Recommended Brands for NVRAM

I am looking for recommendations on brands of non volatile sram.  I
want a device on the order of 32k - 512k by 16 bits, long non powered
data retention, and a real time clock would be preferable but not
necessary.  Speed isn't very criticical as I am running a 30 MHz bus
clock with controllable wait states, so most memory devices will
probably be a lot faster than my bus.

I have looked at Dallas-Maxim, but their devices are fairly expensive.
Texas Instruments has some, but they are known for supply issues on
their memory.  One manufacturer, ZMD I think, recently got bought up by
another firm, Simtek I think, so they may not be a good choice right
now.

Has anybody here have a good experience with a particular brand, that
is reasonablly priced, that you would recommend?

In article <1137768369.190241.91230@g44g2000cwa.googlegroups.com>, 
no_spam_me2@hotmail.com says...
> I am looking for recommendations on brands of non volatile sram.  I
> want a device on the order of 32k - 512k by 16 bits, long non powered
> data retention, and a real time clock would be preferable but not
> necessary.  Speed isn't very criticical as I am running a 30 MHz bus
> clock with controllable wait states, so most memory devices will
> probably be a lot faster than my bus.

Have you considred the FRAM products by RamTron:

http://www.ramtron.com/doc/Products/nonvolatile/nonvolatile_list.asp?ID=6

I haven't used the parallel versions, but I have used the SPI serial 
versions in a few projects.
> 
> I have looked at Dallas-Maxim, but their devices are fairly expensive.
> Texas Instruments has some, but they are known for supply issues on
> their memory.  One manufacturer, ZMD I think, recently got bought up by
> another firm, Simtek I think, so they may not be a good choice right
> now.
> 
> Has anybody here have a good experience with a particular brand, that
> is reasonablly priced, that you would recommend?

Mark Borgerson

On Friday, in article
     <MPG.1e3aa5486e8dc95c98a0d0@newsgroups.comcast.net>
     mborgerson.at.comcast.net "Mark Borgerson" wrote:

>In article <1137768369.190241.91230@g44g2000cwa.googlegroups.com>, 
>no_spam_me2@hotmail.com says...
>> I am looking for recommendations on brands of non volatile sram.  I
>> want a device on the order of 32k - 512k by 16 bits, long non powered
>> data retention, and a real time clock would be preferable but not
>> necessary.  Speed isn't very criticical as I am running a 30 MHz bus
>> clock with controllable wait states, so most memory devices will
>> probably be a lot faster than my bus.
>
>Have you considred the FRAM products by RamTron:
>
>http://www.ramtron.com/doc/Products/nonvolatile/nonvolatile_list.asp?ID=6
>
>I haven't used the parallel versions, but I have used the SPI serial 
>versions in a few projects.

I have used the parallel ones in a few projects and the larger sizes
are 32KB and upwards these days. Two devices will give 16bits, don't think
they do x16 in one package.

-- 
Paul Carpenter          | paul@pcserviceselectronics.co.uk
<http://www.pcserviceselectronics.co.uk/>    PC Services
<http://www.gnuh8.org.uk/>              GNU H8 & mailing list info
<http://www.badweb.org.uk/>             For those web sites you hate

It looks like from their web page that they offer a 128K and a 32K.  I
don't see any reason why I couldn't get away with using two of them in
parallel.

They claim that they have fixed the limited write cycle issue and now
offer "virually unlimited" rewrite capacity.  If this is true, I don't
have a problem with them.

I am presently waiting for the local Rep to contact me to discuss the
application.

Thank You for the suggestion!

Noway2 writes [about FRAM chips from Ramtron]:
> It looks like from their web page that they offer a 128K and a 32K.  I
> don't see any reason why I couldn't get away with using two of them in
> parallel.

Yes, that works fine.

> They claim that they have fixed the limited write cycle issue and now
> offer "virually unlimited" rewrite capacity.  If this is true, I don't
> have a problem with them.

Note that the endurance of FRAMs is not just write endurance like EEPROM
and Flash.  It is read AND write endurance.  They work like core memory;
readout is destructive but the part internally does a restore operation
so this is transparent to the user, other than the fact that reading the
data enough times can make it fail.

However, for the 3.3V parts, the endurance is so high that it is not
possible to exceed the endurance in the operating life of the part.
Thus "virually unlimited".  (When I first saw that phrase, I thought
they were using "virtually" as some kind of weasel-word, but it's not.)

> I am presently waiting for the local Rep to contact me to discuss the
> application.

Unless you want to buy lots of them, no need to talk to a rep.  Just
buy the parts from Newark.  (Unfortunately Digikey and Mouser don't
distribute Ramtron.)

It good to hear from someone who is knowledgable about the subject.  I
am always a bit leary of what the company and its reps tell you, or
rather what they don't tell you.

I had a similar reaction when I saw the word "virtually" as that it was
a form of handwaving.  I supposed the actual number depends on a lot of
variables that are outside of the manufacturer's control.

I also appreciat  you suggesting that I look at Newark.  I don't
normally look there so I didn't even think about them.  Digikey and
Mouser are my two biggest places to look, followed by Arrow and then
Allied (if needed).  Ocassionally I will hit nu-horizons or something
but not very often.

Since my last post, I did speek with a rep who gave me a price quote of
about $10 per 32k device, which is about 1/3 of what I am paying for a
Dallas NVRAM.  They are sending me a couple of each of the 128K and 32K
devices and I am thinking I may integrate them into the next pass of
the board.  The downside to this approach is that I need to add an RTC
chip, but those are fairly plentifull.

Noway2 wrote:
> I also appreciat  you suggesting that I look at Newark.  I don't
> normally look there so I didn't even think about them.

Years ago dealing with Newark was a big hassle.  I've been relatively
happy with them in the past two years, though.  I think the competition
from Digikey and Mouser has really forced them to improve their customer
service.

> Since my last post, I did speek with a rep who gave me a price quote of
> about $10 per 32k device, which is about 1/3 of what I am paying for a
> Dallas NVRAM.  They are sending me a couple of each of the 128K and 32K

Yes, in samll quantities they're about $10 for the 32K*8 part and $25 for
for the 128K*8 part.

I'm told that the Simtek parts are less expensive, but I haven't tried
them.

The Freescale MRAM part also looks interesting, but AFAIK it is still
unobtanium.

Eric Smith wrote:
> Noway2 wrote:
> > I also appreciat  you suggesting that I look at Newark.  I don't
> > normally look there so I didn't even think about them.
>
> Years ago dealing with Newark was a big hassle.  I've been relatively
> happy with them in the past two years, though.  I think the competition
> from Digikey and Mouser has really forced them to improve their customer
> service.

It's not so much "improve customer service" but "give appropriate
customer service".

Newark and Allied and the like all dealt well with big institutional
buyers by assiging them a dedicated salesperson(s) who attempted to
support them by dealing through the buyer's purchasing deparment. The
model does not map well to small customers where the engineer does not
want to deal with purchasing, the salesperson, etc. but just wants the
chips here the next day!

Digikey (first) and Mouser (a little later) figured out that not every
customer wants this hand-holding.  Newark started getting the religion
in the very late 90's. Arrow and Avnet mostly followed. Allied still
assigns a salesperson to each and every customer... (like me, even
though I buy like $50 a year through them!)

Tim.

I wrote:
> Years ago dealing with Newark was a big hassle.  I've been relatively
> happy with them in the past two years, though.  I think the competition
> from Digikey and Mouser has really forced them to improve their customer
> service.

Tim wrote:
> It's not so much "improve customer service" but "give appropriate
> customer service".

Yes.  It's just that in this case, "appropriate customer service" is in
fact very much improved over what they used to provide to small
customers, in at least four ways:

1.  Online ordering on the web.  I don't need to talk to a salesperson,
    who might not be available when I am.  In particular, they can
    accept orders 24x7, though I don't expect them to process them
    outside normal business hours.

2.  Accepting credit cards as payment.  Years ago they insisted that
    I had to submit a credit application; they didn't want to take
    plastic or ship C.O.D.

3.  Fast order fulfillment for in-stock items.  This is the area in
    which I suspect competition from Digikey and Mouser have had the
    most effect.

4.  On the one recent occasion when I actually needed to talk to a
    person, it was easier to do so than when I tried to deal with them
    years ago.  I think this is a side-effect of #1; since few of the
    small customers need to talk to a person, it's not a major expense
    for them to do it.  Years ago when all transactions involved
    speaking to a person (or at least a person having to read a
    mailed-in order), the average cost of processing a transaction must
    have been much higher, thus giving them an incentive to discourage
    small orders from other than well-established customers.

If I were a large customer, I would still expect them to assign me
a salesperson.

Eric