It could be because the 52x drive is streaming the data out faster than the computer can deal with it. This will mean that the computer has to keep stopping/restarting the data transfer, which has issues for the drive as it has to keep slowing down and speeding up, shifting the laser and controlling the transfer rate. It will also mean that the computer spends more time managing its memory and hardware, and less time actually doing the conversion.
Viper Zero is correct in that you should match devices of similar speeds onto IDE channels, but mixing devices on channels results in the faster device dropping to the speed of the slower device, which would cause the whole channel to drop to CD speed, rather than HD speed. If that were happening, I would doubt that it would manifest itself in the form of the reported behaviour.