RAM IC for SX28

Keith M

Hi all,

I'm looking for a serial ram device that can store 32k bytes that I can write to from an SX28 at ~600kbps.· I'd like to read/write one byte at a time.· Besides the basic specs, ease of use is probably the next most important feature I need.· I have zero experience using memory although I understand most of the basic concepts.

Parallel memory chips with 20+ pins·are out of the question as I just don't have the pins, obviously.

Thanks.

Keith
·

Paul Baker

www.ramtron.com has what your looking for, they have 32K serial FRAM availible in SPI and I2C interfaces. I received a few samples of thier 32Kx8 SPI serial FRAMs, but I haven't had the opportunity to play with them yet.·www.newarkinone.com is an authorized distributor of Ramtron.

FRAM is functionally equivalent to SRAM, but it is non-volatile and has near unlimited number of write cycles (measured in many years when a block is perpetually written to, it think·accessing it·at a rate you are looking to do,·the chip will last about 15-20 years).

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
·1+1=10

Post Edited (Paul Baker) : 8/24/2005 2:20:44 PM GMT

LoopyByteloose

Being overseas and a hobbyist, I have stayed away from requesting samples.

I started to make a request at MAXIM's site, but they wanted a lot of occupational information and a business address.

Does the industry frow on hobbyist scrounging for parts via their samples or is this something that they accept as a 'good will' gesture on their part?

I am a bit old school and feel you don't really get anything for free.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
G. Herzog in Taiwan

Paul Baker

Ramtron was the first sample request I made, they never questioned my application as an "Independent Inventor", though I did wait for a couple months to receive them. Perhaps they would have been a little more speedy if they thought they were sending the samples to a large company with potential of selling reels of components. They didn't fullfill my main request for the 128Kx8 parallel FRAM, but that may be due to the part being preliminary status (newarkinone did not have the part stocked, which is why I did a sample request). It doesn't hurt to ask for samples, at worst they will deny the request. I typically just order a part from the distributor if possible, because I see the sample system being mainly for companies who have concrete commercial applications with large scale production.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
·1+1=10

pjv

Hi Keith;

As Paul stated, RAMTRON is your answer. My experience with the SPI type shows it's easy to use and plenty fast.

Cheers,

Peter (pjv)

Keith M

Thanks for the replies.

I think I've narrowed down my choice to the FM25256, 256kbit, 15mhz (should be plenty fast), 4-5.5v. I selected it mainly on capacity and the fact that it runs on 5v.

Paul, where'd you get your samples? Ramtron direct? Newark?

I'm already having trouble locating a source, however:

Ramtron's distributors are:

All American: Indicates they have engineering samples, and Im waiting on a reply email to see if they are available
Future: Minimum order quantity is 100.
Newark: Closest one I could find was low voltage FM25L256, but a month+ leadtime

Digi-key, Mouser: Ramtron isn't listed as a manufacturer

Jameco: Although I can't readily tell, it appears they might sell Ramtron, so waiting on email....

Thanks.

Keith

Paul Baker

I received samples of the FM25256, I placed the request on RAMTRONs website which was forwarded to a distributor. Newarkinone has the FM24C256 in stock which is the I2C version of the same memory. They also have the FM30C256.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
·1+1=10

dkemppai

Kramer said...
Being overseas and a hobbyist, I have stayed away from requesting samples.

I started to make a request at MAXIM's site, but they wanted a lot of occupational information and a business address.

Does the industry frow on hobbyist scrounging for parts via their samples or is this something that they accept as a 'good will' gesture on their part?

I am a bit old school and feel you don't really get anything for free.

I always look at it as being part of their advertising budget. I get a lot of samples. I probably sample a different part every week. I'm just very honest about the quantities, if <1K is an option, that's what I put. I do a lot of samples as personal, and a lot more as buisness (of course, now I have hobby business)

In reality, if you learn how to use a part, and tell a friend who uses it in a commercial application, that still accomplishes the purpose of the sample.

Dan

(Just received 16 samples of ramtron parts at work!)



·

Keith M

I was looking at the datasheet(link here)for the FM25256 and wondered if someone can help here.

Page 7, under Write Operation, the datasheet linked above says, "[noparse][[/noparse].....]Subsequent bytes are data and they are written sequentially. Addresses are incremented internally as long as the bus master continues to issue clocks."

As long as I don't raise CS to terminate the Write operation, can I simply stop sending a clock on SCK until I have more data to write, and then start the clock back up to continue writing where I left off?· I did read the errata concerning suspending SCK in the middle of the data byte --- but that won't apply to me.· I'm writing whole bytes, so I'd never suspend the clock in the middle of the write of a single byte.

My main concern is that the WREN plus the write op-code including the address is really a ton of data, relatively speaking, and I'd hate to have to resend that for each and every byte I needed to write just because they were separated in time.

Page 2, under the SCK pin Description, "Since the device is static, the clock frequency may be any value between 0 and 15 MHz and may be interrupted at any time."·seems to confirm what I'm guessing, but I really have no experience here.

And the related question is, if I can do this, what's HOLD for? [noparse]:)[/noparse]· I'm trying to make this as simple and straightforward implementation as possible using the fewest number of pins.· I'd be happy with the 3-pin minimum!

I think I'm going to be bit-banging here, not using the ISR for the clocks.

Thanks.

Keith

Paul Baker

I too have pondered this, and came to the conclusion that if you know there will be a fair amount of time between writes, its best to apply the hold signal. This should prevent spurrious voltage swings on the line (due to noise etc) causing corrupt data from being written. The main point would likely be if the power is lost, if the hold is asserted the chip knows that fluctuations occuring are not valid data.

Perhaps you can devise a testbed to experiment what the chip will do under various situations.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
·1+1=10

pjv

Hi Guys;

Just back from a week of R&R.

Indeed, things are arranged so you may be temporarily stall the clock/data (while hi or lo) to do something else, and then carry on where you left off, provided you did not raise CS. Every time CS is lowered, a new sequence is started, requiring issue of a command, address, etc.

The stalling technique works very well, as that is exactly what happens when I'm accessing the FRAM, and then the RTOS fires to do someting else, and on task suspension by the RTOS, things merrily continue as if no interruption had occurred.

As for HOLD, I have not used it, so no opinions.

Cheers,

Peter (pjv)

Keith M

Thanks guys for the response.

pjv, I suspected that might be the case based on how the datasheet read. In case you were wondering, I'm still working on my MFM floppy project. No real progress, but I'm trying to take the PC out of the equation by using RAM or a faster USB xfer.