Reference documents and datasheets

Reference documents, datasheets and pinouts for the Atari ST range.

External document archives

dev-docs.org — https://docs.dev-docs.org/
The main archive these days. Broad collection of Atari developer documentation, hardware references and manuals, kept in one place and reasonably well organised. Start here.

exxosforum document thread — https://www.exxosforum.co.uk/forum/viewtopic.php?t=100
Basic copies of documents, largely mirroring what is on dev-docs.org. Useful as a backup if the main archive is unavailable.

exxosforum pinout thread — https://www.exxosforum.co.uk/forum/viewtopic.php?t=149
Pinouts for the various chips used across the ST range. Handy alongside the memory map when you are tracing a signal rather than writing to a register.

Datasheets held on this wiki

Where a datasheet exists for a common part it is worth holding a copy here directly, so it does not disappear when someone else's site goes down.

Nothing uploaded yet. This section is a placeholder for the files listed below.

Parts wanted

These are the chips that appear in the memory map, so a datasheet for each is directly useful when working from these pages. If you have a good clean copy of any of them, please upload it or post a link.

Processors

• MC68000 — ST, STE, Mega ST, Mega STE
• MC68030 — TT, Falcon
• MC68881 and MC68882 floating point coprocessors — Mega STE, TT

Custom Atari chips

• Shifter, GLUE, MMU, DMA — no public datasheets exist, but reverse engineering notes and pinouts are worth collecting
• Blitter
• VIDEL — Falcon
• SCU — TT, Mega STE

Standard parts

• WD1772 floppy disk controller — see WD1772 Floppy Disk Controller for the full consolidated reference, and note the caveat below. See also WD1772 16MHz Myth for the 2026 investigation into the "only the 02-02 does 16MHz" overclocking claim, with bench test results.
• YM2149 programmable sound generator, and the General Instrument AY-3-8910 it derives from
• MC68901 multi function peripheral (MFP)
• MC6850 asynchronous communications interface adapter (ACIA), used twice, keyboard and MIDI
• HD6301 keyboard controller (IKBD)
• NCR 5380 SCSI controller — TT
• Zilog 8530 serial communications controller (SCC) — TT, Mega STE, Falcon
• LMC1992 audio processor — STE, TT
• RP5C15 real time clock — Mega ST
• MC146818 real time clock and NVRAM — TT, Falcon
• Motorola DSP56001 — Falcon

A note on copy quality

A lot of what is in circulation is poor. Many are scans of photocopies, some are missing pages, and a few have been re-scanned so many times that tables are unreadable. Where there is a choice, a clean original beats a familiar copy.

Two specific cautions:

The widely circulated DrCoolZic edition of the WD1772 datasheet contains several transcription errors that are not present in Western Digital's original preliminary WD1770/1772 combined datasheet: the Read Track opcode is printed as 1101, which collides with Force Interrupt and cannot be right (correct is 1110); the write precompensation flag is described as "Unable" where it means Enable; the S5 status table gives the alternate normal data mark byte as $FC where it is $FA; and the sector length code in the 512 byte format listing is given as $01 where it should be $02, apparently left over from adapting Western Digital's original 256 byte format.

But Western Digital's own preliminary datasheet is not reliable either: it is explicitly stamped PRELIMINARY, and its WD1772 step rate table gives 2, 3, 5 and 6ms against the wrong bit codes. EmuTOS's own source comments reject exactly these figures ("not 2, but 6 msec steprate", and so on for the other three), and Hatari implements the corrected table. The corrected values, along with the WD1772-specific 15ms settling delay that the preliminary sheet also gets wrong (it gives 30ms, the WD1770 figure, throughout), come from a third document: the Acorn/BBC Micro "WD1770/2/3 data" reference file, which was written independently for the same family of chips and has been confirmed against both EmuTOS and Hatari. See WD1772 Floppy Disk Controller for the full comparison of all three documents, register by register, and the real-world 16MHz reliability findings.

Some YM2149 documentation is actually AY-3-8910 documentation. The two are register compatible so this is usually harmless, but the envelope resolution differs, 5 bit on the YM against 4 bit on the AY, and the AY returns zero for unused bits on readback where the YM returns what was written.

Related

• Memory Map for Atari ST,STE,TT and Falcon — the register listing these documents support