On Nov 30, 2:07=A0am, m...{at}deichjugend.de wrote:
> On Nov 29, 11:23=A0pm, Tarkin  wrote:
>
>
>
> > The CPU mine came with is:
> > Goldstar Z8400A PS
>
> > Which I replaced with:
> > Zilog Z84C0008PEC
>
> > The system is far, far, _far_ more stable now.
> > I don't know if it is the fact that it's a CMOS
> > part, or if that the chip is faster, or both.
> > In any event, it's a cheap fix with immediate,
> > practical results.
>
> Of course the old chip can have a defect (or some other glue
> chip on the card) and I'm glad you got rid of your problem.
>
> However, I'm thinking of two explanations for the old chip
> to fail:
> - oxydized pins that just needed better contact

The pins were pretty oxidized- noticeable blue-white tinge
on the pins, and the chip was a dickens to remove...

> =A0 Did you try putting the old CPU back in? Perhaps several
> =A0 extractions/insertions are necessary?

I buffed the Goldstar chip with plumber's cloth, wiped away the
dust, reinserted. System seemed more stable, but lock up after
any strenuous activity. See below for details / definition of
'strenuous activity'

> - A CMOS part needs less power than a HMOS one. Perhaps you
> =A0 already maxed out your power supply (lots of card?) or the
> =A0 power supply is on its way out?

I don't think so. My unit is a stock Laser128, base model, that
operates flawlessly (once I calibrated the drive, that is) for
extended
periods fo time under Apple Basic or ProDos 8 or DOS 3.3. The
power draw _may_ be affecting the card, though- it's a clone of
the original MS-SC design, with a baker's dozen logic IC's. A
higher current draw _may_ be causing higher transients during
I/O-intensive tasks.

I loosened and re-inserted the Goldstar chip a couple of times,
still fails my stress test. On a lark, I put in another chip, a
SGS Z8400BB1, which should be a 6MHz HMOS(?) part.
It performs flawlessly as well.

Stress test / strenuous activity:
Boot with 'P' key held down- activates L128 config screen
Select 7E2 9600 baud w/ echo + LF + no zap
CTL+RST to boot into MS-CPM 2.23
execute RAMDRIVE.COM (copied from CP/AM 4.0b)
(RAMDRIVE.COM make a default C: drive)
PIP C:=3DA:PIP.COM
PIP C:=3DA:LOAD.COM
C: (switches to C: drive for execution)
PIP C:CAMEL80.HEX=3DRDR:
(download B. Rodriguez's Camelforth in hex from a hypertem
session on a Win98SE box- using Transfer->Send Text File)
(~14 min of furious blinkylight on the SC. Good indicator things
are going well. if the light goes solid red for more than 5 secs,
the system is of in lala land...)
LOAD CAMEL80
(makes CAMEL80.COM)
CAMEL80
(screen output quoted below)
-----------------------------------------------------------------
Z80 CamelForth v1.01  25 Jan 1995
: HELLO CR CR ." HELLO, WORLD!!" CR ;

ok HELLO


HELLO, WORLD!!

ok BYE

C>A:
A:
A>DIR
-----------------------------------------------------------------

The Goldstar part never finishes the download. It's
lasted as long 10 minutes. The the Zilog and SGS
parts just take whatever I throw at it.

> They do not need apply, though, just came to my mind because
> I had similar problems once.
>

Yeah, it's bog-standard system diagnosis- the chip might not
have been bad. I just had a hunch and prejudice- I've seen
Goldstar appliances, and some 74XXX logic chips, but never
a Goldstar Z80 cpu. I didn't trust it once I got to thinking about
it....

> bye
> Marcus

Thanks for the suggestion! I now know for certain that it's
the chip, and that the chip is bad.

TTFN,
  Tarkin
--- SBBSecho 2.12-Win32