EEProm Problem in new Bascom Version ?

chr-mt · Joined: 25 Feb 2005 Posts: 128

Hi,
just compiled an older Program with the new Version 1.11.9.4
I found that EEprom Data is sometimes not correct, i get 255 as Value when i read out the Variable from the EEprom.
Chip is Mega32
I tried several times using 1.11.9.4 and 1.11.9.3 and the problem only occured using 1.11.9.4.
Using 1.11.9.3 everything is OK.

Can someone confirm this ?

The program is something like this:

KrasserMann · Joined: 24 Sep 2007 Posts: 48

i can confirm this on atmega 128l

since my update to .4 eeprom reading does not work right.

after reinstalling .3 erverything is fine!

Martin

chr-mt · Joined: 25 Feb 2005 Posts: 128

Hi KrasserMann,
thank you for confirming this.
I will send a Bugreport to MCS.

Regards
Christopher

albertsm · Joined: 09 Apr 2004 Posts: 6031 Location: Holland

I tried it but it works as expected.

$regfile = "m162def.dat"
$crystal = 4000000

$baud = 19200
Dim memory as eram Byte
Dim Value As Byte
Dim V2 As Byte

'Store
Value = 55
Memory = Value

'Recall
Do
V2 = Memory
Loop Until V2 <> Value
Print "ERROR " ; V2
'Print Value
End

Can somebody send a small program that demonstates the problem?

chr-mt · Joined: 25 Feb 2005 Posts: 128

Hi,
i tried again this morning for many times.
I used the same hardware, the same program but the problem was gone!
Very strange.

I also wrote some testprograms, all in Version 1.11.9.4, but also these programs did run for hours writing and reading the EEprom on the same hardware and also on a STK500 with a Mega2561 without any problems.

I have no idea what caused the problems on last friday.

Regards
Christopher

AdrianJ · Joined: 16 Jan 2006 Posts: 2483 Location: Queensland

You are aware that eeprom address 0 can have corruption problems ? AFAIK Atmel has never fixed this. And since they dont say anything about it on newer chips, nobody knows what chips or processes it still applies to.

chr-mt · Joined: 25 Feb 2005 Posts: 128

Hi AdrianJ

Duval JP · Joined: 22 Jun 2004 Posts: 1173 Location: France

I try the problem and ...I have no problem
I try to a program using EEPROM
my program use 1/2 of the EEProm space of a M32

all values was good.
JP

Ps
I use the lines :
Dim Memoryzero As Eram Byte At 0 ' bug Atmel
' in fist line of dim for EEPROM variable
' Then

Jbyte = 1
If Memoryzero <> Jbyte Then 'Note however that address 0 can be trashed during reset or start up. This is a fault in the AVR design
Memoryzero = Jbyte
End If

KrasserMann · Joined: 24 Sep 2007 Posts: 48

I can try again tomorrow!

Martin

AdrianJ · Joined: 16 Jan 2006 Posts: 2483 Location: Queensland

I may have struck the same problem. I sent a small test program to a customer, which had several constants read from eeprom and used in a conversion from ADC count to human readable numbers, worked fine on my bench setup, but he said the numbers were garbage after loading the hex and eep files, even after a reload. Brought the unit here, and I verified that the output he had was definitely wrong. Fired up AVR Studio, and set up for debugging, but the program worked perfectly after reloading the hex and eep files in the debugger.

Ingo · Joined: 28 Jan 2005 Posts: 46 Location: Germany

Hi guys,

concerning the EEPROM byte 0 issue: I guess that this has been solved for the current parts. A chip manufacturer like Atmel can not afford to fool their customers (biggest volumes go to the industry) and to hide a bug like this. Anyway I have sent this question as a service request to Atmel to be sure. Usually they are responding very quickly.

Reply from Atmel can be found here: http://www.mcselec.com/index2.php?option=com_forum&Itemid=59&page=viewtopic&t=7499

albertsm · Joined: 09 Apr 2004 Posts: 6031 Location: Holland

When you look in the errata, you can see what is fixed and what not.
For some bugs they simple say ' do not use it' Smile

eeprom can be corrupted when you do not have a brown out.
I found that the only 100% reliable way is using i2c eeprom.
But i also do use eeprom. In fact i use it a lot in production.

Until somebody can send me some more info (by email) and possible, a way to duplicate, i close this 'case'.
The eeprom code was not changed anyway between .3 and .4.