TMS320F28335: undefined ADC tolerances

Dirk,

Given the numbers, it is difficult to say for certain if there is a problem with the ADC.  Is this at room temperature?

Am I correct in understanding that +0.95% error with the 1.5V voltage sample is equivalent to about +20LSB of absolute error?  The error is within the datasheet error of +/-30LSB Gain error plus +/-15LSB Offset error, but I would typically expect better performance than this.

Would the customer be able to perform an in-system offset error correction procedure to see how it compares against the factory trim value?  Does changing to internal reference mode affect the conversions?

Every ADC for this device is trimmed at wafer level.  Most of the datasheet error range is expected to be from temperature drift.

-Tommy

Tommy,

customer comes back, pls refer to excel attachment:

As an ADC input we used either an AA battery (around 1.5V) or short-circuited the ADC input to GND (0V) and cooled or heated the controller.

In a first step an external reference voltage of 2.048V is used (REF3020) and the factory calibration data for offset and gain are used.

In a second step the internal reference voltage of the controller is used.

The factory calibration data is used here, as well. Both measurements show that the temperature drift is relatively small.

However, even at room temperature a deviation of 10…13LSB appeared.

This is caused by a gain error, as the third measurement shows that the offset correction at room temperature should be at +2LSB but is at -2LSB.

The fourth and fifth measurement are taken with an ADC input of 0V and a higher ADC offset (ADCOFFTRIM is changed to higher value so that a negative drift can be measured). These measurements show that the offset has almost no temperature drift.

According to measurements the variation in temperature has a small influence on the accuracy of the ADC. However, there is a deviation, even at room temperature.

If there are any questions about these tests, feel free to contact me.

Here Excel attachment:

ADC of TMS320F28335ZJZQ.xlsx

Dirk,

Thank you for the thorough experimental data. From these values, it does appear that the gain error exceeds datasheet specs at high temperature.

Can you provide some context as to how common the problems are? For example, has there been acceptable performance for X number of boards built, but now a set of Y are abnormal? Or is this widespread for a new board design?

I am trying to determine if these are potentially defective units or if there might be something in the system that needs additional attention.

-Tommy

Dirk,

Are there any updates with respect to the quantities involved?

-Tommy

Dirk,

Is this still an ongoing issue?

-Tommy

Tommy,

customer is still about evaluating different options wie discussed offline.

Pls let case open.