Mac Mini A1176 shorted capacitor repair

[Jeff Liebermann]

Another day in repair hell. I can easily fix my customers machines,
but when one of mine blows up, it's always something complex or messy.

Today's nightmare is my Apple Mac Mini A1176. Push the on button and
nothing happens. No lights, no power, no nothing. Power supply tests
fine with other Mac Mini's. It's not the fuses, not a broken
connector, and not a bad on/off button. Googling for help, I found:
<http://blog.helpmymac.ru/?p=3585>
which matches the symptoms and probable culprit exactly. There are a
mess of ceramic capacitors that sometimes short. The author
recommends that I apply power with an adjustable power supply, and
look for which cap gets hot. That will probably work, but I want to
save that procedure for when I'm desperate and out of other options.

I tried an ohmmeter, which showed 1-2 ohms across all the capacitors.
That's not much help unless I want to replace all the capacitors.
Next, I tried an ESR meter, which showed about 0.5 ohms everywhere.
The problem here is that it can't distinguish between a proper ESR and
a dead short. I tried a capacitance meter, which declared that the
impedance was too low to produce a usable result.

Any advice on what to do next? I'm tempted to just remove all 21
capacitors and test them out of the circuit. With SMT caps, the
effort involved in testing and replacing is about the same.

Any other possible culprits for a dead Mac Mini?

Any other forums inhabited by techs with Mac experience? Not the
official Apple forums, which I've found to be useless. I couldn't
find any that had info on component level repairs.

--
Jeff Liebermann je...@cruzio.com
150 Felker St #D http://www.LearnByDestroying.com
Santa Cruz CA 95060 http://802.11junk.com
Skype: JeffLiebermann AE6KS 831-336-2558

[Michael A. Terrell]

Which ESR meter did you use? Analog or digital? Analog is useless
for the tiny changes. If digital, attack the problem from each end of
that rail to ground and you might find a minor difference, but without a
four lead probe, it's doubtful.


I use a current & voltage limited supply to inject a low current into
the circuit, then follow it across the board with a 4.5 digit voltmeter
and measure the voltage along the bus. (Typically 100 mA.) You will
see a consistent drop, till you reach the bad part, then smaller drops,
if any past that point. If nothing else, it narrows it to a small area
of the board, and the number of parts you need to remove for testing.

I've also used a .1 ohm (Source impedance, and at 100 mV.) 1 kHZ
signal and a Fluke 8920A True RMS voltmeter to make even finer tests
when working at the factory to salvage new circuit boards with over
$8,000 worth of components. Most of these were whiskers under SMD
ceramic caps.


In either case, you want a decent sized conductor from the voltage
source, or you'll lose most of the voltage in the leads. I prefer 16
AWG or heavier, and if possible, solder them to the circuit board at the
power connector.



BTW, I just lost a bid for a nice presentation camera on Ebay. I
want to make some videos of these, and other methods I developed while
working in the manufacturing end of the business. That camera had USB,
DVI & NTSC outputs so it would be handy to use with a monitor on the
repair bench. It had a 16X manual zoom, and several flexible lights.

[N_Cook]

Jeff Liebermann <je...@cruzio.com> wrote in message
news:eod5c8t4341tapimi...@4ax.com...

With a shorted TVS I've used a milli-ohm meter to zero-in, but that of
course that is a proper short. Perhaps a milli-ohm meter can find the
minimum DC ohms, as pretty fine pcb traces I presume

[N_Cook]

If non-zero onms , would localised heating each cap in turn divine the
problem one/s ?

[Uffe Bærentsen]

Den 08-12-2012 04:57, Jeff Liebermann skrev:

> Another day in repair hell. I can easily fix my customers machines,
> but when one of mine blows up, it's always something complex or messy.

Murphy's Law proven once more ;-)

> I tried an ohmmeter, which showed 1-2 ohms across all the capacitors.
> That's not much help unless I want to replace all the capacitors.
> Next, I tried an ESR meter, which showed about 0.5 ohms everywhere.
> The problem here is that it can't distinguish between a proper ESR and
> a dead short. I tried a capacitance meter, which declared that the
> impedance was too low to produce a usable result.
>
> Any advice on what to do next? I'm tempted to just remove all 21
> capacitors and test them out of the circuit. With SMT caps, the
> effort involved in testing and replacing is about the same.

A milliohmmeter would be a possibillity.
Lowest resistance = failed capacitor.
Have used that method with success :-)


--
Uffe

[Doug White]

Jeff Liebermann <je...@cruzio.com> wrote in

news:eod5c8t4341tapimi...@4ax.com:

><snip>

> Today's nightmare is my Apple Mac Mini A1176. Push the on button and
> nothing happens. No lights, no power, no nothing. Power supply tests
> fine with other Mac Mini's. It's not the fuses, not a broken
> connector, and not a bad on/off button. Googling for help, I found:
> <http://blog.helpmymac.ru/?p=3585>
> which matches the symptoms and probable culprit exactly. There are a
> mess of ceramic capacitors that sometimes short. The author
> recommends that I apply power with an adjustable power supply, and
> look for which cap gets hot. That will probably work, but I want to
> save that procedure for when I'm desperate and out of other options.

I'd go with the current limited power supply, with an added wrinkle.
Hose the suspected parts down with spray freeze & look for the one that
defrosts the fastest.

Works very quickly, and is far more sensitive than waiting for smoke or
explosions. I've even found PCB layout errors where there was a dead
short between the power & ground. Takes all of about 5 seconds to find
the fault.

Doug White

[Michael A. Terrell]

That's OK, as long as the fault doesn't open under brute force
methods, and that the short's impedance is high enough to generate all
the heat. If not, you may open traces on or in the board. Not an
approved method at the factory level, on complex boards with multiple
power rails.

[William Sommerwerck]

How many caps need replacement? Why not just replace them and get it over
with?

Remember W Edwards Deming? "You cannot /inspect/ quality into a product."

http://en.wikipedia.org/wiki/W._Edwards_Deming

Ceramic caps /ought/ to be reliable. The fact that even /one/ of these has
(apparently) failed, suggests the others are of questionable quality. I'd
replace all of them with high-grade caps.

[Plai...@yawhoo.com]

On Fri, 07 Dec 2012 19:57:32 -0800, Jeff Liebermann <je...@cruzio.com>
wrote:

We had a similar problem about 30 years ago where I worked in
producion test. Initially the caps would test good, however when
operating voltage was applied one (or more) would short. The problem
was even after a full set of tests, including burn in period there was
no assurance that no further caps would decide to short.

Our approach was pretty much as others have suggesed - apply low
voltage with a current limited supply and look for the point where
there is a 'knee' in the voltage drop. We found and replaced a LOT of
shorted caps that way.

Management decided to purge all of the suspect lots of these caps from
inventory, rely on test procedures to catch any already used to build
products, and cover those that failed in the field under warranty.
The company went out of business several years later. In retrospect,
replacing ALL of the suspect caps would have been a better approach.

PlainBill

[whit3rd]

On Saturday, December 8, 2012 8:46:05 AM UTC-8, William Sommerwerck wrote:
> How many caps need replacement? Why not just replace them and get it over
> with?

Probably only one; the board has perhaps a dozen or so, of multilayer
surface mount ceramics (MLCC), not easy to remove with hand tools.

> Ceramic caps /ought/ to be reliable. The fact that even /one/ of these has
> (apparently) failed, suggests the others are of questionable quality. I'd
> replace all of them with high-grade caps.

Three implementation details make that impractical; first, the capacitors
are not likely labeled with voltage/capacity . Second, the original
manufacturer is not known (no labels). Third, I'm not aware of any
particularly good manufacturer of these (high-capacity, low voltage MLCC).


As capacity of these little multilayer capacitors has risen, the ceramic
insulators have got VERY thin; old ceramic disks were very reliable, but
these MLCC jobbies, with voltage ratings down to 2.5V, are exploring
new failure modes.

[Baron]

William Sommerwerck Inscribed thus:

> How many caps need replacement? Why not just replace them and get it
> over with?

These are also glued down !

> Remember W Edwards Deming? "You cannot /inspect/ quality into a
> product."
>
> http://en.wikipedia.org/wiki/W._Edwards_Deming
>
> Ceramic caps /ought/ to be reliable. The fact that even /one/ of these
> has (apparently) failed, suggests the others are of questionable
> quality. I'd replace all of them with high-grade caps.

--
Best Regards:
Baron.

[William Sommerwerck]

"Baron" wrote in message news:ka0btq$41p$1...@dont-email.me...

William Sommerwerck Inscribed thus:

>> How many caps need replacement? Why not just replace them
>> and get it over with?

> These are also glued down!

I am so living in the past that I'm thinking of these caps as conventional
disk ceramics with wire leads! Forgive me.

[Baron]

William Sommerwerck Inscribed thus:

In general surface mount components are not straightforward to remove
and replace !

--
Best Regards:
Baron.

[Michael A. Terrell]

William Sommerwerck wrote:
>
> How many caps need replacement? Why not just replace them and get it over
> with?

Shotgunning takes longer than finding the real problem with proper
techniques. It isn't like a 50 year old tube radio where the paper
capacitors are all well past their expected useful life. He would have
to remove & test every unmarked ceramic capacitor on that rail, then
solder in new ones of the same value. they are very small, and easy to
lose, especially if you are older and have less than perfect eyesight.
The parts are too small to mark with the proper two character code in
most cases, and marked capacitors cost more than unmarked so they are
rarely used in high volume production where the assembly is done by
automated pick-n-place machines where they only read the value on the
reel when it's loaded.

> Remember W Edwards Deming? "You cannot /inspect/ quality into a product."
>
> http://en.wikipedia.org/wiki/W._Edwards_Deming

He never worked with fine pitch surface mount, lead free technology.
The entire lead free concept is flawed from the beginning.

[stra...@yahoo.com]

On Friday, December 7, 2012 7:57:32 PM UTC-8, Jeff Liebermann wrote:
> which matches the symptoms and probable culprit exactly. There are a
>
> mess of ceramic capacitors that sometimes short. The author
>

I've had to hunt down shorted ceramics. I use a Fluke 8060A that can resolve down to 0.01 Ohm. It's tedious to check all the caps but you will find that the resistance is the power traces is greater than the shorted cap. First touch the probes together and press 'relative' to subtract out the lead resistance and start measuring caps. when you get to the lowest one, clip it out and see if that was all. ESR on ceramics really isn't appropriate as you have a DC short.

Alternatively you could do the variable power supply but don't go for heating and possible trace fusing. Measure the Voltage and clip out the cap with the lowest reading.

Good luck

G²

[JW]

On Fri, 07 Dec 2012 19:57:32 -0800 Jeff Liebermann <je...@cruzio.com>
wrote in Message id: <eod5c8t4341tapimi...@4ax.com>:

>I tried an ohmmeter, which showed 1-2 ohms across all the capacitors.
>That's not much help unless I want to replace all the capacitors.
>Next, I tried an ESR meter, which showed about 0.5 ohms everywhere.
>The problem here is that it can't distinguish between a proper ESR and
>a dead short. I tried a capacitance meter, which declared that the
>impedance was too low to produce a usable result.

When I'm tracking a short I use a Keithley 2000 in four wire mode. Even a
5.5 digit DMM such as an HP 3478A should work pretty good.

[amdx]

Hi Jeff,
Back in Feb. Malcolm Moore posted a circuit for finding shorts.
It used a 741 and had a low ohm trace used as a resistor to compare your
unknown to. It had an audio osc. that changed frequency as you got
closer to your short.
The original thread was on sed.
https://groups.google.com/forum/?fromgroups=#!topic/sci.electronics.design/lbf6e1n2Nok

You might be able to glean something from that link.

The circuit was posted on abse.
My search didn't locate it, but it is there somewhere

Mikek