I2C and 24LC512 memory devices

[RichardTXD]

Sorry false alarm.
The issue is NOT fixed.
The i2c.readfrom_mem command only works after power up then it fails every other time.
I'm using a pre-written serial eeprom - I can read it the first time after power up sometimes. There after it appears to be locked up or similar.
Spent hours on it.
Anyone played with I2c serial memroies please?????

[RichardTXD]

@RichardTXD
Updated to 1.7.4.b1 and made no difference.

Reading first 16 bytes of I2C occasionally works from power up. Thereafter a re-read returns rubbish.

Must power down the WiPy and retry to read I2C memory again.

[livius]

@RichardTXD
can you post code used for init and read data from i2c?
because i use it extensively and no isssues with reading from memory. Example i2c memory of gps.

[RichardTXD]

@livius
Sure.
The writing code is done in an Arduino app, fairly large, that I won't post. It is a work in progress. If you have the memory devices and an arduino I can send you something - it just uses the 'wire' library and does page writes etc.
This is the reading code. Absolutley nothing clever.

-----begin----------
from machine import I2C, UART
import time

# Configure 2nd uart port

uart1 = UART(1, baudrate=19200)
uart1.write('hello')


 # read up to 5 bytes
# configure the I2C bus
# i2c = I2C(0, I2C.MASTER, baudrate=100000,  pins=('P9', 'P10'))
i2c = I2C(0, I2C.MASTER, baudrate=100000)

print('scanning')
print(i2c.scan()) # returns list of slave addresses

loops = 0
while 1:
  result = i2c.readfrom_mem(80, 0, 32)
  print(str(loops) + ' ' + str(result))
  # print(str(loops))
  time.sleep(1.0)
  loops = loops + 1
  # i2c.scan()
  uart1.write('hello')

------ end -----

[livius]

@RichardTXD said in I2C and 24LC512 memory devices:
looks ok

but what if you add
time.sleep(1)
after i2c.scan?

but i see that @jmarcelino found the reason of that? Post here findings :)

[RichardTXD]

@livius
Ta. Fair comment. Shouldn't be needed but I'll try it.
In the post on the new version of firmware it was suggested that the Python code does not send the proper lead in code to address a memory chip. It is just generic code which is guaranteed to fail.
The 24LC512's I use demand an A1 or A0 for read or write then the address then the data.
I read all the PyCom/Python doc I could find and assumed it was the same as Arduino Wire library which works.
A poster showed me a command variation that allows "stop" to be conditional which would allow exaclty what I need to create the correct addressing.
I'll try that tomorrow - gotta leave for the day.
Thanks for the comments and taking an interest. I've used I2C memory chips for last 18 years and quite like them. We've found them to be very robust and reliable and low current though capacity is low by modern standards and somewhat slow.

[jmarcelino]

@RichardTXD
I only have old Atmel 24C32 EEPROMs in my parts bin, I mostly use much larger Cypress etc SPI NOR Flash or F-RAM, but I'll see if I can order some similar to yours and try it

Let us know if you get it working in the meantime.

[jmarcelino]

@RichardTXD
Can you try the new firmware 1.7.5b2 and see if it solves the problem?

I ordered a 24LC512 should get it tomorrow to try as well.

[RichardTXD]

@jmarcelino
Thank you so much for taking an interest and your effort.
Doesn't seem to change anything with same code.
I'm operating under the assumption that I need to manually send the leadin and addressing bytes for read and write.
As below. I tried a few variants but didn't get it working.

from machine import I2C
import time

# configure the I2C bus
# i2c = I2C(0, I2C.MASTER, baudrate=100000,  pins=('P9', 'P10'))
i2c = I2C(0, I2C.MASTER, baudrate=100000)

print('scanning')
print(i2c.scan())       # returns list of slave addresses, tells me chip is on the bus

loops = 0
addressing = bytearray(3)
addressing[0] = 0xA0            # Write command, chip 0 on bus
addressing[1] = 0x00            # MSB address
addressing[2] = 0x00            # LSB address
print(str(addressing))

i2c.writeto(80, addressing, stop=False) # setup chip, no stop means more to follow
i2c.writeto(80, '123hello world456')    # this should be a "page write" (assume)
time.sleep(1)                           # crazy long. 5mS should be enough

addressing[0] = 0xA1            # make it a read command, chip 0 on bus

while 1:
  i2c.writeto(80, addressing, stop=False)   # setup chip, no stop means more to follow
  time.sleep(0.1)                           # not necessary. "try anything"
  result = i2c.readfrom(80, 32)             # clock out 32 bytes in page mode (assume)
  print(str(loops) + ' ' + str(result))
  time.sleep(1.0)
  loops = loops + 1

[jmarcelino]

Hi @RichardTXD,

I received the 24LC512 today and just had to do a quick test now before heading to bed :)

I adapted some existing EEPROM code to support the Pycom board and it all seems to be working very well, please have a look at my version:

https://github.com/jmarcelino/MicroPython/blob/master/EEPROM.py

I don't have time to package this into something better at the moment but can return to it later next week if no one else does it first :)

[RichardTXD]

@jmarcelino
Thank you so much - it is surprising though.
I had got majorly confused with if the memory comand actually sent out the proper sequence or not etc. It seems it does.
I've been away with other things this last month or so (earning money). Now it is time to come back to it. I'll get the breadboard happening again shortly.