7 Oct
2005
7 Oct
'05
12:29 p.m.
From: "Hans Franke" <Hans.Franke at
siemens.com>
---snip---
First thing would be
to connect a logic analyzer to see if the CPU is still running
a programm in ROM or not.
Hi
What is it with logic analyzers. Why not just an
oscilloscope. In most cases, one can be farther along
with an 'oscope in finding what is wrong by the
time one can get an analyzer connected and setup.
I've only had one time that I ever needed an analyzer
and even that time, it didn't work well because
of the complexity of the problem ( design not failure ).
I'll admit that I've often thought of making one
of those address compare circuits to trigger the 'scope
but by the time I'd get serious, I'd found the problem.
Am I alone here or does everyone else think that an
analyzer is the ultimate tool?
Dwight
7 Oct
7 Oct
12:56 p.m.
On Fri, 7 Oct 2005, Dwight K. Elvey wrote:
I have one but rarely use it:
1. With modern circuits it is nearly impossible to connect.
2. Usually by spending a little more time thinking and doing simple
measurements (address and data toggling, clocks running, reset doing the
right thing) I can avoid using the logic analyser at all.
3. for repairing older circuits, often the problem will
maniifest itself as a marginal level, the LA may miss this completely
or give hard to interpret results.
4. Just having a simple oscillator to toggle the reset line lets me trace the
first few CPU instructions with nothing more than a scope (and a piece of
paper).
5. I do have some pre-configured ROM socket --> logic analyser adapters so I
can trace startup code when bring up a new board.
Basically I only use it as a last resort when nothing else works...
Peter Wallace
From:
"Hans Franke" <Hans.Franke at siemens.com>
---snip---
First thing would be
to connect a logic analyzer to see if the CPU is still running
a programm in ROM or not.
Hi
What is it with logic analyzers. Why not just an
oscilloscope. In most cases, one can be farther along
with an 'oscope in finding what is wrong by the
time one can get an analyzer connected and setup.
I've only had one time that I ever needed an analyzer
and even that time, it didn't work well because
of the complexity of the problem ( design not failure ).
I'll admit that I've often thought of making one
of those address compare circuits to trigger the 'scope
but by the time I'd get serious, I'd found the problem.
Am I alone here or does everyone else think that an
analyzer is the ultimate tool?
Dwight
3:47 p.m.
--- "Dwight K. Elvey" <dwight.elvey at amd.com> wrote:
Well I do, but it sounds like everyone is complaining
about the LA setup time (bringing it out of storage).
Mine is a permanent fixture on my bench, with loads of
various sized test clips (and a full set of dip test
clips), in 30 seconds I can have it powered up and
tracing on 8 timing signals, little longer if I want
16 bits of address/data and a clock, can't remember
the last time I used a scope.
But I suppose it like comparing a table saw with a
band saw, what they can do overlaps, and one does
certain things much better then the other, but you use
what you are more familar with and have available and
solve the problem whatever its limitations.
__________________________________
Yahoo! Mail - PC Magazine Editors' Choice 2005
http://mail.yahoo.com
From:
"Hans Franke" <Hans.Franke at siemens.com>
---snip---
First thing would be
to connect a logic analyzer to see if the CPU is
still running
a programm in ROM or not.
Hi
What is it with logic analyzers. Why not just an
oscilloscope. In most cases, one can be farther
along
with an 'oscope in finding what is wrong by the
time one can get an analyzer connected and setup.
I've only had one time that I ever needed an
analyzer
and even that time, it didn't work well because
of the complexity of the problem ( design not
failure ).
I'll admit that I've often thought of making one
of those address compare circuits to trigger the
'scope
but by the time I'd get serious, I'd found the
problem.
Am I alone here or does everyone else think that an
analyzer is the ultimate tool?
Dwight
4:01 p.m.
First thing
would be
to connect a logic analyzer to see if the CPU is still running
a programm in ROM or not.
Hi
What is it with logic analyzers. Why not just an
oscilloscope. In most cases, one can be farther along
with an 'oscope in finding what is wrong by the
time one can get an analyzer connected and setup.
I've only had one time that I ever needed an analyzer
and even that time, it didn't work well because
of the complexity of the problem ( design not failure ).
I'll admit that I've often thought of making one
of those address compare circuits to trigger the 'scope
but by the time I'd get serious, I'd found the problem.
Am I alone here or does everyone else think that an
analyzer is the ultimate tool? 
10:36 p.m.
On Fri, 7 Oct 2005 21:01:51 +0100 (BST)
ard at p850ug1.demon.co.uk (Tony Duell) wrote:
The logic analyser is not the ultimate faultfinding device. In fact
the ultimate faultfinding device isn't made by Tektronix, Agilent or
Lecroy, it doesn't come from RS components, Farnell, or Digikey. As
I've said many times before, it's called a brain :-).
Back in the day, techs didn't get to use Logic Analyzers much in any
case. They were targeted as development tools, not for troubleshooting,
and only the high mucky-muck engineers had them. They were too
expensive (always in the five figures, usually the mid five figures) for
mere 'troubleshooting' and really not that well suited to such pursuits.
A Logic Analyzer is for things like developing a Data General mini.
First thing would be
to connect a logic analyzer to see if the CPU is still running
a programm in ROM or not.
Hi
What is it with logic analyzers. Why not just an
oscilloscope. In most cases, one can be farther along
with an 'oscope in finding what is wrong by the
time one can get an analyzer connected and setup.
I've only had one time that I ever needed an analyzer
and even that time, it didn't work well because
of the complexity of the problem ( design not failure ).
I'll admit that I've often thought of making one
of those address compare circuits to trigger the 'scope
but by the time I'd get serious, I'd found the problem.
Am I alone here or does everyone else think that an
analyzer is the ultimate tool?
8 Oct
8 Oct
12:34 a.m.
--- Scott Stevens <chenmel at earthlink.net> wrote:
They were too
expensive (always in the five figures, usually the
mid five figures)
those same analyzers are about $50 on ebay now
__________________________________
Yahoo! Mail - PC Magazine Editors' Choice 2005
http://mail.yahoo.com
9:49 a.m.
On Fri, 7 Oct 2005 21:34:38 -0700 (PDT)
steve <gkicomputers at yahoo.com> wrote:
--- Scott Stevens <chenmel at earthlink.net> wrote:
Yes. $50 for the analyzer, which is too high a price without the pods,
then you spend 3 years and ultimately another $300 finding the pods.
They were too
expensive (always in the five figures, usually the
mid five figures)
those same analyzers are about $50 on ebay now
11:19 a.m.
From: "Scott Stevens" <chenmel at earthlink.net>
steve <gkicomputers at yahoo.com> wrote:
I got mine on eBay. It wasn't quite as easy as Steve implied, nor as
hard as Scott implied. For about $200 or so, I got a pair of logic
analyzers, a 1630g and a 1630d. Each has a full set of pods, but an
incomplete set of grabber-adapters, so between them I have more than
one complete set of pods and adapters. Plus, I can decide whether I want
the better asynch stuff on the 1630d, or the higher number of inputs on
the 1630g :-). (Or I could put one of them back out on eBay and recoup
some of my $$. That would contribute to the problem Scott is talking
about, though.) It took me perhaps a month of watching auctions (and
getting outbid) before I got what I needed.
Scott is correct that there is a lot of stuff on eBay that isn't worth
much, because the hard-to-find bits have been removed.
Oh, and I have a lead set here that doesn't fit the 1630 series. Anyone
need a set of Gould grabber adapter clips :-)?
Vince
those same analyzers are about $50 on ebay now
Yes. $50 for the analyzer, which is too high a price without the pods,
then you spend 3 years and ultimately another $300 finding the pods. 
10:32 p.m.
Scott Stevens declared on Saturday 08 October 2005 08:49 am:
On Fri, 7 Oct 2005 21:34:38 -0700 (PDT)
steve <gkicomputers at yahoo.com> wrote:
A couple years ago, Purdue's salvage op was selling complete (with pods)
10MHz HP logic analyzers for $50 each. I ended up asking them how much
they wanted for one some day late in December, and they said "Merry
Christmas". Of course, I do buy quite about of stuff from them, mostly
things that they'd tear apart as scrap. And then there was the 1989
Chevy boxvan I bought from them[1] to haul things back from there to my
house. :)
Another reason why it pays to befriend your local scrapper...
[1] One of the few things I have paid a large amount from them (at
auction), but probably a lot less than it was really worth.
Pat
--
Purdue University ITAP/RCAC --- http://www.rcac.purdue.edu/
The Computer Refuge --- http://computer-refuge.org
--- Scott Stevens <chenmel at
earthlink.net> wrote:
Yes. $50 for the analyzer, which is too high a price without the
pods, then you spend 3 years and ultimately another $300 finding the
pods. They were too
expensive (always in the five figures, usually the
mid five figures)
those same analyzers are about $50 on ebay now 
4:15 a.m.
On Fri, 7 Oct 2005, Dwight K. Elvey wrote:
> > What is it with logic analyzers. Why not
just an
> > oscilloscope. In most cases, one can be farther along
> > with an 'oscope in finding what is wrong by the
> > time one can get an analyzer connected and setup.
> > Am I alone here or does everyone else think
that an
> > analyzer is the ultimate tool?
To be perhaps overly general but a distinction I think worth making:
Logic analysers are intended to analyse logic, not electronics. The systems we
deal with are logic systems implemented in electronics, when a functioning system
develops problems it's the electronics which have failed (granted that some problems
manifest themselves in such a way as be to observable as pure logic failures).
I like an o'scope for analysis, it shows you analog characteristics of
signals .. those characteristics still matter.
7 Oct
7 Oct
4:40 p.m.
Am 7 Oct 2005 9:29 meinte Dwight K. Elvey:
From:
"Hans Franke" <Hans.Franke at siemens.com>
---snip---
> First thing would be
>to connect a logic analyzer to see if the CPU is still running
>a programm in ROM or not. Hi
What is it with logic analyzers. Why not just an
oscilloscope. In most cases, one can be farther along
with an 'oscope in finding what is wrong by the
time one can get an analyzer connected and setup.
I've only had one time that I ever needed an analyzer
and even that time, it didn't work well because
of the complexity of the problem ( design not failure ).
I'll admit that I've often thought of making
one
of those address compare circuits to trigger the 'scope
but by the time I'd get serious, I'd found the problem.
Am I alone here or does everyone else think that an
analyzer is the ultimate tool?
It is after all a nice big hammer. Why waste time with
other measures. Shure, when I originaly had my first
KIM (does this sound like a book title or what?), all
I had was a logic probe. I had an oscar at work, but
somehow my superiors didn't feel comfortable with my
intention to take it home. And yes, somehow I solved
all my problems.
And again yes, one could do it again like back then,
but why? You already pointed out one of the real traps
in ding it with just a hairpin and your brain: it takes
time. Quite a lot of time.
Now for the KIM, I'd just add an edge connector and clip
on the probes, key in the important signal names, and
at maximum 5 Minutes later I can go ahead and interpret
what I'm seeing. construct some trap conditions and usualy
the problem gets visible quite fast.
Of course, as with every strategy to find a problem, I
need to have an idea what I'm doing and what I'm looking
for. I mean, checking simple things like power, certain
outputs and clock signals are things to be done before.
Again, without an idea what to do, a bigger hammer is
only good for more pain when he hits your thump :)
And there ar bigger hammers available - the next step
after a LA are some ICE box, which in case of a dead
KIM with an already socketed CPU might be my other
coice, since it's standing right on the shelf ... and
a programm run listing plus logic signal diagrams is
a nice thing.
In both cases, the more in information you get, the more
you need to know what you're doing ... did I mention this
already? To me it's the most important part. You need to
know the basics, you need to know what your're doing.
Otherwise all reading are senseless.
A great tool doesn't make a good mechanic, and a good
mechanic can still do a good job with poor tools. Still
if you combine knowledge with good tools, you get a good
result in less time with less effort.
Comeing back to logic analyzers, I compare them on my
own scale to modern programming environments. I mean,
25 years ago, I did read my flag lists (already on screen)
and worked thru as many lines as possible before restarting
the assembler. Today I just go for the first few lines,
and when I see that there are several flags connected
to the same mistake, I don't scan any further for other
errors, but restart the assembler and look into the new
flag list (if there's one). The new pass is shorter than
the time I need to reload the output files.
Now, for run time errors, I did sit down (in the old
time) and tried to figure out what happens. Today, I
do the same, but if I can't find a result fast, I add
some debug output and see whats happening inside...
Isn't that the same as with a logic analyzer?
Interactive development.
Not as ivory tower like as if you do everythin in your
head, but usualy faster.
Oh, and did I mention that you need to know what you're
doing? I've seen more than enough 'programmers' adding
debug lines and debug lines and searching for hours,
just because they didn't have a glimps what the task
was about? Always the same.
Now, since I already looked onto as a whimp by most on
this list, I can even top this:
I do hardware development the same way. If some CPLD
doesn'T work as I want, I just add debug output and
attatch a logic analyzer:)
Have a nice weekend
Hans
--
VCF Europa 7.0 am 29/30.April und 01.Mai 2006 in Muenchen
http://www.vcfe.org/
8:24 p.m.
On Fri, 7 Oct 2005 09:29:36 -0700 (PDT)
"Dwight K. Elvey" <dwight.elvey at amd.com> wrote:
Or you can use the "Poor Man's Logic Analyzer" which is a pair of D/A
converters hooked to address/data lines and plugged into the appropriate
address lines. You hook the outputs of the DACs into an oscilloscope
configured as an X/Y display and adjust gain so that it makes a grid on
the screen. It shows you dynamically where the software is branching in
the memory map. If you use this method regularly, it can be a
'signature analyzer' of sorts. You'll get to know kind of dynamic
display to expect and/or you can even write short diagnostic programs to
'draw' specific patterns on the display. This sort of 'analyzer'
definitely shows wether a processor is in a 'lively' mode and running
around on the bus, or is stuck in a loop somewhere. If you want to be
fancy, you can latch the bus fed to the DAC with your chosen enable
signals.
From:
"Hans Franke" <Hans.Franke at siemens.com>
---snip---
First thing would be
to connect a logic analyzer to see if the CPU is still running
a programm in ROM or not.
Hi
What is it with logic analyzers. Why not just an
oscilloscope. In most cases, one can be farther along
with an 'oscope in finding what is wrong by the
time one can get an analyzer connected and setup.
I've only had one time that I ever needed an analyzer
and even that time, it didn't work well because
of the complexity of the problem ( design not failure ).
I'll admit that I've often thought of making one
of those address compare circuits to trigger the 'scope
but by the time I'd get serious, I'd found the problem.
Am I alone here or does everyone else think that an
analyzer is the ultimate tool?
Dwight
8 Oct
8 Oct
4:14 a.m.
Scott Stevens wrote:
Or you can use the "Poor Man's Logic
Analyzer" which is a pair of D/A
converters hooked to address/data lines and plugged into the appropriate
address lines. You hook the outputs of the DACs into an oscilloscope
configured as an X/Y display and adjust gain so that it makes a grid on
the screen. It shows you dynamically where the software is branching in
the memory map. If you use this method regularly, it can be a
'signature analyzer' of sorts. You'll get to know kind of dynamic
display to expect and/or you can even write short diagnostic programs to
'draw' specific patterns on the display.
... Such displays originated back in the late-'40s/early-'50s with processors
using CRT-storage-tube memory systems. The digital-address-to-analog-deflection
circuits were already there for the storage-tubes, they simply had to be brought
out to a monitor tube. I recall reading the SWAC had such a display on one of the
two CRTs at the console. Whirlwind might/could have had as well.
6866
days inactive
6868
days old
12 comments
9 participants
participants (9)
-
ard＠p850ug1.demon.co.uk -
chenmel＠earthlink.net -
dwight.elvey＠amd.com -
gkicomputers＠yahoo.com -
Hans.Franke＠siemens.com -
hilpert＠cs.ubc.ca -
pat＠computer-refuge.org -
pcw＠mesanet.com -
vrs＠msn.com