Pretty much plug and go. It's very much like a SCSI interface. Each device has a set of commands documented in its Functional Characteristics manual: read, write, get status, etc. You'd find this in the device's documentation (e.g. 3420) rather than the controller's (3803). The commands do follow a common pattern although there are some variations between devices. The Bus and Tag protocol communicates these commands from the channel (the interface in the CPU) to the device controller, and returns the results. There are several different transfer modes (byte at a time, burst, etc), different ways that the device can signal the CPU for attention, and various ways that either the channel, the controller or the device can say "hang on, I can't do that right now," so it's fairly complex. The logic frames for a typical channel interface built with 1960s vintage SLT are about the same size as the IBM 1130's CPU, for example. I suspect that data buffers (RAM) are a significant part of that. But from the programmer's standpoint, it's very simple. An IO command points to a "channel program" which is usually just a device command word and a data transfer address. The channel takes it from there, and an interrupt is issued when the operation is complete (usually). Channel to controller cabling is standardized, two thick cables. Controller to device cabling is device specific. Some devices have an integral controller. brian =-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=- _| _| _| Brian Knittel _| _| _| Quarterbyte Systems, Inc. _| _| _| Tel: 1-510-559-7930 _| _| _| Fax: 1-510-525-6889 _| _| _| Email: brian at quarterbyte.com _| _| _| http://www.quarterbyte.com