RetroGamer4Ever wrote on 2022-02-02, 14:27:

https://www.transcend-info.com/embedded/product/ssd/psd330

That's what you use. Not an mSATA to IDE adapter. It is designed to do EXACTLY what you need it to, with full performance. The 128GB version isn't even $200 dollars.

I just pray you were being ironic. It's an obsolete toy, for crying out loud!

Cuttoon wrote on 2022-05-20, 10:25:

RetroGamer4Ever wrote on 2022-02-02, 14:27:

https://www.transcend-info.com/embedded/product/ssd/psd330

That's what you use. Not an mSATA to IDE adapter. It is designed to do EXACTLY what you need it to, with full performance. The 128GB version isn't even $200 dollars.

I just pray you were being ironic. It's an obsolete toy, for crying out loud!

Apparently some people can afford to burn money.

mockingbird wrote on 2022-02-02, 18:29:

Here are my preliminary findings.

The Ableconn/Delock Marvell 88SA8052 does not work properly with DMA enabled on the ICH2. I tested it with two drives, one with a Linkmedia chipset and one with a Sandforce. The test platform was the ICH2 southbridge with the latest Windows 9x Intel INF installed for this platform (6.3.0.1007). I tested in both APM and ACPI mode. The symptom exhibited was that the mouse movement was slow, but I couldn't point that behavior specifically to read/write operations, since this did not happen during ATTO benchmarking, and ATTO benchmarking was perfectly fine with good numbers, despite this issue.

I then moved on to testing with a Startech IDE2SAT which uses the same Marvell chip. The chips look different though, the Ableconn has a 2021 datecode and the Startech a 2020 datecode. The Startech chip also has the Marvell logo, while the Ableconn does not, but they are both QFP and exactly the same shape, so I don't think this is really that important or pertinent. I interfaced the IDE2SAT using an "ST6006C" mSATA to SATA adapter. I booted the system, and the problem disappeared completely, and the computer ran perfectly. No more mouse choppiness, and ATTO benchmarks were up to spec.

I finally got to swapping the chips. The replacement chip is an old stock week 48 of 2009 datecode part.

The preliminary result is that it solved the issue. I tested with an ICH7 IDE port in Windows 98 with DMA enabled and everything worked fine. I still have yet to test with ICH2 (PIIX4E which is the southbridge counterpart to the 440 series Intel northbridges), but I believe that it is fixed. I don't know that Marvell is really producing these logo-less chips for Startech or Ableconn (Delock), but if they're not, then they're a pin-compatible substitute. The only way to know for certain is to de-cap one and look at the silicon.

One tip I can offer: When you prepare the drive, do so with the drive in the adapter and not connected to a SATA port but an IDE port. Also, make certain the partition type is set to 0x0C (FAT32 LBA) and not 0x0B (FAT32 CHS).

I haven't been following the many threads about these converters. Has anyone tried one of the unbranded 88SA8052 chips with an older VIA chipset (socket 7 or VT82c686/VT8231)?

The Ableconn converter is still ridiculously expensive by my standards. I think it exceeds the value of my entire old computer collection. I think I'm stuck with 20MB/s through a JMicron chip (and 6MB/s write speed if you align the partitions - that's supposed to improve performance, not destroy it).

Did anyone try installing a Drive with more than 128gb on These Adapters?
The issue is that I have 2 different Adapters and both cause the Same Problem on a Dell Precision D70 and on a Dell Inspiron 9100 - The BIOS of the Systems Display the Drive as a 512GB Drive (I have a 512GB MSATA) but dos/windows9x/windowsXP Displays the Maximum size of 128GB.

If I Format the Drive on those Laptops, it works - just not on a normal PC (to Transfer some data into the Drive) as it says the filme System is corrupt.
And the exact Same issue Happens if I pre-partition and Format it on a PC - after inserting it back into the laptop, WinCP install only detects like 1 partition with some weird GB, and maybe 2 additional ones without FS.

What am I missing here? I don’t get why the BIOS finds a 512GB HDD, but the Operating System doesn’t.

Does anyone of you have a clue?

TheRoss wrote on 2025-11-12, 18:13:

Does anyone of you have a clue?

Sounds like something in your stack is limited to LBA-28 addressing, which has a limit of 128GB
In DOS, you need Free DOS, IBM DOS 7, or MSDOS 7.1 and a BIOS that supports LBA48 for int13h calls. Just because the BIOS can correctly identify the size of the drive doesn't mean it is able to use all the space. If your BIOS can't handle it, you can install an XTIDE Universal BIOS option ROM or a 2004 vintage drive overlay, or maybe get a patched BIOS from Jan if you are really lucky.
In windows 98, you need to install a patch-- https://vogonsdrivers.com/getfile.php?fileid=635 And you want to keep your boot drive in the first 128GB.
Windows XP needs SP1 to do LBA48
Windows 2000 needs SP3 to do LBA48

douglar wrote on 2025-11-12, 19:30:

Sounds like something in your stack is limited to LBA-28 addressing, which has a limit of 128GB In DOS, you need Free DOS, IBM D […]
Show full quote

TheRoss wrote on 2025-11-12, 18:13:

Does anyone of you have a clue?

Sounds like something in your stack is limited to LBA-28 addressing, which has a limit of 128GB
In DOS, you need Free DOS, IBM DOS 7, or MSDOS 7.1 and a BIOS that supports LBA48 for int13h calls. Just because the BIOS can correctly identify the size of the drive doesn't mean it is able to use all the space. If your BIOS can't handle it, you can install an XTIDE Universal BIOS option ROM or a 2004 vintage drive overlay, or maybe get a patched BIOS from Jan if you are really lucky.
In windows 98, you need to install a patch-- https://vogonsdrivers.com/getfile.php?fileid=635 And you want to keep your boot drive in the first 128GB.
Windows XP needs SP1 to do LBA48
Windows 2000 needs SP3 to do LBA48

Hey douglar.

Thank You very much for your answer Sand the time you took looking into my case. It might in deed be an LbA-28 issue I am facing.
I will try to throw in an XPSP2 CD and See if it detects more than 128GB.
Since the laptop is from 2006, I have the feeling that the custom bios rom would probably not work for me since it is a tad beyond the Pentium1 Era.
But I will look into all these suggestions carefully and report back here. With luck I will be able to solve this with the hints you gave me, and this could help someone out in the future who might experience the same issues.

TheRoss wrote on 2025-11-12, 22:18:

Hey douglar. […]
Show full quote

douglar wrote on 2025-11-12, 19:30:

Sounds like something in your stack is limited to LBA-28 addressing, which has a limit of 128GB In DOS, you need Free DOS, IBM D […]
Show full quote

TheRoss wrote on 2025-11-12, 18:13:

Does anyone of you have a clue?

Sounds like something in your stack is limited to LBA-28 addressing, which has a limit of 128GB
In DOS, you need Free DOS, IBM DOS 7, or MSDOS 7.1 and a BIOS that supports LBA48 for int13h calls. Just because the BIOS can correctly identify the size of the drive doesn't mean it is able to use all the space. If your BIOS can't handle it, you can install an XTIDE Universal BIOS option ROM or a 2004 vintage drive overlay, or maybe get a patched BIOS from Jan if you are really lucky.
In windows 98, you need to install a patch-- https://vogonsdrivers.com/getfile.php?fileid=635 And you want to keep your boot drive in the first 128GB.
Windows XP needs SP1 to do LBA48
Windows 2000 needs SP3 to do LBA48

Hey douglar.

Thank You very much for your answer Sand the time you took looking into my case. It might in deed be an LbA-28 issue I am facing.
I will try to throw in an XPSP2 CD and See if it detects more than 128GB.
Since the laptop is from 2006, I have the feeling that the custom bios rom would probably not work for me since it is a tad beyond the Pentium1 Era.
But I will look into all these suggestions carefully and report back here. With luck I will be able to solve this with the hints you gave me, and this could help someone out in the future who might experience the same issues.

The option rom solution requires a socket for an option rom on an ISA card. trying to get it into a laptop isn’t impossible but would require some free space on your existing rom, which would likely be too much to ask for.

This software can fix the issue if you have > 16MB RAM… https://www.vogonsdrivers.com/getfile.php?fileid=2056

mockingbird wrote on 2023-10-19, 00:04:

I don't know that Marvell is really producing these logo-less chips for Startech or Ableconn (Delock), but if they're not, then they're a pin-compatible substitute. The only way to know for certain is to de-cap one and look at the silicon.

Most probably a HiSilicon/JMicron rebranded cr*p part which is being mistaken by the manufacturer or they are plainly commiting fraud, as the adapters with Marvell chips sell for a premium.
Why? Because Marvell themselves don't seem to be producing that converter chip anymore, so your only bet getting the chips are old stock unused rolls. The compatibility with only NT OSs is a frequent behavior on Hisilicon parts. Isn't the first and the last case were is known chinese sell counterfeit parts which have scammed not only users building stuff, but manufacturers as well. As you said, would be nice if someone could do a decap on these supposed "new" Marvell parts to confirm.

hyoenmadan wrote on 2025-11-13, 06:07:

Most probably a HiSilicon/JMicron rebranded cr*p part which is being mistaken by the manufacturer or they are plainly commiting fraud, as the adapters with Marvell chips sell for a premium.
Why? Because Marvell themselves don't seem to be producing that converter chip anymore, so your only bet getting the chips are old stock unused rolls. The compatibility with only NT OSs is a frequent behavior on Hisilicon parts. Isn't the first and the last case were is known chinese sell counterfeit parts which have scammed not only users building stuff, but manufacturers as well. As you said, would be nice if someone could do a decap on these supposed "new" Marvell parts to confirm.

I believe it is JMicron silicon underneath that has been re-worked to make it pin compatible with the Marvell IC.

Either way, I am fed up with both JMicron and Marvell for older systems... Constant weirdness like the device stops responding error in Windows with the PC looping benchmarks after many days. Also sporadic non-detection on boot at times. On My P4 with ICH2, it's ok, but with older platforms like BX and VIA 686B and older, they are nothing but headaches.

Moved on to SM2236. Pricey, and you also lose DRAM cache which you usually get on a decent SATA SSD, but native PATA and MLC NAND make up for it. So far so good.

mockingbird wrote on 2025-11-13, 16:48:

Either way, I am fed up with both JMicron and Marvell for older systems... Constant weirdness like the device stops responding error in Windows with the PC looping benchmarks after many days. Also sporadic non-detection on boot at times. On My P4 with ICH2, it's ok, but with older platforms like BX and VIA 686B and older, they are nothing but headaches.

Sometimes it's not the knock-off silicon that's the problem. Often the problem is that the companies willing to use knock-off silicon for the main course are also willing to use whatever is left over in the back of the closet as the side dishes. "Data sheet says 10K resistor, but this is what I found in the drawer" kind of thing. So the bridge ends up slower and more quirky and JMicron takes the blame for it.

While I still like to play around with different storage devices, I've come to appreciate the Sintechi SD adapters more over time. They have their quirks & limits, but to date they have been predictable quirks & limits for me. Ends up being fewer headaches compared to attempting to get the arbitrary sata device & bridge set up to run fast and stable on a legacy pata controller of uncertain heritage & firmware. Predictability has value. If I feel like I need sata in a system that has an imbedded controller that's <=ATA-4, promise PCI sata adapters are a big time savers.

douglar wrote on 2025-11-13, 17:35:

If I feel like I need sata in a system that has an imbedded controller that's <=ATA-4, promise PCI sata adapters are a big time savers.

Your post is well taken. Did you also find that Promise PCI controllers hog a lot of conventional memory because of their boot ROM? Otherwise it worked pretty nicely in a PCI 486 system.

mockingbird wrote on 2025-11-13, 23:46:

douglar wrote on 2025-11-13, 17:35:

If I feel like I need sata in a system that has an imbedded controller that's <=ATA-4, promise PCI sata adapters are a big time savers.

Your post is well taken. Did you also find that Promise PCI controllers hog a lot of conventional memory because of their boot ROM? Otherwise it worked pretty nicely in a PCI 486 system.

I don’t usually feel a strong need to put sata devices in DOS builds, but I hear what you are saying. There is less free upper memory with the larger option rom. And the promise cards can be on the thin side, so sometimes they sit a little loose in the older pci slots. And they are certainly not an option in a laptop.

However most 486 builds are PIO, and my experience is that most pata-sata bridges do OK when using using PIO unless you have an unusual Sata device or a super cheap sata bridge. And most bridges are usually OK with UDMA too, with the Marvel and Jmicron clones being the best performers. It’s mainly the MWDMA modes that you find in very late 486’s through early pentium 2 systems that will make you go crazy. Seems like few bridges handle them. Or maybe the problem is partly the sata devices’ firmware. Dunno. There are a lot of variables and I never found a fool proof combo. So on MWDMA systems, getting an industrial CF or a sintechi sd bridge might be the way to go.

If I wanted to put an m2 or an ngff sata device in a pata laptop that isn’t going to want mwdma, just about all of those 2.5” enclosures have worked for me.. They seem to have a better build quality that the bridges that sit on the end of 3.5 inch drives. But if there are incompatiblities, it could be something external to the pata bridge chip. It could be issues unique to that pata controller, or the sata device, or the quality of the other components used in the bridge besides the chip.