I am currently using an old AT24C32
to buffer data logger records, before periodic copying of the data to an SD card, and noticed that Adafruit has an I2C breakout board for the MB85RC256V:
So I have been wondering what kind of "real world" power savings I would see in a datalogger with a year long deployment if I switch from using the EEprom to the FRAM. The data writes are supposed to take 1/10 less power than the EEprom, but the communication will be limited to the 100 kHz I2c bus. So there might not be that much difference in the "system active time", other than the removal of the 5ms delay I have in there after the eeprom page writes.
To give some sense of the volume, my loggers generate about 50k characters/bytes of data in 6 months.
My gut feeling is that the standby currents are more important for the long haul, but I have only just ordered the FRAM to test it out, and was wondering if anyone else has tried a swap like this before, and recorded the power consumption stats.
You could just power this FRAM with the same pin that powers Vcc for the DS3231. This device only draws 140 μA for I2C at 400 kHz.
I have use Fujitsu SPI 2 Mbit FRAM. The 2 Mbit part is nice but draws lots of current since it is rated at 25 MHz.
Too bad the 1 Mbit I2C part isn't available yet, only engineering samples are available. You could just store your data in the 128 KB of FRAM.
I use four 2 Mbit parts for a 1 MB RamDisk GitHub - greiman/RamDisk: RAM Disk library for Arduino.
Now that I am pin powering the Ds3231 on one of these cheap modules from ebay (LED and charging resistors removed):
I am able to monitor that humble AT24c32 separately...and even with always on power the eeprom goes down to around 1uA when the mcu sleeps. So I am not even sure if I need to worry about pin powering the eeproms for a long term deployment.
then I came across these 32K eeprom units for a buck:
http://www.electrodragon.com/product/at24c256-256kb-eeprom-data-module-i2c-bus/
and these are a drop in replacement (just have to change the I2C address and the rest of the code is identical) that lets me buffer up to 512 of my two page-write sensor records before I have to dump the data to the SD card. This chip also has a standby mode at only 1uA. So again, probably not worth worrying about pin powering.
The Frams draw 1/10th the power at writing (140uA:Fram vs 2mA:eeprom) , and would allow me to get rid of a couple of 5ms delays at 5mA because the eeproms are slow. But the Adafruit part lists a standby current of 27uA. At that level, I think I would pin power makes more sense.
Where I get muddled on the duty cycle estimates is that I don't know which power level the chips are at during all the 100 kHz I2C bus traffic involved in a typical sensor reading event. I use a slow bus because my builds still look like spaghetti monsters with lots of excess wire. I have not yet jumped into the speedier SPI bus as I have those lines for my SD card. (and I would have to learn new coding). This is why I asked if someone else had actually done head-to-head testing of eeproms vs Frams.
I have a cheap usb scope, but I have not yet figured out how to capture these events in real time to look at what's actually happening because the current is really brief & very low. If you have any suggestions on how I might use a low end scope to do this, I would love to hear them.