TMS320F28386D: DCC configuration

Hi Clement,

You can refer to the device TRM for the DCC value calculations. You can also use the C2000ware driverlib which provides APIs that calculates these values configures the DCC registers. DCC_verifyClockFrequency function for one-shot verification and DCC_continuousMonitor for continuous monitoring.

Regards,

Veena

I obviously already refered to the TRM before writing and the tolerance part is where I am not sure (both the tolerance of the INTOSC1/INTOSC2 and our crystal one), the TRM isn't clear enough about that.

I don't see the two functions you are refering to in the dcc.h/dcc.c files of C2000Ware_3_02_00_00.

But using those functions is anyways not an answer as I need to justify the computation (we are developing for the aeronautical domain so we don't use your code base).

Best regards,

Clément

Hi Clément,

The functions are present C2000ware v3.04.00.

In the TRM steps, the Tolerance is desired tolerance for DCC error. Lower the tolerance, longer the window.

Along with the DCC error, there is a possibility of error due to frequency variance of the clock used for DCC comparison. In case of using INTOSC1/2, the frequency variance is +/-3%

From device datasheet:

I believe other steps are clear.

Regards,

Veena

Just to be sure when configuring the DCC, the clock source 0 is the element to be monitored or is it the clock source 1 one or can it be both ?

Where would I put into the equation the error on the clock source 0 and the one on the clock source 1 exactly ?

Regards,

Clément

Hi Clément,

Basically, DCC monitors the ratio of the frequencies of clock1 and clock2, whether it maintains a constant value (within allowed tolerance). 

Lets say you want to monitor/measure XTAL frequency using INTOSC1 as  reference, you can configure clock0 as XTAL and clock1 as INTOSC1. Since these are different clock sources, you need to take into consideration, the freq variance of INTOSC1. The steps documented in the TRM mentions adding the frequency variance of the clock which acts as the reference clock. I will check with the module experts whether we need take in freq variance of both the clocks.

Lets say you want to monitor PLL output. PLL is using XTAL as the source and you want to make sure PLL/XTAL maintains a constant frequency ratio. In this case, since both are from the same clock source, you need not add the freq variance error in the DCC calculations.

Regards,

Veena

Hi Veena,

Veena Kamath said:

Lets say you want to monitor/measure XTAL frequency using INTOSC1 as  reference, you can configure clock0 as XTAL and clock1 as INTOSC1. Since these are different clock sources, you need to take into consideration, the freq variance of INTOSC1. The steps documented in the TRM mentions adding the frequency variance of the clock which acts as the reference clock. I will check with the module experts whether we need take in freq variance of both the clocks.

Thanks.

My understanding is that variance of clock source 0 is to be put n the "Window" formula and variance of clock source 1 should be in "Frequency error allowed" formula. But I might be wrong ?

Veena Kamath said:

Lets say you want to monitor PLL output. PLL is using XTAL as the source and you want to make sure PLL/XTAL maintains a constant frequency ratio. In this case, since both are from the same clock source, you need not add the freq variance error in the DCC calculations.

Yes but in that case, you would integrate that as the tolerance in the "Window" formula no ? Or is the tolerance in that formula rather the expectations in terms of monitoring tightness ?

Regards,

Clément

Yes it cleared all remaining doubts thanks !

I must admit your answer is pretty clear, in particular the end examples, you should consider adding these to the TRM it could help others.

Best regards,

Clément