21 Jul
2010
21 Jul
'10
5:17 p.m.
Hey;
I was reading something the other day about being able to hand-clock a
Z80, that it was so stable (due to not using dynamic registers,
apparently) that with the appropriate debounce circuit you could literally
manually step it through instructions.
Is this as rare as it sounds?
Has anyone -tried- hand-clocking a Z80?
- JP
Hampton, Iowa
21 Jul
21 Jul
5:43 p.m.
On 7/21/10, JP Hindin <jplist2008 at kiwigeek.com> wrote:
Hey;
I was reading something the other day about being able to hand-clock a
Z80, that it was so stable (due to not using dynamic registers,
apparently) that with the appropriate debounce circuit you could literally
manually step it through instructions.
Is this as rare as it sounds?
I suppose that depends on your definition of rare. It's an uncommon
feature in modern processors, but probably not uncommon in the 8-bit
world.
A friend of mine, the older brother of the guy who taught me machine
code on an Elf long, long ago, was debugging his own Elf (RCA CDP1802)
and being a high school student in the 1970s, did not own much in the
way of "fancy" test gear. He certainly didn't own an oscilloscope.
So when he was having problems figuring out what was wrong with his
Elf, he built a debounced push button and connected it in place of the
1MHz crystal. The timing diagrams in the RCA manuals are very clear
and very complete. My friend stepped through the
8-clock-pulses-per-machine-cycle and watched his machine cycle through
its various states (conveniently brought out on a pair of lines) and
was able to debug his machine with an analog meter (used as a logic
probe, essentially).
Has anyone -tried- hand-clocking a Z80?
Me? No.
-ethan
6:57 p.m.
On 07/21/2010 01:43 PM, Ethan Dicks wrote:
On 7/21/10, JP Hindin<jplist2008 at
kiwigeek.com> wrote:
Me, yes. Z80, 1802, 80C85, 6100 and 6120 are static by design and clock
can be any
debounced signal.
The 8080, 8085 are dynamic, there is a minimum clock speed where
internal registers
(actually capacitors) may loose their charge. However I've run them
very slow
successfully.
Can't speak for others like 6800 and 6502 (65C02 is static).
In all cases running either very slow (say clock of 10,000hz or so for
dynamic parts)
or right down to stopped for static parts makes certain debug operations
much easier.
However, they can fool the unwary as loogic that works at very slow
speeds can fail
horribly when run at full speeds due to propagation delays. This was
especially true
for the 1802/5100/6120 when run near the maximum rated speeds as the CMOS
logic of the day was slow (very long propagation times).
However for hobby use the ability to slow or stop the cpu and observe
the levels using
a voltmeter, logic probe, or simple led and resistor, was both appealing
and made
access to CPU construction and debug approachable.
Allison
Hey;
I was reading something the other day about being able to hand-clock a
Z80, that it was so stable (due to not using dynamic registers,
apparently) that with the appropriate debounce circuit you could literally
manually step it through instructions.
Is this as rare as it sounds?
I suppose that depends on your definition of rare. It's an uncommon
feature in modern processors, but probably not uncommon in the 8-bit
world.
A friend of mine, the older brother of the guy who taught me machine
code on an Elf long, long ago, was debugging his own Elf (RCA CDP1802)
and being a high school student in the 1970s, did not own much in the
way of "fancy" test gear. He certainly didn't own an oscilloscope.
So when he was having problems figuring out what was wrong with his
Elf, he built a debounced push button and connected it in place of the
1MHz crystal. The timing diagrams in the RCA manuals are very clear
and very complete. My friend stepped through the
8-clock-pulses-per-machine-cycle and watched his machine cycle through
its various states (conveniently brought out on a pair of lines) and
was able to debug his machine with an analog meter (used as a logic
probe, essentially).
Has anyone -tried- hand-clocking a Z80?
Me? No.
-ethan
6:49 p.m.
On 7/21/10 1:17 PM, JP Hindin wrote:
I was reading something the other day about being able
to hand-clock a
Z80, that it was so stable (due to not using dynamic registers,
apparently) that with the appropriate debounce circuit you could literally
manually step it through instructions.
Is this as rare as it sounds?
Has anyone -tried- hand-clocking a Z80?
Umm yes, it's called single-stepping, and it's very common. The first
Z80 system I had that had support for that built-in was my IMSAI 8080.
I used single-stepping to debug my BIOS modifications in the mid-1980s.
And yes my IMSAI "8080" is actually a Z80 system...It has a CCS 2810
CPU board which is compatible with the IMSAI front panel circuitry. :)
A single-stepping circuit for a Z80 is pretty simple.
-Dave
--
Dave McGuire
Port Charlotte, FL
7:03 p.m.
Dave McGuire wrote:
On 7/21/10 1:17 PM, JP Hindin wrote:
Um, no. You don't understand how the front panel of the IMSAI
works. It uses memory wait to stop the processor. It doesn't
have anything to do with the CPU clock.
I was reading something the other day about being
able to hand-clock a
Z80, that it was so stable (due to not using dynamic registers,
apparently) that with the appropriate debounce circuit you could literally
manually step it through instructions.
Is this as rare as it sounds?
Has anyone -tried- hand-clocking a Z80?
Umm yes, it's called single-stepping, and it's very common. The first
Z80 system I had that had support for that built-in was my IMSAI 8080.
I used single-stepping to debug my BIOS modifications in the mid-1980s.
And yes my IMSAI "8080" is actually a Z80 system...It has a CCS 2810
CPU board which is compatible with the IMSAI front panel circuitry. :)
7:08 p.m.
On 7/21/10 3:03 PM, Bill Sudbrink wrote:
Um, no. You don't understand how the front panel of the IMSAI
works.
Actually yes I do. :) I've had to dig into it several times over the
years.
I was reading something the other day about being able
to hand-clock a
Z80, that it was so stable (due to not using dynamic registers,
apparently) that with the appropriate debounce circuit you could literally
manually step it through instructions.
Is this as rare as it sounds?
Has anyone -tried- hand-clocking a Z80?
Umm yes, it's called single-stepping, and it's very common. The first
Z80 system I had that had support for that built-in was my IMSAI 8080.
I used single-stepping to debug my BIOS modifications in the mid-1980s.
And yes my IMSAI "8080" is actually a Z80 system...It has a CCS 2810
CPU board which is compatible with the IMSAI front panel circuitry. :) It uses memory wait to stop the processor. It
doesn't
have anything to do with the CPU clock.
No, I didn't say it did, I was talking about single-stepping in
general in the context of the IMSAI system. It is indeed possible to
single-step a Z80 by manually pulsing the clock. One of my SBCs does
exactly that, based on a circuit in one of Steve Ciarcia's books.
-Dave
--
Dave McGuire
Port Charlotte, FL
7:51 p.m.
Dave McGuire wrote:
On 7/21/10 1:17 PM, JP Hindin wrote:
The only problem I could see is if the RAM memory is dynamic and one
made the mistake of single stepping the board master clock rather than
the Z-80 CPU clock. Static RAM wouldn't care...
John :-#)#
I was reading something the other day about being
able to hand-clock a
Z80, that it was so stable (due to not using dynamic registers,
apparently) that with the appropriate debounce circuit you could literally
manually step it through instructions.
Is this as rare as it sounds?
Has anyone -tried- hand-clocking a Z80?
Umm yes, it's called single-stepping, and it's very common. The first
Z80 system I had that had support for that built-in was my IMSAI 8080.
I used single-stepping to debug my BIOS modifications in the mid-1980s.
And yes my IMSAI "8080" is actually a Z80 system...It has a CCS 2810
CPU board which is compatible with the IMSAI front panel circuitry. :)
A single-stepping circuit for a Z80 is pretty simple.
-Dave
9:30 p.m.
On 7/21/10 5:24 PM, Eric Smith wrote:
Yes, I suppose you're right. It's much nicer (IMO) to do the full
trick by asserting /WAIT with /M1, a flip-flop, AND a debounced
pushbutton, though. Then you get true instruction stepping, rather than
cycle stepping.
-Dave
--
Dave McGuire
Port Charlotte, FL
A
single-stepping circuit for a Z80 is pretty simple.
True, but it is not quite as simple as hooking up a debounced pushbutton
to the clock, which was how I interpreted the OP's question. 
10:16 p.m.
On 21/07/2010 22:30, Dave McGuire wrote:
On 7/21/10 5:24 PM, Eric Smith wrote:
Yes, I suppose you're right. It's much nicer (IMO) to do the full
trick by asserting /WAIT with /M1, a flip-flop, AND a debounced
pushbutton, though. Then you get true instruction stepping, rather than
cycle stepping.
Depends on what you want. I've used a debounced switch, and sometimes
it's more useful to step through the various T-states within
instructions, rather than stepping an instruction at a time. Other
times, it's just extra tedious.
--
Pete Peter Turnbull
Network Manager
University of York
A
single-stepping circuit for a Z80 is pretty simple.
True, but it is not quite as
simple as hooking up a debounced pushbutton
to the clock, which was how I interpreted the OP's question.
7:41 p.m.
On 21/07/2010 18:17, JP Hindin wrote:
Hey;
I was reading something the other day about being able to hand-clock a
Z80, that it was so stable (due to not using dynamic registers,
apparently) that with the appropriate debounce circuit you could literally
manually step it through instructions.
Is this as rare as it sounds?
Has anyone -tried- hand-clocking a Z80?
Yes, using a pair of NAND gates forming an RS flip-flop to debounce a
switch. I used to have a 7400 taped to a microswitch with three leads
coming off it (+5V, 0V, and "clock") for that very reason.
I think the last time I used it was to demonstrate to a student that it
was his Z80-to-memory interface that was broken, not my RAM test :-)
Single-stepping can be a useful tool. I remember being disgusted that a
6502 can't do that (it's dynamic so there's a minimum clock speed which
is a lot faster than I can click a microswitch - somewhere in the kHz
range IIRC). Of course on most microprocessors you can devise a circuit
using WAIT states to achieve a similar (but not identical) effect; you
just can't stop at arbitrary parts of the machine cycle. With a Z80,
you can stop in any T-state you want.
--
Pete Peter Turnbull
Network Manager
University of York
8:12 p.m.
Hey;
I was reading something the other day about being able to hand-clock a
Z80, that it was so stable (due to not using dynamic registers,
apparently) that with the appropriate debounce circuit you could literally
manually step it through instructions.
Single stepping at the bus cycle level is actually quite common, and can
be done on many CPUs (even those with dynamic registers). Rather than
slow the clock down, you effectively include a lot of wait states so that
each bus cycle is held for as long as you want.
Is this as rare as it sounds?
Not really.
Has anyone -tried- hand-clocking a Z80?
I think I have. I've certaijnly hand-stepped a PDP11/45 at the T-state
level, buit that doesn't really count in that the machine was designed to
let you do that...
-tony
9:22 p.m.
JP Hindin wrote:
Has anyone -tried- hand-clocking a Z80?
Won't work on an original NMOS Z80 CPU, they were definitely dynamic.
You could get away with somewhat slower than the minimum spec (about
246.3 kHz for a 4MHz NMOS Z80), but not at finger speed.
Some of the later CMOS Z80 cores are static, some are not. In
particular, the original Z180 was CMOS but dynamic. It may be the case
that all CMOS Z80 CPUs (vs. Z180 and other derivatives) are static, I'm
not sure. Check the data sheet for the specific part number you're using.
If you hand-clock a static CPU, remember to debounce your button or
switch. Switch bounce can have glitches faster than the minimum clock
pulse width spec; I've seen them on an oscilliscope many times. Even
if you don't get fast glitches, you'll still get multiple pulses when
only one is desired.
Eric
10:20 p.m.
On 21/07/2010 22:22, Eric Smith wrote:
JP Hindin wrote:
Hmm.. I've definitely done it, and on quite old parts on occasion. I
thought they were NMOS but they may have been CMOS, I suppose. And I
recall being told in the late 70s or possibly around 1980/81 that Z80s
were static -- that would be about the time I started poking at the
hardware in that sort of detail.
Has anyone -tried- hand-clocking a Z80?
Won't work on an original NMOS Z80 CPU, they were definitely dynamic.
You could get away with somewhat slower than the minimum spec (about
246.3 kHz for a 4MHz NMOS Z80), but not at finger speed.
Some of the later CMOS Z80 cores are static, some are not. In
particular, the original Z180 was CMOS but dynamic. It may be the case
that all CMOS Z80 CPUs (vs. Z180 and other derivatives) are static, I'm
not sure. Check the data sheet for the specific part number you're using. If you hand-clock a static CPU, remember to debounce
your button or
switch. Switch bounce can have glitches faster than the minimum clock
pulse width spec; I've seen them on an oscilliscope many times. Even
if you don't get fast glitches, you'll still get multiple pulses when
only one is desired.
That is definitely true.
--
Pete Peter Turnbull
Network Manager
University of York
5275
days inactive
5275
days old
13 comments
9 participants
participants (9)
-
ajp166＠verizon.net -
ard＠p850ug1.demon.co.uk -
eric＠brouhaha.com -
ethan.dicks＠gmail.com -
jplist2008＠kiwigeek.com -
mcguire＠neurotica.com -
pete＠dunnington.plus.com -
pinball＠telus.net -
wh.sudbrink＠verizon.net