working with external memory

[Dave]

malc-c , Why on earth would you need 6 pins for external eeprom storage? I currently use 4, 24LC1025's to store data for a little over a month @ 1 minute intervals. The data is stored for approx. 750 minutes a day during daylight. The data stored is 16 bits per variable of: pointer,azminuth,elevation,up,down,east west and status words. The amount of data you are trying to store isn't much. besides, I only use 2 pins for the 4, 24lc1025's giving me a little over 4 Migabits...

Dave Purola,
N8NTA

[mackrackit]

Six pins for an SD card.

[BrianT]

Your suggested data output string needs comma delimiters since you cannot be certain of the character length of each value. You could serout each word as Dec3 mYvAR to force them all to a known length but that can complicate the end use of the data. If you comma delimit each value, terminate the line with <cr><lf> and label the file as MyLog.csv then Excel can open the file and graphing the data is a breeze.

I use word variables for TemperatureLogged = 1000 + (10*MeasuredTemperature) and depending on actual temperatures the number can have two or three digits. By adding 1000 I can send negative values without needind a sign bit. e.g 890 means -1.1C and 1234 +23.4C

My data loggers record date & time, temperature, pressure, accelerometers for an activity indication and heart rate. I log the data to M25P128 serial memory chips.

HTH
BrianT

[malc-c]

Brian, thanks for the input.

As Dave has already mentioned, the 6 pins were for the SSD card shown in his link.

I gather that using I2C type memory chips I could use the same CLK and DTA lines that hook up to the DS1307 RTC chip, and just give the memory chip a different address, so that would save me a couple of pins?

Given that all I really want to store is date, time, temp1, temp2, temp3, temp4 probably every two minutes what would be the ideal memory chip to get to hold a minimum of 7 days worth of data (if longer so much the better). I was also thinking of a rolling data process, so that when full it would simply delete the oldest data to make space for the most current set, ie so it retails the past 7 days (or what every it amount to) worth. Any suggestions of code to do this would be welcome.

[Ruben Pena]

Brian:
Do you can share the read routines for the M25P128 ?. I am using the M25P16 but my reading
is a little erratic.
Thanks in advance...
Ruben de la Pena V.

[BrianT]

Hi Ruben,

Try this. I have several methods for reading and writing to the M25Pxxx family. All work but the ones not commented out are the fastest. In some infrequent calls I have left the slow routines in place as they don't add any significan time to the download of data from the M25Pxxx memory.

[code]
ReadFromM25P128:
memclk = 0 'data change point
memclk = 1 : reply.7 = memq : memclk = 0
memclk = 1 : reply.6 = memq : memclk = 0
memclk = 1 : reply.5 = memq : memclk = 0
memclk = 1 : reply.4 = memq : memclk = 0
memclk = 1 : reply.3 = memq : memclk = 0
memclk = 1 : reply.2 = memq : memclk = 0
memclk = 1 : reply.1 = memq : memclk = 0
memclk = 1 : reply.0 = memq : memclk = 0
'takes 50 uSecs per character
debug #reply, ", " 'Debug takes 6.7 mSecs
return


RDSRWIP: 'reads the Status Register - waits for WIP bit to go low
'Data from M25P128 changes on fall of clock so sample on rise of clock.
code = $05 'Read Status Register code
gosub sendcode
WIPclr:
for shift = 7 to 0 step -1 'read 8 bit reply
high memclk : reply.0[shift] = memq : low memclk : pauseus 1
next shift
if reply.0 = 1 then wipclr
return

SendCode: 'Serial MSB first sends the CODE byte from PIC to M25P128
TRISE.0 = 0 : PortE.0 = 0 : memclk = 0 'MemClk
TRISC.0 = 0 : PortC.0 = 0 : memd = 0 'MemD
' for shift = 7 to 0 step -1 'send MSB first
' memclk = 0 : memd = code.0[shift] : memclk = 1: pauseus 1 : memclk = 0
' next shift 'this is SLOW ~500 uSecs per character
memclk = 0
memclk = 0 : memd = code.7 : memclk = 1
memclk = 0 : memd = code.6 : memclk = 1
memclk = 0 : memd = code.5 : memclk = 1
memclk = 0 : memd = code.4 : memclk = 1
memclk = 0 : memd = code.3 : memclk = 1
memclk = 0 : memd = code.2 : memclk = 1
memclk = 0 : memd = code.1 : memclk = 1
memclk = 0 : memd = code.0 : memclk = 1
' much faster, takes about 52 uSecs per character
return
/[code]

You can see here I have several methods for reading from the M25P128. The decrementing loop "... for shift = 7 to 0 step -1..." works and is simple BUT it takes about 500 uSecs per character.

HTH
BrianT

[Ruben Pena]

Brian:
Thanks for your fast response...
I am reading wav files stored in the M25P, and using this to send the address:
cs1 = 0 'SEND COMMAND AND ADDRESS
SHIFTOUT MOSI,SCK,1,[$03\8] 'read command $03
SHIFTOUT MOSI,SCK,1 ,[PAGE.BYTE1\8,PAGE.BYTE0\8,00\7]'addr 000000
I send the command and then the 24 bits address,but note in the last address byte I send
only 7 bytes. then I go to a loop to read the data and send it to the port to convert to
analog with a R2R network,until I find a special end of file char to stop the reading.
If I send the 8 bits, I get shifted data. With 7 bits I got the correct data and I can identify
the end of file byte. I do not know why.
the read routine is this:
ASM
MOVLW 8
MOVWF _CNT
SIGUE
BTFSS _MISO ;MOV BIT TO CHR,0
BCF _CHR,0
BTFSC _MISO
BSF _CHR,0
BSF _SCK ;READ AFTER SENDING THE CLOCK
RLF _CHR,1
BCF _SCK
NOP
NOP
NOP
NOP
NOP
DECFSZ _CNT,1
GOTO SIGUE
ENDASM

Am I missing some previous command to set up the chip for reading..?

I will test your routines to see if I got an improvement.
Thanks again...
Ruben de la Pena V.