Hi.

I have a number of (very inexpensive) PATA-SATA adaptors I bought a while ago. They plug into the back of a SATA drive and make it appear as a PATA one, for use in older machines without SATA ports.

The things are actually bidirectional, so could be used the other way around assuming you could physically connect them. They use a Sunplus SPIF-223A sata bridge chip, and work very well on almost every machine I have tried them with.

However, last night I ran into a machine that they DON'T work with, for an odd reason, and I was hoping someone might have an idea how to resolve the problem. I bought the things because I have a number of older shuttle boxes that both only have PATA ports and have only one or two PCI slots, which are full with required cards. This precludes the option of simply installing a PCI SATA card. However, the machines are still very useful, and surprisingly fast. I just want to add larger drives.

On my oldest shuttle, some 8 years old, the things work perfectly. I can simply attach any random SATA drive via an adaptor and it just works.

However, on my old games machine (a shuttle SN41G2 with an athlon XP3200 processor) when the drive is connected, I get the message "no 80 conductor cable connected" and it appears to hang at that point. The bios sees the drive correctly, though. And there definitely IS an 80-conductor cable connected

The detection of an 80 cable IDE cable appears to be done via detecting a low on /PDIAG, pin 34 on the IDE connector, during boot. Allegedly this is grounded internally on the 80-pin cable. On the adaptor, it is also connected to ground. On a real PATA drive, it isn't, but is presumably driven low at the correct point during boot.

I tried disconnecting the adaptor end pin 34, but it made no difference. If it was as simple as the signal needing to be actively driven at a specific time I would expect the adaptors to not work at all, as it's hard-wired to ground. However, as I said, in everything else I've tried, including USB to IDE adaptors they work fine. A friend of mine also has the same machine and it does the same thing, so it's not just a fault with my particular one.

Does anyone have any idea of why this specific series of shuttle get all bent out of shape about these adaptors? Or any ideas on how to persuade the thing that it really does have the correct cable?

pca

See section 8.2.11 of the attached document. Detection of 80W cable is done both by the host (looking for a grounded pin 34), and by the attached drives. The host queries the drives for what they saw after reset.

Basically, Pin 34 in the Host Connector shall not be attached to any cable conductor, and shall be attached to Ground within the connector. This signal is called PDIAG, or CBLID.

The conductor for PDIAG (conductor 66, 67, or 68 .. see Table 15/16 in the document) should not itself be grounded -- the drives use it to detect/communicate the cable type among themselves.

Oh -- if just hooking up a single drive (rather than a pair on the cable), ensure it is using the MASTER connector (end of cable), not the SLAVE connector (middle of cable). They're different.

And jumper the drive(s) (SATA converters) to use CSEL (cable select), rather than hard-jumpering them as master or slave.

Cheers

Another solution to the problem, might be to just use a real 40-wire cable. This will limit the transfer speed to 33MB/sec, but tends to be more compatible in some instances.

Cheers

One possible option to free up a PCI slot (for a SATA controller), is to replace a PCI ethernet card with a USB-ethernet dongle.

Just an idea, if all else fails.

Another way you could do is by getting some SATA to USB adaptors. That would work except unless you needed to boot of the drive and the motherboard doesn't support booting from USB.

FWIW I got a similar thing for my VIA EPIA PD (8235/82C586): a pair of SPIF-223A chips on a little board which plugs into the motherboard PATA connector and provides two SATA connectors on for connecting the drives -- and it plain didn't work.

It's possible I just got a faulty one, but while the BIOS would correctly list the attached device, it always hung solid trying to boot off it, and if I booted Linux off real PATA, the connection to the SATA disk would just be lost interrupts and CRC failures left, right and centre.

The disk was fine on a USB-to-SATA dongle, and I eventually solved the problem by ditching the system's PCI DVB card in favour of a USB DVB tuner I already had (and which anyway had better reception), and getting a very cheap Silicon Image PCI SATA card, which has worked flawlessly ever since.

Which is maybe not very helpful, especially as your Shuttle is presumably Nvidia MCP IDE, not VIA IDE, but I just thought I'd share some SPIF-223A scepticism.

Peter

I recently picked up one of these gadgets from DealExtreme, same chip family as above.

It won't work with an 80-conductor cable at all, but does work with a 40-conductor cable (a short one, perhaps 10" long; I didn't try a longer cable).

Weird. But useful.

Cheers