Mick, I genuinely wouldn't know where to start when trying to identify a good physical connection from a bad one. Presumably a shorter line has reduced resistance/impedance, are there any other readily measurable factors that can be taken into account? The BT engineers were apparently surprised at how good the quality of the line turned out to be... I can now confirm that this was definitely something to do with the configuration of the pipeline... I've made sure that at both ends STAC compression was turned off, swapped to make sure PPP was the only encapsulation type that could be negotiated (not MPP etc...) then made sure bridging was disabled everywhere in the configurations. I saved the configurations to flash and reset the pipeline with my fingers crossed. On coming back up, the pipeline said: Port Up: 00:00:04 Rx signal present Line Q: 15db Good It looked like something on the pipeline was hogging processor time and causing large packets to be dropped or errored. The reported line quality now drifts between 7db and 8db. There are fewer CRC errors by a huge factor: 20-300 WAN Stat >Rx Pkt: 4685 Tx Pkt: 4756 CRC: 237