There can be few among those of us who produce printed circuit boards, who have not at some point placed a component the wrong way round, or with the wrong footprint. Usually this can be rectified with a bit of rework and a fresh board spin, but just occasionally these mishaps make it into the wild undetected. It seems nobody is immune, as [Doug Brown] is here to tell us with a tale of an Apple product with a misplaced capacitor.
The LC series of Macs came out through the early 1990s, and their pizza-box style cases could be found slowly turning yellow in universities and schools throughout that decade. Of them there was a persistent rumor of the LCIII had a misplaced capacitor, so when he received an unmodified original machine he took a look. The investigation is quite simple, but revealing — there are three power supply rails and one of the capacitors does have a significant leak.
The explanation is simple enough, the designer had placed a capacitor on each rail, with its negative side to the ground plane, but one of the rails delivers -5 volts. Thus the capacitor is the wrong way round, and must have failed pretty early in the lifetime of each LCIII. We’re curious then since so many of them went through their lives without the component being replaced, how the circuit remained functional. We’re guessing that there were enough other capacitors in the -5 volt line to provide enough smoothing.
Ah, the joys of a cap designed in with polarity reversed. In the late 80’s I worked at SMC [Standard Micro Systems], and since they produced a chip that implemented the ArcNet protocol, they also made an interface card for the IBM PC. The circuit needed a one-shot chip to stretch the system reset pulse to the width that the ArcNet chip needed, and in tern, the one-shot needed an electrolytic cap to produce the pulse. These boards would fail after a year or so, and the cap was the cause. The design engineer swore up and down that the circuit was correct, and there was no reason for the cap to fail prematurely – he had copied the circuit directly from the one-shot’s spec sheet. Looking at the example circuit and the internal design of the chip, it was discovered that the negative side of the cap was connected to an external timing resistor that was connected to +5V. He had blindly followed the sample design, and ignored the obvious weirdness of the circuit. A +5, current, even thru a large resistor, with the positive of the cap connected thru an internal ground inside the one-shot chip, caused the cap to slowly fail.
Hmm, the world is small, I think I have two of those cards… :) They worked when I last used them in my home network setup at the end of the ’90s. But I only had 2 cards, got more computers, and replaced them with ethernet cards quite soon after. I wonder if they would still work.
This is why you should strive to always make the wrong configuration look wrong and the correct one look right.
Somebody saw three capacitors in a row with one inverted and blindly changed it assuming that because it looked wrong it must be wrong, without actually checking the circuit.
Or they just made a mistake and placed it mis-oriented without it even registering.
No, the cap that was leaking wasn’t the backwards one. Read the article: C22 is the -5V cap, C21 was the leaking one. You can see the trace going straight to the -5V pin. The leaking cap was just a coincidence.
Hooray for bis.., uh, bipolar condensers! 😁🥳
If you follow Louis Rossman then this would feel more like a common occurrence than a rarity.
I’ll say it’s a feature. Based on Louis findings in many videos, I am almost sure that it was done with purpose so it will die sooner, or it will die if you go to an independent repair shop for a fix (video cable having over 30 volts for backlight next to the pin with the video signal, put the ribbon slighty off and off with the screen).
And the fan buyers still throwing money at the fruit company.
I found a similar thing by Tektronix.
The TDS500/600/700 series oscilloscopes have an ‘Option 05’ that provides TV signal triggering functions and involves an additional PCB installed inside the scope.
On one the -ve rail, there is a capacitor that is marked backwards on the silkscreen, I bet due to similar reasons as this Apple computer.
Luckily it seems Tek must have realised somewhere between getting the boards made and assembly, and the capacitor is actually populated ‘backwards’ according to the silkscreen, so in reality it is actually the correct way around in the end.
A bit of a trap when recapping the board though if you just look at the silkscreen…
Interestingly, this backwards cap has persisted across multiple board revisions.. Go figure….
However, in the same scopes, they appear to have underrated a cap on the front control panel. IIRC they used a 10V electrolytic, but it is across what measures as up to 15V. Surprisingly that cap doesn’t seem to fail sooner that the rest, at least in my limited current collection of about 15 scopes.
I replace the caps regardless.
The Tek AWG arb func gens of the same era also have a cap on the front panel that is ALWAYS leaking. It’s only used to make the beeper sound more like a nice chime by prolonging and ‘fading out’ the input signal. Maybe there is some reverse polarity going on or it’s just a cheap part?
In any case, it seems to be a little bit of vanity that ends up potentially killing the instrument….
“Interestingly, this backwards cap has persisted across multiple board revisions”
If the machine was already notorious among technicians for having C33 marked backwards, then there might have been a case for keeping it wrong on purpose. rather than adding more complication.
There’s lots of cases like that. Notably, when people found out electrons were negatively charged, they didn’t revise the definitions of units etc.; everyone just had to learn (for the rest of time) that electrons move in the opposite direction to current.
But then, electrons being negatively charged is an arbitrary choice too.
I mean, it’s the same choice. Electrons are “negative” in terms of the units that happened to already be in use in black & white times.
Like, there were papers saying the half-cell potential of copper(II) was 0.33V and what have you; if the coin toss had gone the other way, and they had called it -0.33V, then Thomson would have found electrons to be “positive”.
“they didn’t revise the definitions of units”
They did! Well, kind of. Siemens in Germany was known for doing this (afaik).
Their schematics followed the physical current direction, rather than the technical one.
So ground was marked positive..
Also, we learned in school (physics) about both directions in 5th class already.
And how earlier researchers had messed up big times here. ;)
I found a similar thing by Tektronix.
The TDS500/600/700 series oscilloscopes have an ‘Option 05’ that provides TV signal triggering functions and involves an additional PCB installed inside the scope.
On one the -ve rail, there is a capacitor that is marked backwards on the silkscreen, I bet due to similar reasons as this Apple computer.
Luckily it seems Tek must have realised somewhere between getting the boards made and assembly, and the capacitor is actually populated ‘backwards’ according to the silkscreen, so in reality it is actually the correct way around in the end.
A bit of a trap when recapping the board though if you just look at the silkscreen…
Interestingly, this backwards cap has persisted across multiple board revisions.. Go figure….
However, in the same scopes, they appear to have underrated a cap on the front control panel. IIRC they used a 10V electrolytic, but it is across what measures as up to 15V. Surprisingly that cap doesn’t seem to fail sooner that the rest, at least in my limited current collection of about 15 scopes.
I replace the caps regardless.
The Tek AWG arb func gens of the same era also have a cap on the front panel that is ALWAYS leaking. It’s only used to make the beeper sound more like a nice chime by prolonging and ‘fading out’ the input signal. Maybe there is some reverse polarity going on or it’s just a cheap part?
In any case, it seems to be a little bit of vanity that ends up potentially killing the instrument….