Usually these kind of incompatibilities come down to interaction between the controller, driver, storage device and the bridge.

This board uses a VIA VT82C686B and a HPT 370a, yes ?

Does it make a difference if you use the ports connected to the Via south bridge vs the High Point Raid?

What storage device are you using?

The StarTech ones use a Marvell chip and most of the generic ones use a JMicron one. The Marvell chip is often considered better quality or more compatible.

Possibly unpopular opinion: it really doesn't matter. What you're describing is the exact experience I have had with any kind of "to IDE" converter including CompactFlash which is supposed to be nearly passive.

Compared to end of the road 40GB/80GB drives that just worked in any system (32GB BIOS limitation aside), these converters are half-baked and very poor quality.

The biggest issue I've had with the StarTech is that when using that converter on the primary interface on certain Pentium-class motherboards (don't recall the exact models), the CD-ROM on the secondary interface (different cables) becomes inaccessible, even though they should be completely unrelated.

The biggest issue I've had with the JMicron is massive data corruption when attempting to use DMA through it, and massive data corruption when running a WD 1TB drive through it. A Toshiba 1TB drive is fine (other than DMA).

I have a PNY SATA SSD and it does not work through these converters, period.

Two drives on one cable may have problems also, if you care.

The best way to sum up these things is if you want the performance of IDE and frustrations with SCSI. I really hope ZuluIDE turns these things into e-waste someday.

It's a AMD 761 chipset, and 120GB Lexar SSD, though I also pulled out an old 80GB mechanical sata drive for testing. Raid controller made no difference in detection of the drive.

So I made a bit of progress. I asked a friend nearby if I can come borrow the one out of his PC, IDE2SATA2, identical to the ones I bought originally. Guess what? It just worked. I think I bought 6 bad ones in a row, haha. 4 of them were from ebay so... you know. Maybe it's down to a specific firmware that causes havoc with this chipset. But then again, they should work in the original xbox, but they don't there either. I guess we can just chalk this up to being very unlucky with the "good" adapters, and very lucky with the "bad" ones.

I am starting to agree - maybe I should just find a different storage solution.

jakethompson1 wrote on 2025-06-23, 18:12:

Possibly unpopular opinion: it really doesn't matter. What you're describing is the exact experience I have had with any kind of "to IDE" converter including CompactFlash which is supposed to be nearly passive.

Wellllll-- To be precise, there is no "to IDE" converter with CompactFlash devices. CompactFlash devices are IDE devices. (Or at least they are supposed to have a true IDE mode that works just like IDE when attached to a properly grounded Pata cable) The Pata 50 to Pata 40 adapters that people use with CF's are not any more complicated than the Pata 44 to Pata 40 adapters you can use with laptop drives. They are not nearly passive. They are completely passive.

The problem people run into with CF's is that CF's are often much newer than their retro PC. All the IDE history stuff comes into play. Stuff like PIO vs MWDMA vs UDMA / CHS vs ECHS vs LBA22 vs LBA28 vs LBA48 / ATA Removable Flag / ATAPI / Weird early IDE electrical stuff, etc. It's the kind of unexpected stuff that you can expect when pairing two devices from different decades. It's not CF specific either. You can get issues if you use a last-gen PATA spinning disk in really old PC or if you try to use a really old spinning disk in a last gen PATA PC. Too many different standards and these were not areas that product testers would have focused on during development back in the day.

So with that in mind, there is no 100% PATA / SATA bridge because there is no 100% standard IDE.

• If you really want to use SATA and you have a PCI >= 2.2 slot, getting a PCI SATA adapter with a bootable ROM is often the path of least resistance and best performance.
• If your system is older than PCI 2.2, a CF or a SD2IDE can be an easier path than a PATA / SATA bridge. CF's can be complicated because there are a lot of different generations & firmware. The SD2IDE devices are simplest because there's not a lot of different firmware
• If you really want to do MWDMA, then be prepared to try out a couple different storage devices.

halexudson wrote on 2025-06-23, 19:35:

So I made a bit of progress. I asked a friend nearby if I can come borrow the one out of his PC, IDE2SATA2, identical to the ones I bought originally. Guess what? It just worked. I think I bought 6 bad ones in a row, haha. 4 of them were from ebay so...
<snip>

Does the Marvell chip have the logo on your friend's adapter? What about your adapters?

For BX, use JMicron. For everything else, use Marvell.

halexudson wrote on 2025-06-23, 19:35:

It's a AMD 761 chipset, and 120GB Lexar SSD, though I also pulled out an old 80GB mechanical sata drive for testing. Raid controller made no difference in detection of the drive.

So I made a bit of progress. I asked a friend nearby if I can come borrow the one out of his PC, IDE2SATA2, identical to the ones I bought originally. Guess what? It just worked. I think I bought 6 bad ones in a row, haha. 4 of them were from ebay so... you know. Maybe it's down to a specific firmware that causes havoc with this chipset. But then again, they should work in the original xbox, but they don't there either. I guess we can just chalk this up to being very unlucky with the "good" adapters, and very lucky with the "bad" ones.

I am starting to agree - maybe I should just find a different storage solution.

Glad to hear that you are making progress. 6 bad controllers in a row randomly seems unlikely though. Could happen I guess but the only bad ones I have are ones that I broke.

Looking at your board on TRW, it looks like your board has two IDE controllers, VIA VT82C686B & HighPoint HPT370. Is that correct? The VT82C686B can be touchy. I've seen people report issues with VT82C686B and the Jmicron Pata/Sata bridges in the past. The HighPoint HPT370 is just kind of rare.

There is more than one version of the Marvell Para/Sata bridge chip. Can you post a picture of the adapter?

douglar wrote on 2025-06-23, 20:22:

Wellllll-- To be precise, there is no "to IDE" converter with CompactFlash devices. CompactFlash devices are IDE devices. (Or at least they are supposed to have a true IDE mode that works just like IDE when attached to a properly grounded Pata cable) The Pata 50 to Pata 40 adapters that people use with CF's are not any more complicated than the Pata 44 to Pata 40 adapters you can use with laptop drives. They are not nearly passive. They are completely passive.

The problem people run into with CF's is that CF's are often much newer than their retro PC. All the IDE history stuff comes into play. Stuff like PIO vs MWDMA vs UDMA / CHS vs ECHS vs LBA22 vs LBA28 vs LBA48 / ATA Removable Flag / ATAPI / Weird early IDE electrical stuff, etc. It's the kind of unexpected stuff that you can expect when pairing two devices from different decades. It's not CF specific either.

My specific experience was:
- Generic CF to IDE adapter from China
- Transcend 512MB CF card
- Generic ISA Multi I/O Card with IDE connector
- 386DX-40 motherboard with UMC 481 chipset

The issue was: unless the IORDY pin on the CF adapter was snipped to sever its connection to the ISA bus, the board would soft lock up (Ctrl-Alt-Del still worked) during POST.
In an IRC conversation, I was told that this issue is well-known in the Acorn RiscPC community--that CF-IDE adapters fail to put diodes on lines that are supposed to be open collector.
Fair, even with those diodes, the adapter would still be "passive," my point is, they aren't the set it and forget it solution they're made out to be. If you have a quiet IDE drive in your drawer, for your own sake, use it!

douglar wrote on 2025-06-23, 20:22:

You can get issues if you use a last-gen PATA spinning disk in really old PC or if you try to use a really old spinning disk in a last gen PATA PC.

That hasn't been my experience at all with late 90s to early 2000s IDE disks (1GBish up to 40GB).
I mess with 286 through Pentium systems. They work in any system I throw them at, at worst, using a 1024/16/63 configuration, or with XT-IDE Universal BIOS.

I can't expect them to be 100% perfect (those are mostly PIO or at most, Intel chipset DMA) but, they are 10 times better than these converters/adapters, as if making a working "IDE" (ATA version whatever, technically, yes) interface is lost like the recipe for medieval stained glass or Roman concrete.

That's why I get so frustrated with the adapters/converters--I know it doesn't have to be that way.

halexudson wrote on 2025-06-23, 19:35:

It's a AMD 761 chipset, and 120GB Lexar SSD, though I also pulled out an old 80GB mechanical sata drive for testing. Raid controller made no difference in detection of the drive.

So I made a bit of progress. I asked a friend nearby if I can come borrow the one out of his PC, IDE2SATA2, identical to the ones I bought originally. Guess what? It just worked. I think I bought 6 bad ones in a row, haha. 4 of them were from ebay so... you know. Maybe it's down to a specific firmware that causes havoc with this chipset. But then again, they should work in the original xbox, but they don't there either. I guess we can just chalk this up to being very unlucky with the "good" adapters, and very lucky with the "bad" ones.

I am starting to agree - maybe I should just find a different storage solution.

Just to offer my own data point: I've bought a number of the Startech adapters (10+) over the past few years and used them in as many systems and never once had a problem.

When you tried your friend's adapter that worked: did you match everything else about your original setup? Cable, jumper setting, SSD, etc.

And you're sure you were getting legit adapters, right? In the Startech box, with the Marvell chip labeled. We all know how ebay can be.

Going to throw a contradictory experience here compared to everyone else replying, and recommend that you're better off using JMicron controllers for SATA->IDE. I've had pretty bad write I/O corruption with Marvell controllers and all VIA chipsets, which goes away when I use JMicron controllers and disable IDE Block Write and IDE Prefetch in the BIOS.

douglar wrote on 2025-06-23, 20:52:

Looking at your board on TRW, it looks like your board has two IDE controllers, VIA VT82C686B & HighPoint HPT370. Is that correct? The VT82C686B can be touchy. I've seen people report issues with VT82C686B and the Jmicron Pata/Sata bridges in the past. The HighPoint HPT370 is just kind of rare.

There is more than one version of the Marvell Para/Sata bridge chip. Can you post a picture of the adapter?

Yes that's right. Attached are pictures of the Marvell chip on the working and a bad one. The first line is the same - 88SA8052-NNC2. To note, the working one is a IDE2SAT2, and the nonworking is a IDE2SAT. But, I had and returned 2x IDE2SAT2's not working already so I cannot test or take photos of those. I sourced my 6 in 2 batches - 2 from amazon brand new, 4 from ebay from the same seller as "open box new old stock". The seller does a lot of used electronics so I don't think he intentionally sent me faulty ones. I actually remembered I have a windows XP machine that though it does have sata, I think it has an IDE controller as well. I will try to test one of these on there tomorrow.

halexudson wrote on 2025-06-24, 02:31:

Yes that's right. Attached are pictures of the Marvell chip on the working and a bad one. The first line is the same - 88SA8052-NNC2. To note, the working one is a IDE2SAT2, and the nonworking is a IDE2SAT. But, I had and returned 2x IDE2SAT2's not working already so I cannot test or take photos of those. I sourced my 6 in 2 batches - 2 from amazon brand new, 4 from ebay from the same seller as "open box new old stock". The seller does a lot of used electronics so I don't think he intentionally sent me faulty ones. I actually remembered I have a windows XP machine that though it does have sata, I think it has an IDE controller as well. I will try to test one of these on there tomorrow.

Can you swap the chips to rule out the PCB? I have successfully repaired one I purchased years ago that was burnt by doing this. I ordered a 2011 datecode part from China.

mockingbird wrote on 2025-06-24, 15:47:

Can you swap the chips to rule out the PCB? I have successfully repaired one I purchased years ago that was burnt by doing this. I ordered a 2011 datecode part from China.

Theoretically, but, that's a lot of soldering for a $18 part 🤣. I'm not great with the hot air gun yet. I'd probably melt the chip.

halexudson wrote on 2025-06-24, 16:03:

Theoretically, but, that's a lot of soldering for a $18 part 🤣. I'm not great with the hot air gun yet. I'd probably melt the chip.

Ah, in that case, I wouldn't advise it.

I posted before about the newer versions of 88sa8052 without the logo. I also swapped one of those on a DeLock mSata adapter and it fixed an issue...

Definitely great adapters and they work great with 90% of platforms with DMA enabled. For VX (MDMA) and BX (UDMA2), I use JMicron and that works fine with DMA enabled. But JMicron is problematic for me as others have reported with DMA on other platforms.

mockingbird wrote on 2025-06-24, 22:32:

Ah, in that case, I wouldn't advise it. […]
Show full quote

halexudson wrote on 2025-06-24, 16:03:

Theoretically, but, that's a lot of soldering for a $18 part 🤣. I'm not great with the hot air gun yet. I'd probably melt the chip.

Ah, in that case, I wouldn't advise it.

I posted before about the newer versions of 88sa8052 without the logo. I also swapped one of those on a DeLock mSata adapter and it fixed an issue...

Definitely great adapters and they work great with 90% of platforms with DMA enabled. For VX (MDMA) and BX (UDMA2), I use JMicron and that works fine with DMA enabled. But JMicron is problematic for me as others have reported with DMA on other platforms.

JMicron chips have issues with DMA on ICH4 and probably newer and on Via 686B (and possibly other Via controllers/south bridges), AFAIK.

EDIT: JMicron and Marvell chips support TRIM, which can be useful for use with SSDs.