On Jul 23, 8:33 am, Spehro Pefhany <speffS...@interlogDOTyou.knowwhat>
wrote:
> On Tue, 22 Jul 2008 09:30:33 -0700 (PDT), the renowned rickman
>
>
>
> <gnu...@gmail.com> wrote:
> >I have a small module that is shorted between the +12 volt plane and
> >ground.  I am having a hard time finding where the short is so it can
> >be fixed.
>
> >The bare boards were supposed to be tested, so I don't suspect the
> >board itself.  I have visually inspected everything I can including
> >looking under the chips as much as I can see and found no sign of a
> >problem.
>
> >My bench supply current limits (foldback actually) and I am seeing
> >about an Amp into the 12 volt rail.  Probing with a volt meter I can
> >see 10 mV at the point where I connect the power to the board.  This
> >drops to about 1 mV on the other edge of the board.  But I can't find
> >a particular point where the voltage says "here it is"!
>
> >Any ideas on how to find and fix this short?
>
> >Rick
>
> Do you have a blank board to check? I hope so.  Can you see the
> clearance where vias etc. pass through the power planes by holding the
> board up to a strong light?
>
> As others said you can follow the gradient. If the gradient does not
> change then you've found a direction where the plane is not carrying
> current. But with two internal planes this won't be all that easy.
>
> Hate to say it, but it kinda sounds like it might be internal to the
> board from what you're saying. Like maybe they made two gnd planes or
> something like that.
>
> Just a couple of times I've taken almost every single component out of
> a board to find the problem with 100% certainty (failure analysis,
> just to ensure we didn't get more like that). Whiskery shorts on
> boards (sometimes intermittent so it could pass e-test) were the
> hardest to find, but you could see them under a microscope. OTOH,
> sometimes it's huge and right in front of your face.

When you say you found various shorts, were they design flaws (which
would be repeated on each board) or manufacturing flaws (which would
be individual)?  This is one board out of a run of 110.  I have found
one other board with a short on the 3 volt rail, but this one is on
the 12 volt rail.  So there is no design flaw.

I am currently suspecting a decoupling cap at this point.  The
gradient is small and points toward one end of my 4.5 x 0.85" board.
The 12 volt plane is only on about half the board and power has been
applied to the end near the middle of the board.  The gradient points
to the opposite end.  When I get some more time to work on this, I
will test with power on the "opposite" end and test toward the power
connector.  If I see the same gradient, I will start removing
components.  If the gradient slopes the other way, I will suspect
multiple shorts which are likely an internal board problem (which I
expect is unlikely).

BTW, thanks for all of the suggestions everyone!  Some of the ideas
were obvious to me, others at least made me think a bit.  I appreciate
the different viewpoints.

I may try the freeze spray thing.  I don't have a supply that will put
out 10 A, but I have an AC transformer that puts out 1 VAC at some
huge current.  That with a diode might actually do the job, who
knows?

Rick

Paul Keinanen wrote:
> On Wed, 23 Jul 2008 04:39:39 -0700 (PDT), Didi <d...@tgi-sci.com> wrote:
>
> >larwe wrote:
> >> ....
> >> Of course a Real Man(sm) would just blow the short out with AC line
> >> voltage.
> >
> >Yahoo, that would be it! In 220V countries even better than in the
> >US... :-)
>
> The problem is that in 230 V countries, the mains fuses are only 10-16
> A, so if there is a stubborn short on the PCB, the 50-100 A short
> circuit current may blow the mains fuse but leave the PCB short
> unaffected :-(.
> ....

Yes, that's a clear advantage of the 110V  (one can expect twice the
current at 220 is needed to blow the fuses), but 220 is still more fun
for
this application - sparkles, smoke, danger etc. (I recently got 100n
at 3kV
discharged left to right hand fingertips... was making a new 3kV 60W
power
supply, would not have been a success without any of that :-).
 For practical purposes a car battery should be the best source to
blow PCB areas.

Didi

------------------------------------------------------
Dimiter Popoff               Transgalactic Instruments

http://www.tgi-sci.com
------------------------------------------------------
http://www.flickr.com/photos/didi_tgi/sets/72157600228621276/

Original message: http://groups.google.com/group/comp.arch.embedded/msg/19187ceba73c1bf2?dmode=source

In article <8ea6b8c3-2bed-40ec-8734-2378efc3e2b0
@l42g2000hsc.googlegroups.com>, g...@gmail.com says...
> On Jul 22, 8:02 pm, CBFalconer <cbfalco...@yahoo.com> wrote:
> > rickman wrote:
> >
> > ... snip ...
> >
> > > The bare boards were supposed to be tested, so I don't suspect
> > > the board itself.  I have visually inspected everything I can
> > > including looking under the chips as much as I can see and found
> > > no sign of a problem.
> >
> > Have you taken an unstuffed board and tested that assumption?
> 
> No, but I have over a hundred boards that are working just fine.
> Actually, I have 103 working boards... unfortunately, I need to
> deliver 104, no kidding!  I have three other boards where I have
> identified the problems and can be fixed.  One of those has at least
> two open vias.  These boards are from the batch I got from Sunstone
> where they had a 25% X-out rate due to plating problems in the vias.
> I ordered 6 panels.  They made 7 and I still didn't get enough
> boards.  They had to make a seventh panel with 16 working boards to
> complete the order.

What was the size of your vias?   I've been using 8 mil trace and space
and 21 mil vias for many years without problems.   However, some of
the larger fine-pitch stuff I use now will soon force finer traces
and spacing.
> 
> I was very concerned about vias opening in the field.  So I am glad
> that I have only found one board with open vias after assembly.  I
> just got 56 boards out of a four day bake which I will be retesting.
> 
> Rick


Mark Borgerson

On Wed, 23 Jul 2008 07:50:06 -0700 (PDT), Didi <d...@tgi-sci.com> wrote:

>Yes, that's a clear advantage of the 110V  (one can expect twice the
>current at 220 is needed to blow the fuses), but 220 is still more fun
>for this application - sparkles, smoke, danger etc. 

By the way, can you make a hot sausage in a 110 V country by just
sticking the neutral at one end of the sausage and live conductor into
the other end ?

Paul

On Jul 23, 10:02 am, Stef <stef...@yahooI-N-V-A-L-I-D.com.invalid>
wrote:
> In comp.arch.embedded,
>
>
>
> rickman <gnu...@gmail.com> wrote:
> > On Jul 23, 8:45 am, Dombo <do...@disposable.invalid> wrote:
> >> Paul Keinanen schreef:
>
> >> > On Wed, 23 Jul 2008 04:39:39 -0700 (PDT), Didi <d...@tgi-sci.com> wrote:
>
> >> >> larwe wrote:
> >> >>> ....
> >> >>> Of course a Real Man(sm) would just blow the short out with AC line
> >> >>> voltage.
> >> >> Yahoo, that would be it! In 220V countries even better than in the
> >> >> US... :-)
>
> >> > The problem is that in 230 V countries, the mains fuses are only 10-16
> >> > A, so if there is a stubborn short on the PCB, the 50-100 A short
> >> > circuit current may blow the mains fuse but leave the PCB short
> >> > unaffected :-(.
>
> >> The mains fuse doesn't immediately blow at 16A. I've seen copper
> >> evaporate from the PCB with a short on the the mains side with the mains
> >> fuse still left intact.
>
> > Yes, this sounds like an excellent idea!  Maybe I can even evaporate
> > the entire power plane completely eliminating any possibility of a
> > short!
>
> For that purpose I can recommend one of our machines with 40,000uF caps
> at 350 Volts, they're ideal for evaporating copper from boards!
>
> --
> Stef    (remove caps, dashes and .invalid from e-mail address to reply by mail)

If I get to that point, would you like to provide a unit for
evaluation and I will post the results.  Maybe this will generate a
lot of sales???

Rick

On Jul 23, 11:17 am, Mark Borgerson <mborger...@comcast.net> wrote:
>
> What was the size of your vias?   I've been using 8 mil trace and space
> and 21 mil vias for many years without problems.   However, some of
> the larger fine-pitch stuff I use now will soon force finer traces
> and spacing.

I used 6/6 and 10 mil vias in 24 mil pads.  Sunstone originally said
they could make 10 mil vias, but after I got the boards they said they
used a 13 mil drill since that was within their spec of +- 3 mil.  As
long as they got the holes on the pads, I didn't care, but some of the
holes are right at the edge of the pads.  I've never been able to tell
if this is a problem as long as the trace is not cut from the pad.

Rick

Paul Keinanen wrote:
> Didi <d...@tgi-sci.com> wrote:
> 
>> Yes, that's a clear advantage of the 110V  (one can expect twice
>> the current at 220 is needed to blow the fuses), but 220 is still
>> more fun for this application - sparkles, smoke, danger etc.
> 
> By the way, can you make a hot sausage in a 110 V country by just
> sticking the neutral at one end of the sausage and live conductor
> into the other end ?

No, that operation here requires an adaptor and half-length
sausages.  :-)

-- 
 [mail]: Chuck F (cbfalconer at maineline dot net) 
 [page]: <http://cbfalconer.home.att.net>;
            Try the download section.

On Jul 23, 10:27 am, "dkel...@hotmail.com" <dkel...@hotmail.com>
wrote:
> On Jul 22, 9:30 am, rickman <gnu...@gmail.com> wrote:
>
>
>
> > I have a small module that is shorted between the +12 volt plane and
> > ground.  I am having a hard time finding where the short is so it can
> > be fixed.
>
> > The bare boards were supposed to be tested, so I don't suspect the
> > board itself.  I have visually inspected everything I can including
> > looking under the chips as much as I can see and found no sign of a
> > problem.
>
> > My bench supply current limits (foldback actually) and I am seeing
> > about an Amp into the 12 volt rail.  Probing with a volt meter I can
> > see 10 mV at the point where I connect the power to the board.  This
> > drops to about 1 mV on the other edge of the board.  But I can't find
> > a particular point where the voltage says "here it is"!
>
> > Any ideas on how to find and fix this short?
>
> > Rick
>
> Hi Rick
>  I've described this methode before. Most people seem to think
> that a method of finding the short requires feeding current through
> the short. My method uses a different idea.
>   Put your supply across the 12v line, from one end to the other,
> in current limit. The line then has a small voltage drop across it.
>  Place a DVM lead on the ground. The location isn't important
> since there is no current in the ground.
>  Use the 200mv scale. This works best with a 4 or 5 digit meter.
> With the other lead, probe along the 12V line. When the meter
> reads closest to zero, you are at the location of the short.
>  I have used this method many times when other methods
> seem to fail.
>  The idea is to consider the 12 v line to be like a resistor with
> a tap on it. The ground it the tap. With the second lead as
> another tap, when both taps are at the same location along
> the resistor, there is 0 volts.
> Dwight

Good idea!  When I get some time I will try this.  Thanks.

Rick

On 23 Jul., 16:31, rickman <gnu...@gmail.com> wrote:
> On Jul 23, 8:33 am, Spehro Pefhany <speffS...@interlogDOTyou.knowwhat>
> wrote:
>
>
>
> > On Tue, 22 Jul 2008 09:30:33 -0700 (PDT), the renowned rickman
>
> > <gnu...@gmail.com> wrote:
> > >I have a small module that is shorted between the +12 volt plane and
> > >ground.  I am having a hard time finding where the short is so it can
> > >be fixed.
>
> > >The bare boards were supposed to be tested, so I don't suspect the
> > >board itself.  I have visually inspected everything I can including
> > >looking under the chips as much as I can see and found no sign of a
> > >problem.
>
> > >My bench supply current limits (foldback actually) and I am seeing
> > >about an Amp into the 12 volt rail.  Probing with a volt meter I can
> > >see 10 mV at the point where I connect the power to the board.  This
> > >drops to about 1 mV on the other edge of the board.  But I can't find
> > >a particular point where the voltage says "here it is"!
>
> > >Any ideas on how to find and fix this short?
>
> > >Rick
>
> > Do you have a blank board to check? I hope so.  Can you see the
> > clearance where vias etc. pass through the power planes by holding the
> > board up to a strong light?
>
> > As others said you can follow the gradient. If the gradient does not
> > change then you've found a direction where the plane is not carrying
> > current. But with two internal planes this won't be all that easy.
>
> > Hate to say it, but it kinda sounds like it might be internal to the
> > board from what you're saying. Like maybe they made two gnd planes or
> > something like that.
>
> > Just a couple of times I've taken almost every single component out of
> > a board to find the problem with 100% certainty (failure analysis,
> > just to ensure we didn't get more like that). Whiskery shorts on
> > boards (sometimes intermittent so it could pass e-test) were the
> > hardest to find, but you could see them under a microscope. OTOH,
> > sometimes it's huge and right in front of your face.
>
> When you say you found various shorts, were they design flaws (which
> would be repeated on each board) or manufacturing flaws (which would
> be individual)?  This is one board out of a run of 110.  I have found
> one other board with a short on the 3 volt rail, but this one is on
> the 12 volt rail.  So there is no design flaw.
>
> I am currently suspecting a decoupling cap at this point.  The
> gradient is small and points toward one end of my 4.5 x 0.85" board.
> The 12 volt plane is only on about half the board and power has been
> applied to the end near the middle of the board.  The gradient points
> to the opposite end.  When I get some more time to work on this, I
> will test with power on the "opposite" end and test toward the power
> connector.  If I see the same gradient, I will start removing
> components.  If the gradient slopes the other way, I will suspect
> multiple shorts which are likely an internal board problem (which I
> expect is unlikely).
>
> BTW, thanks for all of the suggestions everyone!  Some of the ideas
> were obvious to me, others at least made me think a bit.  I appreciate
> the different viewpoints.
>
> I may try the freeze spray thing.  I don't have a supply that will put
> out 10 A, but I have an AC transformer that puts out 1 VAC at some
> huge current.  That with a diode might actually do the job, who
> knows?
>
> Rick

use car battery? ;)

-Lasse

In comp.arch.embedded,
rickman <g...@gmail.com> wrote:
> On Jul 23, 10:02 am, Stef <stef...@yahooI-N-V-A-L-I-D.com.invalid>
> wrote:
>> In comp.arch.embedded,
>> rickman <gnu...@gmail.com> wrote:
>> > On Jul 23, 8:45 am, Dombo <do...@disposable.invalid> wrote:
>>
>> >> The mains fuse doesn't immediately blow at 16A. I've seen copper
>> >> evaporate from the PCB with a short on the the mains side with the mains
>> >> fuse still left intact.
>>
>> > Yes, this sounds like an excellent idea!  Maybe I can even evaporate
>> > the entire power plane completely eliminating any possibility of a
>> > short!
>>
>> For that purpose I can recommend one of our machines with 40,000uF caps
>> at 350 Volts, they're ideal for evaporating copper from boards!
>
> If I get to that point, would you like to provide a unit for
> evaluation and I will post the results.  Maybe this will generate a
> lot of sales???

No extra sales I'm affraid. It is not their intended use but was their
self-destruct mechanism. :-)

-- 
Stef    (remove caps, dashes and .invalid from e-mail address to reply by mail)

There are two kinds of pedestrians... the quick and the dead.
		-- Lord Thomas Rober Dewar