[SI-LIST] Re: DDR stubs and non- stuff options

Hello Maheshwari,

I did a design using daisy chain to drive two DDR2 ( above 200 MHz ) DRAM part 
on a 4-layer PCB. The part closer to the controller can be loaded or no-load. 
So I believe it is feasible to implement your design to support MDDR at 133 MHz.

I agree with all the recommendations from the mail-list. Something I did to 
implement the daisy chain design include:

1) use SI tool for pre-layout simulation to develop design guideline 
2)  strictly enforce the typical design guideline such as balancing trace 
length within the DQ group. Use SI tool to understand design guideline priority 
and trade-offs due to tight routing channel on 4 layer board.
3) After getting the prototype board, refine SI model with post layout 
simulation to achieve correlation with the measurement on prototype board.
4) After getting correlation between simulation and measurement, use SI tool to 
develop the optimized settings on driver strength, ODT, delay etc for different 
read/write cases to maximize design margin. Then program the controller with 
the optimized settings.


A balanced T layout daisy chain design will likely yield more timing margin 
when longer stub length is needed. I opted for daisy chain with short stub to 
the 1st DDR part ( longest stub length <100 mils ) instead of balanced T 
because of other design constraints on component placement.

Have fun in your design !


Catherine Xu
__________________________________
Sr. HW Design Engineer, Ph.D
HP - San Diego
_________________________________ 


-----Original Message-----
From: si-list-bounce@xxxxxxxxxxxxx [mailto:si-list-bounce@xxxxxxxxxxxxx] On 
Behalf Of Hermann Ruckerbauer
Sent: Thursday, May 07, 2009 8:49 AM
To: Doug Brooks
Cc: ranjith; si-list@xxxxxxxxxxxxx
Subject: [SI-LIST] Re: DDR stubs and non- stuff options

Hello Maheshwari,

I agree with Doug that a 0.1" stub is most likely not going to cause any
issue. I guess the double loading in the full populated case is going to
be worse in SI than the single loaded case with the stub.
In the double loading case this stub is still there. In this case you
will have the motherboard stub PLUS the DRAM Package stub (I guess this
is DDR1-266 ==>  no OnDie Termination for DQ or CA)  and the Cin  of the
DRAM which is causing much more reflections than just the unloaded
motherboard stub in the non populated case.

regards

Hermann Ruckerbauer
www.eyeknowhow.de








Doug Brooks schrieb:
> Think in terms of the "critical length." You don't say what the rise time
> of the clock is, but at 133MHz the critical length is likely not less than
> 1.5" to 2" (quite likely more). Thus, .1" is such a small fraction of the
> critical length that the stub reflections will not be an issue.
>
> Doug Brooks
> http://www.ultracad.com
>
>
>
>   
>> Thanks for the feedback.
>> We are using mobile DDR. The Maximum clock frequency is 133MHz.
>> While routing dasisy chain, we are estimating 100 mils stub in DDR(1)
>> which close to the processor (see below diagram). If we no mount this
>> DDR(1) this 100 mils additional stub will be there. Is this stub will
>> create an issue in 133MHz DDR operation?
>>
>>       TLA = 1"      stub = 0.1"
>> CPU ----------------o------- DDR(1)
>>                     |
>>          TLB = 0.7" |
>>                     |
>>                     |
>>                   DDR(2)
>>
>> Thanks
>> Ranjith
>>
>>
>> Stubs are never a good idea.  Whether your stubs will be enough to wipe
>> out your margins depends on how fast you intend to signal and what your
>> layout looks like.  You should arrange your layout so that the
>> conditionally stuffed component is the middle one in the chain.  Is this
>> DDR2, or DDR3?
>>
>> Steve
>>
>> maheshwari palanisamy wrote:
>>     
>>> hi team,
>>> I am currently designing a processor module. In this case, I have a
>>> processor
>>> which can support 16bit DDR and supports 2 chip selects.
>>>
>>> To get the benifit of price, I am planning to use both the chipselects.
>>> But
>>> stuff only one DDR for lower memory requirements and will not stuff the
>>> second DDR. Both DDR will be stuffed only for highest memory
>>> configuration.
>>>
>>> In the case of only one DDR been stuffes, will the stubs in the DDR
>>> address &
>>> data create issue because of reflections? And Is this recommended to
>>> stuff
>>> one alone?
>>>
>>> Thanks in advance.
>>> Maheshwari
>>>
>>>       
>> ------------------------------------------------------------------
>> To unsubscribe from si-list:
>> si-list-request@xxxxxxxxxxxxx with 'unsubscribe' in the Subject field
>>
>> or to administer your membership from a web page, go to:
>> http://www.freelists.org/webpage/si-list
>>
>> For help:
>> si-list-request@xxxxxxxxxxxxx with 'help' in the Subject field
>>
>>
>> List technical documents are available at:
>>                 http://www.si-list.net
>>
>> List archives are viewable at:
>>              http://www.freelists.org/archives/si-list
>> or at our remote archives:
>>              http://groups.yahoo.com/group/si-list/messages
>> Old (prior to June 6, 2001) list archives are viewable at:
>>              http://www.qsl.net/wb6tpu
>>
>>
>>
>>     
>
> ------------------------------------------------------------------
> To unsubscribe from si-list:
> si-list-request@xxxxxxxxxxxxx with 'unsubscribe' in the Subject field
>
> or to administer your membership from a web page, go to:
> http://www.freelists.org/webpage/si-list
>
> For help:
> si-list-request@xxxxxxxxxxxxx with 'help' in the Subject field
>
>
> List technical documents are available at:
>                 http://www.si-list.net
>
> List archives are viewable at:     
>               http://www.freelists.org/archives/si-list
> or at our remote archives:
>               http://groups.yahoo.com/group/si-list/messages
> Old (prior to June 6, 2001) list archives are viewable at:
>               http://www.qsl.net/wb6tpu
>   
> ------------------------------------------------------------------------
>
>
> Eingehende eMail ist virenfrei.
> Von AVG überprüft - www.avg.de 
> Version: 8.0.238 / Virendatenbank: 270.12.20/2100 - Ausgabedatum: 05/06/09 
> 06:04:00 
>
>   
------------------------------------------------------------------
To unsubscribe from si-list:
si-list-request@xxxxxxxxxxxxx with 'unsubscribe' in the Subject field

or to administer your membership from a web page, go to:
http://www.freelists.org/webpage/si-list

For help:
si-list-request@xxxxxxxxxxxxx with 'help' in the Subject field


List technical documents are available at:
                http://www.si-list.net

List archives are viewable at:     
                http://www.freelists.org/archives/si-list
or at our remote archives:
                http://groups.yahoo.com/group/si-list/messages
Old (prior to June 6, 2001) list archives are viewable at:
                http://www.qsl.net/wb6tpu
  

------------------------------------------------------------------
To unsubscribe from si-list:
si-list-request@xxxxxxxxxxxxx with 'unsubscribe' in the Subject field

or to administer your membership from a web page, go to:
http://www.freelists.org/webpage/si-list

For help:
si-list-request@xxxxxxxxxxxxx with 'help' in the Subject field


List technical documents are available at:
                http://www.si-list.net

List archives are viewable at:     
                http://www.freelists.org/archives/si-list
or at our remote archives:
                http://groups.yahoo.com/group/si-list/messages
Old (prior to June 6, 2001) list archives are viewable at:
                http://www.qsl.net/wb6tpu

Other related posts: