_sum_val = otpREG->S3TEMP1VAL + otpREG->S3TEMP2VAL + otpREG->S3TEMP3VAL;
float s3_temp_x_temp = ((otpREG->S3TEMP1 * otpREG->S3TEMP1) + (otpREG->S3TEMP2 * otpREG->S3TEMP2) + (otpREG->S3TEMP3 * otpREG->S3TEMP3));
float s3_temp_x_val = ((otpREG->S3TEMP1 * otpREG->S3TEMP1VAL) + (otpREG->S3TEMP2 * otpREG->S3TEMP2VAL) + (otpREG->S3TEMP3 * otpREG->S3TEMP3VAL));

s3_slope = ((s3_temp_x_temp * 3) - (s3_sum_temp * s3_sum_temp)) / ((s3_temp_x_val * 3) - (s3_sum_temp * s3_sum_val));
s3_offset = (s3_sum_val - (s3_sum_temp / s3_slope)) / 3;

float offset = 0.0;
float slope = 0.0;
uint16_t val = 0U;


switch(sensor)
{
case (TEMPERATURE_SENSOR_2):
adcREG2->GxSEL[1U] = 0x80000000U; /* start the temperature sensors */
offset = s2_offset;
slope = s2_slope;
break;
case (TEMPERATURE_SENSOR_3):
adcREG2->GxSEL[1U] = 0x40000000U; /* start the temperature sensors */
offset = s3_offset;
slope = s3_slope;

break;
default:
return -1;
}

/* wait for conversion */
while (!(adcREG2->G1SR & 1)) { /* Spin Wait */ }

val = adcREG2->GxBUF[1U].BUF0 & 0xFFFU; /* Get Value */

adcREG2->GxSEL[1U] = 0x0; /* set up the two temperature sensors */


return ((float)val - offset) * slope;

Are there any updates on this?

Should we be doing in house temp calibrations and not trusting OTP?

Thanks!

Hello Dmitri,

I went to the top of this thread to get more information on the original problem. If your calibration values from OTP are reading 0xFFFF they were never programmed. This is most likely due to you having an early device sample rather than a production device. Does the device have TMP instead of TMS on the part number marking on top of the device? TMP parts are pre-production where there is a certain percentage of production tests that have been implemented and some that have not. In this case, it would appear the temp sensor calibration hasn't been done.

My recommendation is if the values in OTP are 0xFFFF then preform your own calibration, but if not, then they can be used as they are. This could be a workaround just for your development and taken out when full production parts are in use. In fact, if you read 0xFFFF from OTP for a production device, I would fail it during manufacturing test and return to TI for evaluation since this means part of the production process was skipped.

Thanks for the reply Chuck,

The problem is that the values are not 0xFFFF on the chip I am using, only the HDK supplied by TI.

The values out of OTP are:

S2TEMP1 303K
S2TEMP1VAL 1957

S2TEMP2 233K
S2TEMP2VAL 1533

S2TEMP3 398K
S2TEMP3VAL 2520


Using those values and the reading Read Value: 1331, the Converted Value is: 198.877K which is not realistic. This makes sense from the OTP values if you look the value 1533 reads 233K and my ADC is reporting 1331.

I tried it again to double check on the other temperature sensor with values:

S3TEMP1 303
S3TEMP1VAL 1882

S3TEMP2 233
S3TEMP2VAL 1464

S3TEMP3 398
S3TEMP3VAL 2456

I get a Read Value: 1281, Converted Value: 202.751K, again this makes sense looking at the OTP values as a 1464 translated to 233K.

The Temperature sensor read values definitely change in a cool or warm environment as expected during testing, so the sensors work. The OTP values programmed are not letting me sanely decode the temperature. This can be mitigated by doing an inhouse calibration of temperature readings rather than trusting TI OTP values. Is that the recommended procedure? Is it just the first few batches are flawed?

Ive tried on a few RM57 chips with the same results, the HDK chips did not have the values programmed in flash so I could not test on TI supplied boards.

I think this may be an issue with the ADREFHI being 3.3v during OTP Calibration while the system I am on uses 5v ADREFHI.

this could be the case. I don't know under what condition they have calculated the OTP values. I will check this with our OTP experts and get back to you.

Thank you!