• State TI Thinks Resolved
  • Locked Locked
  • Replies 5 replies
  • Answers 1 answer
  • Subscribers 42 subscribers
  • Views 494 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.

Original question:

TIDA-01168: Poles&Zero compensation strategy for digital outer voltage control loop

TIDA-01168: computing digital compensation coefficients

Ponsonby Rivers
Intellectual 620 points
Part Number: TIDA-01168
Other Parts Discussed in Thread: LM5170, LM5170-Q1, SFRA

We have stability issues with our converter which is based on the TIDA-01168. What is the correlation between the coefficients used in the software (Buck mode shown below) and the LM5170 design calculator? What parameters were entered in the LM5170 calculator to obtain these coefficients?

/* Buck mode digital compensation constants */
#define B0_BUCK (2.602812)// (0.443017) /* B0 constant */
#define B1_BUCK (0.040566)//(0.264689) /* B1 constant */
#define B2_BUCK (-2.562246)// (-0.17832) /* B2 constant */
#define A1_BUCK (0.918233)// (0.466008) /* A1 constant */
#define A2_BUCK (0.081767)// (0.533991) /* A2 constant */

  • 0
    TI__Expert 5120 points

    Ponsonby, hello. We have fielded your inquiry and our design owner will review and respond to your inquiry within one to two business days.  It is currently late Friday evening in Germany (where this design is owned) so please allow us until Tuesday at the latest to respond.

    Best Regards,

    John Fullilove

    Reference Design Operations

  • 0
    TI__Intellectual 1450 points

    Hello Ponsonby, 

    I asked the product line for help with this request. It has been a while since I worked on the design and I need to refresh what I did there three years ago.

    You should first investigate if the inner current loop is stable. This is easily done using an electronic load operating in constant resistance mode and keeping the voltage on the ISET pin constant (e.g. pushing 12A to the load set to 1ohm). The current loop must be stable and the step response has to be free of overshoots.

    I believe the most complete answer is available in this excellent videotraining from Garrett Roecker:

    https://training.ti.com/average-current-mode-control-bidirectional-dcdc-systems

    Alternatively, some materials from Biricha power are available:

    https://e2e.ti.com/cfs-file/__key/communityserver-discussions-components-files/171/Presentation_5F002D005F00_Mr._5F00_Ali_5F00_Shirsavar.pdf

    Starting from the slide 36 the document discusses how to get the linear differential equation from the analog transfer function.

    Best regards, 

    Jiri Panacek 

  • 0 in reply to Jiri Panacek85
    Intellectual 620 points

    Jiri,

    Can you please answer the following questions:

    (1) What is the correlation between the coefficients used in the software and the LM5170 design calculator?

    (2) You've given a method on how to obtain stability however, how did you tune the Converter? For example, how did you deduce the A2 coefficient to be 0.081767 instead of say 0.091767

  • 0 in reply to Ponsonby Rivers
    Intellectual 620 points

    Jiri,

    Can you please provide an update with my previous post?

    Thank you in advance.

  • 0 in reply to Ponsonby Rivers
    TI__Intellectual 1450 points

    The LM5170-Q1 calculator returns the frequency of poles and zeros in the type-2 compensator. A digital equivalent of the type-2 compensator is a 2p2z controller. A one of the widely adapted design process has following steps:

    • design a stable controller in the analog domain, express the transfer function H(s)
    • Convert the transfer function to digital domain using the bilinear transformation (H(s) to H[z])
    • Convert the discrete transfer function H[z] to the linear diference equation.

    The material below shows a step-by-step example starting from the page (27). This example covers unfortunately Type-3 compensator.

    https://processors.wiki.ti.com/images/7/78/Seminar_manual_v1_0rl.pdf

    Attached is the analysis for the type-2/2p2z controller.

    The SFRA tool should be able to this for you:

    https://www.ti.com/tool/SFRA

    For the second question - there is always some trial and error. Do not expect these coefficients matching your design or any calculator. Before you start tuning the loop it is important that you understand the analog counterpart in detail. (gain+phase margin and their effects). The math behind this is quite exact but the development process uses a lot of simplifications and assumptions. For instance, we don't expect the ESR of the output capacitor changing. In reality it changes a lot therefore it moves the ESR zero, etc.

    I personally find the analysis in the time domain more practical (evaluating undershoot/overhsoot/settling time). Nevertheless, the frequency part is important to get into the right region of values.

    e2e-lm5170-q1-compensator-coefficients.docx