I had good success with a variant of #2 on a very large digital board that had
a short between the +5V and ground -- both of which were planes due to the
high current and multiple loads.

I energized the +5V bus with a supply set for 1 V and current limited to 10 A.
Then went around the board measuring between adjacent +5 and ground points
[like across a bypass cap or an IC]. Where I found the lowest drop was the
short.

Dr. G.



In article , "Bob Shuman"
wrote:
When I worked as a circuit test engineer that produced fairly complex
multi-layer PCBs many years ago, there were three primary methods we used to
find shorts across the power rails.

1. Use a calibrated micro-ohm meter, fixing one lead to the PCB ground and
taking various resistance readings by moving the other lead from the board
edge connector along the traces to locate the point where the meter provided
the lowest resistance reading. You could then fix the lead to that point
and then take readings by moving the other (previously fixed) lead along the
grounds to again find the minimal reading. This usually led you to the area
of the board where the short or defective component was located. If you had
sensitive enough equipment and some good test leads, this procedure usually
worked pretty well when there was an actual hard short.

2. Apply a current limited, voltage limited (lower than the nominal design
voltage, for instance 5V DC) power source across the power rail at the PCB
edge connector. Start with a fairly low current limit and increase this as
needed, but keeping the current reasonable (you don't want a defective
component to explode - been there done that). We then used either a thermal
sensitive plastic sheet (material it contained was like the stuff used in
"mood rings" from the 1970's) or an infra-red camera to find the "hot spot"
on the PCB. This technique had the advantage of working for soft shorts,
such as were created by defective components (transistors, other
semiconductors, ICs, capacitors, etc.) or even resistive type shorts that
were created by contamination from foreign materials (conductive growths,
moisture, salt water contamination, etc.) We were even able to "see"
internal shorts on 8 layer circuit boards using the camera.

3. Visual observation and path tracing coupled with selective unsoldering of
legs of suspect components (assumes through hole mounting, not surface
mounting) and use of isolation rings (small plastic rings that slid over the
unsoldered leg of an IC isolating it from the multi-layer solder
pad/circuit. This technique was generally used as a last resort and usually
in combination with procedures 1 and 2 above, prior to scrapping
"difficult", but costly product that had been diagnosed with a power rail
short.

The above techniques are what I used about 20+ years back when I had
engineering responsibility in a large electronics factory. I'd imagine that
there are likely better approaches today due to improvements in technology
so would be interested to hear what others recommend.

Good luck!

Bob

"Henry Kolesnik" wrote in message
...
Sometime ago I think I recall someone posted or wrote an article on a neat
way to isolate a shorted component on a pcb using common test eqpt but I
can't recall the methodology. I'm trying to find a shorted component on
a
Wavetek 188-S-1257 signal generator. The B+ line reads about 0.4 ohms and
I'm not having much luck disconnecting componets. I don't have a
schematic
and my eyes ain't what they used to be for tracing and I want to minimize
the unsoldering. Does anyone recall the article or have a good way?
tnx
hank wd5jfr

"Henry Kolesnik" wrote in message
...
Right now it looks like 2 might be the problem marked AVX 103 and AVX 849.
I'm guessing that they're tantalums but not sure. The Wavetake was mfg in
1989. I need help on the values and voltages. There's too many to just
start changing.
tnx
hank


The tantalums in WaveTeks were the weak links. I was continually chasing
shorted tantalum bypasses when I owned two WaveTek 3000 signal
generators.

Pete

Wild Bill wrote:

I haven't tried one, but floppy disk drive heads are sensitive. I was
reminded of a circuit in a magazine (many years ago) that used a floppy head
for a pickup.

Some questions. How are you going to position it against the trace
accurately? What about double sided, or multilayer boards?

The problems with using AC to find shorts is that you get false peaks
and dips from the inductance of the traces, and the characteristics of
the components. Another problem is that some parts self destruct with
only a small reverse voltage so you can damage a lot of parts while
troubleshooting the board.

I used the DC voltage drop & sensitive digital meter method on boards
that people couldn't fix with AC, then had to find the parts they
damaged. The whole idea is to find and fix a problem quickly, and
reliably.

--
We now return you to our normally scheduled programming.

Take a look at this little cutie! ;-)
http://home.earthlink.net/~mike.terrell/photos.html

Michael A. Terrell
Central Florida

I found the bad tant cap, it was a 22 ufd, 20 volt, not any AVX but a
yellowish tan blob with a big L. I replaced it with an electrolytic. I
found it using my Fluke 87 that measures to the nearest tenth of an ohm and
only the suspect cap would flicker between 0.3 and 0.2 ohms, all the rest
were 0.4 or 0.3. Now I'm kicking myself for not buying an HP meter that
could read 1/100ths maybe 1/1000ths because I could see no use for it. Now
I can see a use and one is on my list but nevertheless my Fluke saved a lot
of desoldering. The Wavetek still doesn't work as something else is
keeping the voltage low and I think it might be a regulator. Now I wish I
could find a schematic for the Wavetek 188-S-1257, as it would keep me from
wasting so much time.
Thanks for all the tips. guys.
73
hank wd5jfr
"Henry Kolesnik" wrote in message
...
Sometime ago I think I recall someone posted or wrote an article on a neat
way to isolate a shorted component on a pcb using common test eqpt but I
can't recall the methodology. I'm trying to find a shorted component on
a
Wavetek 188-S-1257 signal generator. The B+ line reads about 0.4 ohms and
I'm not having much luck disconnecting componets. I don't have a
schematic
and my eyes ain't what they used to be for tracing and I want to minimize
the unsoldering. Does anyone recall the article or have a good way?
tnx
hank wd5jfr

On Thu, 15 Jan 2004 16:54:48 -0500, kenneth l wright
put finger to keyboard and composed:

Change all those little tear drop bypass caps connected to the shorted line.
Won't hurt to replace them all, might save a future short. Ken

I agree. I saw a *lot* of shorted tantalums when I was troubleshooting
multilayer minicomputer PCBs during the 80s.

Henry Kolesnik wrote:

Sometime ago I think I recall someone posted or wrote an article on a neat
way to isolate a shorted component on a pcb using common test eqpt but I
can't recall the methodology. I'm trying to find a shorted component on a
Wavetek 188-S-1257 signal generator. The B+ line reads about 0.4 ohms and
I'm not having much luck disconnecting componets. I don't have a schematic
and my eyes ain't what they used to be for tracing and I want to minimize
the unsoldering. Does anyone recall the article or have a good way?
tnx
hank wd5jfr


- Franc Zabkar
--
Please remove one 's' from my address when replying by email.

Some quick, off hand comments imbedded below:

Steve N.

"Michael A. Terrell" wrote in message
...

Some questions. How are you going to position it against the trace
accurately?

First you do some controlled experiments where you put the test signal
through known runners to characterize the given pickup. How to orientate it
& where it picks up from.


What about double sided,

I don't see a problem. If you get it set up so it detects at sufficient
distance (.070 or so).

or multilayer boards?

Should still work, but you'll have to follow the target trace by "braile"
(sp) since you might not be able to see it.

The problems with using AC to find shorts is that you get false peaks
and dips from the inductance of the traces, and the characteristics of
the components.

At audio this can't be a problem. I'm convinced it will just be current
defined by the generator.

With a runner short you won't have much current in anything but the
runners anyway, no?

Another problem is that some parts self destruct with
only a small reverse voltage so you can damage a lot of parts while
troubleshooting the board.

Also, there won't be much voltage (remember the DC method?). If you do
have this, you're in risk of burning up runners. That's too much current.

I used the DC voltage drop & sensitive digital meter method on boards
that people couldn't fix with AC, then had to find the parts they
damaged. The whole idea is to find and fix a problem quickly, and
reliably.

Hard to believe a 600 ohm audio gen will blow up anything, but it certainly
is possible... of course you just don't go in there blazing away with
power..

I don't mean to say the DC method is bad, or that AC is better. AC is
just another option, especially if you don't have a good enough DVM. Seems
to me I saw a commercial system which did use AC.

My Fluke has problems under an ohm and at fractions of a volt even though it
is a 5 digit. Should be good down there. May try it to see how DC works
just for info.

73

--
Steve N, K,9;d, c. i My email has no u's.

Henry, dont laugh at this method..it works....read it thru before nixing it.
I have had this happen before too. simply put about 5 volts at 500 mils on
the b+ line, regulated at the 500 mils. Let it sit for a few minutes and
then go looking for the part to be running warm. The part will be
dissipating 5*.5= 2.5 watts of heat. sooner or later the bad part WILL get
warm. It will NOT lift traces unless they are VERY small.
If this approach fails, the next thing I do is go in with a new(sharp)
razorblade and start as far away from the power supply and cut B+ traces one
at a time until the short goes away. This isolates the short to a smaller
area.
Suspect Tantalum caps as they usually fail in this mode of low ohms
shorts...let us know when you find it...Eddie

"Henry Kolesnik" wrote in message
...
Sometime ago I think I recall someone posted or wrote an article on a neat
way to isolate a shorted component on a pcb using common test eqpt but I
can't recall the methodology. I'm trying to find a shorted component on
a
Wavetek 188-S-1257 signal generator. The B+ line reads about 0.4 ohms and
I'm not having much luck disconnecting componets. I don't have a
schematic
and my eyes ain't what they used to be for tracing and I want to minimize
the unsoldering. Does anyone recall the article or have a good way?
tnx
hank wd5jfr

See my previous post about finding the culprit but I had another problem and
it was the 7815 3 term reg which I replaced with NOS and the Wavetek came to
life but not great. I measured the 15 volt terminal and it was 23 volts so
I may not have much left. I have question on your method. For a shorted
component with very low resistance, 0.2 I can't see much heat created since
P=I squared R. A short of 0.2 ohms wouldn't dissapate much power and all
you'll do is heat the traces, other components and perhaps blow short if
you're lucky!
73
hank wd5jfr
"Eddie Haskel" wrote in message
m...
Henry, dont laugh at this method..it works....read it thru before nixing
it.
I have had this happen before too. simply put about 5 volts at 500 mils on
the b+ line, regulated at the 500 mils. Let it sit for a few minutes and
then go looking for the part to be running warm. The part will be
dissipating 5*.5= 2.5 watts of heat. sooner or later the bad part WILL get
warm. It will NOT lift traces unless they are VERY small.
If this approach fails, the next thing I do is go in with a new(sharp)
razorblade and start as far away from the power supply and cut B+ traces
one
at a time until the short goes away. This isolates the short to a smaller
area.
Suspect Tantalum caps as they usually fail in this mode of low ohms
shorts...let us know when you find it...Eddie

"Henry Kolesnik" wrote in message
...
Sometime ago I think I recall someone posted or wrote an article on a
neat
way to isolate a shorted component on a pcb using common test eqpt but I
can't recall the methodology. I'm trying to find a shorted component on
a
Wavetek 188-S-1257 signal generator. The B+ line reads about 0.4 ohms
and
I'm not having much luck disconnecting componets. I don't have a
schematic
and my eyes ain't what they used to be for tracing and I want to
minimize
the unsoldering. Does anyone recall the article or have a good way?
tnx
hank wd5jfr

On Fri, 16 Jan 2004 08:53:49 +0800 budgie wrote:

On Thu, 15 Jan 2004 16:44:35 -0600, "Steve Nosko"
wrote:

Seems to me there's a way to use a moderate current and sense the
_magnetic_ field. Follow the field around the board along the runners.
However, I don't remember what was used to sense the field.

HP used to make a hall effect (IIRC) probe for current tracing. One of their
Bench Briefs technotes described the probe and the process.

It was the HP 547A Logic Current Tracer combined with the HP 546A
Logic Pulser. These worked only with TTL or CMOS circuits. The pulser
would drive current into any point you wanted and the tracer could be
used to follow the current path.

You would set the pulser to put out a continuous pulse train, and then
adjust the sensitivity of the tracer so that it would just trigger on
the current pulses sent out by the pulser. It would thus ignore other
current that was flowing in the same trace as long as that current was
either DC or consisted of pulses which were smaller than those put out
by the logic pulser. I think the logic pulser could put out an amp or
so, but they were short enough that they didn't damage anything.

They work well, but the current tracers are rather hard to come by now
and they are rather expensive when you find them.

-
-----------------------------------------------
Jim Adney
Madison, WI 53711 USA
-----------------------------------------------

The 7815 needs only 3 volts of headroom, so 23 volts is plenty. Any more
than that will be dissapated as heat. Your right that the traces will also
get warm but it's still safe. The other post that advised the use of freeze
spray was a good one too. I have used both...Eddie
"Henry Kolesnik" wrote in message
...
See my previous post about finding the culprit but I had another problem
and
it was the 7815 3 term reg which I replaced with NOS and the Wavetek came
to
life but not great. I measured the 15 volt terminal and it was 23 volts
so
I may not have much left. I have question on your method. For a shorted
component with very low resistance, 0.2 I can't see much heat created
since
P=I squared R. A short of 0.2 ohms wouldn't dissapate much power and all
you'll do is heat the traces, other components and perhaps blow short if
you're lucky!
73
hank wd5jfr
"Eddie Haskel" wrote in message
m...
Henry, dont laugh at this method..it works....read it thru before nixing
it.
I have had this happen before too. simply put about 5 volts at 500 mils
on
the b+ line, regulated at the 500 mils. Let it sit for a few minutes and
then go looking for the part to be running warm. The part will be
dissipating 5*.5= 2.5 watts of heat. sooner or later the bad part WILL
get
warm. It will NOT lift traces unless they are VERY small.
If this approach fails, the next thing I do is go in with a new(sharp)
razorblade and start as far away from the power supply and cut B+ traces
one
at a time until the short goes away. This isolates the short to a
smaller
area.
Suspect Tantalum caps as they usually fail in this mode of low ohms
shorts...let us know when you find it...Eddie

"Henry Kolesnik" wrote in message
...
Sometime ago I think I recall someone posted or wrote an article on a
neat
way to isolate a shorted component on a pcb using common test eqpt but
I
can't recall the methodology. I'm trying to find a shorted component
on
a
Wavetek 188-S-1257 signal generator. The B+ line reads about 0.4 ohms
and
I'm not having much luck disconnecting componets. I don't have a
schematic
and my eyes ain't what they used to be for tracing and I want to
minimize
the unsoldering. Does anyone recall the article or have a good way?
tnx
hank wd5jfr