• State Resolved
  • Locked Locked
  • Replies 10 replies
  • Subscribers 82 subscribers
  • Views 1630 views
  • Users 0 members are here
Support feedback
Options
Options
Similar topics

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.

Power Suite SFRA Library

Alexandre Jousse
Prodigy 100 points
Other Parts Discussed in Thread: SFRA, CONTROLSUITE, TMS320F28335

Hi,

I am currently trying to perform a frequency analysis of my power converter and I have a few issues.

First of all, why does the gain transmitted to the SFRA module is a ratio ( Vout / 4096 ) and do we get the real transfer function of the system?

Also, I am stuck with the way the frequency steps are calculated. I tried to go from the definition of step multiply (10^(1/step per decade)) to get my frequency array but the bode diagram I have does not match my theoretical bode calculation.

Thank you very much for your help

Alexandre Jousse

  • 0
    TI__Mastermind 43730 points

    1. "Gain transmitted to the SFRA module" do you mean the injection amplitude by this?? 

    2. SFRA measures the frequency response of the plant, this is a technique from system identification field. Once you have the frequency response you can curve fit the response to a transfer function. Scripts for which are available in SFRA package.  

    3. Steps per decade is the number of steps in a decade starting with the start frequency, for example if you have start at 100 and steps per decade as 30 then you will have 30 points on which SFRA will compute the frequency response from 100 to 1000. You can verify this by looking at the FreqArray variable in the watch window. SFRA init routine will populate with the correct points at which the frequency response is computed. 

    4. What kind of system are you analyzing? buck, boost, inverter? In our experiments we have been able to validate many systems using SFRA. 

  • 0 in reply to Manish Bhardwaj
    Prodigy 100 points
    First thing, thank you for this quick response.

    1. SFRA COLLECT function has two inputs: SFRA Duty with the small signal injected and the output voltage for exemple if you're analyzing the frequency response of this voltage. In the documentation example in open loop, it says :
    "//Read ADC and computer Fbk Value
    Vout1_Read = (float32)Vout1R/(4096.0);"
    Thus I believe that by looking at this Voltage you can calculate the magnitude of the frequency response and trace the Magnitude Bode Diagram.

    But why do we have to divide it by 4096 (ADC max converted number) ? Doesn't it make the magnitude a normalized one ? (from 0 to 1) and not the real magnitude of the system?

    3. Thank you for this explanation and it matches what I thought about this topic

    4. I am analyzing the output voltage of a boost converter and by giving the result a quick look, my resonant frequency is not the one I find theoretically.
  • 0 in reply to Alexandre Jousse
    TI__Mastermind 43730 points

    1. Yes the reason to divide by 4096 is to normalize the values, it's easy to design control loop software this way, we follow this approach on all our designs and collateral as it enables scalability in software design. I will recommend using normalized values. For example today you are designing a 20KW supply and tomorrow a 1 KW , your software change will be minimal is you follow normalized concept. Additionally in power system  analysis this is a common technique for analysis as well. 

    4. You will need to pay attention to the model, whether DCM CCM will all matter when using SFRA. I have used it on the boost converter (PFC) and it has worked fine. Also note there will be some margin of error. 

  • 0 in reply to Manish Bhardwaj
    TI__Mastermind 43730 points
    Just to give you some inputs, typically verifying model with measured SFRA will take a few iterations.. make sure you are accounting for all the gains in the system including ADC scaling etc.. as the SFRA measures the plant as seen by the digital controller
  • 0 in reply to Manish Bhardwaj
    Prodigy 100 points

    Thank you very much for these informations and I will take a deeper look at all the gains of the system.

  • 0 in reply to Alexandre Jousse
    Prodigy 100 points

    Hi,

    I have a slightly different problem now. I would like to use the GUI and need to connect the GPIO 28-29 (SCI com) to my XDS100v2 but I don't know which pins are the FTDI Port B TX and RX on the XDS... 

    Thank you in advance,

    Alexandre Jousse

  • 0 in reply to Alexandre Jousse
    TI__Mastermind 43730 points

    You can refer to any of our EVMS

    C:\ti\controlSUITE\development_kits\~controlCARDs\TMDSCNCD28377D_v1_3\R1_1_A

  • 0
    Intellectual 310 points
    Alexandre Jousse, I am needing to do exactly you are doing. May you send the ccs files in .zip to this my email: evertonsx@hotmail.com? I thank you so much. I wait your return.
  • 0
    Intellectual 310 points
    Sirs, I would like by some way get a simple example to operate the dsp tms320f28335 experimenters kit (that cost more than $99). We have a team with this dsp. Every one in our team needs do the frequency response analyze for the power converter. So, If someone here give us at least one way to get a simple code to f28335 with the SFRA GUI, we will thank so much. Perhaps, in the future, we could buy more pieces for this dsp.
  • 0 in reply to Francisco Reis
    TI__Mastermind 43730 points
    Francisco, I have replied to your post on the other thread. Let's use just one thread for the conversation.