Re: Suggestions please for amplifier and speakers

In article <NB2Ui.94028$lV4.32186@xxxxxxxxxxxxxxxxxxxxxxxxx>, Ian Iveson
<IanIveson.home@xxxxxxxxxxxxxxxx> wrote:
Jim Lesurf said:

Have you come across the 'bathtub curve' of reliability?
Manufacturing engineers know it well. :-)

Upside-down bathtub, presumably, if you're talking about reliability
rather than expected failure.

I was thinking of failure rate as that is what you can measure at
each stage as the population thins out. The 'reliability' would be an
integral of one minus this to show a lifetime reliability
distribution.

Right, I guess. Just about anything that turns the bathtub upside down
would be an indication of expected reliability, in the general sense of
the word, i.e. "not prone to failure these days". If you want a measure
of lifetime reliability then I suppose you integrate.

The problem when comparing "new with secondhand" is where you are starting
from.

1) If the item is already X years old, you have to start with the
statistics for that age (assuming anyone knows them!). Since the item is
'selected' by having survived the first X years these may be different to
the total population.

All else being the same, your chance of living to 70 years old is *higher*
if you are 65 than if you are 15. So much for 15 year olds being fitter.
8-]

2) If the items are 'new', how do you know the curves you have are correct
for *this* batch of items, and reliable for the *one* in front of you?
Maybe these have been soldered poorly. e.g. someone handling the boards had
been eating oranges for lunch, and this means the soldered joints on some
of these particular boards will decay. (Nope, I didn't make that one up.
Has happened in my experience.)

Good manufacturers presumably shake and burn their products to remove
the extremity at the left of the bathtub, so the risk of buying new
shouldn't be too much greater than at the prime of life.

You can test sets and run them for a time. However the front end of the
curve may be many days/weeks long, so it become impractical to run them for
that long. And any "shake and burn" might actually introduce problems which
cause a temporary rise in the risk of failure in the near future - i.e. you
are building in more chance of an early failure just after you stop
testing.

Also, whatever you do, some customers are going to do something else. 8-]

e.g. here being a customer who connected mains via the loudspeaker socket.
(Having carefully made up a lead with a speaker plug on the end. Yup, I've
encountered this one as well. ;-> )

Or more likely, they use the speakers from hell with the most demanding
waveforms of weirdo music you've never encountered. Played loud, of course.

The right slope of the bathtub has always perplexed me. As with that of
the "product life cycle". Is there a formal rule of inevitable demise?
Can nothing last forever? I guess that's a big problem for religious
statisticians.

Well I suppose "nothing" may well last... :-)

You may recall the ads, "Nothing works faster than Anadin".

....to which people in my house always used to chorus, "... so use
'nothing'!" Cheaper, as well as faster. 8-]

There are some well know deterioration mechanisms for some components.
Things like doping drag in the base region of bipolars or the equivalents
in valves. These rarely produce problems early on unless there is some
other problem, or the device is being seriously over-driven. But they will
get the device eventually. Although remember that units that have worked
for ages may then have this right-end further away than a new set of units
as the aged devices are now 'selected' by not yet having failed.



You must assume that the item has not already failed, of course.

The curves would be statistical, done on a population of devices. Any
predictions are therefore statistical estimates unless you are
dealing with specific wear or degradation factors.

I was thinking simply that any service or overhaul has its own bathtub.
Hence be wary of "recent overhaul" or "rebuilt as new", which add risk
but also offer some chance of a better outcome.

Indeed. As with demanding tests to 'soak and rattle' which may cause
problems which not bothering would have avoided. There is always a risk
that a repair may have introduced a new problem to the unit. This is why I
tend to be of the "Tain't broke => Don't fix" school myself. This includes
things like electrolytics which other people feel they just have to change
automatically after N years. Check the performance. If it performs OK and
shows no problem, leave it. Not like it was going into orbit or being used
at the bottom of the ocean for years. But then when I check my own kit I'm
not needing to justfy charging anyone. Yes Sir, we had to change the
clutch. :-)

A few years ago I had an intermittent fault with odd symptoms on one of the
power amps I use. Crackles plus small changes in dc offet at the output.
Tried all the obvious suspects indicated by the symptoms. Wiggling or
freezer spray didn't show it. Even replaced the 20+ year old caps for new
just in case. ;-> But changing the caps had no effect on performance or
the problem. The caps were fine.

Turned out to be an intermittent connection in one of the bonds *inside*
the pack of one of the TO92 packed transistors in the middle of the amp.
The transistors were being used well in spec, and are of a decent
manufacture. All the others I use in the same way are fine (so far).
Replaced it. But of course for all I know my re-soldering will give
problems in future years, or the new device - despite being the same part
number - may be less reliable than the originals. But then I may have
weeded out the weak one and the others may carry the set on for decades.

I am afraid life does have risks. If someone is trerrified that what they
buy might go wrong, then don't buy anything. :-) New or second-hand, it
might go wrong. Fact of life, alas. But at least it keeps repair bods and
makers in business.

Slainte,

Jim

--
Electronics http://www.st-and.ac.uk/~www_pa/Scots_Guide/intro/electron.htm
Audio Misc http://www.audiomisc.co.uk/index.html
Armstrong Audio http://www.audiomisc.co.uk/Armstrong/armstrong.html
.

Relevant Pages

  • Re: PC antivirus software question
    ... Get and read the latest issue of Consumer Reports. ... They gather annual reliability data from ... "engineering, publication, or investigative reporting". ... Reliability talks in terms of specific failure rates, ...
    (alt.home.repair)
  • Re: PC antivirus software question
    ... Get and read the latest issue of Consumer Reports. ... They gather annual reliability data from ... "engineering, publication, or investigative reporting". ... Reliability talks in terms of specific failure rates, ...
    (alt.home.repair)
  • Re: Why is the CEV so heavy?
    ... the SRB-J2S launch vehicle ... Recall Richard Feynman's evaluation of SRB reliability in 1986, ... "An estimate of the reliability of solid rockets was made by the range ... (Since there are two rockets on the Shuttle, these rocket failure rates must ...
    (sci.space.policy)
  • Re: Thinking About Laptop Computers (yet) Again
    ... reliable that I don't expect it to fail on a trip. ... Last time I had a car fail on a trip was ... I've been involved in reliability testing for a long time ... failure rate of the PCMCIA card, the PCMCIA slot and its associated ...
    (rec.audio.pro)
  • Re: Shuttle proves being superb manned spacecraft
    ... It reduces reliability and introduces more ... How many times has a life support failure ended a mission? ...
    (sci.space.shuttle)