SATA? In an old world PCI Power Mac?! IMPOSSIBRU! (Updated 11/22/23)
- Thread starter XodiumRetro
- Start date
There's a build of my flashrom fork at https://forums.macrumors.com/thread...in-a-beige-power-macintosh-g3.2303689/page-32 but I haven't tried it with the
-p satasii option. Of course, no PCI method is going to work for a flash chip that needs a voltage change. It should at least work as well as the linux flashrom.exe and maybe the Mac OS 9 firmware updater.
Many thanks Joe! I downloaded "flashrom-joevt-10.4-ppc" and will explore it as soon as I can. If it writes or can be made to write firmware to PCI card resident PROMs under Mac OS it'd be a significant advancement for Macintosh SATA card accessibility in my view. Presumably it would ignore and overwrite preexisting PROM data, or lack of data.
The logic behind the apparent inability of the SilG3114101UpDriveR1 and SilG351210UpDriveR1 firmware installers to overwrite zero or alien data puzzles me, though maybe the idea was to prevent inadvertent overwrite of unintended PCI card PROMs or some similar safety precaution. Maybe one of us could rather easily modify the command structures in those scripts to eliminate that barrier - perhaps with a clear warning for the user to be certain to target the intended PCI card.
And everyone: I substantially refined my SATA section of my Nerd Cave notes again, aiming for as clean, clear, and accurate a description of this subject as I can. Comments are very highly appreciated of course.
If it doesn't annoy Collin I encourage verbatim replication of that information elsewhere to insure it's preserved, including siting independent copies of all the linked files within a reliable domain which won't disappear like an individual's domain might when they die. TinkerDifferent would be a good location for example... Just attribute us properly please. And please add a link to the original page so people can check for updates.
My first efforts to revise my last Sil3512 card, of unique design, and my two port HighPoint HPT372NLFI for Macintosh PCI use both failed, alas. I suspect a Sil3512 card hardware flaw, possibly an open Sil3512 IC pin solder joint. I'll explore further. Sans any review of the bridge IC I installed, as best I recall, Collin's 1S2_512.rom, which I vaguely suspect is his uncompressed Sil3112 firmware, into the HighPoint HPT372NLF card's native 2 Mb PROM as a blind experiment, to no avail. My efforts to put these two cards into service aren't important though of course...
Cheers!
The logic behind the apparent inability of the SilG3114101UpDriveR1 and SilG351210UpDriveR1 firmware installers to overwrite zero or alien data puzzles me, though maybe the idea was to prevent inadvertent overwrite of unintended PCI card PROMs or some similar safety precaution. Maybe one of us could rather easily modify the command structures in those scripts to eliminate that barrier - perhaps with a clear warning for the user to be certain to target the intended PCI card.
And everyone: I substantially refined my SATA section of my Nerd Cave notes again, aiming for as clean, clear, and accurate a description of this subject as I can. Comments are very highly appreciated of course.
If it doesn't annoy Collin I encourage verbatim replication of that information elsewhere to insure it's preserved, including siting independent copies of all the linked files within a reliable domain which won't disappear like an individual's domain might when they die. TinkerDifferent would be a good location for example... Just attribute us properly please. And please add a link to the original page so people can check for updates.
My first efforts to revise my last Sil3512 card, of unique design, and my two port HighPoint HPT372NLFI for Macintosh PCI use both failed, alas. I suspect a Sil3512 card hardware flaw, possibly an open Sil3512 IC pin solder joint. I'll explore further. Sans any review of the bridge IC I installed, as best I recall, Collin's 1S2_512.rom, which I vaguely suspect is his uncompressed Sil3112 firmware, into the HighPoint HPT372NLF card's native 2 Mb PROM as a blind experiment, to no avail. My efforts to put these two cards into service aren't important though of course...
Cheers!
I substantially modified my SATA section of my Nerd Cave notes yet again, this time with recognition that the SilG3114101UpDriveR1 and SilG351210UpDriveR1 firmware installers appear to have no practical use in their current form, so I regelated them to a rhetorical sideline. I now suggest that FlashROM is rather likely to be effective so it's now indicated as the best firmware installer for all SATA cards with Silicon Image SATALiink bridge ICs except the Sil3112. But I've not proven this to myself through direct experience yet, so I describe it with that caveat.
Sorry about my serpentine path, but I hope I'm approaching an accurate and cleanly presented review of this realm. Comments are very highly appreciated of course! Cheers!
Sorry about my serpentine path, but I hope I'm approaching an accurate and cleanly presented review of this realm. Comments are very highly appreciated of course! Cheers!
I've bought two batches of the Sil3112 cards from Aliexpress and have had zero luck with flashing them. I've tried flashing in both a Mac (Dual G4/500) and a DOS PC.
When I try and flash them on a Mac, the interface freezes when I try and update the firmware and when I try via DOS, it tells me that "No EEPROM/Flash device was found"
These cards appear to have 128K ROMS (AM28F010-150JC). I'm not sure what I'm doing wrong here. If anyone has any tips or tricks before I try and desolder the ROM and figure out how to program it separately, I would appreciate it.
EDIT: I think I see the problem. The EEPROM on my card requires 12V to flash, so the only option is to desolder and flash externally. Is it worth buying a larger ROM to use the uncompressed firmware if I'm going through the trouble of soldering/desoldering?
When I try and flash them on a Mac, the interface freezes when I try and update the firmware and when I try via DOS, it tells me that "No EEPROM/Flash device was found"
These cards appear to have 128K ROMS (AM28F010-150JC). I'm not sure what I'm doing wrong here. If anyone has any tips or tricks before I try and desolder the ROM and figure out how to program it separately, I would appreciate it.
EDIT: I think I see the problem. The EEPROM on my card requires 12V to flash, so the only option is to desolder and flash externally. Is it worth buying a larger ROM to use the uncompressed firmware if I'm going through the trouble of soldering/desoldering?
Attachments
Last edited:
JAG, it should work - there should be no need to remove the 1 Mb PROMs. I used Collin's cool SeriTek1S2 firmware installer 5.3.1 (or here or here) and his cool Sil3112-1S2 patched compressed PROM firmware to reprogram AliExpress Sil3112 and thus probably the same cards sans removal of their original 1 Mb PROMs, though in my case in my Power Macintosh 9500/120 with a G3 445 MHz processor card running OS 9.1. The process was superbly easy and efficient.
But I don't recall which PCI slot I used to do so, and later discovered that the card's not universally functional in any slot under all slot card neighbor mixtures. I don't know which specific Macintosh or Mac OS you're using, nor could I discern whether a particular slot or community of cards is likely to cause a dysfunction anyway, so I suggest removing all nonessential cards temporarily. And if you still experience a dysfunction try slot experiments - presumably you'll discover which work and which don't, at least in a mostly unpopulated PCI environment.
And run some version of Mac OS 9 (maybe OS 8 is fine too but my instinct would be to run OS 9 until your glitch is resolved).
And don't move the PROM voltage selection shorting bar - for the original PROM leave it in its original R25 position.
Addendum: Nice catch JAG, and sorry I failed to notice the PROM difference between the cards you received and those I did, or am guilty of some other personal memory failure.
Collin knows better than I do and perceives no need to upgrade to 4 Mb. But personally I view the extra storage capacity as wise because it causes no harm and adds the option of using the uncompressed version of firmware. And it might possibly prove very useful if some cool new firmware which requires more that 1 Mb arises later. The opportunity to add capacity is free now, but would come at the expense of another round of component ordering and soldering tasks later if a glorious new firmware option arises. Just my personal view of course...
Other news everyone: Due to time pressures forcing me to defer exploration of FlashROM to later, I replaced all the PROMs in my seven AliExpress Sil3114 cards with externally reprogrammed 4 Mb AM29LV040B PROMs, moved the shorting bar to position R4, replaced all the electrolytic capacitors with tantalum capacitors, drilled two holes and cut a slot in each panel, and installed a short SATA cable with eSATA panel mount connector to provide an eSATA port in the panel as Robin-Fo illustrated. No glitches arose, and I'm populating all my Mac OS 10 systems with them. It's sooooo nice having three ports of wicked fast and solidly reliable internal SATA and one port of external eSATA in all these systems, and I look forward eagerly to replacing all the original SCSI and PATA storage devices with solid state SATA storage devices.
I plan to continue work with FlashPROM as soon as I can and will provide updates here and in my SATA section of my Nerd Cave notes whenever new information arises. Cheers!
But I don't recall which PCI slot I used to do so, and later discovered that the card's not universally functional in any slot under all slot card neighbor mixtures. I don't know which specific Macintosh or Mac OS you're using, nor could I discern whether a particular slot or community of cards is likely to cause a dysfunction anyway, so I suggest removing all nonessential cards temporarily. And if you still experience a dysfunction try slot experiments - presumably you'll discover which work and which don't, at least in a mostly unpopulated PCI environment.
And run some version of Mac OS 9 (maybe OS 8 is fine too but my instinct would be to run OS 9 until your glitch is resolved).
And don't move the PROM voltage selection shorting bar - for the original PROM leave it in its original R25 position.
Addendum: Nice catch JAG, and sorry I failed to notice the PROM difference between the cards you received and those I did, or am guilty of some other personal memory failure.
Collin knows better than I do and perceives no need to upgrade to 4 Mb. But personally I view the extra storage capacity as wise because it causes no harm and adds the option of using the uncompressed version of firmware. And it might possibly prove very useful if some cool new firmware which requires more that 1 Mb arises later. The opportunity to add capacity is free now, but would come at the expense of another round of component ordering and soldering tasks later if a glorious new firmware option arises. Just my personal view of course...
Other news everyone: Due to time pressures forcing me to defer exploration of FlashROM to later, I replaced all the PROMs in my seven AliExpress Sil3114 cards with externally reprogrammed 4 Mb AM29LV040B PROMs, moved the shorting bar to position R4, replaced all the electrolytic capacitors with tantalum capacitors, drilled two holes and cut a slot in each panel, and installed a short SATA cable with eSATA panel mount connector to provide an eSATA port in the panel as Robin-Fo illustrated. No glitches arose, and I'm populating all my Mac OS 10 systems with them. It's sooooo nice having three ports of wicked fast and solidly reliable internal SATA and one port of external eSATA in all these systems, and I look forward eagerly to replacing all the original SCSI and PATA storage devices with solid state SATA storage devices.
I plan to continue work with FlashPROM as soon as I can and will provide updates here and in my SATA section of my Nerd Cave notes whenever new information arises. Cheers!
Last edited:
I appreciate the encouragement but the EEPROM on my cards requires 12V to flash and as such can't be flashed on the card.
For anyone else struggling with this, removing the EEPROM and flashing in a programmer did work just fine and my card is now booting a Dual G4/500 without any trouble and will try in my Beige G3 next.
If anyone out there has bought a SIL3112 card and can't seem to flash it have a look at what EEPROM is on it. That was the key piece of information I was missing.
Thanks for the consolidated flashing guide. Unfortunately, my eBay card has an AMD AM28F010 ROM chip (12V write) and I was unable to flash it with the utilities (OS 9 or DOS). I don't have either an air soldering kit nor an EEPROM programmer, so I'm trying my luck buying another card on Amazon which has a clearly marked SST chip. If that one doesn't actually have the right chip, then perhaps I will make the case for buying more tools.
Do these cards work with, and are bootable from optical drives? I have a few PowerMac towers, two have the early ATA controllers that are limited to 128GB, and my B&W G3 has a dead optical drive too. Another machine I could use these with is a Power Computing machine. I have a SCSI2SD in it, but all my SCSI CD-ROMs are dead. If optical drives work , that could be a good application for one of these cards.
Do these cards work with, and are bootable from optical drives? I have a few PowerMac towers, two have the early ATA controllers that are limited to 128GB, and my B&W G3 has a dead optical drive too. Another machine I could use these with is a Power Computing machine. I have a SCSI2SD in it, but all my SCSI CD-ROMs are dead. If optical drives work , that could be a good application for one of these cards.
If I had my act more together I'd say hey, you could send your cards to me and I could flash 'em for you. To anyone proficient with hot air, it's a really quick and easy job. Though that doesn't account for the programming part...
And yes, optical drives do work. I've actually got a SATA optical drive in my 9600 and I've booted from/installed Mac OS from it many a time. Old Macs see the SATA card as a SCSI interface so they work pretty much natively in my experience.
(The only time things start getting really weird is New World Macs. Some require a voltage regulator swap to work properly.)
I've done lots of soldering over the years, but never with hot air. I wouldn't mind adding to my skillset. Come to think of it, I might have a programmer, but I don't know it's capabilities. It is part of a device to interface with GM OBD1 fuel injection systems. I used to burn chips to change the calibrations. This is the device, the company is now out of business. https://web.archive.org/web/2021042...toprom-package-usb-version-p-54.html?cPath=97
The second card I got had an ST chip, but it was also a 12V write chip so I was unable to flash it. Perhaps it is time to invest in some more tools.
If I am removing the ROM anyway, is there any advantage to replacing it with a 512k and using the full un-reduced ROM?
If I am removing the ROM anyway, is there any advantage to replacing it with a 512k and using the full un-reduced ROM?
I got a PowerMac 7600. Nice condition but the original hard-drive was deceased; the CDROM was seized; the floppy needed service, and more RAM would be nice. Obviously this was a candidate for the PCI-SATA treatment.
So I took a SeriTek-reflashed board from my B&W G3 and stuck it in the 7600. No-go. The ROM didn't load and the card reported itself as (1095,3110). MacOS9 wanted nothing to do with it and neither did Linux (YDL). Hhmm. The card was perfectly fine in the B&W G3.
I also have a reflashed Sil3114 card and this correctly reported itself as (1095,3114) in the 7600 although MacOS can't use it. As an aside, Linux patched with https://www.kernel.org/pub//linux/k....6.1-mm3/broken-out/ide-siimage-sil3114.patch works with SATA drives attached to channel 0 or 1 of a SiL3114. [As another aside, I suspect that it may be possible to flash a SiL3114 with SeriTek/1S2 and have two channels seen by Classic Mac.] Anyhow, the PCI bus itself seemed good. Yet the 3112 fails??
Finally, I swapped the card with an almost identical one in my Beige G3. Surprise .. that card worked in the 7600 and the troublesome card worked in the G3!
Explain that. Supply rails iffy on the 7600 and the bad card misreads its eeprom???
So I took a SeriTek-reflashed board from my B&W G3 and stuck it in the 7600. No-go. The ROM didn't load and the card reported itself as (1095,3110). MacOS9 wanted nothing to do with it and neither did Linux (YDL). Hhmm. The card was perfectly fine in the B&W G3.
I also have a reflashed Sil3114 card and this correctly reported itself as (1095,3114) in the 7600 although MacOS can't use it. As an aside, Linux patched with https://www.kernel.org/pub//linux/k....6.1-mm3/broken-out/ide-siimage-sil3114.patch works with SATA drives attached to channel 0 or 1 of a SiL3114. [As another aside, I suspect that it may be possible to flash a SiL3114 with SeriTek/1S2 and have two channels seen by Classic Mac.] Anyhow, the PCI bus itself seemed good. Yet the 3112 fails??
Finally, I swapped the card with an almost identical one in my Beige G3. Surprise .. that card worked in the 7600 and the troublesome card worked in the G3!
Explain that. Supply rails iffy on the 7600 and the bad card misreads its eeprom???
There might be a problem with different versions of Open Firmware. You would have to compare the ROMs to see what's different. The ROMs can be converted from fcode to Forth, modified, then tokenized back to fcode. The ROMs might have compressed parts which complicates the process. Maybe you can find discussions about the cards on 68kMLA and MacRumors.
I was initially suspicious of the early OFW version of 1.0.5 in the 7600 but people had reported success with the SeriTek reflash in similar machines with this version. And I know that the ROMs are identical in the cards because I flashed them both myself.
Link at top of page still works. Is this the same card and has anyone gotten it to work with OS9? Does RAID work in Macs under OS9 or its horrid replacement? Wondering if it'll work in Classic under same.
Curious about Linux as well, eventually building a new tower. New toy in stocking awaiting Ubuntu install.
Curious about Linux as well, eventually building a new tower. New toy in stocking awaiting Ubuntu install.
Not sure what link you're referring to here. In fact, the actual 3112-based card that works for me in the PM7600 is this that I posted about much earlier - https://tinkerdifferent.com/threads...c-impossibru-updated-11-22-23.1494/post-10750 The card that doesn't work in this machine is identical to except the eprom is soldered. The 3114 4-port card doesn't woth for Classic MacOS.
The 3112 card works fine for OS7/8/9/X and Linux. I hace no experience of using RAID, sorry.
I use this card with Debian 8, YellowDog 3, and NetBSD 9 -- though YDL3 does need a kernel built with the siimage driver.