Mac Mini A1176 shorted capacitor repair

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.

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

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

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

Murphy's Law proven once more ;-)

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

Jeff Liebermann wrote in news: snipped-for-privacy@4ax.com:

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

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.

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."

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.

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

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

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.

William Sommerwerck Inscribed thus:

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.

William Sommerwerck Inscribed thus:

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

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.

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

I've had to hunt down shorted ceramics. I use a Fluke 8060A that can resolv e down to 0.01 Ohm. It's tedious to check all the caps but you will find th at the resistance is the power traces is greater than the shorted cap. Firs t touch the probes together and press 'relative' to subtract out the lead r esistance and start measuring caps. when you get to the lowest one, clip i t 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 heati ng and possible trace fusing. Measure the Voltage and clip out the cap with the lowest reading.

Good luck

G²

When I'm tracking a short I use a Keithley 2000 in four wire mode. Even a

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.

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