As for metal film resistors - just the fact that they
exist on the board
makes them unreliable and yield killing. Old saying - the most reliable
parts are the ones that are not there.
That has to rate as one of the most stupid thigns I have ever read.. It
is wrong on so many levels...
I cna think of many times when removing components reduces reliability.
Setting aisde the 'sillies' -- like removing a zener diode so the PSU
output goes sky-high and ruins all the ICs, consider compoents designed
to protect the device from transients on input lines. Without them, the
device might well work fine for many months, then ther'es an
inductively-coupled spike on one of the cables and bye-bye expensive IC.
Alternatively, you may add extra compopnents so that they all work within
their design ratings. I will accept that 2 resistors in series are less
reliable than a single reissotr. But the startup resisotr in most SMPSUs
does consist of a 2 components in seires. They're often around 100k a
time. This is not for power disipation reasons, it's because there is a
maximum voltage tyhat resisotrs will withstand. And it would be exceeded
by the votlage acrss the startup rsistor. I gues there are speical HV
reisusotrs, but they are not common and rarely used here.
In fact a logical corrollery of minimising the component count would be
to remoove all crowbar circuits. After all, they do nothing most of the
time, and they can malfucntion and trip in error, But I am sure glad
they're there when needed.
Or remocving other safety circuits. My laser printer has a fairly complex
circuit to prevent the fuser from overheating if soemthing fails inthe
control system. In fact, it's malfunctioning at the momnet, and in that
sense it's unrelaible, but I am certainly not going to improve
reliability by permanently disabling it.
The other heing is that not all components are equally reliable. I think
we would all agree that thigns with mechanical contacts --connectors,
switches and relays are a lot less reliable than semiconductors. And that
electrolytic capacitors, for all I don't beleive the witch-hunt normally
directed against them, are lses reliable thean polystyrene capacitors.
Now, my expeirenice, gained form working on many thousands of devices
over the last 40 years is that complex ICs are noticeably less reliable
than simpler ones. Microprocessors and microcontrollers do fail more
often than 555s. And metal film reissotrs hardly ever fail.
-tony