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.

TM4C129XNCZAD: SDRAM selection for TM4C129X

Mastermind 8695 points

Replies: 10

Views: 1782

Part Number: TM4C129XNCZAD

The TM4C129X comes with strong suggestions for using SDRAM with and LCD and the EPI.

What features are required for achieving speeds necessary to drive a 800x480 color LCD? This LCD controller uses one byte to define pixels. It will need bytes at a rate half the pixel clock. Suppose the pixel clock is 30MHz, the LCD controller reads bytes at 15MHz.

Access time does not seem to matter since the EPI only runs at 60MHz. The max access time at those speeds is 17ns; slower than most SDRAM specs I see.

16 bit data vs 8 bit data is obviously twice the speed, but the EPI bus only has 31 signals, so there seems to be limits on how many data and address lines you get.

I have been told "the buffer access is limited the SDRAM burst mode is needed to provide enough data quick enough". Does Tivaware automatically make use of SDRAM burst, or is special configuration necessary? The TIvaWare Users guide only discusses burst in relation to Host Bus 16 mode.

10 Replies

  • Here is a timing diagram of SDRAM read timing using EPI. It is from page 833 of the datasheet. It shows reading 4 bytes in 7 EPI clock cycles.

    Best Regards,
    Bob Crosby

  • In reply to Bob Crosby:

    Bob Crosby
    It shows reading 4 bytes in 7 EPI clock cycles.

    Thanks for the drawing & explanation.     I note: first the row data being presented, then the column data, a "NOP", and then 2, 16 bit (AD[15:0] MCU reads!    The two 16 bit transfers read (32 bits, total) is the equivalent of 4 bytes (as you state) ... (also 32 bits) - is this understanding correct?      Again thank you...

  • In reply to cb1_mobile:

    The address lines are also connected to the data lines this example TI schematic for the TIDM-TM4C129XSDRAM www.ti.com/.../tidrdx1.pdf

    I believe we get 2 bytes here assuming you pick a RAM chip with 16 data IO.
  • In reply to Peter Borenstein:

    In TI's example schematic, a 0Ω 0402 resistor is in series with the clk signal. What is this for? Do I need it?

    www.ti.com/.../tidrdx1.pdf
    www.ti.com/.../tidrdx2.pdf
  • In reply to Peter Borenstein:

    I would suspect that resistor - in some (non-zero) resistance - provides a "better impedance (or level) match" between MCU's clock & RAM.    (and such provision (saves) a pcb "hack" should the resistor be required - perhaps w/"certain" RAM chips.)

    When you write, "get 2 bytes (via 16 bit data I/O)" vendor's diagram shows, "back to back" such reads - which (I believe) yields "2 bytes, 2 times - thus 4 (total) bytes... (I asked for vendor confirmation as I do not use this MCU)

  • In reply to cb1_mobile:

    Here's a diagram from the same datasheet that shows a few more reads in "burst mode". AD[0:15] are changed to be driven in when data is on the bus. DQML and DQMH are signals which enable SDRAM output data pins.

  • In reply to cb1_mobile:

    cb1_mobile
    I would suspect that resistor - in some (non-zero) resistance - provides a "better impedance (or level) match" between MCU's clock & RAM.    (and such provision (saves) a pcb "hack" should the resistor be required - perhaps w/"certain" RAM chips.)

    Or the same footprint accommodates a chip w/o that input, or there was a layout limitation and physical jumper was used to avoid increasing the number of layers.

    It's also possible that some chips need an inductive filter and the resistor is a placeholder, although that's an awfully small inductor.

    Robert

    Perhaps TI could enlighten?

  • In reply to Robert Adsett72:

    It looks like this was added as a series termination resistor to avoid reflections on the clock line. In the end it looks like they did not need to add resistance, hence the value is zero Ohms.

    To answer the original question, is it required? In the case of the TI layout, no. However, it is not a bad idea to add it for a new layout, particularly if the CLK trace is long. It is easy to put in a zero Ohm resistor if no additional series resistance is needed.

    Best Regards,
    Bob Crosby

  • In reply to cb1_mobile:

    cb1,
    Yes, that is two 16-bit reads, thus 4(total) bytes. Also, I "LIKE" that you added the burst mode timing diagram. Thank you!

    Best Regards,
    Bob Crosby

  • In reply to Bob Crosby:

    That make sense Bob, thanks.

    And as you say it's easy to put in a zero ohm resistor (and cheap). Certainly easier than adding after the fact.

    Robert

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.