Unstable S-ATA-connection when reading.

I have an old PC running as a file server:
- PIII 600 MHz
- Gigabyte i810 mainboard
- 256 MB SD-RAM
- IBM DeskStar 80 GB hard drive P-ATA (system drive, /dev/hda1)
- Debian 4.0r1 Etch without Gnome
And i have recently installed a couple of new parts:
- SDM PCI SATA RAID 2P (SiL3512) controller
- Western Digital Caviar SE16 500 GB S-ATA (/dev/sda1)
- D-Link DGE-528T Gigabit Ethernet
(I have also attached the output of # lshw)

But the system freezes when I read large files (200 MB+) from the Caviar disk. If I transfer a (large) file from the DeskStar to the Caviar, nothing bad happens. But if I transfer a (large) file from the Caviar to the DeskStar, the transfer freezes after a couple of seconds, and I have to kill the programs. The same happens if I copy a (large) file internally on the Caviar disk. So something nasty happens when I read a remarkable amount of data from the Caviar disk.

/var/log/syslog gives me something like this each time:
Nov 13 16:05:11 server kernel: ata1.00: exception Emask 0x0 SAct 0x0 SErr 0x280000 action 0x2 frozen
Nov 13 16:05:11 server kernel: ata1.00: (BMDMA stat 0x64)
Nov 13 16:05:11 server kernel: ata1.00: tag 0 cmd 0xc8 Emask 0x2 stat 0xff err 0xff (HSM violation)
Nov 13 16:05:11 server kernel: ata1: soft resetting port
Nov 13 16:05:12 server kernel: ata1: SATA link up 1.5 Gbps (SStatus 113 SControl 310)
Nov 13 16:05:42 server kernel: ata1.00: qc timeout (cmd 0xec)
Nov 13 16:05:42 server kernel: ata1.00: failed to IDENTIFY (I/O error, err_mask=0x4)
Nov 13 16:05:42 server kernel: ata1.00: revalidation failed (errno=-5)
Nov 13 16:05:42 server kernel: ata1: failed to recover some devices, retrying in 5 secs
Nov 13 16:05:47 server kernel: ata1: hard resetting port
Nov 13 16:05:47 server kernel: ata1: SATA link up 1.5 Gbps (SStatus 113 SControl 310)
Nov 13 16:05:47 server kernel: ata1.00: configured for UDMA/100
Nov 13 16:05:47 server kernel: ata1: EH complete
Nov 13 16:05:47 server kernel: SCSI device sda: 976773168 512-byte hdwr sectors (500108 MB)
Nov 13 16:05:47 server kernel: sda: Write Protect is off
Nov 13 16:05:47 server kernel: sda: Mode Sense: 00 3a 00 00
Nov 13 16:05:47 server kernel: SCSI device sda: drive cache: write back

The S.M.A.R.T-values on the hard drive is good.

Does anyone have a clue of what can be wrong? Is it a hardware or software problem?