A group of small flash devices was built on the JEDEC EPROM pinout so that a programmer, and in some cases a socket, could treat them as the memory they replaced. They erase electrically. They are not ultraviolet EPROMs, and a 27C programming algorithm does not write them. The EPROM organisations they line up with are on JEDEC EPROM densities and packages.
28-pin, 256 Kbit and 512 Kbit
SST's cross-reference puts its SuperFlash parts on the same row as the EPROMs of that density.
| Density | Flash on the EPROM pinout | EPROMs on the same SST row |
|---|---|---|
| 256 Kbit, 5 V, byte-wide | SST27SF256 | AM27C256, M27256, AT27C256 |
| 512 Kbit, 5 V, byte-wide | SST27SF512 and Winbond W27E512 | AM27C512, M27512, AT27C512 |
The SST note against a 27C512 replacement says the SST27SF512 matches the pinout and that program and erase need a programmer which supports SST. That is the practical distinction. The chip can sit in the outline of a 27C512. The algorithm is the SST algorithm. Winbond's W27E512 is the part that table lists as the electrically erasable equivalent at 512 Kbit.
32-pin, 5 V, byte-wide
The same cross-reference continues at 1 Mbit and above, still at 5 V and still byte-wide. These are the devices Moates' Ostrich page is pointing at when it lists the 29F040 as a 32-pin target beside the 27C010. The Ostrich emulates that outline. It does not make a 29F040 and a 27C010 the same programming algorithm.
| Density | SST | AMD | ST | Winbond in the same row |
|---|---|---|---|---|
| 512 Kbit | SST39SF512, SST29EE512 | AM28F512 | M29F512 | W29EE512 |
| 1 Mbit | SST39SF010, SST29EE010 | AM29F010, AM28F010 | M29F010 | W29EE011, W29EE012 |
| 2 Mbit | SST39SF020, SST29EE020 | AM29F002 | M29F002 | W29C020, W49F002 |
| 4 Mbit | SST39SF040, SST29SF040 | AM29F040 | M29F040 | W29C040 |
The cross-reference comments distinguish a uniform-sector device from a page-write device even when the package matches. Against the Winbond W29EE512 it says the SST39SF512 needs a software change if the 128-byte sectors of the other arrangement are required, and that the SST29EE512 is the drop-in it offers for the page-write part. Sector size is part of the device choice. Matching the package is the first check, not the last.
These 32-pin flashes are also what a tuning chip was often made from once windowed EPROMs became hard to buy. The controller sees a read-only memory of that outline. The blank device on the programmer is whichever row above matches the marking. Intel's 28F010 and 28F020 appear in the 1 Mbit and 2 Mbit cells of the same table, as further 5 V byte-wide names, and they belong to that identification problem rather than to the 16-bit 28F800 on the burst-flash page.
For the byte-wide 4 Mbit device and reported diesel-controller examples, see Am29F040 flash in engine controllers.
Reported ECU examples
An ECU-file listing identifies two AM29F010 files, labelled high and low, for a Bosch EDC15M 0 281 010 152 in an Audi A6 2.5 TDI. It also records the marking B58399. A separate service listing identifies AM29F010 for Siemens Simtec 56, 5WK9073 / 90506365, in a 1997 Opel Vectra. These are reports for identified units, not a universal fitment rule. Preserve the distinction between high and low files and check the actual chip and board before programming. ECUfile: EDC15M; DumpTools: Simtec 56.