Big IDE and SATA chipset info thread.. thing..

Came across a post by someone called Jeff Walther about flashing Promise ATA cards. From what I've read in the past, the dedicated Mac versions have a larger bios, just like some of the SI cards.

In article , Don Bruder
wrote:

> I just picked up a Promise "Ultra66" PCI ATA controller card for a
> horribly bad song and a couple outdated dance steps, with the full
> understanding that it probably won't work in a Mac "as is". I'm wanting
> to cram it into an elderly (but still spry!) PowerMac 7500 that's
> currently running MacOS 9.1 on a G4/350 CarrierZif processor with 376
> megs of RAM.
>
> I've googled enough to find several messages and web pages referencing a
> procedure for converting/diverting/swiping/whatever-you-want-to-call-it
> this card for use with Macs instead of PCs, but so far, that's all
> they've been: References to the procedure, with no actual sighting of
> the procedure that's being referred to.
>
> Seeing as I'm at least decently competent at electronic hardware repair,
> modification, and so on at the component-on-the-board level, "I ain't
> skairt" of going after a card with a soldering iron, as some of these
> mentions seem to suggest is involved, and I'm wanting to get this baby
> up and running in the 7500. Anybody got a pointer to the actual
> procedure, rather than just something that says "If you follow the
> directions, it'll work", without ever actually giving the directions?
>
> If it can be googled, I either haven't asked the right question, or I've
> looked straight at the answer without realizing it, so a little help
> here would be appreciated.

Try the archive for the lowendmac email list "supermacs" over the last
month or so. There are two emails that describe the procedure. The
first is a little vague but covers the difficulties of flashing the
converted card. The second actually tells which resistors to move. LEM
is here: http://www.lowendmacs.com. It'll take you a bit of surfing to
get to the list archive.

Oh, but in practice, you really need to have a working VST UltraTek/66 on
hand in order to convert the Promise card. You need to borrow the VST
card's flash chip for a few minutes in order to make the conversion of the
Promise card. Given this fact, there's not much point in doing the
conversion unless 1) you need several cards, or 2) it will be so much fun
it's worth it just for the entertainment.

You need the VST card, because as far as I know no one has managed to hack
the VST firmware updater. I had some discussions with someone either in
this news group or by email, but who I found in this news group who was
trying to extract the ROM resource from the flasher and program it into a
non-flashable EPROM, but he ran into problems. Apparently there are some
extra bytes in the ROM resource that don't belong in the firmware on the
chip. Using a cheap EPROM will work, but then the card won't be
flashable so it's stuck with whatever firmware you program it with.

Oh what the heck, here's a copy of the message. Note that the second
message quotes relevant bits of the first message:

> >>(snipped)
> >>
> >>> On the VST/Promise cards, the Promise cards require moving several
> >>> resistors and some other fiddling.
> >>
> >>Promise was the first manufacturer who announced the intention of producing
> >>Mac-compatible IDE cards, but somehow never got around to it. Now that
gets > >>me interested: which model is capable of Mac conversion, and how
does one do
> >> it?
> >
> >Hi Sam,
> >
> >Unfortunately, I've never written the details down.
>
> A friend asked me to convert a few of these cards, so this time I wrote
> the details down. The following Rnn refer to surface mount resistors.
> There's some good info on Marc Schrier's Clock Chipping Home Page on how
> to remove surface mount resistors using the two pencil method.
>
> Move R12 to R16. These are near the top at the left middle.
> Move R48 to R46. These are on the right side of the card near the bottom
> and the top respectively.
> Move R10 to R9. These are under the flash chip.
>
> If you have the version of the Promise card that comes with a 66.667 MHz
> half-size metal can oscillator in the center of the chip, then these
> additional steps are required:
>
> Remove the 66.667 MHz half-size metal can oscillator.
> Move R21 to R19. Also near top at left middle.
> Add a 1000 ohm (1K ohm) resistor to R27 next to the osc. position. The
> R27 is a ways to the right of the pads for the resistor.
>
> All the resistors in question are 1000 ohm, so you don't actually have to
> move R12 to R16. You could move R48 to R16 and R12 to R46 for example.
> But the locations I paired are near each other, and doing them one at a
> time makes it less likely you'll lose the tiny resistors.
> >
> > One of the resistors is under the Flash chip. So that must be removed.
> > The easiest way would be to clip the pins and then desolder the (32) pins
> > one by one, but then you'd need a replacement chip and some way to program
> > it. So I usually do this the hard way and get all 32 pins to come loose at
> > once so I can reuse the chip. A replacement chip can run $10 or more, which
> > kills any economy in this conversion.
>
> I use an adjustable heat gun for this. I remove the rear metal bracket
> from the card, use a large spring loaded clip from Home depot (the metal
> type with the rubber on the snout) to support the board by clipping it
> near the opposite edge from the flash chip and then heat the back of the
> board where the flash chip's pins poke through. Every so often I test
> the solder around a few of the pins by poking the solder with a dental
> pick. When it's soft, I push gently on the pins of the chip until they go
> through/out of the board and the flash chip comes off.
>
> Before beginning I cover all the tiny components near the flash chip in
> modeling clay so that they don't disappear in the process. It takes 1 to
> 3 minutes to get the chip off this way (not counting the set up time).
> This part requires a bit of coordination and forthought. It also helps
> to be prepared to gently pry the chip from the front if the last few pins
> get wedged in the holes, but one must be careful not to pry on the chip if
> the pins are stuck because the solder isn't melted. If the solder isn't
> melted you'll pull the gromets out of the holes along with the chip's pins
> and that will ruin the board.
> >
> >One version of the Promise card has a 66 MHz half can oscillator on the board
> > which must be removed. Another version does not.
> >
> > Once all the resistors are in place, getting the firmware in place is still
> > a pain. The firmware updater from VST will not update the card unless it
> > already has VST firmware on it.
> >
> > So, what I do is to install a socket where the Flash chip goes. Then plug
> > in a programmed Flash chip from a real VST card. Then install the card in
> > a computer and run the flash updater, but stop at the last step before
> > it updates the firmware.
> >
> > Then oh so carefully, with the machine turned on, and the Promise card in
> > the PCI slot, pull the flash chip from the socket and install the flash chip
> > that you desoldered from the card. Then click on the last button in the
> > firmware updater, and it will program the VST firmware into the chip which
> > formerly had PC firmware on board.
> >
> > You can see that in addition to the difficult desoldering, you really need
> > to have an original (or copy) VST card already in hand, because you need
> > a programmed flash chip in order to fool the firmware updater into working.
>
> Actually, what I do these days is use my chip programmer to program the
> flash chip while it's off the card. But this still required me to have a
> legitimate VST card at one time in order to read the flash chip contents.
> And chip programmers range from about $200 up to umpteen thousand dollars
> (mine is about a $500 model), so unless you want one for other reasons,
> it's kind of pointless to get one just for this.
>
> But for anyone who wants to try this, there's the complete information, in
> a not very well organized fashion.

The only tricky part of the above is getting the Promise card to flash.
All the other stuff you can figure out by setting a Promise card next to a
VST card and moving components until they're identical. That's how I
figured it out. PeterH wrote in another thread that he originated the
conversion (minus the Flash chip moving trick) and others copied him, and
that's true in some cases, but a few of us did figure this out on our
own. Still, he did some very interesting work in identifying the
functions of some of the pins on the Promise chip. Makes me wonder if
he's got access to the Promise chip datasheets.