One or two SDRAM chips for a DM640 32-bit memory implementation?
Started by ●April 7, 2005
Hi all,
Reference designs with a 32-bit wide EMIF usually implement the
solution with two SDRAM chips.
Is there any particular reason why a single 32-bit wide SDRAM is not
used?
I'm trying to reduce part count / PCB space as much as possible, and
am wandering if there are any caveats with 32-bit wide SDRAM chips
connected to EMIF...
Thanks!!
Jean-Pierre
Reply by ●April 7, 20052005-04-07
>Hi all,
>
>Reference designs with a 32-bit wide EMIF usually implement the
>solution with two SDRAM chips.
>
>Is there any particular reason why a single 32-bit wide SDRAM is not
>used?
>
They're probably easier to get and in nicer packages for probing :-)
>I'm trying to reduce part count / PCB space as much as possible, and
>am wandering if there are any caveats with 32-bit wide SDRAM chips
>connected to EMIF...
>
If they are very finepitch or (F)BGA packages, probing will be hard.
Otherwise, it's probably a win all the way - less devices on the nets
means less capacitance, easier routing, no stubs to worry about, smaller
PCB, less power consumption...
Cheers,
Martin --
Martin Thompson CEng MIEE
TRW Conekt, Stratford Road, Solihull, B90 4GW. UK
+44 (0)121-627-3569 : martin.j.thompson@mart...
http://www.conekt.net/
>
>Reference designs with a 32-bit wide EMIF usually implement the
>solution with two SDRAM chips.
>
>Is there any particular reason why a single 32-bit wide SDRAM is not
>used?
>
They're probably easier to get and in nicer packages for probing :-)
>I'm trying to reduce part count / PCB space as much as possible, and
>am wandering if there are any caveats with 32-bit wide SDRAM chips
>connected to EMIF...
>
If they are very finepitch or (F)BGA packages, probing will be hard.
Otherwise, it's probably a win all the way - less devices on the nets
means less capacitance, easier routing, no stubs to worry about, smaller
PCB, less power consumption...
Cheers,
Martin --
Martin Thompson CEng MIEE
TRW Conekt, Stratford Road, Solihull, B90 4GW. UK
+44 (0)121-627-3569 : martin.j.thompson@mart...
http://www.conekt.net/
Reply by ●April 7, 20052005-04-07
--- In c6x@c6x@..., "jeanpierrepoulin"
<Jean_Pierre_Poulin@s...> wrote:
>
>
> Hi all,
>
> Reference designs with a 32-bit wide EMIF usually implement the
> solution with two SDRAM chips.
>
> Is there any particular reason why a single 32-bit wide SDRAM is
not
> used?
>
> I'm trying to reduce part count / PCB space as much as possible,
and
> am wandering if there are any caveats with 32-bit wide SDRAM chips
> connected to EMIF...
There are no real disadvantages technically. But I think you will
find that you can not get as large a capacity with the 32 bit wide
chip as with the 16 or 8 bit wide chips. Typically the 32 bit wide
chips are specialty devices for telecom and embedded apps where space
is at a premium and so the price can be a premium as well. The 8 bit
wide chips are pretty much commodity chips and will have the best
price. Also, supply may be tighter and lead times longer. Finally,
as soon as demand picks up for any reason, even if it is just in the
other size chips, you will see supply tighten up and prices rise on
the specialty parts first. They are all made on the same fab, so fab
capacity is limited.
