TMS320F280049: SFRA question

Three more questions:

5. matlab script file, read the data from excel sheet, it will read to 102th row, but there is only 101 rows in excel file, is it a typo?

6. Regarding to use matlab to compare the Curve filtted vs measure result, there are much differences for the phase vs frequency, why?

7. The application note says it's better to run SFRA in ram to habe the best performace, but the example project didn't run from RAM, what's specific purpose of example project?

5. Yes please adjust it to be accurate to the no of rows you have in the excel. 

6. MATLAB has a setting to enable phase wrapping, you may want to check with that enabled, 360=0degrees 

7. The project is FLASH and we copy the SFRA routines to ramfuncs to enable faster execution. I am not sure what the discrepancy here is it aligns with the document. 

1. In the configuration you have shown the SFRA will measure 2p2z * Td where Td is the digital delay which will affect the phase at high frequency, if your control /ISR rate is 100kHz you can observe the effect of this digital delay. Hence the effect you are seeing is expected. 

2. Typically you need to add injection that the system can sustain and give good results, it is empirically determined. If the gain in the system is high typically a lower injection is enough, otherwise, a higher injection may be needed to improve SNR. However, a higher SNR can cause the system to vary significantly from the operation point and this is also not desirable. It needs to be adjusted empirically for each system , I have use INJ from 0.5% to 2% in typical systems.

3. Seems to me something in SW is saturating when you inject during Ref 0.5, 

4. We just use polling method in background for the SCI hence you do not see PIE setting for it.  I think you can disable this interrupt without any issue. 

Hi Manish,

Thanks for your great support here.

3. I have done some test on it, and call DCL_runDF22_C1 without clamp feature. And set injection amplitude to 0.002, 0.005, 0.008, and 0.02. The result is related with injection amplitude when set the Ref to 0.5. Please see below result. Here is my testing code.example1_stb_f28004x_f32_coff.zip

7. From the SFRA example (C:\ti\c2000\C2000Ware_DigitalPower_SDK_2_00_01_00\libraries\sfra\examples), I see the SFRA_F32_inject and SFRA_F32_collect are located at FLASH instead of RAM, therefor those 2 functions are not run from ram, is it?

Address and symbols in map file.

0 00081d1c __SFRA_F32_collect
0 00082029 __SFRA_F32_inject

After this you should see the below happen to the code

.TI.ramfunc 
*          0    00086000    000004b7     RUN ADDR = 0000c000
                  00086000    0000020e     sfra_f32_coff.lib : sfra_f32_background.obj (.text)
                  0008620e    000001bc                       : sfra_f32_config_reset.obj (.text)
                  000863ca    00000064                       : sfra_f32_collect.obj (.text)
                  0008642e    00000026                       : sfra_f32_inject.obj (.text)
                  00086454    00000020                       : sfra_f32_config_reset.obj (SFRA_F32_Data) [fill = 0]
                  00086474    0000003f     driverlib_coff.lib : flash.obj (.TI.ramfunc:_Flash_initModule)
                  000864b3    00000004                        : sysctl.obj (.TI.ramfunc)

Hi Manish,

Thanks for your great support here. I think the 7th could be closed.

How about the 3rd question? I set the Ref = 0.5, and use different injection amplitude, 0.002, 0.005, 0.008, 0.2, the sweeping result seems have differences. The result seems not very good when Ref = 0.5 with using 0.002 INJ, but Ref = 0.03 with 0.002 INJ is good. What cause the differences?

Hope everything is fine with you and your family. Best wishes in current special situation.

Hi Manish,

May I have further comments from yours for the 3rd question? Thanks...