ahh, so you're the one who grabbed that... i was eyeing it but it was a tad too pricy. glad it ended up in a good place
-
Hello MLAers! We've re-enabled auto-approval for accounts. If you are still waiting on account approval, please check this thread for more information.
Mathey MSATA-13UMAC IDE/SATA card
- Thread starter finkmac
- Start date
Interesting insights, thanks joevt!
Thanks. You may have noticed that a few of the Kuroutoshikou SATA cards have gone missing as well.ahh, so you're the one who grabbed that... i was eyeing it but it was a tad too pricy. glad it ended up in a good place
It provides the 2.5V core voltage for the controller chip. The cheaper cards just use a big diode in series with the 3.3V regulator. 3.3V - 0.7V = 2.6V; close enough, eh!
Some card designs use a 2.5V fixed LDO instead of the adjustable regulator.
Ah I see it now, interesting. I have a few cards on the way to experiment with. I'd like to benchmark and compare the VT6421A with the SIL3112 cards.
Did you encounter any sleep related issues, or is the AMS1117 3.3V LDO only problematic with the SIL3112?
Great! Glad to hear you're going to try it for yourself. I would love to see some benchmarks. I did a back-to-back QuickBench with a SIL3112 and the VT6421A just as a quick sanity check. Performance looked about the same, but I was only using a 500GB Hitachi HDD. I don't have any spare SATA devices fast enough to let either card really stretch its legs.
I tried sleep/wake on my Quicksilver. No problems in 10.4, works properly every time. In 9.2.2, it only seems to wake properly from sleep if the machine was booted into Tiger, then 9.2.2 selected as the startup disk, then restarted into 9.2.2. If cold booted into 9.2.2, or restarted in 9.2.2, the machine starts to wake from sleep but the video is never restored, and it's not responsive to keyboard commands; power button + return does not shut it down while it's sitting at a black screen. Not sure what going on there but it doesn't seem related to the LDO since it does work under specific circumstances. I'll try 8.6 on my 5400 later.
Strange. Although I would expect the sleep/wake issue to manifest if the OS was booted from a disk handled by this controller.
Does OpenFirmware see the card?
Okay, the first time I didn't give it a fair shot; the partition size on the SATA drive was too large. I made a smaller partition and copied my 8.6 system folder to it and verified that the 5400 will boot from that with a SIL3112 card. Then switched to the VT6421A card and it boots to a brief flashing question mark, followed by a happy Mac, then back to a flashing question mark indefinitely. Not sure what that indicates.
Is there anything that would be useful to see in OF in this case? I understand that it needs to be accessed through the serial port on this machine, which I am not set up for currently, but could do if necessary.
Is there anything that would be useful to see in OF in this case? I understand that it needs to be accessed through the serial port on this machine, which I am not set up for currently, but could do if necessary.
I got impatient and posted another thread to try troubleshooting the boot issue.
I've modded two different cards:
- Conceptronic CSATACOMBO 6421 V1.3
- Delock 70146
I used a Winbond W49V002AP for the flash chip and an Atmel AT93C46A EEPROM flashed with the binary finkmac uploaded. I tried both the Mathey and the XRack firmwares on the flash chip.
I tried them in my G4 MDD as well as my G4 Digital Audio. With or without the EEPROM, the subsystem IDs are the default ones (1106, 3249), hence the MacOSX mkext won't load and the card doesn't show up in the System Profiler on Tiger. It shows up on Leopard with the right device name but the driver is still not loaded.
Interestingly booting OS9 works and it also sees other disks in the system, both SATA and IDE.
I've ordered the AT93C46A EEPROMs from AliExpress and they are clearly not new. Only a few of them was readable with Xgpro, but I was seemingly able to program a few of them, and they read back fine. I wired them up to the VIA VT421A according to the datasheet. I tried getting VCC and GND from different spots, connecting ORG, leaving it disconnected, but even when I wired it up exactly as obsolete did, the card didn't work properly.
What am I missing?
- Conceptronic CSATACOMBO 6421 V1.3
- Delock 70146
I used a Winbond W49V002AP for the flash chip and an Atmel AT93C46A EEPROM flashed with the binary finkmac uploaded. I tried both the Mathey and the XRack firmwares on the flash chip.
I tried them in my G4 MDD as well as my G4 Digital Audio. With or without the EEPROM, the subsystem IDs are the default ones (1106, 3249), hence the MacOSX mkext won't load and the card doesn't show up in the System Profiler on Tiger. It shows up on Leopard with the right device name but the driver is still not loaded.
Interestingly booting OS9 works and it also sees other disks in the system, both SATA and IDE.
I've ordered the AT93C46A EEPROMs from AliExpress and they are clearly not new. Only a few of them was readable with Xgpro, but I was seemingly able to program a few of them, and they read back fine. I wired them up to the VIA VT421A according to the datasheet. I tried getting VCC and GND from different spots, connecting ORG, leaving it disconnected, but even when I wired it up exactly as obsolete did, the card didn't work properly.
What am I missing?
Last edited:
I was wondering why the EECS pin was connected to GND, so I disconnected the EEPROM related pins of the VT6421A from the board and wired up the AT93C46A directly, no difference.
I tried measuring the voltage between CS and GND when I powered on the Mac, and it didn't go up. My understanding is that is should rise up to the VCC voltage to activate the EEPROM and read its contents, so I'm very puzzled now. It would be nice to know how it works for obsolete.
I tried measuring the voltage between CS and GND when I powered on the Mac, and it didn't go up. My understanding is that is should rise up to the VCC voltage to activate the EEPROM and read its contents, so I'm very puzzled now. It would be nice to know how it works for obsolete.
Last edited:
Hi @haramia, just catching up on this after a busy weekend, I responded to your PM but I see you posted newer updates here. Sounds like you lifted the EECS pin on the VIA chip off the board like I suggested. Weird that you never see it go high. The AT93C46 datasheet states "A valid instruction starts with a rising edge of CS" so I think you are on the right track for the reason it is not working for you.
What are you using to probe EECS? Unfortunately the earliest I could get an oscilloscope trace for you would be 2 weeks from now, since I will be traveling.
AT93C46A does not have the ORG pin; it's 16-bit only, so you should leave pin 6 disconnected on those. I have only tested the plain AT93C46 but I don't see a reason why AT93C46A shouldn't work.
What are you using to probe EECS? Unfortunately the earliest I could get an oscilloscope trace for you would be 2 weeks from now, since I will be traveling.
AT93C46A does not have the ORG pin; it's 16-bit only, so you should leave pin 6 disconnected on those. I have only tested the plain AT93C46 but I don't see a reason why AT93C46A shouldn't work.
I was able to source a few PC cards that had the pads for the EEPROM. I soldered it on and it's working properly. I've tested 10.2.8 with a Kingston A400 SSD, and it seems to be working fine. To make it work, I had to install it onto a PATA disk, upgrade to 10.2.8, then take an image and restore it onto a partition on the SSD.
This way it boots, but the interesting thing is the same SSD boots and works with my SIL3112 card as well, even though that card is supposed to only support 10.3+.
I've also encountered the sleep/wake issue. I was booted into Leopard on one of the PATA HDDs when the VT6421A card was installed. When my G4 MDD went to sleep I wasn't able to wake it up.
I'll check if I can trigger the issue when booted into an OS from a disk connected to the controller card.
This way it boots, but the interesting thing is the same SSD boots and works with my SIL3112 card as well, even though that card is supposed to only support 10.3+.
I've also encountered the sleep/wake issue. I was booted into Leopard on one of the PATA HDDs when the VT6421A card was installed. When my G4 MDD went to sleep I wasn't able to wake it up.
I'll check if I can trigger the issue when booted into an OS from a disk connected to the controller card.
Last edited:
Update:
The sleep/wake issue may have been caused by an unrelated factor. I disconnected all disks except for the SSD containing 10.2.8 and it goes to sleep and wakes up properly.
Unlike an older HDD, it seems this SSD is not happy with my SIL3112, so it seems the VT6421A's compatibility with modern SSDs are better at least.
Another note, I was able to install 10.2.6 with the Apple Internal Edition DVD directly onto a SATA disk and boot from it.
I still want to do some benchmarks to compare the controllers.
I also have an ACARD AEC-6290M on the way which uses a PATA only controller chip with the addition of two SATA-PATA bridges and theoretically it should also be capable of 150MB/s data transfer speed while retaining better compatibility for older OSs.
The sleep/wake issue may have been caused by an unrelated factor. I disconnected all disks except for the SSD containing 10.2.8 and it goes to sleep and wakes up properly.
Unlike an older HDD, it seems this SSD is not happy with my SIL3112, so it seems the VT6421A's compatibility with modern SSDs are better at least.
Another note, I was able to install 10.2.6 with the Apple Internal Edition DVD directly onto a SATA disk and boot from it.
I still want to do some benchmarks to compare the controllers.
I also have an ACARD AEC-6290M on the way which uses a PATA only controller chip with the addition of two SATA-PATA bridges and theoretically it should also be capable of 150MB/s data transfer speed while retaining better compatibility for older OSs.
Last edited:
I have tried the VT6421 cards in a 5400 and a 7600. I'm ready to declare that they just don't work well in Old World machines. They show up in System Profiler, and connected drives will appear and work to some extent, but file transfers will hang, OS installs will fail, and booting always fails.
It's a bummer, as I was hoping this chip would be a good basis for a multi-function PCI card for 5x00 machines. The cards work great in G4s, at least. I haven't tried one in a G3 yet.
It's a bummer, as I was hoping this chip would be a good basis for a multi-function PCI card for 5x00 machines. The cards work great in G4s, at least. I haven't tried one in a G3 yet.
Do these cards have compatibility with newer SATA 300 and 600 devices? I know back in the day these VIA6421 cards hated newer drives and you have to force the drive to 150Mbit mode (via jumper) for it to work on these cards. Later SATA drives dropped the SATA-150 mode jumpers and won't work at all with this chip in many cases.
It mainly depends on the controller on the drive I think. Some had to be forced, some was compatible with the older protocol by default.
