Thursday, October 6, 2016

Unsuccessful day testing with loaner Diablo drive


I set up the Diablo disk drive loaned to me by Al Kossow, with a disk cartridge containing an Alto image. My first steps were to carefully clean the disk drive heads and the surface of the disk cartridge. Next was to cycle the drive up but stop before it loaded heads, repeating this quite a few times to blow as much dust as possible out of the drive and cartridge.

I set up my disk tool, the logic analyzer, and the oscilloscope so that I could validate the behavior of the tool with a real drive. It is critical that I have well formed pulses arriving from the disk drive in order to detect them reliably. My tool should tell me if I have any checksum errors when reading sectors, which will be an instant way to validate whether signal integrity is sufficient. 

There is a smell from the rear of the Diablo that I can only describe as overheating wire or resistors. It is coming from the rotary actuator area or the flat motherboard that runs along the rear face of the drive. I will keep my eye on that as I test. 

With everything hooked up, I fired up the drive, loaded the heads and attempted to use it. It remained inert, not seeking or reading at all. First thing to check is whether the unit is wired to be selected as Unit 1 - but it was.

Next thing to check is whether I am seeing the File Ready status coming back from the drive - if that isn't happening, I have a serious issue with the interface. It did light up, but the problem seems to be in the other direction, recognition of commands from the interface.

I did quite a bit of diagnostic work before I discovered that my 3.3V input to the level shifter board had a bad solder joint. With that corrected, I could again see the index and sector mark pulses. Data bits are streaming in as well.

Still, it does NOT obey seek instructions. Time to shift to checking the validity of the 'strobe' signal which is what kicks off the actual activity. I am concerned because I see the command, whether the seek command or reading a sector, sitting in the command register with no sign that the FSM completed. 

My scope shows that the strobe signal is activated by the fpga but it does not get to the strobe reception circuits inside the Diablo. I am tracing wiring and testing to figure out where this is failing. 

It turns out that the output drops from 5V only down to 1/9V, in spite of the input to the level shifter dropping to virtually 0V. I have to do more investigating to sort out why this is occuring.

No comments:

Post a Comment