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.

TPS75003: Is TPS75003 suited for XILINX, Spartan7?

Part Number: TPS75003
Other Parts Discussed in Thread: TIDA-050000, TIDA-00716, TPS65023B, TPS650250, TPS62097, TPS568215, TPS62067, TPS65023

 One of my customers is looking for a PMIC can be used for XILINX, Spartan7 for thier net products.

 Could you reply me whether TPS75003 is suited for XILINX, Spartan7 or not?

 Also could you please tell me other TI PMIC device if TPS75003 is not suited?

 Your reply would be much appreciated.

 Best regards.

 Kazuya. 

  • Kazuya-san,

    Officially, our PMIC recommendation is that you follow TIDA-050000 for powering Spartan-7.

    However, if you plan on implementing Spartan-7 without DDR memory (many designs do not require DDR with Spartan-7) then it is possible you will only need to power 3 rails:

    • VCCINT = 0.95V or 1.0V
    • VCCAUX, VCCADC = 1.8V
    • VCCIO = 3.3V

    The issue with using TPS75003 is that VFB1,2 = 1.22V for the 2 DC-DC controllers, meaning it cannot generate voltages lower than 1.22V at the output.

    TPS75003 is designed to power Spartan-3, and according to UG394 (and TIDA-00716) it may be suitable for powering Spartan-6, but it will not work for Spartan-7.

    As a result, you can either modify TIDA-00716 to use TPS650250 for powering Spartan-7, or you can modify TIDA-050000 as needed to power Spartan-7 completely from the TPS65023B. These 2PMICs are very similar, and either one will work in your design. You will know for sure after you use the Xilinx Power Estimator (XPE) to determine the load current on the VCCINT rail:

    • If the current required by VCCINT is <1.6A, TPS650250 can provide power to all the rails of Spartan-7
    • If the current required by VCCINT is <1.7A, TPS65023B can provide power to all the rails of Spartan-7
    • If the current required by VCCINT is >1.7A, then you will need an external DC-DC regulator in addition to either PMIC

  •  Hello Brian,

     Thank you very much for your reply and I'm sorry to be late respose due to a long national holiday.

     I see. I will tell them both of those reference designs.

     Thank you again and best regards,

     Kazuya Nakai.  

  •  Hello Brian,

     The customer downloaded schematic data from TIDA-050000 site. 

    https://www.ti.com/tool/TIDA-050000

     They found the following 3 pdf files.

    TIDA-050000(006)_Sch.PDF

    TIDA-050000(007)_Sch.PDF

    TIDA-050000(008)_Sch.PDF

     They think that (007) file is for Spartan-7.

    But they couldn't find any difference between these files except title name.

     Is (007) for Spartan-7?

     Thank you and best regards,

     Kazuya Nakai.

  • Nakai-san,

    Yes, Variant 007 is for Spartan-7.

    The schematic (copy attached) is specific for Spartan-7 and matches TIDA-050000 Design Guide, diagram for Variant 007 on page 4.

    TIDA-050000E2(007)_Sch.PDF

    What your customer is probably seeing is that the cover page of each schematic (006, 007, and 008) all show the Artix-7 Block Diagram. This is because you cannot create a Variant for a cover image in Altium. However, the rest of the pages for each schematic PDF are different, matching the name of the Variant in the documentation (Integrated power supply ref design for Xilinx Artix-7, Spartan-7, and Zynq-7000 Design Guide).

  • Also, it should be noted that Variant 006 (for Artix-7) and Variant 007 (for Spartan-7) use the same building blocks (TPS62067, TPS568215, and TPS65023B) to power the FPGA. Note that TPS62067 can be replaced by TPS62097, but TPS62097 was not released until after we released this TI Design.

    Depending on the number of I/O rails needed (1.8V, 2.5V, 3.3V, etc), it is possible your can reduce Variant 007 further and only use TPS65023B + 1 External DC-DC. Most Spartan-7 designs do not require DDR memory, so this rail may also be unnecessary.

    If the Xilinx Power Estimator (XPE) says the current required by the core rail (VCCINT) is <1.7A, it is possible Variant 007 can be reduced to only use TPS65023B by itself.

  •  Hello Brian,

     I checked TIDA-050000E2(007)_Sch.PDF you attached. On page 3 in the document, VDCDC1_A, VDCDC2_A, VDCDC3_A of TPS65023B

    are set to 1.2V(as V1A), 1.8V(as V2A), 3.3V(as V3A). These divider resistor value are same as TIDA-050000E2(006)_Sch.PDF

    and TIDA-050000E2(008)_Sch.PDF.

     As the result, I couldn't find any different point between TIDA-050000E2(006)_Sch.PDF, (007) and (008). These schematic are same, don't there?

     Does the customer need to calculate divider resistor value to adjust output voltage for Spartan 7 power rail (3.3V, 1.8V and 2.5V)?

     Thank you very much and best regards,

     Kazuya.

       

  • Kazuya,

    You are correct, the TPS65023 output voltages are fixed and there are not variants for the resistors in the different designs. This is to reduce complexity and there is a note showing the resistor calculation for generating output voltage.

    If you refer to this webinar we did with Xilinx, we are positioning the TIDA-050000 as a philosophy for using building blocks to create a scalable power design that will work for many Xilinx FPGA and SoC families.

    You are saying that on page 3 of the schematic:

    • DCDC1 (V1A) = 1.2V
    • DCDC2 (V2A) = 1.8V
    • DCDC3 (V3A) = 3.3V

    So to modify the design for Spartan-7:

    • DCDC1 = 3.3V (copy resistor divider from DCDC3)
    • DCDC2 = 1.8V (copy resistor divider from DCDC2)
    • DCDC3 = 2.5V (re-calculate resistor divider)

    The only resistor divider you need to calculate is for 2.5V, and the rest you can copy from the existing resistor dividers (but re-assign the net name for the rail as needed). 

  •  Hello Brian,

     Thank you for your reply. I understood that R17 and R24 are needed to change for 2.5V output.

     Could you please tell me what is recommended value for R17/24?

     Thank you again and best regards,

     Kazuya Nakai. 

  • R24 = R17*(Vout/Vref - 1) = R17*(2.5V/0.6V - 1) = 3.167*R17

    R24 = 100k, R17 = 316k

           -OR-

    R24 = 49.9k, R17 = 158k

    The tolerance of these resistors should be 1% or better.

  •  Hello Brian,

     Thank you very much for your many help!

     Best regards,

     Kazuya.