Retrocomputing Stack Exchange is a question and answer site for vintage-computer hobbyists interested in restoring, preserving, and using the classic computer and gaming systems of yesteryear. It only takes a minute to sign up.
Sign up to join this community
The Amiga computers would by default automatically detect a disk being inserted and respond by reading from it and displaying its icon. Polling the drive(s) for a new disk would produce a clicking sound every 1-2s.
If one had more than one drive or used the computer hard drive only with no disk(s) inserted, the drive checking sound would keep going.
Patches existed to stop this without loss of the auto-detection of inserted disks - how did these work?
The sound came from the fact that drives would only report on the presence of a disk if the heads were moved. Thus by default, the OS would move the heads back and forth.
Alternatively, the heads could be moved fully to one side and then asked to step further to that side (by default the heads would be centered to minimise seek time or read the central rootblock immediately).
Most drives would then refuse to move the heads, but report on disk presence anyway. (Some older drives would not, therefore if a noise was still heard after trying the patch, it was time to reboot and not try again!)
Since at least Kickstart 2.0, the OS has built-in code that allows for drive checking without making the click noise. To enable it, just set the TDPB_NOCLICK flag in the tdu_PubFlags of trackdisk.device's unit structure (of type struct TDU_PublicUnit) for the drive you want to silence.
Don't know about earlier. I guess trackdisk.device was patched with custom code, as it may be that some of the internal functions of the device appear - undocumented - in the LVO table.
Just a small addition from the technical department:
trackdisk.device (TDD) polls the disk status in regular intervalls. In order to detect a quick disk change in between two polls, the drive latches a DC status until the head is moved. So, TDD needs to move the head to update the status.
Moving the head outwards usually stops once track 0 is reached by gating the step pulse within the drive. However, some drives don't gate the outward step - so the head would hit the physical end stop endlessly, possibly disaligning the heads (remember the 1541?). Since C= sourced different types and there's no way for TDD to tell whether the head actually does try to step outwards from track 0, TDD steps outward until the end sensor triggers and then alternates between tracks 0 and 1. So, the clicking on an empty drive never stops. If you listen very closely (or open the drive) you can tell the two directions apart.
As already mentioned, the v33 trackdisk.device introduced the TDPB_NOCLICK flag that simply tells TDD to keep stepping outward, stopping the clicking - unless one of the non-gating drives is used. Then, a muffled click is heard and TDPB_NOCLICK should not be set for that drive.
The sound is caused by the drive head hitting the backstop. By default trackdisk.device periodically steps it inwards, causing it to hit the backstop every time.
From version 33.7 of trackdisk.device an alternative method is available that steps the head out, and then in again alternately. That way it doesn't hit anything. Other utilities that do this either make use of that feature or patch trackdisk.device to add it.
From the trackdisk.device source code:
* ;------ Step the drive. We will recalibrate it eventually
* ;------ anyway, so don't worry about where it is.
* ;------ if noclickstart flag is set, always step out.
* ;------ to save space, clear it then change it to set it.
* ;------ D3 is a flag to tell if I changed direction.
MOVEQ #0,d3 ; assume changed (d3=0)
BTST.B #TDPB_NOCLICK,TDU_PUBFLAGS(A3)
BEQ.S CC_Click
BCLR #CIAB_DSKDIREC,TDU_6522(A3) ; will get changed!
SNE d3 ; d3=1 if bit was 1
CC_Click:
BCHG #CIAB_DSKDIREC,TDU_6522(A3)
MOVE.B TDU_6522(A3),D0
MOVE.B D0,D1
BCLR #CIAB_DSKSTEP,D1
MOVE.B D1,(A2)
MOVE.B D0,(A2)
