Recently I got a pile of Advantech SQF-SM8 640 M.2 SATA SSDs at work. Some were small enough to be interesting for retro builds. So I went onto AliExpress and purchased an PLX PEX8111 PCI-PCIe bridge and an NFKH N-1061A SATA to PCIe controller. I wasn't expecting much, but first results were perfect: on boot on a P3 system, the extension ROM of the N-1061A wakes up, shows its v4.27 firmware, detects the SQF-SM8M2-64G-SBC SSD and declares it's in AHCI mode.

Here's the card:

Just one problem: legacy OS don't understand AHCI. Now, this chipset supports IDE mode and after some digging I found a firmware images for the ASM1061 here.

Now, booting from the floppy image works fine, it flashes the IDE firmware fine - but then things go less well:
- no more extension ROM activity at POST
- the system becomes unable to boot from floppy (it hangs when FDD boot should be initiated), even though it will still boot from IDE HDD and allow access to the floppy.

Fortunately, I was able to roll back to the AHCI firmware which behaved as before, but multiplie attempts to flash two different IDE firmware versions (4.20 and 4.27) resulted in the same issues. I then tried a different motherboard on the off chance that was causing the issues (first board: FIC KC19+ (i820) with P3-500, second board Tekram P5MVP-A4 (MVP3) with K6-2 350) but behaviour was identical. Incidentally, this latter board has 5V only PCI 2.1 and the PCI to PCIe adapter still worked - not only is the PEX8111 5V tolerant, but the card manufacturer actually hooked up the 5V lines. Nice. The P5MVP-A4 is also running a K6Plus patched BIOS (1.07 patch j.2) that also solves 64GB HDD limit (confirmed with 80GB PATA drive) so SSD size is definitely not the problem.

So what's going wrong with IDE mode? Well, if I read the instructions next to the fw images I posted above I see:

change the toggle switch on the adapter over to IDE mode

The toggle switch... right. No toggle switch on my N-1061A adapter...

My suspicion was that a toggle switch like that would do something with a pin on the controller which - lacking a pin - I could influence by messing around with an SMD resistor or two. So I looked into the [url=https://semiconductors.es/semiconductors.php?id=949526]datasheet for the ASM1061[/] but even though the pins are described in detail on pages 10 and 11, there's nothing there that looks to me like a pin that would toggle these modes.

Anyone have an idea what I could try?