Guide to flashing PC SIL3112 SATA cards for Mac

[kahunalx]

Not reallly, myself and many others had issues with the card, it only worked somewhat and was pretty much unusable.

I "fixed" the issue by soldering an OEM regulator on the card, but that makes it unusable in my QS.

If you get the 64bit 3124 cards working with OS 9 I'd be interested to test those too!

Hi there,
I also bought this card and I'm experiencing issues with it on my G4 MDD. Basically, sometimes it works, sometimes it doesn't. I tested with different SSDs, I'm getting pretty much the same results.
Could you please let me know what regulator did you solder in? I'd like to give it a try.
Thanks!

 

[Ataman Honcho Honchev]

Hi there,
I also bought this card and I'm experiencing issues with it on my G4 MDD. Basically, sometimes it works, sometimes it doesn't. I tested with different SSDs, I'm getting pretty much the same results.
Could you please let me know what regulator did you solder in? I'd like to give it a try.
Thanks!

The correct regulator is the one on the picture. Apologies, I am at my home #1 and there I have only the "batch 0" of SeriTek/1S2, with the original (incorrect) regulator. I think, I am the only person in the world who has the pre-production SeriTek/1S2 without the correct regulator soldered (or the PCB revised).

The later SeriTek/1SE2 with the proper regulator is at the home #2.

Thus, I am just re-posting the relevant fragment of the card "made" by the HK pirate: that's the correct regulator.

On the other side, the entire 1S2 or 1SE2 is soon to become completely obsolete, one way (see the new 3112/3114/3512 driver) or the other (ParSATA driver).

I am more betting on the ParSATA at this time: so far no regulator problems and it has both SATA (2x) and PATA (1x) ports. Looking at the spec any EPROM is fine: while the registered EPROM size is only 64K, it looks the FCode residing in that part can "suck" the drivers from the "hidden" part of the EPROM.

An other upcoming solution is the 4x and 6x SATA port cards from LSI and Adaptec, featuring 3x3112 (2x3112) or 3x3512 (2x3512). I already posted the results.

The problem of disk-resident mass storage SCSI (SCSI Manager 4.3) drivers not being installed when the SIM is a system extension is already solved. It's very non-trivial, I don't think that was ever made before.

 

[Ataman Honcho Honchev]

BTW: due ParSATA project and these Adaptec / LSI cards the 3124 was postponed.
I verified: even that the LSI cards feature 3112 and the Adaptec cards feature 3512, the regulators are fine and there is no problem with the sleep on G4 QS.

 

[Ataman Honcho Honchev]

Could you please let me know what regulator did you solder in? I'd like to give it a try.

You asked that ("how to pirate the SeriTek/1S2?") from the co-founder of FirmTek and the original author of all FirmTek drivers. 🤣

But I am used to such requests.😃

The brief answer is: it's not just the regulator! The driver I wrote in 2003 having in mind the HDD drives made at that time pretty much sux in 2023 with SSD-s made today. Apologies for that: it was 20 years ago. Not only I was much more stupid back than - we had much less tools and the SATA standards were in infancy.

The best solution is to completely re-do the entire driver from the top to bottom.

 

[for this]

Hi there,
I also bought this card and I'm experiencing issues with it on my G4 MDD. Basically, sometimes it works, sometimes it doesn't. I tested with different SSDs, I'm getting pretty much the same results.
Could you please let me know what regulator did you solder in? I'd like to give it a try.
Thanks!

IME, for my MDD, the regulator has nothing to do with the intermittent malfunction. The original regulator on the Chinese card is AMS1117 3.3V. I have tried replacing it with an MIC29150 3.3V (I know it's for QS/DA but still...) and found no improvement.

In my case, it was the SATA and power cables not making a good contact. After changing the cables, the card has been working much more reliably for a year now. The issue still occurs, but rarely.

AYA 16-Inch 4-Pin IDE Molex to 5 SATA Power Cable 18AWG

AYA 16-Inch 4-Pin IDE Molex to 5 SATA Serial ATA Power Cable 18AWG

www.amazon.com

(cut the last three connectors out)

Page Not Found - Aliexpress.com

(no longer available but to give you an idea)

Both are snug-fit and can be used to connect two 3.5 HDD drives in the MDD's front bay where space is very limit.

PS. Some SSDs don't work with the card, probably because of the outdated driver.

 

[Ataman Honcho Honchev]

PS. Some SSDs don't work with the card, probably because of the outdated driver.

Absolutely - and I am guilty as charged. My excuse: these SSD weren't available back than.
If you look at the new drivers (I posted betas for 3112, 3114, 3512) you will be happier.
In the new release (hope the coming week) you will see that the extension will always mount the attached drive.

I am not releasing it yet because there will be hopefully a driver for VIA6421A, that should end the regulator misery.

 

[weckart]

Hi there,
I also bought this card and I'm experiencing issues with it on my G4 MDD. Basically, sometimes it works, sometimes it doesn't. I tested with different SSDs, I'm getting pretty much the same results.
Could you please let me know what regulator did you solder in? I'd like to give it a try.
Thanks!

Just as a confirmation, here are a couple of snaps of the Sonnet Tempo SATA card, basically a rebranded 1S2 card with the same firmware. Sonnet used the same MIC29150 3.3v regulator chip as the HK pirate.

 

[Ataman Honcho Honchev]

Just as a confirmation, here are a couple of snaps of the Sonnet Tempo SATA card, basically a rebranded 1S2 card with the same firmware. Sonnet used the same MIC29150 3.3v regulator chip as the HK pirate.

View attachment 2186724View attachment 2186723

Some correction. There is a small batch of SeriTek/1S2 cards which was made before the problem was discovered.
All these cards - except one - were modified by hand. Except ONE.
I am curious - if the Sonnet card is offered for a humble $800 on eBay... how much that guy would sell the "original" card for? IMO, the original Sonnet card - even for collectors - isn't worth more than, say, $100. The "original" one maybe an other $10 extra. The $800

Sonnet Tempo Serial ATA / SATA PCI Card For Macintosh Powermac Computers. | eBay

If your looking at this then you already know what it is, and so I won't bore you with the details. This is a working pull from a machine I acquired and I am passing it along. Sonnet Tempo Serial ATA / SATA PCI Card.

www.ebay.com

is a joke. The picture of my card is here, you can clearly see the original regulators.

After that first batch the PCB was changed and Sonnet got the updated version.

 

[Ataman Honcho Honchev]

One more... so you can see: it's 2023, not 2003.

 

[ojfd]

BTW: due ParSATA project and these Adaptec / LSI cards the 3124 was postponed.
I verified: even that the LSI cards feature 3112 and the Adaptec cards feature 3512, the regulators are fine and there is no problem with the sleep on G4 QS.

Looks wild.

Is that a miniSwap enclosure on top of G4? Does eSata enclosures work in OS9 and with your cards + eSata breakout cable?
Out of curiosity - did you also wrote the firmware for those SeriTek enclosures?

 

[kahunalx]

You asked that ("how to pirate the SeriTek/1S2?") from the co-founder of FirmTek and the original author of all FirmTek drivers. 🤣

But I am used to such requests.😃

The brief answer is: it's not just the regulator! The driver I wrote in 2003 having in mind the HDD drives made at that time pretty much sux in 2023 with SSD-s made today. Apologies for that: it was 20 years ago. Not only I was much more stupid back than - we had much less tools and the SATA standards were in infancy.

The best solution is to completely re-do the entire driver from the top to bottom.

My most sincere apology 🙇‍♂️🙇‍♂️🙇‍♂️
I obviously have no idea to who I'm talking to!
I just saw a post that described pretty much what I was seeing on my G4 and... hey I wanted that solution for me as well!

Right now is working way better, just failed a couple of times out of several power-off/power-on cycles. On the next reboot I just pressed "option" and then it allowed me to select my "Tiger" volume on the card.

In case it helps:
- First of all, I'm using the HK pirate card, it comes with the 29150 regulator. I have another 3112 PC vanilla card, exactly the same but with the standard regulator, I haven't tried with this one.

- I'm using a Transcend SSD230S 512GB SSD drive.

- I have OS 9 running on another SSD drive connected to the ATA100 IDE bus with an adaptor. I booted it up and flashed the card with the firmware dosdude1 posted on the 68kmla.org forum and reddrag0n referenced here.

- I'm using now shorter SATA cables. I have tested with all the SATA cables I have and none of them really click and secure in place, especially on the card side. So, I used a home-made approach with some kapton tape to keep them connected (don't you dare to slip out of the connector 😝). I also bought new cables, I'll be probably testing with them later.

(this is with the old long SATA cables)

(here is how it shows up now on Tiger)

Again, excuse my broad ignorance in all of this.

I'm not sure what is the default firmware the HK pirate uses. I saw dosdude1 posted that firmware a couple of months ago, so I went ahead and flash it directly.

Looking for other more reliable alternatives! 😃

Thanks all!

 

[DearthnVader]

Hi there,
I also bought this card and I'm experiencing issues with it on my G4 MDD. Basically, sometimes it works, sometimes it doesn't. I tested with different SSDs, I'm getting pretty much the same results.
Could you please let me know what regulator did you solder in? I'd like to give it a try.
Thanks!

I and many others had issue with the regulator the HK Pirate uses that are supposed to make the PowerMac G4 Quicksilver compatible with this PCI Card.

Some people do not have these issues, so I have not been able to figure way some people don't have issues.

I checked that all the connections were good in the regulator mod, and they seemed to be, but the issue was tho I could sometimes format the connected SATA SSD's they just didn't really work reliably. Thus they were really unusable.

I had a dead 3112 card with the standard regulator so after trying other drives and cables I just switched the regulator from the dead card and now the card is 100% reliable tested in a PowerMac 9600, PowerMac G3 Beige, PowerMac G4 MDD, and PowerMac G5 PCI.

It does not work with my PowerMac G4 QS, as we already know it needs the regulator mod.

The SSD's I'm using are the cheap but fast SSD's you can get off eBay not my auction )

Netac Internal SSD 256GB Solid State Drive SATA III 6GB/s lot | eBay

P ackage include: Netac 256GB internal SSD 1. Backwards compatibility with SATA 2.0 devices. Storage temperature: -40 ° C-85 ° C. Read Speed: Up to 500MB/s. Write Speed: Up to 400MB/s. TRIM: Support (Requires system support).

www.ebay.com

They work really well but the PCI Bus and model of Mac do make a difference in the speeds. The Mac's that are faster work faster, but I don't really get near the 133MB/s in peak drive speeds for a PCI card. I think the best peak speeds I've seen were in the G5 at about 90MB/s.

There is always overhead.

 

[Ataman Honcho Honchev]

The regulator problem has not much to do with the controller not recognizing certain drives.
The regulator problem affects only (and all!) QuickSilver and Digital Audio G4 Macs. Unfortunately these are probably the majority of G4 Macs running today.

To check the problem:
- disconnect all drives from the controller
- Install the controller in the G4
- In some cases (said G4 + some 3114 controllers) the machine won't even turn on properly, even without any driver
- In all cases if the affected G4 turns on, but when you invoke the "sleep" (be it "X" or "9" does not matter) the computer won't wake up.

With the only correct regulator I am aware of (see above) that problem disappears. That problem is due the ASIC.
We (= Apple DTS engineers, both FirmTek engineers, the manufacturer of the SeriTek cards) spent countless hours to verify this. The original design of Silicon Image uses the correct regulator. The manufacturers use a cheaper alternative. Which does not work because it wasn't made as Silicon Image designed the reference controllers.

I still do have reference design of 3112, 3512, 3114. All these use the correct regulator. The manufacturers try to be cheaper and it does not work.

The exceptions I am aware so far:

- the second (and later) batches of SeriTek cards
- the three-chip and two-chip RAID cards made by LSI and Adaptec
- the engineering samples of Silicon Image (largely ignored back than and totally ignored today).

All other cards, including the Adaptec 1210SA are baaaaaaaaaaaad!!!

Once the regulator is correct, the sleep problem is gone.

But:

- The SATA spec of 2003 is not the same as the SATA spec of 2023. Do not expect the software of 2003 support all SSD-s of 2023.

- Sometimes it's the SSD, which requires a workaround.

- The widely pirated SeriTek/1S2 set of drivers is my first attempt at SATA. It was OK in 2003, in 2023 it sux, sux and sux! And sux horribly. Apologies for that, it's my fault. True, the SeriTek/1SE2 is much better - but it wasn't pirated.

- The quality of the SATA L-to-L cables is patchy to say the least.

- The "quality" of the SATA "L" connectors sux even more horribly than my SeriTek/1S2 driver. These "L" connectors are designed to last only few plug-in-plug-out cycles. There are very good exceptions - see the Adaptec 2610SA card.
The middle name of the "L" connectors of that card is spelled "QUALITY". Much of the rest is the best just to provide a safe connection via PERMANENT patch cable. Once installed please don't move it.

In addition, SeriTek/1S2 does not support hot-swap. So unless you don't have a SATA bus analyzer and / or don't have the debug build of SeriTek/1S2 sources you can't really identify the problem: is it my driver, the loose cable, the loose "L" connector or a drive being not up to the spec. One is sure: if the computer wakes up properly after the sleep, than it has nothing to do with the regulator. And the regulator (once the computer is up and running) does not have much effect on drive being recognized or not.

 

[Ataman Honcho Honchev]

https://www.ebay.com/itm/3642131481... at about 90MB/s. There is always overhead.

The peak with 3112 is around 100MB/Sec, the controller is just that slow.
As far as the hardware: the 3112 is not much, but a SATA shell built around a glorified CMD 680 PATA controller "core".

The link you provided is very interesting.

- The seller is "netac-official-it", from Hebron, Kentucky. They sell the 256 GB drive for $16.50.
- An other seller, "netac-official-store" sells the same drive for $17.40

Netac 256GB SSD 2.5'' SATA III 6Gb/s Internal Solid State Drive 500MB/s PC/Latop | eBay

Rugged & anti-impact: Netac internal SSD supports shock-resistant solid state core, and it makes Netac's solid state resistant to shock and vibration. With Netac SSD performance in your PC, you can upgrade laptop / desktop computer directly to enjoy super-fast OS Boot-up, application loads...

www.ebay.com

They are located in Hebron, Kentucky.
- But $19.60 will buy you the same drive, in red. Also from Hebron, Kentucky.

Netac 256GB SSD 3D NAND 2.5'' SATA III 6GB/s Internal Solid State Drive 500MB/s | eBay

Netac 256 GB SSD 3D NAND 2.5" SATA III 6GB/s Internal Solid State Drive 500MB/s for PC/Laptop/Mac. B rand: Netac. Interface: SATAIII 6.0 Gb/s. Compatible with SATA 3.0 devices. Backwards compatibility with SATA 2.0 devices.

www.ebay.com

And so on. It looks like it's the same seller - but the price does vary, under what nick he sells the product.
The SSD controller is like on other no-brand controllers.
There is a secret mode on it. The link was mentioned here:

Info - Netac N530S 256GB SATA III SSD Review - Just how good can a $21.99 256GB SSD be?

I was on the market for an inexpensive SSD to use in older systems for short term testing, and I came across this listing, a Netac 256gb SATA III 6Gbp/s 2.5in SSD for just $21.99 at time of purchase. Similar budget drives from more reputable brands start at $28-30. Out of curiosity I purchased...

forums.tomshardware.com

The link itself is here (again, from tomshardware), mostly about that "Yeestor" SSD controller:

https://www.usbdev.ru/f/index.php?topic=5312.0

Don't be afraid of the ".ru" extension and all kind of warnings you get - just proceed. I have a major advantage reading all what they write - so I did.
As expected, nothing harmful there. Few links to "secret" SSD microcode, provided by Chinese manufacturer to whom they want to see to benefit from that:

Яндекс

Найдётся всё

disk.yandex.ru

xfjportal share MPTOOL – Smi AU IS Phision … controller

In any case seeing the price confusion on eBay and reading the problems reported by Russian users did not make me very exited about that product. I did see a 240GB no-name SSD here (FR) in Auchan yesterday for 40 Euro. First grabbed a pair, than gave them to the cashier for return. I may buy few of these through your link later and let my agent in the US ship them to me. But first I want the driver to be more compatible with what I have. And finish that VIA 6421A driver.

 

[flyproductions]

With the only correct regulator I am aware of (see above) that problem disappears.

As i mentioned in this post in some other thread, there seems to be at least one more.

I butchered this FS8860 from some dead TV-tuner-card and it works like a charm in my QS. With the stock AMS1117 on the card, the QS didn't even bong! Sure i can give no guaratee that this works for everybody. But the part has one main advantage over the MIC29150: It nicely fits to the solder pads.

 

[kahunalx]

Quick question if I may.
With either the AMS1117 or the 29150 regulator. Do you have the issue where sometimes the SSD drive is not recognized on boot?
That is happening on my MDD G4.

As for speed/throughput. I have the same SIL3112 chip integrated on an Abit P4 PC motherboard and pretty much maxes out at 100MB/sec on the PC

 

[DearthnVader]

Quick question if I may.
With either the AMS1117 or the 29150 regulator. Do you have the issue where sometimes the SSD drive is not recognized on boot?
That is happening on my MDD G4.

As for speed/throughput. I have the same SIL3112 chip integrated on an Abit P4 PC motherboard and pretty much maxes out at 100MB/sec.

I don't have that issue in any of the Macs I've tested with, listed #462 with drive type.

Is the drive just not auto mounting or it doesn't show up at all?

What brand and model of drive?

What OS or OS's?

 

[kahunalx]

Thanks for the reply!

A prohibited sign appears on the screen. Similar to:

The MAC is not booting.

Then I power cycle it, and it boots just fine. This is happening only sometimes, perhaps 2 out of 10 times I power on my MDD or something like that.

It's MacOS X Tiger what I have installed on the Transcend SSD230S 512GB SSD drive connected to the SIL3112 SATA PCI controller.
The system is a PowerMac G4 1.25 dual processor that can run OS9 natively. I believe it's this one:

Power Macintosh G4 1.25 DP (MDD) Specs (Mirrored Drive Doors, M8573LL/A, PowerMac3,6, M8570, 1914): EveryMac.com

Technical specifications for the Power Macintosh G4 1.25 DP (MDD). Dates sold, processor type, memory info, hard drive details, price and more.

everymac.com

More on what I have and done here:

https://forums.macrumors.com/threads/guide-to-flashing-pc-sil3112-sata-cards-for-mac.1690231/post-32095310

Thanks!

 

[Ataman Honcho Honchev]

Anyone can explain what is on this screenshot? It looks VERY scary. 😱

 

[Ataman Honcho Honchev]

Things here progress even more scary way.

 

[Ataman Honcho Honchev]

Thanks for the reply!

A prohibited sign appears on the screen. Similar to:

View attachment 2190905


The MAC is not booting.

Then I power cycle it, and it boots just fine. This is happening only sometimes, perhaps 2 out of 10 times I power on my MDD or something like that.

It's MacOS X Tiger what I have installed on the Transcend SSD230S 512GB SSD drive connected to the SIL3112 SATA PCI controller.
The system is a PowerMac G4 1.25 dual processor that can run OS9 natively. I believe it's this one:

Power Macintosh G4 1.25 DP (MDD) Specs (Mirrored Drive Doors, M8573LL/A, PowerMac3,6, M8570, 1914): EveryMac.com

Technical specifications for the Power Macintosh G4 1.25 DP (MDD). Dates sold, processor type, memory info, hard drive details, price and more.

everymac.com

More on what I have and done here:

https://forums.macrumors.com/threads/guide-to-flashing-pc-sil3112-sata-cards-for-mac.1690231/post-32095310

Thanks!

I bet, you got caught by one of the bugs of the widely pirated SeriTek/1S2.
The possible reasons:

1) My SeriTek/1S2 code of 2003 is far worse than what I would write today. It has a host of bugs, apologies for that.
2) The SSD you are using was made in 2020-es, about 20 years after the old SeriTek build.
It complies with SATA-III. Back than SATA-I was the latest and greatest.


If you pay attention to previous screens, in 2023 I am at 6.0.0, introducing new chipsets (like 3114, VIA6421 and more coming). You actually see the VIA6421, as "VIADuct3".
But why do I talk, talk, talk - and nothing, just screenshots (and few "beta" builds weeks ago)?

Here you see something like "6.0.0b14" or 14th "Beta" of 6.0.0 as, basically a teaser.
The reason: debugging a MacOS 9 "SIM" is far more difficult than a "KEXT" for macOS X. So it takes a lot of time!
I am reluctant to release (even as a demo) something I am not very confidential with.

It takes forever with a SIM. Once the "9" part of the 6.0.0 is done, things will progress much faster.

 

[Ataman Honcho Honchev]

I can testify, the VIA6421 has absolutely no regulator problem.
It could have other problems, but so far so good. "Sleep" works just fine.

 

[DearthnVader]

I bet, you got caught by one of the bugs of the widely pirated SeriTek/1S2.
The possible reasons:

1) My SeriTek/1S2 code of 2003 is far worse than what I would write today. It has a host of bugs, apologies for that.
2) The SSD you are using was made in 2020-es, about 20 years after the old SeriTek build.
It complies with SATA-III. Back than SATA-I was the latest and greatest.


If you pay attention to previous screens, in 2023 I am at 6.0.0, introducing new chipsets (like 3114, VIA6421 and more coming). You actually see the VIA6421, as "VIADuct3".
But why do I talk, talk, talk - and nothing, just screenshots (and few "beta" builds weeks ago)?

Here you see something like "6.0.0b14" or 14th "Beta" of 6.0.0 as, basically a teaser.
The reason: debugging a MacOS 9 "SIM" is far more difficult than a "KEXT" for macOS X. So it takes a lot of time!
I am reluctant to release (even as a demo) something I am not very confidential with.

It takes forever with a SIM. Once the "9" part of the 6.0.0 is done, things will progress much faster.

Well, I mean the SSD's I use are SATA III, but they work just fine with the SiL3112. I think is is just a matter of how well the drives support backwards compatibility.

 

[Ataman Honcho Honchev]

Well, I mean the SSD's I use are SATA III, but they work just fine with the SiL3112. I think is is just a matter of how well the drives support backwards compatibility.

Exactly. One SATA-III drive could be fine, the other not. In most cases this is due one of the problems in the driver.
I am not going to fix it, I am rather working on a new driver.

 

[Ataman Honcho Honchev]

New driver names, please vote!

- SImagicP133
- SImagic2
- SImagic2LP
- SImagic2ADPT
- SImagic4
- VIADuct3
- Vitamin4
- SImagicII
- Frodo8

Feel free to guess, who is who!

(Frodo8 to stay anyway because 'Frodo' is the semi-official name of the controller. Not all codename is known to me.
Missing: VIA6421, Silicon Image or CMD 0680, Vitesse 7174, Silicon Image 3114, Silicon Image 3512)

The numbers at the end usually correspond to the number of supported channels.
Frodo8 will be funny, with 8 channels. And probably one of the easiest: in form of KEXT it is already in G5 as G5 has that chip. So on "X" side it is natively supported by Apple, just not in all machines.

Their Frodo KEXT suffers from even bigger SATA-III problem than the SeriTek/1S2: as far as I remember only OCZ Vertex2 is compatible with Frodo KEXT. Not because Frodo won't support more, it's just the bug in Apple OF and Apple KEXT.