Interesting RK8E fault

Josh Dersch derschjo at gmail.com
Sat Dec 8 00:02:11 CST 2018


On Fri, Dec 7, 2018 at 9:58 PM Josh Dersch <derschjo at gmail.com> wrote:

> Hi all --
>
> Finally got all the parts together (and my act together) to actually get
> an RK05 lashed up to my PDP-8/e -- only took a decade or so :).  I fixed a
> few problems with the RK05 and it appears to be behaving very nicely.
>
> The RK8E controller is mostly working properly but fails interestingly
> when running the formatter, and during the exerciser -- on cylinder 128 and
> 192 and very infrequently on cylinder 64 it will get a cylinder mismatch
> when doing the seek.  When running the formatter during the verification
> pass, on cyls 64 and 128 if I retry the read it'll continue without issues,
> but it's never successful on a retry on cylinder 192.  I tried hooking it
> to the RK05 in my 11/40 and it exhibits the same behavior, so I'm guessing
> the drive isn't at fault.  And the error is consistent across packs (of
> which I have only two).
>
> Apart from that fault the drive and controller seem to work fine -- I
> wrote out an OS/8 pack with Adventure on it (or at least the first 191
> cylinders of it) and it works without issue.
>
> Reading the RK8E service docs and schematics, the cylinder address compare
> is done by reusing the CRC buffer, so I suspect the issue is in or around
> there -- the big problem is that debugging it is rather painful since that
> logic is in the middle board of a three board set, with jumper blocks on
> top -- so bringing it out on an extender isn't an option.  I'm curious if
> anyone's seen this issue or is so very familiar with the logic that the
> fault is obvious.
>
> I suspected the 7496 shift register at E14 which takes in the cylinder
> address to be compared w/the header on disk, and I went ahead and replaced
> it in the hopes that I'd get lucky, but no go.
>
> Anyone have any advice?
>
> Thanks,
> Josh
>
>
I'll add that during the format/verification the drive seeks properly (i.e.
it's not missing a step or overstepping), which I've confirmed by watching
the thing walk through the tracks with the cover off.

- Josh


More information about the cctalk mailing list