• State Resolved
  • Locked Locked
  • Replies 16 replies
  • Subscribers 83 subscribers
  • Views 1779 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.

TMS320F28379D: High Resolution PWM simultaneous duty and phase control

TWeinert
Prodigy 65 points
Part Number: TMS320F28379D
Other Parts Discussed in Thread: TIDA-010054

Hello e2e Community,
for a current project we need a PWM which we can adjust continuously with high resolution both in the duty cycle and in phase shift. For the high-resolution adjustment of the duty cycle it is necessary to adjust only one edge at each channel of the PWM (A: falling (FEP), B: rising (REP)). To adjust the phase shift, both edges (BEP) of both channels (A&B) of the PWM must be adjusted simultaneously in the same height. It is therefore essential for the conversion that the edges of each PWM channel can be adjusted independently of each other, since we want to combine both operating modes.

I tried many different combinations and read the reference manual several times. As I see it all configurations (no matter if with CMPAHR & CMPBHR or TPPHSHR) lead to the fact that there is only one MEP for each channel of the PWM module. During the initialization you have to decide if this MEP will affect the rising, the falling or both edges. If both edges are selected, they can only be adjusted by an identical value. However, we would need a total of 4 MEP per PWM module (2 per channel) so that we can adjust each edge of both channels individually.

As I see it, the desired PWM cannot be implemented. Switching during operation from BEP to REP and FEP (duty cycle control mode) causes the high-resolution part of the phase shift to be lost. A change from REP and FEP to BEP (phase shift mode) leads to the loss of the high resolution part of the duty cycle. This is not an option because there are jumps in the PWM during the changeover.

Therefore, the question arises whether there is an understanding problem on our part or whether there is this limitation in the HRPWM? Do you possibly have an example program or a configuration that should work?

  • 0
    TI__Mastermind 43730 points

    Hi, 

    Can you confirm if the below is correct

    1. Using Up-down count mode?

    2. High resolution period is not required ?

    3. high resolution dead time is required? i.e. xB is complimentary of xA always?  

    4. We do have SPS for DAB converter here http://www.ti.com/tool/TIDA-010054, but we do not show duty cycle control on that, it is fixed to 50%. 

  • 0 in reply to Manish Bhardwaj
    TI__Mastermind 43730 points

    Is the goal to get waveforms similar to https://ieeexplore.ieee.org/stamp/stamp.jsp?arnumber=6658916 fig 5, c and d ?

    Is duty cycle itself fixed to 50% or that is also modulated? 

  • 0 in reply to Manish Bhardwaj
    Prodigy 65 points

    Thanks for the response.

    1. I tried both

    2. HR Period ist not needed

    3. Yes, B is always complimentary to A. HR Deadband is needed.

    Both links you posted are not showing what we like to achieve. I attached a picture of the resulting bridge voltages and currents.

    To achieve these voltages we need to control duty and phase of all PWMs. A and B in each module are always complementary. Duty is not fixed to 50 %.

    Is the source code for TIDA-010054 freely available? I can't find it.

  • 0
    Prodigy 65 points

    My configuration for HR duty would be:

    EPwm1Regs.HRCNFG.bit.EDGMODE = HR_FEP;
EPwm1Regs.HRCNFG.bit.CTLMODE = HR_CMP;
EPwm1Regs.HRCNFG.bit.EDGMODEB = HR_REP;
EPwm1Regs.HRCNFG.bit.CTLMODEB = HR_CMP;

    And my configuration for HR phase would be:

    EPwm1Regs.HRCNFG.bit.EDGMODE = HR_BEP;
EPwm1Regs.HRCNFG.bit.CTLMODE = HR_PHS;
EPwm1Regs.HRCNFG.bit.EDGMODEB = HR_BEP;
EPwm1Regs.HRCNFG.bit.CTLMODEB = HR_PHS;

    It seems that both configuration are not compatible.

  • 0 in reply to TWeinert
    TI__Mastermind 43730 points

    Hi, 

    Thanks for the additional explaination. We have a lot of legacy on this module and hence some of the fileds can be confusing. 

    CTLMODE shows this as a binary option between phase high resolution and compare / duty high resolution. Which is indeed the case with TBPHSHR usage.

    But we do have a workaround to implement a high resolution phase shift using TRREM register. the TIDA-010054 page i pointed to was using this TRREM register but we moved this to TBPHSHR because only phase and deadband were needed to be high resolution. 

    http://www.ti.com/lit/ug/spruhm8i/spruhm8i.pdf

    see 15-40.

    The write to TRREM directly bypasses the TBPHSHR, so you can use CTLMODE HR_CMP.

  • 0 in reply to Manish Bhardwaj
    TI__Mastermind 43730 points

    Yes, the TIDA-010054 is available in limited release right now, the SW link should be live soon (either by end of this week or early next week).

    We had it live yesterday but had to take it down due to some logistics. 

  • 0 in reply to Manish Bhardwaj
    TI__Mastermind 43730 points

    Just updating the Software for TIDA-010054 is available for request on the TID page. 

  • 0 in reply to Manish Bhardwaj
    Prodigy 65 points

    Thanks for letting me know. At the moment I have not yet come to test the operation with TRREM. I still don't understand how the register works. Is there any further literature?

  • 0 in reply to TWeinert
    TI__Mastermind 43730 points

    Please request the software for TIDA-010054, and we will share a previous version with you.

  • 0 in reply to Manish Bhardwaj
    TI__Mastermind 43730 points

    The TRREM is documented in the datasheet

    http://www.ti.com/lit/ug/sprui33c/sprui33c.pdf

    the SW we will share will help you with the context and application. 

  • 0 in reply to Manish Bhardwaj
    Prodigy 65 points

    I'm gonna go check out the SW soon. The data sheet description doesn't really clarify what is going on in the register.  At least I don't quite understand it (Table 18-42). I understand that the register has 11 bit and the lower 8 bit can be used as an alternative to TBSPHSHR.  But from which calculation does the register represent the remainder? What happens in the upper 3 bit? Is the "Translator" the background calculation from the HR registers for the MEP and normally not open to the user?

  • 0 in reply to TWeinert
    TI__Mastermind 43730 points

    The upper three bits are used to determine the mode, up count, down count etc..  

    If you use the TBPHHR register then all these are hidden to the user, but for the use case you have TBPHHSR cannot be used and hence we need to directly write to TRREM. 

    I will see if we can provide the description of the bits 10,9 and 8  in a more clear manner.

  • 0 in reply to Manish Bhardwaj
    TI__Mastermind 43730 points

    Please see below

                    Asymmetrical Mode:

                                            TRREM[7:0]     = TBPHSHR[15:8]

                                            TRREM[10,9,8] = 0,0,0

                    Symmetrical Mode:

                                            TRREM[7:0]     = TBPHSHR[15:8]

                                            TRREM[10,9,8] = 0,0,1

  • 0 in reply to Manish Bhardwaj
    Prodigy 65 points

    I have now also tried the TRREM register and cannot reach the desired PWM. It seems that the selection of the edge can't be outsmarted by this either. You have to decide whether both edges or only one edge is adjusted. Also the DAB example does not help me, because only the phase is adjusted. It seems like the hardware is limited to use only one MEP per channel.

  • 0 in reply to TWeinert
    TI__Mastermind 43730 points

    Hi, 

    I will arrange for this to be handled offline, once you local representative is back in office. 

    -Manish

  • 0 in reply to Manish Bhardwaj
    TI__Mastermind 43730 points

    This request has been completed offline. Closing this thread.