• State TI Thinks Resolved
  • Locked Locked
  • Replies 4 replies
  • Answers 2 answers
  • Subscribers 99 subscribers
  • Views 1053 views
  • Users 0 members are here
Support feedback
Options
Options
Related

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.

DM365 DVEVM: 22 vs 33Ohm series R's for DDR2 / NAND

NickAllen
Intellectual 525 points

Hi guys,

The DDR2 SDRAM on Sheet 2 & Sheet 17 of the DM365 EVM schematic contains 22ohm resistor arrays in series with the address lines but 33ohm resistors in series with the data bus lines. My initial gut feel is that this is irrelevant but this is very closely followed by my second gut feel which is don't ever make assumptions. There is also the three differential pairs from the DDR2 where 2 of which are protected by pairs of 22Ohm series resistors but the third by a 33Ohm pair.

The NAND data lines are all protected by 33Ohm resistor arrays.

I suppose my question really is; is the 22 or 33Ohm value in each particular case specifically chosen? Does the value selected have to do with the length of tracks / impedence matching? Reflections? Do I ultimately really need any series resistors here? I would like to reduce my unique component count.

Secondly; are there any other DM365 reference designs available that I can use to cross reference the DVEVM?

Many thanks,
NickA

  • 0
    Intellectual 565 points

    The series resistors are damping resistors used to slow down signal edges. They are used to reduce EMI.

     

    Recall that a pulse with fast rise time has stronger high-frequency components than a pulse with slow rise time.

     

    Most prudent engineers will add damping resistors to a DDR/DDR2 design, even if they are populated with 0 ohm resistors or removed completely in the end. The reason is that if you do not include them and run into EMI issues, it is a real pain to add them and may result in ripping up large (if not all) of the routing.

     

    Another way to slow the rise time is to reduce the strength of the tranceiver drivers. As for some datapoints, the DM355 EVM does not include damping resistors, but all DDR2 traces are routed on internal layers (copper plane acting as shield). On the DM355 Leopard board, the designer added damping resistors. On my own board I have DDR2 traces on inside and outside layers but  still decided to remove the damping resistors because there simply isn't enough room between the DM355 and DDR2. However, I reduced the tranceiver driver strength. I haven't run into any EMI issues on my board yet...

     

    JPM

     

  • 0
    TI__Expert 5955 points

    You are correct. The resitor value selected depends on the trace length. You would have to run board simulation to ensure the impedence matching, reflections minimized, etc. TI recommend to follow the DDR PCB layout guideline which can be found here. You can also find the DDR signal termination resistor value in the section 6.10.3.1.9 of the data sheet.

    There are some DM365 reference designs that it can be found here. The Appro has an IPNC product.

    Hope this helps.

    -Tai Nguyen

     

     

  • 0 in reply to Tai Nguyen
    Genius 4530 points

    Hi Tai,

    By board level simulation, do you mean using IBIS model of the DDR and DM365? And also, what parameters within the DDR/ DM365 define using 22 Ohm and 33 Ohm resistors for different lines?. As you said, if EMI is a defining factor then shouldnt it behave similarly for signals having similar rise times?

    Regards,

    Sid

  • 0 in reply to Sid
    TI__Expert 5955 points

    Hi Sid,

    Yes, it is running the HyperLinx using the IBIS models. SPICE simulation is too slow and difficult to even get an accurate model. For terminations, the speed timings, IO driver strength and test loads define using 22 Ohm or 33 Ohm for termination. For DDR2, we use 22 Ohm for termination on our valuation board. Talking to Spectrum Digital regarding to the resistor value, they said that either one should work on their EVM.

    Regards,

    Tai Nguyen