Philipp Hachtmann wrote:
> The RK8E does not pass the Diskless Diagnostic
(DHRKAE) test in my
Ugly. RK8E is too funny...
Yes, I totally agree that the RK8E is a weird beast. It is definitely
not the easiest thing to troubleshoot.
I once had very similar problems with a (good!) RK8E. I isolated
ICs on the boards which behaved randomly wrong.
But: Nothing was bad.
Check your PSU's 5V!
My quite loaded lab8/e back then passed *all* tests with 5V actually
only 4.3V. All tests mean: All CPU maindecs, hours of
peripheral tests, TD8E tests etc. But *not* that awful
the RK05s did not work properly, of course.
I checked the +5V supply today, both with a DVM and an Oscilloscope to
look for ripple or noise.
The +5 was running at +4.776 V, and there was virtually no ripple or
noise on the line. I checked it on all of the +5V bus lines and it was
So, is +4.776V "good enough", or do I need to tweak the +5 supply up a
This was with everything plugged into the Omnibus:
CPU, EAE, PC8E, RX8E, 2x4K Core Sets, 16K Solid State Board, RK8E,
2xM8650's (Serial), MemExt/Timeshare, Front Panel and Bus Loads
The diskless test is something that frightens me every time I start
The test itself is a crazy piece of code.
I've actually typed the whole
thing in from the source listing PDF, so that I can "instrument" the
code by putting in various typeouts and such as the tests go along.
I've also run the RK8E Drive Control & Instruction Test (DHRKBG). It
runs great until it does the part where it does "WRITE ALL" commands to
format the pack, and then does "READ" commands to verify, and it fails
partway through that, just as the RKLFMT (formatter) fails.
The reason it gets part way through the RKLFMT verify and the
verification pass in the RK8E Drive Control & Instruction test is that
the write (format) part puts the upper bit of the disk address in word
0, and the lower 12-bits of the disk address in word 1 of each sector.
Since the first word of each sector is '0000' for most of the pack, it
verifies OK there. It's when it changes to '0001' that is when the
I didn't have much more time today other than to do the power supply
testing, and, just on a hunch, I moved the RK8E boardset to a couple of
different locations in the Omnibus, and it didn't make any difference.
Of course, if anyone has any other ideas, I'm all ears.