WD1772 Floppy Disk Controller - exxos edition 2026

[exxos]

Following on from the register listing mess, the WD1772 floppy controller datasheet situation turned out to be just as bad. Three different source documents in circulation (Western Digital's own preliminary sheet, the widely used DrCoolZic edition, and an old Acorn/BBC reference), and every single one of them has genuine errors in it. None of them agree with each other, and worse, some of the "corrections" one document made turned out to be wrong themselves.

So I've gone through all three line by line, cross checked everything against Hatari and EmuTOS source, and fixed nine confirmed errors including a wrong step rate table, a wrong data mark byte, and a self contradicting revolution count.

Also finally put to bed the "02-02 chips can be overclocked to 16MHz" myth. There's no official documentation for it anywhere, and real testing shows no WD1772 or VL1772 should be trusted for sustained 16MHz regardless of what's printed on it. The only chip actually built for that is AJAX.

End result is a proper corrected reference, available as a formatted PDF datasheet and as a new wiki page. Marking it PRELIMINARY for now since nobody else has been through it yet, so if you spot anything wrong, shout.

https://www.atari-wiki.com/index.php?ti ... Controller

WD1772-Datasheet-exxos.pdf

[rubber_jonnie]

From what I understand the only 1772 variant that's reliable at 16Mhz is the Ajax, and it was deliberately designed as such, whereas some 1772's will run at 16Mhz but were not designed to do so.

[stephen_usher]

The alternative to buying any of them is to buy this instead, which will do 16MHz reliably:

https://www.vretrodesign.com/products/v ... eplacement

[exxos]

The alternative to buying any of them is to buy this instead, which will do 16MHz reliably:

https://www.vretrodesign.com/products/v ... eplacement

But how much testing has that actually had for Atari disks with copy protection and the rest of it ? A "simple" emulator of the WD1772 is probably easy, but getting it 100% identical, where's the proof and testing to prove it really is an alternative ?

[exxos]

From what I understand the only 1772 variant that's reliable at 16Mhz is the Ajax, and it was deliberately designed as such, whereas some 1772's will run at 16Mhz but were not designed to do so.

Exactly. I go into a lot of details about all this anyway.

Based on "word on the street" and my own tests. This is an approximation.

Capture.PNG

For ST's using my 1.44 floppy kit, its generally not a problem. Its only a problem on machines like the MSTE etc which run at 16mhz constantly. So almost anything will work with one of my floppy kits in a ST/F/M/E.

There seems to be a myth for the 02-02 working fine at 16mhz. It's not technically false, but it's not the full story either.

[exxos]

There's another bit of confusion that never quite dies on the various Atari forums: the WD1772-00 vs WD1772-02 suffix on the chip, and whether it actually means anything for HD or 16MHz work.

Had a dig through Western Digital's own 1986 Storage Management Products Handbook to see if there's a real answer buried in there somewhere. Turns out there is, sort of. WD177X-00 and WD1772-02 are two genuinely separate, officially documented revisions, not just a date code or a marketing suffix someone made up along the way. WD gave them entirely separate datasheets in the same book. The actual documented differences:

- Data separator: -00 uses a basic ring shift register design. -02 replaces it with a proper phase locked loop, with real quoted specs (around 6% capture range, 4 byte lock response). This looks like the actual headline change between the two.
- Write precompensation: 125ns on -00, 187ns on -02.
- Spin-up complete status bit: fires after 6 revolutions on -00, 5 revolutions on -02.
- Sector addressing range: 1 to 240 on -00, 0 to 244 on -02.
- Step rates: identical on both, 2/3/6/12ms, despite what you'd assume from all the noise about "faster" revisions.
- Both are stated as 100% pin compatible with each other, so it's a straight swap either way.

Nowhere in either datasheet does WD say a single word about 16MHz or HD support for either one. So the old "02-02 is the good one for HD" story still doesn't hold up as an official spec, same conclusion as ever. My guess is the genuinely better PLL in the -02 is exactly why that reputation started somewhere along the way, a real improvement that just never actually got tied to any HD claim from WD.

Standard caveat: this was AI assisted research, pulling straight from a scanned WD databook, so treat it as a lead rather than gospel until someone's properly cross checked it. If anyone's got a -00 and a -02 to bench test side by side, would be good to know if the numbers above actually hold up in practice.

TL;DR

00 vs 02 probably doesn't matter :shrug:

If I can find some time, I will dig out a new 00 and 02 have here and see what happens at 16mhz. I suspect there won't be much difference.

[ijor]

There's another bit of confusion that never quite dies on the various Atari forums: the WD1772-00 vs WD1772-02 suffix on the chip, and whether it actually means anything for HD or 16MHz work.

- Data separator: -00 uses a basic ring shift register design. -02 replaces it with a proper phase locked loop, with real quoted specs (around 6% capture range, 4 byte lock response). This looks like the actual headline change between the two.
- Write precompensation: 125ns on -00, 187ns on -02.
- Spin-up complete status bit: fires after 6 revolutions on -00, 5 revolutions on -02.
- Sector addressing range: 1 to 240 on -00, 0 to 244 on -02.
- Step rates: identical on both, 2/3/6/12ms, despite what you'd assume from all the noise about "faster" revisions.
- Both are stated as 100% pin compatible with each other, so it's a straight swap either way.

This can't be right. The only difference between both part is that the 02-02 (that's the actual chip label) has an enhanced digital PLL. It is not accurate that the -00 has a basic simple PLL. It is already an excellent one. But the -02 one is superior.

Among other things, a digital only device simply can't perform 187ns write precompensation from an 8MHz clock. You would need either a much higher clock or analog logic.

Nowhere in either datasheet does WD say a single word about 16MHz or HD support for either one.

I can't say I'm sure. But what I reconstructed from several sources is that the -02 enhanced PLL was designed on purpose to support HD. You won't find this in any WD document. AFAIK it was never supported officially. I assume, but then again I'm not sure, that the target was something similar to the MSTE setup, not running constantly at 16 MHz.

[exxos]

This can't be right. The only difference between both part is that the 02-02 (that's the actual chip label) has an enhanced digital PLL. It is not accurate that the -00 has a basic simple PLL. It is already an excellent one. But the -02 one is superior.

Seems to a contradiction in the datasheet. In features it quotes BUILT·IN WRITE PRECOMPENSATION,
INCREASED TO 187 MS. Then on the same page, write precompensation of 187 nsec from nominal.

I can't say I'm sure. But what I reconstructed from several sources is that the -02 enhanced PLL was designed on purpose to support HD. You won't find this in any WD document. AFAIK it was never supported officially. I assume, but then again I'm not sure, that the target was something similar to the MSTE setup, not running constantly at 16 MHz.

Only post so far I've found..

Starts with a text called "Introduction and Notes for adding High density disk drives to the Atari ST" by Martin in 1989, callsign GW6HVA, circulated on UK packet radio via the BBS GB7OSP. It says the 1772 carries a suffix of 00-02 or 02-02, that "the original chip is designed to run at 10MHz" (wrong, it's 8MHz) and that "only the 02-02 chip will handle the higher clock speed of 16MHz". No mention of heat, duty cycle or per-chip variation, so the unconditional version of the claim is there from the start.

But the 02 has been found unreliable by myself and others. So it seems unlikely at this point that 16mhz was ever possible. It's a overclock. So I'd be very interested if 16mhz was mentioned in any official WD doc. Or where the 16mhz claim come from.

@Darklord posted this on AF..
https://www.atari-forum.com/viewtopic.p ... 9&#p205609

From an article in ST Format, mostly the older WD1772's don't work, and only about 5% of the 02-02
variety doesn't. I had an 02-02 in my STacy, and I guess it was in the that 5%. Sean's HD mod wouldn't
work until I replaced it with an Ajax chip - took right off after that and worked great. :)

So there's a few common users who now state 16mhz isn't reliable on the 02 either. I wouldn't trust STF anyway. I've had older VL1772 and WD 00 work fine in my floppy kits. Just I don't run them at 16mhz all the time which was the issue I think Troed ran into on his MSTE. IIRC it switched into 16mhz and stopped there. A lot of older floppy kits did the same. It's why I created a better switching on mine and never had any problems with any chip, at least on the ones I tested.

The only problem I did find like 30 years ago, which I never figured out, was running from the shifter 16mhz was unreliable. Probably just not a good clock source. I never had problems with using a external osc. In fact it's why I put a jumper on the 16mhz clock on the H5 to "one day" look into it again.

[exxos]

Also posts on startdot
https://stardot.org.uk/forums/viewtopic ... 634#p14634

Whilst messing on the Elk, it occurred to me that due to the way the various 151 control signals operate, the principal state of the 1772 clock is now 16Mhz unless the drives are being accessed with a DD disc inserted when the clock will drop down to 8MHz. A much better arrangement would be to have the clock always sitting at 8MHz unless a drive is accessed with a HD disc present

So a few years after I posted switching to 8mhz, someone else came to the same conclusion as well.

In a February 2019 thread he says standard genuine 1772s all work with his mod, just not the 1770, and that he was buying 02-02s only to future-proof in case he went back to ADFS work:

He also writes that ordinary 1772s are officially rated only up to 10MHz. That's the same wrong 10MHz figure as the GW6HVA text, which suggests he read the same lineage rather than anything from WD.

EDIT:

I went back to the original Western Digital databook on this one, the 1986 Storage Management Products Handbook, rather than relying on what gets repeated around the forums.

Both the WD177X-00 and the WD1772-02 have their own separate datasheets in there, and both carry the same miscellaneous timing table:

Clock Duty (low) 50 to 67 nsec
Clock Duty (high) 50 to 67 nsec

That is the clock input. A 50ns minimum half period works out at a 100ns period, which is 10MHz. The 67ns maximum gives 7.5MHz at the other end. So the "rated at 10MHz" figure that keeps turning up in the old high density write ups wasn't plucked out of thin air after all, it comes straight out of WD's own timing table. Both datasheets also state that the clock input needs a free running 50% duty cycle at 8MHz plus or minus 0.1%, so 8MHz is still the operating spec, but at least the 10MHz number now has a home.

The important bit is that those numbers are identical on the -00 and the -02. There is no separate, faster timing table for the -02. And 16MHz needs a 31.25ns half period, which is nowhere near the 50ns minimum on either of them.

So whatever else the -02 improved, neither revision was ever rated anywhere close to 16MHz. Running a 1772 at 16MHz is an overclock on all of them, and asking a chip to do twice what it was specified for has never been a recipe for a long and happy retirement.

Databook is here if anyone wants to check, the tables are on pages 48 and 74 of the PDF:
http://bitsavers.org/components/western ... ndbook.pdf

The other thing worth pointing out is that not one of the people who made the 16MHz claim ever says how many chips they actually tested. Not the original write ups, not anyone repeating it since. If this started with somebody swapping a single 02 into a single machine and finding it worked, the result would look exactly like what we have. One good chip on the bench becomes a fact on the internet, and forty years later it's still being quoted as a specification.

[exxos]

I went digging for where the "Western Digital stopped making the 1772" story actually comes from, and I think I've found it. It's on the Best Electronics custom chips page, in a history section about the Ajax:

https://www.best-electronics-ca.com/custom-i.htm

There's rather more on there than just "Ajax was an updated chip". The claims are:

The first high density target was the TT, not the STE. Atari Marketing asked Engineering to get the TT internal drives up to 1.44MB without major changes to the motherboards.

Best states plainly that the WD1772 was never designed to run at 16MHz, but that through testing some chips were found to manage it, and that Atari used hand selected and tested WD1772s clocked up from 8MHz to 16MHz. So according to Best it was cherry picking, not a chip designed for the job.

They also say Western Digital hadn't done a new production run of the 1772 in years and couldn't be persuaded to do another. That looks like the origin of the story everyone repeats. Worth noting I have brand new WD 02-02 chips here with a 9427 date code, so week 27 of 1994, which is well after Atari had already moved on to the Ajax. Somebody was still making them.

The VLSI chips get an explanation too. Atari had VLSI take a licence from Western Digital and do a small run of standard 1772s in VLSI labelled packages for Atari.

A couple of other bits from the same page. The Ajax apparently auto senses the input clock speed and switches itself into high density mode, and the C302096 and C302434 came from two different manufacturers.

Last one is interesting for anyone fighting a Mega STE. Best reckon not all Mega STEs can be upgraded to 1.44 successfully, and they put it down to the sheer number of motherboard revisions plus RF interference from the internal power supply causing read and write errors on high density. That's a different failure mode to the chip overheating, so if you've got a Mega STE that won't play ball, there may be more than one thing to rule out.

None of it is sourced, so treat it as one dealer's account rather than gospel. But it does line up with what several of us have found on the bench, which is that some 1772s will do 16MHz and plenty won't, and there's no way to tell which is which until you try it.