This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

TMS320F28035-Q1: Oscillator compensation using temperature sensor ADC value got by unsuitable sampling window.

Part Number: TMS320F28035-Q1

Hi experts, 

In customer design, they use oscillator temperature compensation to maintain high quality communication (SCI and CAN). However, they don't implement the workaround of errata "ADC: Temperature Sensor Minimum Sample Window Requirement", i.e. the sampling window size is not larger than 550 ns. 

They want to check: 

1. If not implement the errata's advisory, the oscillator compensation will be not right, how large the impact will be? 

2. If not compensate the oscillator temperature deviation, is there any serious risk for the communication interface operation? 

Since the communication performance will be more significant for bootloader (or OTA will be failed), and the bootloader can't be updated via OTA, so they want to know the impact, and only if the impact is un-avoidable and proposed to be addressed, they will consider to add the compensation and the workaround. 

Thanks, 

Will 

  • Hi Will,

    To answer their questions:

    1. What sampling window is the customer using? This error will depend on the sampling window.

    2. I will need to reach out to a colleague to ask about the maximum allowable clock deviation for SCI and CAN.

    Best Regards,

    Ben Collier

  • Hi Ben,

    They use 7 ADC clocks, the minimum sampling window size shown in datasheet. And the ADC clock frequency is 60MHz. 

    Regards, 

    Will 

  • Hi Will,

    To answer the original questions:

    1. Since we no longer test the temperature sensors at this short sampling window, we are unable to provide reliable data on how large the impact will be, we are only able to say that the data will be unreliable if you do not follow the workarounds in the errata. It is possible that the error will not be so bad if the voltage that is sampled just before the temp sensor is close to the voltage output of the temp sensor. This would be similar to the double-sampling workaround in the errata. Also, the customer could test the error they see on the material they have purchased, but we could not guarantee that this would be the same for all material. 

    2. The risk to the communication interface will depend on the bit rate, with faster bit rates being more vulnerable to deviations in clock frequency. 

    Best Regards,

    Ben Collier

  • Hi Ben, 

    Thanks for your comments. I think customer wants to know the quantitative analysis to support them evaluate the risk. I told them before the similar qualitative suggestion, but they are not accepted it since it only shows possibility but not know the specific impact. 

    If we truly can't provide the data, I will tell them that they must implement the workaround. What do you think?

    Regards, 

    Will 

  • Hi Will,

    That certainly sounds like the safest option, since we cannot guarantee the performance of the Temperature Sensor when it is used outside of datasheet specifications.

    Just curious, is the customer asking about minimum sample times on the temperature sensor because they are developing a new design, or are they troubleshooting on an old design on which they were already using the minimum sample time? 

    Best Regards,

    Ben Collier

  • Hi Ben, 

    Yes, they are troubleshooting on an old design on which they were already using the minimum sample time. And their design departs into bootloader + app. They are evaluating the necessity of this workaround on bootloader, since bootloader can't be updated with OTA, if they must modify the bootloader, they will need to return the equipment to factory. 

    Regards, 

    Will 

  • Hi Ben, 

    Further, customer wants to know if not compensate the oscillator's frequency, i.e. not use API of osc_comp(), how much will the crystal oscillator frequency change with temperature? +- 5%? Need a quantitative value to evaluate it, thanks. 

    Regards, 

    Will 

  • Hi Will,

    I am working on getting an answer for you. Please expect a response shortly.

    Best Regards,

    Ben Collier

  • Hi Will,

    I think this table in the datasheet answers your oscillator frequency drift question:

     

    Best Regards,

    Ben Collier