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.

TPS54318: TPS54318

Part Number: TPS54318


Chris Glaser recently gave me helpful advice on modern TI buck converters for LPDDR4.

Given our challenging thermal requirements, we're now considering the older TPS54318 buck converter: one for 1.8V and another for 1.1V.

I'd like to ask:

- is this converter recommended for new designs?

- is this converter considered appropriate for LPDDR4 applications?

Best regards,

  - Bren

  • Brendan,

    TPS54318 is a current production device. TPS54318 can generate both 1.8 and 1.1 V outputs from a 3.3 V or 5 V input.
    1.1 V is the VDDQ voltage for LPDDR4. What is the 1.8 V rail used for?
  • Hi,

    Thanks for the quick reply.

    1.8V is also needed for LPDDR4 apparently.

    Given that this is a CM regulator, now a few years old, I was concerned that it would be stable at the kind of light-loads expected from LPDDR4 devices.

    Also, I'd like to confirm that it's possible to construct the PDN entirely from ceramic decaps, to meet LPDDR4 rail tolerances (+/- 5%?), in the presence of the kind of worst-case transients expecte dfor LPDDR4, without the risk of stability problems.

    Best regards,

    - Bren
  • Brendan,

    If the circuit is designed correctly, it will be stable even at no load. TPS54318 does not have any special power saving light load efficiency operating modes though, so at light load efficiency is rather low.

    You can design the converter exclusively with ceramic capacitors. 5% is not a particularly stringent transient requirement, but you will still need to take care to balance loop BW and output capacitance to met your goals.
  • We don't need the special light-load modes, as we're not designing this system with battery-saving sleep modes.

    Can confirm that to keep this device in FCCM at light loads, I should check that the current-ripple-ratio is sufficiently high (changing Fsw and L), to prevent pulse-skipping?

    So concerning the Cout vs B/W question, I'd like to get that just right. Could you recommend an app-note or book? Also, would you recommend the use of an FRA for tuning (like the Bode 100)?

    Thanks for the help!

    - Bren
  • Brandon,

    Technically it is possible to cause pulse skipping with an unusually large inductor value. The COMP pin voltage is proportional to the peak switch current. At light loads the peak switch current will be much lower for an unusually large inductor relative to a smaller (recommended) inductor. In that case the corresponding COMP pin voltage can be below the minimum switching threshold and the device will skip pulses. That is an extreme corner case. The datasheet has an inductor sizing formula with Iout in the denominator. If you specify Iout as much lower than the nominal 3 A (say 100 mA), then you will calculate one of these large inductors. I typically will not let Iout in that equation fall below 20% of the nominal 3 A.
  • Regarding loop measurement, there is lots of info on the and elsewhere. I have not looked at them recently as I have been doing this for a really long time, but I can do a search later to see if there is anything that I would particularly recommend. I personally use and recommend Bode 100 (and this comes from a long time Venable user). It is a little complicated to use relative to the Venable analyzers, but I think less expensive and more powereful.
  • OK, that's been useful - thanks for the help!

    - Bren