[SI-LIST] Re: DDR DRAM
- From: "john lipsius" <johnlipsius@xxxxxxxxx>
- To: <si-list@xxxxxxxxxxxxx>
- Date: Thu, 9 May 2002 01:13:48 -0700
I agree with Kevin. But it's hard to guarantee. The following is only
my 2 cents.
Also, sounds like 200MHz DDR, not something to attempt without a system
timing budget. The appearance is that a total timing budget isn't available.
See links at the bottom.
That budget, with input from Spice results of asic io, incl. clock chain delay
and delay line logic, package, board traces & terminations, is required...Not
to mention
the noise budget.
If it's not available, then you're flying blind when you have to debug.
If a board layout could fix this (sounds doubtful), it would require doing the
Spice
& timing budget, from which the answer would just "fall out", assuming the
modeling
is accurate.
....Quick 'n Dirty....
> Placing parallel termination at ASIC end (680 ohms each to
> 2.5V and Ground) seems to help the working of DRAM "X".
1. Assuming the above is saying it helps but doesn't fix it, then this points
to those 3 bits'
asic timing being problematic. A pin-configurable delay line would help,
if there is
such a thing. Typically, the controller adaptively adjusts the delay on
pwr up self test, I
believe.
2. Lengthening DataStrobe trace might do the trick, but you have to watch out
for the
effect on the write cycle. Again, timing budget comin' atcha.
As Kevin said, any solution needs to be simulated at 4 corners.
Termination for Point-to-Point Systems
http://www.micron.com/products/technote.jsp?path=/DRAM/DDR+SDRAM&fileID=584
DDR SDRAM timing budget.... it's a long read ... good luck in your quest
http://download.micron.com/pdf/pubs/designline/dl399.pdf
http://download.micron.com/pdf/pubs/designline/DL92b.pdf
-John
>
> I think you need to show the timing analysis for sampling
> read data.
>
> Micron has a very good app note that highlights different ways
> of adding up the numbers.
>
> You mention that both DRAMs meet the jedec spec but your part
> doesn't work.
>
> It sounds like you don't have the right delay line in your asic
> to align the data and strobe appropriately, so that you can sample
> correctly, given the jedec-spec'ed variances and variances in your
> board layout.
>
> Before you can decide what you need to do to make your current
> thing "better"...you need to figure out the full read timing analysis,
> I think.
>
> For instance, if your design can tolerate extra read return latency,
> it may be that you need to lengthen the data strobes on your board,
> relative to the data.
>
> Without seeing the numbers, it's hard to say where you have margin
> to play with. It's possible that your design doesn't have enough
> skew tolerance inside the asic, so that you might not be able
> to work for all corner cases, unless you make more drastic asic changes.
>
> I may be off base. Apologies if so.
>
> -kevin
>
> > Delivered-To: si-list@xxxxxxxxxxxxx
> > Subject: [SI-LIST] Re: DDR DRAM
> > To: si-list@xxxxxxxxxxxxx
> > X-MIMETrack: Serialize by Router on D03NM028/03/M/IBM(Release 5.0.9a
> > |January
> 7, 2002) at 05/07/2002 09:26:52 AM
> > MIME-Version: 1.0
> > Content-Transfer-Encoding: 8bit
> > X-archive-position: 2772
> > X-ecartis-version: Ecartis v1.0.0
> > X-original-sender: ajmani@xxxxxxxxxx
> > X-list: si-list
> >
> >
> >
> > Hello SI-LISTers,
> > Thanks to all of you who replied to my post. A number of questions have
> > been raised and I wish to answer them.
> >
> > 1) I used HP 2.5 GHz probes with HP Infinium 1.5 GHz oscilloscope, using 8
> > GHz sampling rate.
> >
> > 2) I had gone through several Application Notes/other material and decided
> > to use source terminations because of the following constraints:
> >
> > a) I do not have enough board space to provide parallel terminations or
> > both.
> >
> > b) I am already exceeding our power budget, and can't afford the
> > additional current drain of parallel terminations.
> >
> > c) The parallel termination used by me (on the failing bits only) were
> > selected by simulations and were also found to work on the actual card.
> >
> > The problem as I see is that brand "X" DRAM puts data after the Strobe
> > (although still within their specs of 500 ps), whereas brand "Y" DRAM puts
> > data before the Strobe. Hence, I have a problem of matching my card layout
> > to work with both DRAMs.
> >
> > Regards, Ravinder
> > PCB Development and Design Department
> > IBM Corporation
> > Email: ajmani@xxxxxxxxxx
> > ***************************************************************************
> > Always do right. This will gratify some people and astonish the rest.
> > .... Mark Twain
> >
> >
> > ------------------------------------------------------------------
> > 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:
> > //www.freelists.org/webpage/si-list
> >
> > For help:
> > si-list-request@xxxxxxxxxxxxx with 'help' in the Subject field
> >
> > List archives are viewable at:
> > //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:
> //www.freelists.org/webpage/si-list
>
> For help:
> si-list-request@xxxxxxxxxxxxx with 'help' in the Subject field
>
> List archives are viewable at:
> //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:
//www.freelists.org/webpage/si-list
For help:
si-list-request@xxxxxxxxxxxxx with 'help' in the Subject field
List archives are viewable at:
//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:
