Forums

Adding RAM to Cypress PSoC

Started by KenLem January 30, 2006
I want to add serial RAM to Cypress PSoC.  I'm looking for something
around 4K Bytes- ( 4096 x 8) 64KBytes (65536 x 8) for < $2.  It doesn't
have to be non-volatile.

I'm planning on adding some RAMTRON FRAM to use a storage but my
appliation needs RAM to run.

Any suggestions?

KenLem wrote:
> I want to add serial RAM to Cypress PSoC. I'm looking for something > around 4K Bytes- ( 4096 x 8) 64KBytes (65536 x 8) for < $2. It doesn't > have to be non-volatile. > > I'm planning on adding some RAMTRON FRAM to use a storage but my > appliation needs RAM to run. > > Any suggestions?
Yeah, in all honesty, if you need more ram on a PSOC, you need a new part. Have you looked at the other processor families?
The part I'm using only has 512 Bytes but PSoC's reconfigurablity and
features are mandatory for what I am doing.  I have no option and do
not wish to change processors.

I'll think of something.

rickman wrote:
> KenLem wrote: > > I want to add serial RAM to Cypress PSoC. I'm looking for something > > around 4K Bytes- ( 4096 x 8) 64KBytes (65536 x 8) for < $2. It doesn't > > have to be non-volatile. > > > > I'm planning on adding some RAMTRON FRAM to use a storage but my > > appliation needs RAM to run. > > > > Any suggestions? > > Yeah, in all honesty, if you need more ram on a PSOC, you need a new > part. Have you looked at the other processor families?
I understand not wanting to change processors.  But the
reconfigurability of the PSOCs is a bit of a red herring.  The idea is
that it can be a cheaper part by letting you reuse the same hardware to
do different things at different times.  But just like with FPGAs,
there are very few designs that can take advantage of that.  Also, the
price advantage of this approach can be very minimal due to the extra
chip area requried to make the features programmable.  But with current
process geometries, processors often are pin count limited; they have
to be a certain size to have the pin connections.  It is not
automatically cheaper to have a smaller processor with the same number
of pins.

In the end it can be just as cost effective to use a processor with
standard peripherals, although it will depend on your unique
requirements.  If you need working ram, it will be very slow to add it
onto the PSOC and require some of that programmable IO which is what
you are trying to optimize.

Digikey has SAM7S32s for $4.  I expect in any quantity that would make
a difference, you can get them for $2.50 or less.  If push comes to
shove, you can add the SAM7 as a smart RAM and you'll get 8 KB!



KenLem wrote:
> The part I'm using only has 512 Bytes but PSoC's reconfigurablity and > features are mandatory for what I am doing. I have no option and do > not wish to change processors. > > I'll think of something. > > rickman wrote: > > KenLem wrote: > > > I want to add serial RAM to Cypress PSoC. I'm looking for something > > > around 4K Bytes- ( 4096 x 8) 64KBytes (65536 x 8) for < $2. It doesn't > > > have to be non-volatile. > > > > > > I'm planning on adding some RAMTRON FRAM to use a storage but my > > > appliation needs RAM to run. > > > > > > Any suggestions? > > > > Yeah, in all honesty, if you need more ram on a PSOC, you need a new > > part. Have you looked at the other processor families?

rickman wrote:

> I understand not wanting to change processors. But the > reconfigurability of the PSOCs is a bit of a red herring. The idea is > that it can be a cheaper part by letting you reuse the same hardware to > do different things at different times. But just like with FPGAs, > there are very few designs that can take advantage of that. Also, the > price advantage of this approach can be very minimal due to the extra > chip area requried to make the features programmable. But with current > process geometries, processors often are pin count limited; they have > to be a certain size to have the pin connections. It is not > automatically cheaper to have a smaller processor with the same number > of pins.
Rickman, Without the reconfigurability, the PSOC would be just another 8-biter, as good as many 8-biters on the crowded market. The reconfigurability is a pure marketing feature created to raise the public interest to the part. There is no cost or performance advantages however this is somewhat different which may seem appealing. Vladimir Vassilevsky DSP and Mixed Signal Design Consultant http://www.abvolt.com
Vladimir Vassilevsky wrote:
> rickman wrote: > > > I understand not wanting to change processors. But the > > reconfigurability of the PSOCs is a bit of a red herring. The idea is > > that it can be a cheaper part by letting you reuse the same hardware to > > do different things at different times. But just like with FPGAs, > > there are very few designs that can take advantage of that. Also, the > > price advantage of this approach can be very minimal due to the extra > > chip area requried to make the features programmable. But with current > > process geometries, processors often are pin count limited; they have > > to be a certain size to have the pin connections. It is not > > automatically cheaper to have a smaller processor with the same number > > of pins. > > Rickman, > > Without the reconfigurability, the PSOC would be just another 8-biter, > as good as many 8-biters on the crowded market. The reconfigurability is > a pure marketing feature created to raise the public interest to the > part. There is no cost or performance advantages however this is > somewhat different which may seem appealing. > > Vladimir Vassilevsky
I belive that was my point. The reconfigurability sounds like a good thing, but in the end it will only be an advantage in a very small percentage of apps. Even then it may not make the part any less expensive than a part that will do the job without reconfiguration.
Cypress do seem to be buying sockets with extra-competitive pricing, in an effort to 
gain market share in the industrial segment. Their fabs are full of Ipod PSOC but they 
are a very minor player in industrial apps.

IIRC, Atmel did this several years ago, just before raising prices 20%

-Andrew M


"rickman" <spamgoeshere4@yahoo.com> wrote in message 
news:1138777219.091505.227920@o13g2000cwo.googlegroups.com...
> Vladimir Vassilevsky wrote: >> rickman wrote: >> >> > I understand not wanting to change processors. But the >> > reconfigurability of the PSOCs is a bit of a red herring. The idea is >> > that it can be a cheaper part by letting you reuse the same hardware to >> > do different things at different times. But just like with FPGAs, >> > there are very few designs that can take advantage of that. Also, the >> > price advantage of this approach can be very minimal due to the extra >> > chip area requried to make the features programmable. But with current >> > process geometries, processors often are pin count limited; they have >> > to be a certain size to have the pin connections. It is not >> > automatically cheaper to have a smaller processor with the same number >> > of pins. >> >> Rickman, >> >> Without the reconfigurability, the PSOC would be just another 8-biter, >> as good as many 8-biters on the crowded market. The reconfigurability is >> a pure marketing feature created to raise the public interest to the >> part. There is no cost or performance advantages however this is >> somewhat different which may seem appealing. >> >> Vladimir Vassilevsky > > I belive that was my point. The reconfigurability sounds like a good > thing, but in the end it will only be an advantage in a very small > percentage of apps. Even then it may not make the part any less > expensive than a part that will do the job without reconfiguration. >
Hi,

Thanks everyone for the responses.  Sometimes it's nice to step back
from ones own solutions to consider alternatives.

>>But the reconfigurability of the PSOCs is a bit of a red herring.
I can see your point however I have used the reconfigurability in a number of appliations. I also like the ability to have 1 processor on my desktop that I can alter as needed instead of having to select a processor with the right mix of features from an entire family (PIC's). Plus, I have uses for the analog portion of the PSoC's such as the low pass filters. In this case, I am using the PRoCl, which is just a PSoC with a built in wireless radio. http://www.cypress.com/portal/server.pt?space=CommunityPage&control=SetCommunity&CommunityID=209&PageID=259&fid=65&rpn=CYWUSB6953 I haven't found another part with the same features for anywhere near the price. What about using the RAM buffers on the dataflash for extra RAM?
KenLem wrote:
> Hi, > > Thanks everyone for the responses. Sometimes it's nice to step back > from ones own solutions to consider alternatives. > > >>But the reconfigurability of the PSOCs is a bit of a red herring. > I can see your point however I have used the reconfigurability in a > number of appliations. I also like the ability to have 1 processor on > my desktop that I can alter as needed instead of having to select a > processor with the right mix of features from an entire family (PIC's). > Plus, I have uses for the analog portion of the PSoC's such as the low > pass filters. > > In this case, I am using the PRoCl, which is just a PSoC with a built > in wireless radio. > http://www.cypress.com/portal/server.pt?space=CommunityPage&control=SetCommunity&CommunityID=209&PageID=259&fid=65&rpn=CYWUSB6953 > > I haven't found another part with the same features for anywhere near > the price. > > What about using the RAM buffers on the dataflash for extra RAM?
What data flash? Have you attached an external flash? You can find serial ram devices with lots of memory. But it will be very slow accessing it. I know Dallas makes "one-wire" devices like this and I think there are a lot of makers of I2C or SPI compatible ram.
Oh, sorry.  I've been mulling over parts and forgot to mention that I
might use 8Mbit Atmel dataflash instead of the Ramtron FRAM.

http://www.digikey.com/scripts/DkSearch/dksus.dll?Detail?Ref=147595&Row=573852&Site=US

I wanted to use FRAM so that I wouldn't have to worry about the number
of writes but if I have enough dataflash, I can just do my own wear
leveling.

Hmmm... Not a big Microchip fan but I see they have RFPics.  I wonder
how they compate to the PRoC?