In article <6...@mid.individual.net>, 
P...@topmail.co.uk says...
.....

> [%X]
> 
> At least I have got some discussion on the non-destructive methods going.
> I know it seems like no fun but there are times when a non-destructive
> approach is needed. After all, having to rebuild a board that is charred
> is not easy and charcoal is a bugger to solder to. ;>

But most engineers are just like children, only really like things that
make noises, flash or better still explode. Just like trying to teach 
science to teenagers.

-- 
Paul Carpenter          | p...@pcserviceselectronics.co.uk
<http://www.pcserviceselectronics.co.uk/>;    PC Services
<http://www.pcserviceselectronics.co.uk/fonts/>; Timing Diagram Font
<http://www.gnuh8.org.uk/>;  GNU H8 - compiler & Renesas H8/H8S/H8 Tiny
<http://www.badweb.org.uk/>; For those web sites you hate

rickman wrote:
> On Jul 23, 4:17 pm, Paul Keinanen <keina...@sci.fi> wrote:
>> On Wed, 23 Jul 2008 19:49:58 +0100, "Paul E. Bennett"
>>
>> <Paul_E.Benn...@topmail.co.uk> wrote:
>>> In reviewing some of the other replies I noticed that no one had yet (at
>>> the top level of the tree) suggested using a signal injector and a trace
>>> probe. By injecting a current driven signal onto the power rail and
>>> ground you can prod around with the trace probe and determine where the
>>> short is by the signal level dropping nearest the location.
>> What kind of signal would that be ?
>>
>> Using a variable frequency RF-generator you should be able to
>> determine when the tracks and the short is resonated e.g. at 1/2 or
>> 1/4 wavelength. After determining the PCB material and hence the
>> velocity factor, you should be able to determine the distance from the
>> feedpoint.
>>
>> One other method would be to inject a sub-nanosecond pulse edge and
>> determine, how long it takes for the reflection from the short to
>> return back to the generator.
>>
>> A different approach would be to inject a significant current into the
>> +12 V line and try to detect the magnetic field on the +12 V trace,
>> until it disappears into the ground plane. This should work even if
>> the +12 V trace is within a multilayer PCB. Some Hall sensor might
>> useful, if DC current is used to create the magnetic field, but if an
>> AC (mains /audio frequency) current is running in the trace, some
>> pick-up coil should be able to pick up the stray magnetic field.
> 
> As a matter of fact, the last day job I had the company had just
> bought a $50,000 TDR.  I am sure fixing a $150 board is exactly the
> job they had planned for it.  Even with that, I'm not sure it could
> show me the short.  I don't know for sure if it is in the boards or on
> top of it.  The TDR won't tell me that.
> 

It most likely wouldn't work very well. A really wide trace of plane 
smears out the location too much.

With most of us being in the Wild West I am surprised nobody came up 
with the obvious: Take a Smith&Wesson and blast holes into it until the 
current drops. That's where the short is. Or rather, was. Just kidding ;-)

-- 
Regards, Joerg

http://www.analogconsultants.com/

"gmail" domain blocked because of excessive spam.
Use another domain or send PM.

On Thu, 24 Jul 2008 23:14:17 +0100, "Paul E. Bennett"
<P...@topmail.co.uk> wrote:

>Paul Keinanen wrote:
>
>> On Wed, 23 Jul 2008 19:49:58 +0100, "Paul E. Bennett"
>> <P...@topmail.co.uk> wrote:
>> 
>>>In reviewing some of the other replies I noticed that no one had yet
>>>(at the top level of the tree) suggested using a signal injector and a
>>>trace probe. By injecting a current driven signal onto the power rail
>>>and ground you can prod around with the trace probe and determine where
>>>the short is by the signal level dropping nearest the location.
>> 
>> What kind of signal would that be ?
>
>The current injection probe I used would put in a square wave at about
>1MHz or 10kHz (switch selectable). The trace probe could find this
>signal on a shorted line and give an indication of how big it was. When
>you got very close to the short you would actually lose the signal.
>Works best if the return is on a ground plane but can cope with other
>layout practices as well. The probes usually come as pairs.

So I assume this uses the inductive reactance (instead of resistance)
to create a voltage gradient when the test current flows along the
line ? That 1 MHz test frequency sounds quite low, since a PCB track
would have XL about 0.01 ohm/mm @ 1 MHz, so the XL for a 100 mm track
would only be about 1 ohm. The signal generator output impedance would
have to be well below the typical 50 ohms in order to create a
significant voltage at the feed point.

Of course the probe can contain a high gain RF-amplifier to get
meaningful reading. Using RF would eliminate the problems of measuring
very low DC voltage gradients along the line (such as amplifier offset
drift and galvanic voltages caused by dissimilar metals).

Paul
 

Paul Keinanen wrote:
> On Thu, 24 Jul 2008 23:14:17 +0100, "Paul E. Bennett"
> <P...@topmail.co.uk> wrote:
> 
>> Paul Keinanen wrote:
>>
>>> On Wed, 23 Jul 2008 19:49:58 +0100, "Paul E. Bennett"
>>> <P...@topmail.co.uk> wrote:
>>>
>>>> In reviewing some of the other replies I noticed that no one had yet
>>>> (at the top level of the tree) suggested using a signal injector and a
>>>> trace probe. By injecting a current driven signal onto the power rail
>>>> and ground you can prod around with the trace probe and determine where
>>>> the short is by the signal level dropping nearest the location.
>>> What kind of signal would that be ?
>> The current injection probe I used would put in a square wave at about
>> 1MHz or 10kHz (switch selectable). The trace probe could find this
>> signal on a shorted line and give an indication of how big it was. When
>> you got very close to the short you would actually lose the signal.
>> Works best if the return is on a ground plane but can cope with other
>> layout practices as well. The probes usually come as pairs.
> 
> So I assume this uses the inductive reactance (instead of resistance)
> to create a voltage gradient when the test current flows along the
> line ? That 1 MHz test frequency sounds quite low, since a PCB track
> would have XL about 0.01 ohm/mm @ 1 MHz, so the XL for a 100 mm track
> would only be about 1 ohm. The signal generator output impedance would
> have to be well below the typical 50 ohms in order to create a
> significant voltage at the feed point.
> 
> Of course the probe can contain a high gain RF-amplifier to get
> meaningful reading. Using RF would eliminate the problems of measuring
> very low DC voltage gradients along the line (such as amplifier offset
> drift and galvanic voltages caused by dissimilar metals).
> 
A possible difficulty with RF is it couples just as well through 
capacitance as through inductance. You might get many false positives.
I seem to recall H-P many years ago, sold a probe with a Hall effect 
device in the tip. You could put a small, controlled DC into the short, 
& see which traces it was flowing through. Maybe such a device could be 
made up today: there are small Hall-effect ICs around.

On Sat, 26 Jul 2008 07:41:41 +0800, David R Brooks
<d...@iinet.net.au> wrote:

>Paul Keinanen wrote:
>> On Thu, 24 Jul 2008 23:14:17 +0100, "Paul E. Bennett"
>> <P...@topmail.co.uk> wrote:
>> 
>>> Paul Keinanen wrote:
>>>
>>>> On Wed, 23 Jul 2008 19:49:58 +0100, "Paul E. Bennett"
>>>> <P...@topmail.co.uk> wrote:
>>>>
>>>>> In reviewing some of the other replies I noticed that no one had yet
>>>>> (at the top level of the tree) suggested using a signal injector and a
>>>>> trace probe. By injecting a current driven signal onto the power rail
>>>>> and ground you can prod around with the trace probe and determine where
>>>>> the short is by the signal level dropping nearest the location.
>>>> What kind of signal would that be ?
>>> The current injection probe I used would put in a square wave at about
>>> 1MHz or 10kHz (switch selectable). The trace probe could find this
>>> signal on a shorted line and give an indication of how big it was. When
>>> you got very close to the short you would actually lose the signal.
>>> Works best if the return is on a ground plane but can cope with other
>>> layout practices as well. The probes usually come as pairs.
>> 
>> So I assume this uses the inductive reactance (instead of resistance)
>> to create a voltage gradient when the test current flows along the
>> line ? That 1 MHz test frequency sounds quite low, since a PCB track
>> would have XL about 0.01 ohm/mm @ 1 MHz, so the XL for a 100 mm track
>> would only be about 1 ohm. The signal generator output impedance would
>> have to be well below the typical 50 ohms in order to create a
>> significant voltage at the feed point.
>> 
>> Of course the probe can contain a high gain RF-amplifier to get
>> meaningful reading. Using RF would eliminate the problems of measuring
>> very low DC voltage gradients along the line (such as amplifier offset
>> drift and galvanic voltages caused by dissimilar metals).
>> 
>A possible difficulty with RF is it couples just as well through 
>capacitance as through inductance. You might get many false positives.

Capacitive coupling can be an issue if the probe has a high input
impedance. A probe using a small pickup coil for measuring the
magnetic field created by the RF_current_ or measure directly the
voltage gradient along the PCB track inductive reactance, could have a
low input impedance and the capacitive coupling should not be an
issue.

Paul

In article <3...@a1g2000hsb.googlegroups.com>,
rickman  <g...@gmail.com> wrote:
>I seem to recall a magnetic liquid with some very unusual properties.
>I don't recall the exact form, but in all of the images I saw it was
>intact and 3D.  It would illustrate some very odd fields, one caused
>it to look like a spiky soccer ball.  Anyone know what it was?

Ferrofluid --- basically a fine suspension of iron particles in oil,
I think. The "spiky soccer ball" is pretty characteristic of what it looks
like near a magnet. I think the spikes are not originally from the shape
of the field, but because the surface is somewhat unstable in a strong
enough field.

-- 
   Wim Lewis <w...@hhhh.org>, Seattle, WA, USA. PGP keyID 27F772C1