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.

TPS62088: Expected thermal requirements, and using TPS62823 for LPDDR4

Part Number: TPS62088
Other Parts Discussed in Thread: TPS62823, TPS62826, TPS62135, TPS54339E, TPS62827

Hi,

I read the Chris Glaser's article on LPDDR4 regulators (great article), and I need to ask a few questions.

1) The TPS62088 datasheet gives a very low value for THETAjc, which AFAIK may be due to measurement difficulties.

Assuming Vin=3.3V, Vout=1.1V, efficiency=0.85, Iout=3A, we'd lose about 0.6W in the device.

Can I meaningfully use the small THETAjc to estimate the case temperature? Is it intended that, for this device, most of the heat is conducted into the PCB?

2) It seems to me that the TPS62823 would fit my LPDDR4 application as well as the TPS62088. I'd like to ask: is there any reason why the TPS62823 wasn't included in Chris Glaser's LPDDR4 article? In other words, is there any reason not to use TPS62823 to make a 1.1V rail for LPDDR4?

3) The TPS62823 has a larger package (QFN). Which package would be considered to have better thermal performance for a fanless system - the TPS62088's WCSP or the TPS62823's QFN?

Thanks,

 - Bren

www.planetanalog.com/author.asp

  • Hi Brendan,

    Thanks for reading my paper!

    Yes, TPS62823 (and TPS62826) are larger, QFN options that can be used for this same application. These weren't yet released when I wrote the paper.

    Comparing the D/S, the thermal performance is better with the QFN and they are more efficient.

    This app note discusses thermals in detail: www.ti.com/.../spra953c.pdf Yes, most heat is dissipated into the PCB and not through the case.

    What is your application?
  • OK thanks for that.

    My application is an embedded system, with loads at:
    - 1.1V (LPDDR4)
    - 1.8V (LPDDR4 and I/O)
    - 3.3V

    We don't require very high efficiency at light-load, as we're not running on batteries. We want good transient response and good efficiency in the 1-3A range. The upstream power-source is a 3.3V DC input.

    So for the 1.1V and 1.8V rails, would you recommend the TPS62823 (or TPS62826) for a new design?

    In your LPDDR4 paper, you discuss 3A regulators. Do you generally find that this is enough for 32b LPDDR4 devices, or also for 64b LPDDR4 devices? Obviously this depends on a lot of things, but what's the common usage?


    We also have a use-case where we want to replicate the same power-tree, but don't have the 3.3V DC input. Instead we have a DC input ranging from 5V to ~20V, and we want to step that down to make an internal 3.3V rail (4A). What would you recommend for that? I saw the TPS54339E, but that looks a little old now. Would you say the TPS62135 has advantages over it?

    Thanks in advance,

    - Bren
  • Thanks Bren.

    Yes, the TPS6282x family is recommended for new designs. It is a higher efficiency, though larger, option for LPDDR4. The TPS62826 offers a pin to pin upgrade to 4A, if needed, with the TPS62827.

    The details of our customer's DDR loads have rarely made it down to me, but I've not heard any issues. I would think the manufacturer would have an estimator tool for current consumption requirements.

    Yes, TPS62135/6 is a good 4A regulator if the Vin is kept below 17V. If it's higher than that, you need a different and bigger device.
  • Chris,

    This has been a great help - thanks!

    - BRen