• State Resolved
  • Locked Locked
  • Replies 4 replies
  • Answers 1 answer
  • Subscribers 63 subscribers
  • Views 408 views
  • Users 0 members are here
Support feedback
Options
Options
Related

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.

Csense Parameter Relevant

Roy Zhang
Prodigy 90 points
Part Number: TPS51727
Other Parts Discussed in Thread: TPS53624

Hi TI experts,

I'm validating and tuning a power design basing on TPS51727 while taking notice of CSENSE parameter relevant. Can you please help to reply the following questions for my better understanding?

1. Let's start from Equation 10 in TPS51727 datasheet (SLUS806B), however, no equation is specified for the calculation of REQ in this datasheet, neither in that of the proximate part TPS53624 (SLUSB66). In the attached spreadsheet for TPS51727 thermal compensation network calculation, deriving from D15 equation on DCR sense calculations sheet against Equation 10, REQ seems to be "RSEQU in parallel with Rnet1 (or RP_N)", the latter is "RNTC in series with RSERIES together, then in parallel with RPAR".

Well, what is the exact definition of REQ? Or is it just like the above derivation from the spreadsheet?

2. The attached spreadsheet is the original copy from TI, in which the entered inputs well meet the specifications of TPS51727EVM (SLUU379). It can be seen the parameters of RSEQU, RSERIES, RPAR implemented in TPS51727EVM are pretty closed to those calculated in the spreadsheet, but, with only one exception that CSENSE implemented value (15nF) is much lower than the calculation (25nF), and the latter seems like a very common value in many cases when calculated by the spreadsheet. 

What is the consideration for reducing CSENSE in actual implementation? Is there any pros and cons by doing this?

3. The last one, what are the uses of the extra resistors (R21, R22, R23, R24) in TPS51727EVM schematics? For noise reduction? If yes, how to pick their parameters?

Many thanks and best wishes,

Roy

5050.TPS51727EVM-361 Design Calculation DCR-sense.xls

  • 0
    TI__Genius 11095 points

    Hello, 

    1- Looking at Fig 31 in the TPS51727 datasheet:

    • R_p_n is the equivalent combination of R_ntc, R_series, and R_par... So, R_p_n = (R_ntc+R_series)||(R_par)
    • R_eq is the division ratio formed by R_Sequ, and R_p_n. So R_eq = (R_p_n)/(R_p_n + R_sequ)

    2- The value of Csense can be tuned for the dynamic response of TPS51727. Lower Csense will result in faster response to the load line setting, but may lead to over-reaction.

    3- R21, R22, R23, R24 provide a series resistance with each of the CSP/CSN inputs. In theory, these can be omitted. I guess they were just added on the EVM for practical considerations and maybe the ability to add a bit more input filtering. 

  • 0 in reply to Matt Schurmann
    Prodigy 90 points

    Hi Matt,

    Thanks so much for your quick reply and useful guidance.

    Speaking of Question 1, yes I definitely agree R_p_n = (R_ntc+R_series) || R_par, while I suppose R_eq should "conceptually" be (R_sequ || R_p_n), because if R_eq = R_p_n / (R_p_n + R_sequ) was correct, R_eq became dimensionless so that Equation 10 in TPS51727 datasheet as well as the C_sense calculation in the spreadsheet would not make sense.

    After further derivation in s-domain, I understand the physical meaning of Equation 10 as getting Vc_sense(s) in phase with Vdcr(s) but with a ratio of R_p_n / (R_p_n + R_sequ) in amplitude. Please comment.

    Thanks again,

    Roy

  • 0 in reply to Roy Zhang
    TI__Genius 11095 points

    I think about it like this. 

    If we didn't use temp compensation with the NTC resistor, we would be trying match the time constants so that L/R_dcr = R_seq * C. When we add R_p_n (the combination of the NTC, series and divider resistors), now we need to match L=R_dcr = R_seq * (Scale) * C. And that scaling factor = R_p_n /(R_p_n + R_sequ). And yes, conceptually (R_sequ*R_p_n)/(R_sequ + R_p_n) is the same as the parallel combination (R_p_n || R_sequ). 

    You may also find this paper helpful https://u.dianyuan.com/upload/space/2013/03/24/1364066287-791321.pdf 

  • 0 in reply to Matt Schurmann
    Prodigy 90 points

    Much appreciated for your comment and info sharing. This topic can be closed, I think. :-) 

    Roy