3Confirmation that this reasoning is not too invalid is that HPPI specify a maximum length of 25m where their fastest signal is the 25MHz clock. Figure 2 suggests that doubling the frequency reduces the length for a given loss by 1.4, so our 100MBaud signal, at 50MHz, should be acceptable at 25/1.4 or 18m. There may be some factors which make the links marginally worse than HPPI, other factors which make them better. The AT&T application note on the 41 series buffers suggests that the 100M Baud links could go 35m in their 1249 cable. As the 1249 cable has a typical loss of half the Madison HPPI cable worst case loss, this would suggest the HPPI cable should have a worst case length limit for the links of 17m. Tests performed with the AT&T buffers by INMOS have shown that they appear to give good margins for transmission over 10m of the 28AWG Madison SCSI2 cable. Further tests have suggested that with good quality cable, it may be possible to go to 30m (see Appendices).