I’ll post further updates to the 1-Wire, I2C, SPI, etc unified EEPROM command in this thread.
TODO:
Partial page writes are now handled properly. Most I2C and SPI EEPROMs can do partial page writes, but not all. 1-Wire eeproms cannot, and a full 8 or 32 bytes need to be written each time. This is all handled below the hood, so the user doesn’t need to worry about anything.
If you have any EEPROMs I’d appreciate some test reports! I have only tested with 24x02, 25x020, and DS2431. It will be a miracle if I managed to add full catalogs of EEPROMs without a single bug.
Bugs aside, it seems like this is the end of the work on EEPROMs for now. Not sure why I decided now was the time to add this, and it took over the last week or so. The results, especially the updated hex viewer, are super spiffy though. Much more professional and easier to read.
To do:
I should have studied some existing programmer instead of brute force building my own list of devices 
Came across one of these universal programmers on Taobao, and reading the compatible device list was really illuminating.
Aside: In an effort to ape the competition this programmer lists “26” devices all over the sales material. No such device is listed in the supported parts, and in fact I can’t seems to find any EEPROM with a 26x designation. They also list “95” which is really just STM’s name for the common 25x. Gotta collect them all!
| Device | Type | Mode |
|---|---|---|
| 25Q | SPI Flash | SPI |
| 25x (& “95x”, by STM) | SPI EEPROM | SPI |
| 24x | I2C EEPROM | I2C |
| 93x | Weird SPI EEPROM | 3WIRE |
Thus far we have:
93x is weird, mainly because it doesn’t use 8 bit bytes. It won’t work with most standard SPI peripherals without some funny business.
In 8 bit mode the command/address is 10 bits, in 16 bit mode it takes 9 bits. Neither align with the standard 8 bit data read/write format of the data word.
Here’s the write sequence. Start bit (1) opcode (01) address (6 or 7 bits), 1 data word. Busy is indicated by raising CS after >tCS (250ns) and waiting while DO is low (so use pull-up resistor).
| Device | Size | Bytes | Organization | Dummy bits | Address | Total bits |
|---|---|---|---|---|---|---|
| 93x46A | 1Kbit | 128 | x8 only | 0 | 7bits | 10 |
| 93x46B | 1Kbit | 128 | x16 only | 0 | 6bits | 9 |
| 93x46C/E | 1Kbit | 128 | x8 or x16 | 0 | 7 or 6bits | 10 or 9 |
| 93x56A | 2Kbit | 256 | x8 only | 1 | 8bits | 12 |
| 93x56B | 2Kbit | 256 | x16 only | 1 | 7bits | 11 |
| 93x56C | 2Kbit | 256 | x8 or x16 | 1 | 8 or 7bits | 12 or 11 |
| 93x66A | 4Kbit | 512 | x8 only | 0 | 9bits | 12 |
| 93x66B | 4Kbit | 512 | x16 only | 0 | 8bits | 11 |
| 93x66C | 4Kbit | 512 | x8 or x16 | 0 | 10 or 9bits | 12 or 11 |
| 93x76A | 8Kbit | 1024 | x8 only | 1 | 10bits | 14 |
| 93x76B | 8Kbit | 1024 | x16 only | 1 | 9bits | 13 |
| 93x76-/C | 8Kbit | 1024 | x8 or x16 | 1 | 10 or 9bits | 14 or 13 |
| 93x86A | 16Kbit | 2048 | x8 only | 0 | 11bits | 14 |
| 93x86B | 16Kbit | 2048 | x16 only | 0 | 10bits | 13 |
| 93x86-/C | 16Kbit | 2048 | x8 or x16 | 0 | 11 or 10bits | 14 or 13 |
x8 or x16 bits is selected by the ORG pin. If ORG is low, the device is x8, if ORG is high, the device is x16.
Total bits = Start Bit (1) + Opcode (2) + Dummy bits (0/1) + Address bits
Here are the potential devices, at least the Microchip and Atmel range: 46/56/66/76/86. A parts are fixed 8 bit words, B are 16 bit words, and C have a ORG pin that determines the organization.
Ah, so that makes sense. 8 bit bytes can be used, but the initial bits are 0, then after the start bit the data is accepted.
93x EEPROM support integrated into the SPI mode eeprom command. This took a rework of how the hardware abstraction layer works, but the result has an improved structure.
This is untested for now as I don’t have any 93x EEPROMs to test against. Some are on the way.
New documentation page with all the information I collected on various memory chips.
| Bus | Chip | Tested |
|---|---|---|
| 1-Wire | GX2431 | OK |
| 1-Wire | DS24B33 | OK |
| SPI | 25x020 | OK |
| SPI | 25x040 | OK |
| SPI | 25x320 | OK |
| SPI | 25x640 | OK |
| SPI | 25x128 | OK |
| SPI | 25x256 | OK |
| SPI | 25x512 | OK |
| SPI | 25x1024 | OK |
Started testing a few EEPROMs at a time.
| Bus | Chip | Tested |
|---|---|---|
| I2C | 24x01 | OK |
| I2C | 24x02 | OK |
| I2C | 24x04 | OK |
| I2C | 24x08 | OK |
| I2C | 24x16 | OK |
| I2C | 24x32 | OK |
| I2C | 24x64 | OK |
| I2C | 24x128 | OK |
| I2C | 24x256 | OK |
| I2C | 24x512 | OK |
| I2C | 24xM01 | OK |
No issues found.
Trying to get the 93x EEPROMs going, but it’s been slow.
From here. So the incoming and outgoing clock phases need to be different, but the RP2040 SPI module doesn’t support that as far as I can tell.
This is technically a Microwire EEPROM, not SPI, but it is really close. RP2040 has a Microwire mode, but it seems WAY more involved than SPI. When enabled it’s sending double the data I intend plus weird changes on MOSI, probably the normal SPI way of using it doesn’t apply.
The data is coming out of the chip, but the RP2040 can’t see any of it.
Maybe I can just change the clock phase before reads?
hwspi_deselect(); // deselect the EEPROM chip NOTE: 93X are CS active HIGH
hwspi_set_cphase(0);
hwspi_write_n(address_array, 2); // send the address bytes
hwspi_set_cphase(1);
hwspi_read_n(buf, read_bytes); // read bytes from the EEPROM
hwspi_select(); // select the EEPROM chip NOTE: 93X are CS active HIGH
hwspi_set_cphase(0);
Got it I just needed to manipulate the clock phase between write and read. Will need to store the current setting and restore after I guess.
What an annoying little thing, and it’s so old demonstration projects for RP chips are non existent.
Now I will continue verifying all the chips in 8 and 16 bit modes.
ETA:
Evidently the app note was clear. I assumed the pic had different settings for read and write clock phase, but that clearly just says to switch it.
| Bus | Chip | Tested |
|---|---|---|
| SPI | 93x46(-16) | OK |
| SPI | 93x56(-16) | OK |
| SPI | 93x66(-16) | OK |
| SPI | 93x86(-16) | OK |
Tested the set of 93x EEPROMs in 8 and 16 bit address mode.
| Bus | Device |
|---|---|
| I2C | 24x1025 |
| I2C | 24x1026 |
| I2C | 24xM02 |
| SPI | 25X010 |
| SPI | 25X080 |
| SPI | 25X160 |
| SPI | 25XM01 |
| SPI | 25XM02 |
| SPI | 25XM04 |
| SPI | 93X76(C) |
I’ve sourced all the missing chips, but I’m going to skip 25xm04 as it only has a single supplier and costs 25RMB+.
I’m happy to tack the ST (M95M04) and Microchip (25CSM04) components onto a Digikey order and test them, if that would be helpful?
Yes! If it’s not a faf. I’m buying a lot of ewaste chips on Chinese websites and this one is not common.
Microchip 25CSM04 just passed eeprom test. Will solder the ST part onto a breakout and test after lunch.
Yep, check 25xM04 off your list, the ST part (M95M04) tested fine as well.
Thank you so much! I’ve updated the docs and will push shortly.
I pushed an organizational update to the docs with pinout illustrations, tables, and info on key pins.