18F45K50 EEPROM problem(solved)

[GTF]

What might cause nothing to be written to the EEPROM? The program does not stall at any of the read or writes, but there is no data stored? I can get it to work with a bare bones EEPROM only test program, and it was working fine in my program up until yesterday. I have disabled any new code and most of the old code and it still doesn't work. Can't get past the initial calibration sequence anyway because the "calibrated" value never gets written to EEPROM. There is a reset at the end of this sequence and a "calibrated" check at startup. All EEPROM addresses remain FF. Tested on 2 different chips.

Don't see much difference in the assembled write code other than bank usage(test code on the right works). Variable list is getting long though.

[kersing]

Please post your flowchart or a simple test flowchart so we can check the calling sequence etc.

[GTF]

The simple test flowcharts attached seem to work, mostly. But then I am only writing/reading to 1 or 2 addresses and have only 3-4 variables. My program should write to 28 EEPROM addresses during calibration, including a final write to address dec 52 of the dec value "126". I can see values written now, but they don't seem correct. The 4 addresses in the top row should all be written "0" during a recalibration. Sometimes there is a "FF or two". Other writes appear to be randomly missed as well. I'm using a fresh PCB today. The correct value "126" appears to have been written at the end of the calibration sequence in this image, but the program always ends up in the flashing LED loop at startup rather than continuing to the main program loop. I have not made any changes to this portion of the flowchart since back when it was a V5 project.

[GTF]

The other small test program.

I think the changes I made just before the problem started was adding a circular buffer and a macro that reads those 5 top row addresses and sends the values out the UART. That macro is currently deleted.

[GTF]

This test program randomly worked a few times, but mostly will not. When it doesn't, nothing appears to be written to EEPROM. Code read in and read out are identical on compare.

[GTF]

The programs work fine if I use code blocks for the EEPROM write/reads rather than the FC EEPROM component. New test program attached.

So I reviewed the CAL and FCD files again. It appears that the incorrect eeprom type is specified in the FCD. It should be type 3 not type 2. Seems to be working with just that correction.

[medelec35]

Hi Grant,
Good spot!
I will move this topic to the bugs section for Matrix to post corrective update.

Martin

[Benj]

Hello,

Yes a good find indeed. I have now applied the fix to the 18F24K50, 25K50 and 45K50 FCDs.

I'll try and do a post 6.1.3 release of the bug fixes some time this week so all the current fixes are in one place.