Email: Password: Remember Me | Create Account (Free)

Back to Subject List

Old thread has been locked -- no new posts accepted in this thread
Hans Van Pelt
05/31/12 03:04
Read: 626 times

#187584 - Hold your horses
Responding to: ???'s previous message
Helo Kurt,

Before you run off to start using EEPROM, I would like to point you two major snags related to non-volatile memory in general and EEPROM in particular.

1) EEPROM (and other floating gate memory types, like flash) has a property called ENDURANCE. That means that after a couple ten thousand to a million writes, depending on your luck, your EEPROM will physically wear out, after which it will seize to be reliable. A guaranteed minimum number is always given in the datasheets, which is usually 20000 to 100000. So depending on how often you write to your EEPROM, this will limit the operational lifespan of your device to no more than (pushing it...) a couple of DAYS!!! (if it relies on the proper operation of the EEPROM chip) But yes, I have actually seen units fail because of designers underestimating EEPROM endurance problems every few months.

2) Non-volatile memory contents survives power cycles. That's good, but also a headache at the same time. Because what if the power cycles in the middle of a write sequence? How do you know, after the power comes back, that the data in your non-volatile memory is incomplete or dodgy? You will face the comparable issue at very first startup?

There are hard- and software answers to these issues which do turn non-volatile memories, even EEPROM, into very practical solutions, but they may not be as easy as you imagine. To put it very short: you can not simply willy nilly replace your RAM with EEPROM, and then expect your project, with little other changes, to now survive power cycles. I do hope you realise that.
I suggest that you look into battery backed up RAM, which at least doesn't have the endurance issue. Most if not all 8051 derivatives have an inherent property which allows you to preserve the internal RAM contents using an external battery delivering just a few volts. See the 8051 "bible" on that.
You WILL need to properly support any non-volatile memory device through an additional layer of software though, much as if it were a mass storage device working as a backup for your RAM than as direct working memory.

"We have it coded to use external memory, and we need that memory to still be there if even if the device is turned off for a while."

Maybe you will understand now that some requirements all too readily point in the direction of a certain simple and convenient solution (in this case: external memory which retains its contents), but that they only do so because they come form a person who is unfamiliar with the ins and out of the subject matter. You will see a lot of customers are like that...

List of 29 messages in thread
More Ports or Non-Volatile RAM      Kurt Adam      05/23/12 17:47      
   Numbers always help      Jim Granville      05/23/12 22:17      
   Trying to read your mind      Michael Karas      05/23/12 22:48      
   Write down your project requirements      David Prentice      05/24/12 04:54      
      Details      Kurt Adam      05/24/12 05:48      
         Most projects do not need external memory      David Prentice      05/24/12 06:20      
            SILabs have 8-porters        Erik Malund      05/24/12 07:30      
               Don't overlook the additional memory ...      Richard Erlacher      05/24/12 09:09      
            Further Details and Thankyou      Kurt Adam      05/24/12 08:44      
               AT89S8253 ?      Jim Granville      05/24/12 16:04      
                  This worked. Thanks      Kurt Adam      05/30/12 17:54      
                     Hold your horses      Hans Van Pelt      05/31/12 03:04      
                        F-RAM      Per Westermark      05/31/12 05:31      
                           Yes! I like 'em, and there's no write delay      Richard Erlacher      05/31/12 06:38      
                        3 things      Erik Malund      05/31/12 07:33      
         some or all?      Erik Malund      05/24/12 09:42      
            Re:some or all      Kurt Adam      05/24/12 09:50      
               how much and can you      Erik Malund      05/24/12 10:07      
               Options      David Good      05/24/12 10:08      
   Had you read the datasheets ...      Richard Erlacher      05/24/12 09:06      
      sorry      Kurt Adam      05/24/12 09:22      
         I doubt it'll hurt you ...      Richard Erlacher      05/25/12 08:16      
            WOW      Erik Malund      05/25/12 09:48      
               Yes      Per Westermark      05/26/12 03:32      
               Well, it is a troll, as someone has pointed out ...      Richard Erlacher      05/26/12 07:04      
                  "his" money      Erik Malund      05/26/12 07:54      
                     You're probably right. It's a shame, but it's so      Richard Erlacher      05/26/12 19:31      
            Is that true?      Andy Neil      06/05/12 05:19      
               I doubt it ...      Richard Erlacher      06/05/12 09:55      

Back to Subject List