Re: still 65802 problem

[David.Empson@bbs.actrix.gen.nz]

In article <27JAN199320341384@zeus.tamu.edu> a0m2504@zeus.tamu.edu (AKITOMO MORITA) writes:
> ...
> Basically, the card uses C0F0-C0FF (when in slot 7) to select either
> a 1k ROM or one of 7k RAMs (8 total possibilities).  It uses just the
> last three bits of the address line to select one of these using a
> 74ls138 3-to-8 decoder.

I assume it also uses DEVSEL to determine when these addresses are valid.

> It also uses 3 of the 74ls75's 4 D-FF to hold the current selection.
> The problem is that the D-FF's are holding the wrong number.  I
> cannot trace the problem back toward the edge connector since the
> traces go under several IC's which I do not wish to take the time 
> to desolder and solder back on later.

> In case you're wondering, the WDC 65c802 is rated for 6Mhz, but this
> doesn't mean kaka since the computer will only run a 1.023 Mhz.

Well, it might mean something.  A 65802 with high clock speed will
have shorter address and data setup times.  The hold times appear to be
constant, independent of the rated processor speed, but is shorter than
that of a 2 MHz 6502/65C02.

If the Rana card is taking slightly too long to latch the address at
the end of the cycle, it could be picking up garbage after the 65802
floats the address bus.  It has a maximum of 10 nanoseconds from the
end of Phase 2 until the address will disappear (compared to 20 ns for
a 2 MHz 65C02).

To find out whether this is the problem, I'd need to know full details
of the circuitry used to latch the addresses.  You mentioned a 74LS138
and 74LS75, but I also need to know what is connected to the various
enable/clock pins of each, and whether there are any other chips
between the Apple II bus and the inputs (e.g. there might be a buffer
chip on the address bus).

If this is the cause of the problem, there isn't much you can do about
it, since all 65802s have the same address hold time.  If there are
any unused logic gates on the card, it might be possible to do a
slight modification to the card, so thatit latches the address earlier
(e.g. when phase 2 is high and Q3 falls, which is about half way
through the CPU phase of the cycle).

It may be possible to replace the chips on the card with ones that
have a faster response time.  74LSxx chips aren't particularly fast.
Unfortunately, the faster chips tend to impose a greater loading on
their inputs, which may cause problems in itself.

> What I want is any news of anybody experiencing trouble one they replaced a
> 65C02 with a 65C802.  I don't care about who manufactured them, I want to 
> know about the problems anyway.  OR if anybody familiar with timing diagrams
> could tell me if it IS a timing problem between a NCR 65C02 and a WDC 65C802.

I haven't dealt with 65802s, but am reasonably familiar with
timing-related problems, since I've done some card designs for the
Apple II, as well as aiding with the design of a couple of
single-board computers at work.  My main computer is a IIgs.

> Now, what I don't want are rumors, suggestions, or guesses.  No offence to
> anybody, but I want to know what the problem IS, not what it could be.

Well, as no-one has replied (in public) with anything definite, you're
stuck with my guesses.

> One more thing, the 65C802 upon power up starts in emulation mode (E=1) and
> I put it in emulation mode and tried the Hard Drive (just in case) and it 
> still does not work.

Yep.  I thought you might have done something like run a program that
accidentally left the CPU in native mode.


It is definitely a hardware problem, so ignore my earlier comments
about Rockwell 65C02s, interrupts, etc.
-- 
David Empson

Internet: David.Empson@bbs.actrix.gen.nz    EMPSON_D@kosmos.wcc.govt.nz
Snail mail: P.O. Box 27-103, Wellington, New Zealand