You guys need to think this through.

Buy a 33-cent I2C 256x8 EEPROM and forget about all that other effort.

--Bob

Jesse Lackey wrote:
> Hi... I'm using the 64J11 right now, and need to store things in NV 
> memory, and yep, this is what I'm doing.
>
> A few things... this doesn't apply to a bootloader, but since I only 
> have to store something like 32 bytes, I only need to copy those to RAM 
> before doing erase & rewrite, of course.  It costs 1024 bytes of flash 
> to save 32, but hey, these chips go to 40Mhz @ 3.3V and are 
> significantly cheaper than the (slightly older) generation (like the F6520).
>
> If the bootloader will be rewriting say everything above 1K in program 
> space (or potentially rewriting it - whatever isn't rewritten is 
> unused), then you can erase a 1K block, then do multi writes to it, 
> sequentially, of course.  No need to save anything.
>
> Now if the bootloader is designed to be able to rewrite just little 
> arbitrary sections of flash, then you've indeed got the 1K RAM problem.
>
> Also note much more limited endurance vs. EEPROM.  Like, 1K rewrite 
> cycles, typical, instead of 100K or 1M.  This is a problem for me, which 
> I'm going to solve by storing my approx 32 bytes in a series of blocks, 
> so when I do a "save" I find the end of the series of blocks, and write 
> the new values into the erased flash.  When I run out of memory (1K, or 
> maybe I'll use several K), erase everything and start over.  So in 
> effect spreading the wear over a larger area.
>
> Cheers,
> J
>
>
> Forrest W Christian wrote:
>   
>> Well, I am working on a bootloader for the 18f46J11....
>>
>> And came across this snipplet:
>>
>> "To maintain the endurance of the program memory cells, each Flash byte 
>> should not be programmed more than once between erase operations. Before 
>> attempting to modify the contents of the target cell a second time, an 
>> erase of the target page, or a bulk erase of the entire memory, must be 
>> performed."
>>
>> Add to that:  "The minimum erase block is 512 words or 1024 bytes."
>>
>> This isn't a big problem for the bootloader (rewriting 1024K at a time 
>> is ok - a bit dangerous in page 0, but not a big issue).    What is bad 
>> is that this particular pic doesn't have any flash other than the 
>> program memory, so I in effect get to use the program memory as flash 
>> data memory to store my nonvolatile settings.
>>
>> Assuming I am reading my datasheet correctly, that means I get to 
>> reserve 1024 of the 3776 bytes of RAM so I can read the entire block of 
>> configuration memory in, make modifications to it, and then re-write 
>> it.   Am I missing something? 
>>
>> -forrest
>>
>>
>>
>>     

-- 
http://www.piclist.com PIC/SX FAQ & list archive
View/change your membership options at
http://mailman.mit.edu/mailman/listinfo/piclist