The mic's CD-ROM no doubt works best on Windows. Looking at the transactions Windows does to the device and what causes it to wake up is probably the best way to reverse engineer its logic.
Having a removable device with no "medium" inserted is a reasonable way to have the host have access to the "medium" at times you feel are convenient. The logic to make the device show up are device dependent. I've worked with devices which successfully use this trick to share storage with the host. It could be selectively set so that the medium was present so the host could sync files to it, or selectively "ejected" so that the device could examine those files for its own use.
Usually plugging the device into the host is enough to make the medium (logically) show up. There may be some particular SCSI command that windows sends that causes this to happen. The command could be as innocuous as test Unit Ready. Hence seeing what happens on Windows.
A host will usually send Test Unit Ready commands to a device periodically, and when one succeeds it will then access the medium. This works for real devices with real media and virtual devices with virtual media. I've never known any device use the Unit Attention mechanism, if there is one.
Same answer for the printer. Send Test Unit Readys to it periodically (about 1/sec is good) until something happens.
This really isn't a question about composite devices, but about MSC devices.
