EEPROM writing - be aware

A word of caution concerning EEPROM writing on 20X2 , 28X2(3V) and 40X2(3V).

I was just doing a bit of K-series PIC programming which involved occasional writing to on-board EEPROM.
Obviously, write endurance is important.

As you all know, I am a bit 'keen' about reading Data Sheets
When I scanned the Microchip Data Sheet for the 18F25K20 45K20 I noticed:
[Erase/write cycle endurance] "up to 10K for program memory and 100K for EEPROM."
(section 1.2 the data sheet. 18F25K20 45K20 = 3V version of 28-X2 and 40-X2)

These values are a tenth of the 'typical' values for the 16F and 18F 'traditional' non-K PICs.

Later on, in table 26.20 it says this is a 'min' value.

The same spec is given for the 18F14K22 (= PICAXE 20-X2) and 18LF14K22.


For 16F and 18F (non K) we have a 'typical' value of 1million write endurance.
I know DPG showed they can do more but I'm just comparing like-for-like from Data Sheets.

I haven't done any further hunting around, but I'm just saying it's something to keep an eye on.
As usual Microchip are probably erring comfortabley on the side of caution, so don't panic.
But I would suggest that until any further is discovered that you be careful.

If someone else could cofirm this that would be helpful as my Ebay spectacles have gone on the blink.
(If there is a new Data Sheet with nicer numbers please let me know)

Those "new" numbers might suggest writing to the EEPROM say for each reprogramming of the 20X2 (and similar) or an occassional in use/operation change would not be an issue but for highly repetitive writing an external EEPROM would be a worthwhile investment (as in, more so than some of the other PICAXE chips)

Umm..

What I was really trying to say was:-
For those who have assumed that all EEPROMs have a typical life of (arguabley) 1 million writes then think again if using these K series.
Esp for those porting a programme where a miilion writes was safe....

Neither of us know what a true 'typical' value is but be aware of what Microchip are suggesting.

I only have preliminary datasheets for the 20X2 and 28X2/40X2 but those ratings seem to be in line with other PICmicros. Looking at the 08M, 18X and a 18F452 datasheet those gives 100K min, 1M typical endurance for Data Eeprom with 10K min, 100K typical for Flash.

For 'normal use' ( ie just downloading ) those ratings shouldn't be a problem, but WRITE frequency will affect Data Eeprom. For 10K endurance of Flash, you'd need to download 100 times every day to wear it out in three months.

Perhaps it's more a case that we've all got this "1M minumum" endurance for Data Eeprom mistakenly burned into our heads - I haven't checked all PICAXE's.

There's another figure which is of interest for X2 users; "Number of total erase/write cycles before refresh", 1M min, 10M typical for 28X2/40X2. If you perform that many writes, any data written before those writes and not updated itself by a write may start to become corrupt. Data can be refreshed by simply READ and WRITE it again to the same location, so for high write fequencies it's worthwhile taking time out to just READ and WRITE the whole Data Eeprom, though in practice that may not be necessary.

Importantly, anyone not using Data Eeprom or only infrequently has little to worry over.

"I only have preliminary datasheets for the 20X2 and 28X2/40X2 but those ratings seem to be in line with other PICmicros."
- can you give me a link to that hippy please. It's not the same in my copies of the Ds.

Oh, have a care with para 4 hippy. That is going to confuse people unless they read the nitty-gritty of the Data Sheet.

I've cut'n'peed from my Data Sheets.
I can't see how you can say they are 'in line' - especially if you read the 'features' headings.
Please let me know if there are some more up to date DS.

and these tables... note how we only have 'Min' for K series , which is in line with the features headline.

20X2 / 18F14K22
http://ww1.microchip.com/downloads/en/DeviceDoc/41365B.pdf

28X2-3V / 18F25K20
http://ww1.microchip.com/downloads/en/DeviceDoc/41303E.pdf

Same as mine then.

Also, any idea which silicon release you are using?

Page 4
http://ww1.microchip.com/downloads/en/DeviceDoc/80425C.pdf

and

Page 7
http://ww1.microchip.com/downloads/en/DeviceDoc/80366F.pdf

That's an easy one ( see attached ) ...

ReadSilicon Result = 1 10 10001

3V
28-pin
Die Revision 17

This may not match what's been put into Rev-Ed production.

Lovely, nothing to worry about or think about then.