17 Apr
2008
17 Apr
'08
12:33 a.m.
Date: Wed, 16 Apr 2008 15:49:42 -0700 (PDT)
From: "Eric Smith"
The most common error I've seen in Z80 designs is
to drive the clock input
directly from a TTL part. The minimum Vih for the Z80 clock input is
higher than the minimum Voh for a TTL part. The Mostek Z80 CPU manual
shows a circuit using a transistor as an active pullup to increase the
clock swing.
With the clock driven by standard TTL, Zilog specified a 330 ohm
pullup. I've seen lots of instances where the Z80/Z80A clock was
driven from an LSTTL or even a "canned" crystal oscillator with no
pullups at all. It might be that later steppings were more tolerant
of clock voltage.
If one had an 8284A in the hellbox, it would also make a good choice;
Voh(min) for the CLK output is 4v--but that might not be considered
to be "vintage". Some early S100 boards used an 8224 to generate the
Z80 clock.
Cheers,
Chuck
17 Apr
17 Apr
1:27 a.m.
Chuck wrote:
With the clock driven by standard TTL, Zilog specified
a 330 ohm
pullup.
For 2 MHz operation. At 4 MHz, it doesn't meet the spec. You can
see that on a scope; the voltage doesn't get to Vih until well into
the high phase, thus violating the clock rise time spec.
I've seen lots of instances where the Z80/Z80A
clock was
driven from an LSTTL or even a "canned" crystal oscillator with no
pullups at all.
I've seen them, and I've seen them act flaky. Even if it works
fine at room temperature and 5.0V supply on one day on with a
particular Z80 CPU, who knows whether it will work next month
at a different temperature with a less well adjusted power supply
and a different CPU chip. A good engineer wouldn't use a TTL
output directly, but a lazy one might.
It might be that later steppings were more tolerant
of clock voltage.
Maybe. More likely of the CMOS parts than the original NMOS. I doubt
they changed the masks of the NMOS part very much over the product life.
I don't think they ever did a die shrink, since with NMOS that required
almost completely redoing the layout. The CMOS parts may well have had
die shrinks that would affect the electrical parameters, as CMOS
would scale over a fairly wide range.
If one had an 8284A in the hellbox, it would also make
a good choice;
Voh(min) for the CLK output is 4v--but that might not be considered
to be "vintage". Some early S100 boards used an 8224 to generate the
Z80 clock.
If you're not concerned with it being "vintage", just use a
74HC(T), 74AC(T), or 74FC(T) part to drive it.
Eric
6094
days inactive
6094
days old
1 comments
2 participants
participants (2)
-
cclist＠sydex.com -
eric＠brouhaha.com