Electronic newbie question regarding components

[George C.]

Do I understand correctly that even 2 identical components (same brand
and part number) can perform not quite as well as the other?
In such a case, what's the way to get the 'ideal' component? Just keep
buying them and testing them to see which one gives the desired
values?

 

[Ralph Mowery]

Do I understand correctly that even 2 identical components (same brand
and part number) can perform not quite as well as the other?
In such a case, what's the way to get the 'ideal' component? Just keep
buying them and testing them to see which one gives the desired
values?

In some circuits that is correct. The only way is to test many devices.
Components are designed to meet certain minimum specifications and certain
tollorances. The circuits are designed so they work within the devices
parameters. Sometimes when looking for the maximum performance you will
have to subistute several devices to get the best performance but that is
seldom necessary. When the first big (12 foot ) dishes were being sold for
tv reception a very low noise high gain amplifer was needed. Some contained
2 transisitors with the same part number but maybe with a 1- or -2 after it.
The devices were selected for the lowest noise figuer for the first device.
I am not sure now, but not too long ago the microprocessors were tested for
the speed they could operate at without missing an instruction and were
priced at the speed they operated at. They were then marked and sold for
the speed they were reliable at. That is why you could overclock a computer
, especially if you raised the voltage on them a tenth of a volt or so. The
processor may or may not burn out sooner, but if it lasts 5 years instead of
10 or more years and you are changing the computer every 2 to 4 years it
will not make any differance.

 

[Don Taylor]

Do I understand correctly that even 2 identical components (same brand
and part number) can perform not quite as well as the other?
True

In such a case, what's the way to get the 'ideal' component? Just keep
buying them and testing them to see which one gives the desired
values?

Perhaps this analogy will help, but my analogies are always dangerous.

Think about a recipe that needs eggs in it. You go to the store
and there are stacks of eggs. Every carton has a size label on it.
But the eggs in every carton are slighly different, no two are
EXACTLY the same weight, down to the million'th of an ounce, even
if the eggs are in the same carton or two cartons have the same
label on them.

Fortunately, your recipe probably won't explode if you have an egg
that isn't exactly the right size, there is flexibility for a little
variation built into the recipe. But some recipes need more accuracy
than others, and you or they plan for this.

The same is true for most components. It is a very rare item that
comes only and exactly in exactly perfectly completely precise sizes.
Actually at the moment I cannot think of one that I could go buy.

So most products that are built from items are supposed to have some
built in flexibility for the variation. In fact many circuits that
are built are specifically designed knowing that there will be this
variation and thus compensate for it.

As you say, it certainly is possible to keep buying and testing
parts until you find the desired values. This is actually done
in some cases, for example where a circuit needs a pair of parts
that are as close to identical as possible. So your idea is used.
But lots of other cases it is easier or cheaper to be flexible.

Does that help?

 

[Matt J. McCullar]

Do I understand correctly that even 2 identical components (same brand
and part number) can perform not quite as well as the other?
In such a case, what's the way to get the 'ideal' component? Just keep
buying them and testing them to see which one gives the desired
values?

This is true. Most of the time, though, the differences are so slight as to
not matter.

I'm guessing that when we're talking about mass-produced components, such as
diodes, transistors, resistors, etc., the larger the number of your
production run, the more identical the parts will be toward one another.
But if you're making only small quantities of something, chances are the
differences will be more noticeable.

I sometimes run into power-control circuits in which output transistors have
to be "matched"; that is, you don't want one transistor carrying more of the
load than the other. I personally don't like this and consider it poor
circuit design. It shouldn't have to be that critical. What's worse,
having to order a large batch of replacement power transistors and being
able to match up only a few out of a couple of dozen runs into money. It's
especially aggravating when you see different logos on different
transistors; each manufacturer has its own special way of making parts, and
you can bet that they won't match exactly with the same part made by
somebody else.

Fortunately, there are not that many instances in which "matching" is really
critical. RF circuits can be, though, particularly when you're dealing with
a balanced modulator; if the diodes aren't matched up just right, you may or
may not be able to get it to balance.

From my experience, the best way to get "matching" parts is to order a large
batch from the same manufacturer, hopefully from the same production run.

 

[Tim Shoppa]

Do I understand correctly that even 2 identical components (same brand
and part number) can perform not quite as well as the other?
In such a case, what's the way to get the 'ideal' component? Just keep
buying them and testing them to see which one gives the desired
values?

You have to quantify what "ideal" is.

If you cannot quantify the qualities you need, then you're not in the
right business.

It's sort of like the fact that there's nobody in the world who is
exactly six feet tall.

Tim.

 

[mike]

Do I understand correctly that even 2 identical components (same brand
and part number) can perform not quite as well as the other?
In such a case, what's the way to get the 'ideal' component? Just keep
buying them and testing them to see which one gives the desired
values?

component datasheets give the tollerance spreads. I guess that you
have to just design with that in mind. If you are testing for a too narrow
performance spectrum you will probably be beaten by father time and
global warming as some , if not all material caracteristics will change
in time, temp, pressure etc.... Although it is a special case, I had
some frustrating time sorting through crystals to get frequencies close
to ideal.