RE: Re[2]: black widow pcb problem no play no picture clock working

From: Alex Yeckley <ayeckley_at_elektronforge.com>
Date: Wed Aug 10 2005 - 12:19:41 EDT

> Can you elaborate a little bit more on this for those of us
> still going up the learning curve? :)

I'll do what I can, but at some point we cross over into the "trade secret"
area.

> What is glue-logic?

There may be an IEEE-approved definition (Institue of Electrical and
Electronics Engineers), but my working definition is "discrete ICs". The
term "glue logic" seems to have developed at the same time as programmable
logic devices (which essentially combine the functions of discrete ICs under
one roof). The "glue" I think implies "that which connects everything
together". AND gates, OR gates, counters, latches etc. to me constitute
"glue logic". CPUs, FPGAs, etc. are not. I'm sure there are fine points to
be made that I'm glossing over, but hopefully you get the idea.

> When you break out the logic analyzer, can you give some
> general information about your troubleshooting procedure, what make and
model you
> prefer to use and what you're looking for?

I'm afraid I can't really discuss much about that, but maybe I can explain
in general terms. A logic analyzer is *sort of* like an oscilloscope,
except with a lot more channels all at once. An oscilloscope is typically
used to measure analog voltages that may or may not correspond with "logic
hi" or "logic low" voltages. An logic analyzer OTOH really specializes in
just the logic high and low states, and doesn't so much care about voltages
less than zero or higher than 5V. It's also usually working at faster
speeds (100Mhz minimum for anything built in the last 20 years, I'd
estimate) than oscilloscopes. Because it has so many channels (typically
multiples of 8, ranging from 32 to literally hundreds) it can be used to
watch the binary data on data busses, address busses, and commmunications
channels (on the circuit under test). Depending on how sophisticated the
analyzer you are using, you can configure it to present that data to you in
a variety of ways, depending on what you are looking for. For example, you
can set it up to recognize the CPU opcodes, so you might be looking at
assembler mnemonics rather than just hexadecimal or binary data. When you
use a logic probe or oscilloscope you're looking at only one bit.

Remember the movie "Porky's"? The shower scene, where everyone's trying to
look thru the peephole at once? That's a logic probe - everything has to go
thru a little peephole and you don't really get the big picture. A logic
analyzer is being in the shower with the girls. Maybe even having them soap
you up. I digress...

With a logic analyzer you're looking at all the data you care to see. As
far as what to look for, you typically (not always) need to know what the
software is trying to do. You might see data going out over the bus, but
you won't really know if it's the correct data unless you know what the
software is making the hardware try and do. Sometimes it'll just be
fetching instructions from the program ROMs, sometimes it'll be the results
of mathematical computations (in the case of the mathbox), sometimes it'll
be vector generator opcodes (like in this case), sometimes it'll be sound
commands going to a POKEY, etc. Just having a logic analyzer will probably
not solve anyone's hardware problems, unless they have spent some time
really understanding the software as well. We (OK - mostly me to be honest)
have spent a lot of time developing test software, test hardware and test
procedures to make various boards easier to diagnose, and that's really our
double-top-secret sauce.

Oh - you could use an emulator too, but that's a whole 'nother topic :-)

Alex

----
ayeckley@elektronforge.com
www.elektronforge.com
---------------------------------------------------------------------------
** Unsubscribe, subscribe, or view the archives at http://www.vectorlist.org
** Please direct other questions, comments, or problems to chris@westnet.com
Received on Wed Aug 10 12:19:47 2005

This archive was generated by hypermail 2.1.8 : Wed Aug 10 2005 - 13:50:02 EDT