[sdiy] dsPIC fun board

[Ingo Debus]

Am 17.01.2009 um 23:10 schrieb Eric Brombaugh:

> This is true. The only caveat is that the write endurance of the on- 
> chip flash in the dsPIC is somewhat limited. MCHP only guaranteed  
> about 1000 cycles for the initial parts, although they may have  
> upped since I last looked. That's plenty for development purposes,  
> but for use as non-volatile storage in the end application it might  
> be a bit shy of what's needed.
>

Oops. Finally I found that parameter in the data sheet of the  
dsPIC33. It's on page 272, called Cell Endurance. 1000 is the typical  
number, the minimum is only 100! (I hope I got this right, I have no  
idea what the unit E/W means)

I had Atmel processors on my mind, where 100k Write Cycles are  
specified for the Flash memory (AT89C51RD2/ED2).
Funny that there's such a big difference.


>> Also, in my experience, EEPROMs are unreliable.
>
> This interests me. What specifically were the pitfalls you  
> experienced in using EEPROM?

It happened very rarely, but sometimes stored data were corrupted. I  
don't know why. The EEPROM wasn't written very often. There was  
additional circuitry that prevented unallowed states of the chip  
enable line during power-up/down.
Of course you can always add redundancy in software since EEPROM are  
often much larger than needed (store the data twice, use checksums  
etc). But this means additional programming effort.

Ingo