Hi,

On february 17'th 2003 I asked an question on the following
subject: "[PIC]: PIC18FXX2 Silicon/Data Sheet Errata, experiences ?"
Anyway strengthened by the follwing fro Olin (and others):

>I thought these were old problems, so I just checked the Microchip web
>site.  The latest errata sheet for version 5 does not mention this bug. =
 I
>don't know what you were looking at, but I was looking at DS90150A
>released January 2003.  This applies to the rev 5 parts, which have =
been
>in production a while.  The samples I received last summer were rev 5.

I was assured that reading the revision and making sure that it was B5
(or later) this errata would not bite me. However the real world =
sometimes
things are not as they are in the ideal world.=20

This errata bit me, hard, suddendly an product started behaving very =
strange,
an conversion routine (bin24->ascii) started flaking out, producing =
jiberrish
at times. Out of ten products tested two had these symptoms, =
investigating
and trying to pinpoint the cause (s/w, h/w, temp etc.) lead me to an =
this beeing
an h/w issue and it was temp related. At startup the output from this =
routine=20
was almost random, after 10 minutes is became correct. Put some cooling =
spray
on thew pic and everything went in the gutter again, leaving it for 10 =
minutes
is came back with normal operation.=20

Worth noting here:=20
The program fills about 80% of an 18F452 (pure assembler code), =
everything seemed to
work flawlessly exept for this particular routine. The routine used 10 =
bytes located
in access ram for the conversion, so the error might have been affecting =
only these
bytes. I can extract data by other means (without acii conversion) and =
hence view=20
the internals and thereby verify where the data gets corrupt. In this =
case the only
time this particular routine was used was for LCD and serial output, all =
other data processing
(filter, stability detection, etc. ) was done on raw binary data and =
seemed unaffected.


Anyway, it turned up that our contactor (for the pcb) ordered this batch =
in march 2003,
got pics with date code 0231. Reading the 'revision' with ICD2 shows:
"PIC18F452 revision 5, found"

This is *not* B5 I repeat the 'revision' ICD2 displays is not related to =
the
errata revision Bx (where x is numeric). I do not know where this 'rev. =
5' fits in
but I'm 100% positive that these chips have revision <B5.

Also it so happens that I have an new batch here with date code 0307, =
talking
to my distributor he indicated that he was almost 100% sure that these =
should be B5
(or later). Reading *these* chips with ICD2 showed:
"PIC18F452 revision 4, found"

Conclusion, it seems like the datecode overrides whatever is in the ID =
memory,
in the errata for B3/B4 is states that all chips with datecode upto 0252 =
can be affected.
I'm not sure whatever is convyed in the ID memory from factory but it =
seems not to be related
to the Bx numbering at all.

Or can anyone enlighten me to how to determine what revision you have ?=20

Be *very* vary for 18F452 with date codes <2003.=20

During these conditions: 5V, 25C=B0, 3.6864Mhz * 4 (pll) =3D 14.7456Mhz
2 out of ten never worked in at room temp, one started working
after about 10 minutes, additional 2 stopped working when I put cooling
spray onto the chip (temp< 5C=B0). An total of 4/10 that was subject for =
the errata.

/Tony

--
http://www.piclist.com hint: PICList Posts must start with ONE topic:
[PIC]:,[SX]:,[AVR]: ->uP ONLY! [EE]:,[OT]: ->Other [BUY]:,[AD]: ->Ads