• State
  • Locked Locked
  • Replies 4 replies
  • Subscribers 65 subscribers
  • Views 337 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.

TPS7A11: Output Impedance

Ivan Perino
Prodigy 10 points
Part Number: TPS7A11
Other Parts Discussed in Thread: TPS74801

One or the most typical method to design a suitable PDN is the target impedance method. The datasheet does not contain any graph/chart related to impedance output.

It is indeed possible to make a profile of the output impedance with AC analysis by using any of the Spice model available (even when it is pre-configured for Transient Simulation).

However, my question is: Is that AC curve something I can trust on? because it seems that the spice model is not initially intended for AC analysis, just transient.

Analyzing the curve, and modeling the LDO as a simple V-R-L circuit (for PDN simulation in Hyperlynx), the resistance is about R = 3.28 mOhm and inductance L= 1.4uH.

The frequency at which the impedance starts sloping up is 290Hz. Even with 22uF as the bulk capacitor, there is a peak of about 358 mOhm around 28.5KHz

----

So, this LDO seems to present a flat response just to 300Hz, then, the bulk capacitor delivers the power for higher frequencies. However, to do so, the bulk must be quite big, in the order of several miliFarads. Trying with other LDO like TPS74801 for example. The performance is quite better, I mean, it provides flat impedance up to 1.12KHz

My question Again. can I trunt in these curves? Should I characterize the LDO in my lab?

Ivan

  • 0
    TI__Genius 12365 points

    Hi Ivan, 

    There is the following application note: Extracting a Lumped Output Impedance Model With SIMPLIS or SiMetrix

    However, not all of our PSpice models are as accurate as we wished and do not model all the characteristics of our devices. I would highly recommend characterizing this in the lab, as this will provide better results and be more suitable for your intended application. 

    Best, 

    Edgar Acosta

  • 0 in reply to Edgar Acosta1
    Prodigy 10 points

    Dear Edgar,

    Thank you so much for your response. I know the document you have indicated, it is a very good explanation to do the same I have done. So, as you pointed out such a document, I assume that the SPICE model for the LDO is correct and accurate enough to get AC analysis. Because the SIMPLIX tool uses the model of the LDO under verification, on the contrary, there would not be any curve at all.

    I have used the SPICE model of the LDO with PSPICE for TI (instead of Simplix) to do the same analysis described in the document. So I guess I can trust on it? And that is my question. Because the same author of the document (Mr John DeFiore) indicates:

    One caution is that the results are only as good as the model. Some LDO models are built on a
    simplified framework and may give poor results.

    Regarding measuring in the lab, it is really a concern, given the time it takes and it is just for only one LDO! (I have many different LDO and many other analysis to run in order to get my design done on time). However, in case the SPICE model is accurate enough and the AC analysis is something trusty, that is all I need to know.

    From my point of view, I believe that such information is really important and should be at least on the datasheet as it is the PSRR and many others. After all, Mr  John DeFiore himself indicates:

    System-level validation engineers are increasingly concerned with the output impedance characteristics of
    the power supplies on their boards

    Thank you so much for your support.

  • 0 in reply to Ivan Perino
    TI__Genius 12511 points

    Hi Ivan,

    Edgar will get back to you tomorrow, thanks for your patience

  • 0 in reply to John Schneider
    TI__Genius 12365 points

    Hi Ivan, 

    Thank you for your patience. 

    Just as John DeFiore stated, the results are as good as the model, which we are also working on getting better and more accurate models. One way to verify that the extracted output impedance is trustworthy is to perform a load transient just as in "Extracting a Lumped Output Impedance Model With SIMPLIS or SiMetrix". 

    When doing so, if the results match or closely match what is shown in the DS, then we can trust the model. 

    Some of our models are not as accurate specially since they only model certain characteristics, therefore, we highly recommend doing bench evaluations. 

    I understand the importance of the output impedance and we have been implementing this in some our newer LDOs as part of the DS. 

    Best, 

    Edgar Acosta