RE: CBM900 hard disk timeout

Hello!

W dniu 2014-08-22 17:35, Martin Hoffmann-Vetter napisał(a):

> So it's possible, the drive can't step to a specified track. This is
> the difference between ST412 and ST506. ST412 drives used buffered
> seek. All step pulses are counted and the drive will moved in one pass
> to the specified track. ST506 drives seek with every step pulse. Is it
> possible to switch the WD1003 or the hard drive between buffered and
> unbuffered seek? In buffered seek you need a SEEK_COMPLETE signal on
> the control port. If the drive don't send it, it's possible the
> controller detect a timeout! Can you step the drive with pulses on the
> control port manually? You need a direction signal (low = inward, high
> = outward - toward track 0), too. After the drive is powered on the
> READY and SEEK_COMPLETE signals must be active.

Well, I played with the disk, it was an interesting exercise :-) The 
disk appears to be working well:

* The LED is ON when DRIVE_SELECT_1 signal is asserted (it must stay 
asserted for the rest of the test).

* The READY and SEEK_COMPLETE signals come active when the drive 
finishes initialization.

* The TRACK0 signal is active, so you were right: the drive seeks the 
head to track 0 during initialization.

* The stepper motor moves the head when I pulse the STEP signal. The 
DIRECTION_IN signal must be asserted since the head is at track 0 and 
must move inward.

* The TRACK0 signal goes to false when the head moves in.

* The SEEK_COMPLETE goes to false when the head moves, and then returns 
to true.

All this means to me that the stepper mechanism is working.

The only problem I found is that the drive does not send any pulses to 
the INDEX line. I don't know why it's that, and whether it is enough to 
confuse the controler into shooting timeout errors. The drive itself is 
able to detect index condition, because when there are no disk 
revolutions, it flashes an error code.

Maybe some simple timing generator could be used to fake this line.

Regards,
Michau.





       Message was sent through the cbm-hackers mailing list