TMS320F28035: questions for internal ADC reference and AIO usage ?

Aki Li said:

I only found the below spec regarding the internal reference: Overall gain error and ADC temperature coefficient.

The ambient temp range for their application is about -25~85C, and in some critical conditions, they need to consider -40~105C. Is there any concern about the temp drift? Besides, as for the accuracy, is there any spec defined? Also, I wonder whether its accuracy is related to the accuracy of the 3.3V VDDA supply.

The ADC datasheet parameters apply to the full supported operating range of the device.  So the Overall gain error should stay within the specified +/- 60 LSB across supported VDDA voltages and across supported temperatures -- it already includes the effects of the temperature coefficient.

Aki Li said:

2. AIO usage

a. Does using AIO as output have less cross-talk effect for adjacent analog signals, compared to using AIO us input?
b. Note that it is not suggested to use AIO with high frequency toggling, so is it okay to use it for 2KHz output signal to control LED?
c. Is there any extra delay to use AIO as a status indicator(input), compared to GPIO?

The AIOs use VDDA + VSSA for their reference voltages (versus VDDIO + VSS for GPIOs).  This increases the potential for noise on the AIOs to reach the internal VDDA + VSSA supplies, which may then contribute noise to the ADC conversions.  I would personally not recommend using the AIO for a 2kHz output if there is another GPIO available, but the decision ultimately depends on the noise tolerance of the end system.  There is no additional delay expected for the AIOs.

Aki Li said:

In the datasheet, it is not suggested to use digital signals with sharp edges(high dv/dt) to connect the AIO. Is there any suggested dv/dt range? Is it safe to use 100us for the rising/falling delay for the digital signals?

This are no specific recommendations for dv/dt.  The layout and noise tolerance of the system will determine what is acceptable.

Aki Li said:

1. How does internal reference 3.3V comes from? Does it take directly from the 3.3V VDDA supply? We needs to know whether the accuracy of VDDA can affect the accuracy of internal reference.

The internal reference is generated using a bandgap circuit powered by VDDA. The datasheet specifications apply to the entire ADC system (including any bandgap error or drift) across supported operating conditions.

In other words, there is no need to worry about how the reference may be affected by VDDA. The ADC is expected to satisfy the datasheet performance as long as the device is regulated within the Recommend Operating Conditions.

In case the intent of this line of questioning is to predict the exact accuracy of the ADC for specific operating points, we do not recommend this approach because it is not reliable and we will not support the effort because it is beyond the bounds of the datasheet.

Aki Li said:

2. Customer is considering whether to use AIO for input or output. Is there any suggestion for this? I am not sure whether there is any difference between using as input or output regarding the noise contribution.

The output function is potentially more prone to noise because it will redirect energy from the analog supplies to drive external loads. This can be mitigated by reducing the drive requirements (for example: higher R, lower C).

High frequency activity (for both input and output) has the potential to couple back into the analog supplies. It is best to limit the AIO activity to relatively static signals. For example, to signal (output) a persistent good or error state, or to monitor (input) an enable signal that indicates when some background task should execute.

Aki,

This is proprietary design information that we cannot share here.  The ADC is designed to reject VDDA noise, but we do not have numbers to share on F2803x.  On newer devices, this is specified as PSRR in the datasheet.

-Tommy