M. Zuiko 9-18 Sample Variation?

Started Mar 11, 2013 | Discussions thread
Beach Bum Senior Member • Posts: 1,055
Double sigh...

tedolf wrote:

jtan163 wrote:

tedolf wrote:

FrankS009 wrote:

Ignorance is not bliss. To think that lens sample variation is just an urban legend is wishful thinking I fear.

I have just never seen any objective evidence of it-just a lot of subjective talk.

Show me some test charts done under quasi-scientific conditions and I will believe it.

Modern quality controlis good but not foolproof.

It is very good.

When a plane falls out of the sky and crashes, that is often an example of modern quality control failing.

No, it is usually pilot error.

Someone gets distracted at some critical juncture in a process (they may not even realise they have been distracted).

Not with automated equipment.

Given that lenses are mass produced it seems almost inevitable that there will be some variation from batch to batch, if not within batch - though I believe that variation will occur within batches too.

Actually the opposite is true.

Mass production leads to less variation within batches.

Conditions in the factory vary as do conditions in the component supplier's factories and their suppliers and so on.

Not in modern factories.

Conditions are very closely controled.

Then of course there is human error, laziness, maliciousnous etc.

Automation eliminates most human error.

Happens all the time.

Happens very rarely.

It is appearent that you have never worked in a factory.

It's one of the reasons we get recalls on some batches of products, because one day/week/month something in the production process was not right and the products came out defective.

Wrong again, most recals are based on design flaws, not manufacturing defects.

The second most frequent cause of recals is materials defects.

Manufacturing variance is way down on the list.

It's also part of the reason we have warranties - because some times individual copies of a product fail for whatever reason e.g. an insect flew in front of a sensor causing a measurement to be out - that's a gross and simplistic example scenario, but you get the idea (I hope) it is the kind of thing that can happen.

Yeah, so?

Not sure what you would consider quasi scientific, or indeed scientific conditions, but Roger Cicala at Len Rentals often talks about and charts sample variation on their blog.

I have read his stuff, he seems to have an agenda to me.

From other blog posts that Roger has written I believe they use imatest to test the lenses on a fairly regular basis (I think whenever they are returned or sent out - one or the other) and I believe that they tend to have reasonably decent sample sizes, e.g. 10s or more lenses of a single model in many cases and sometimes lenses from different batches.

Here's a couple of posts including data that I believe was collected in a reasonably rigorous manner (their business depends on it to some extent) by people who deal with lenses, lens maintenance, repair and calibration for a living.


This chart is taken from the last of the above URLs - it shows the results of an IMATEST run on 100 copies of the Canon 25-70mm.

excpt the chart doesn't show if that variance is out of acceptable tolerance.

Without that information the chart is worthless.

By my eye the range of values scored are something like 345-665. That's some variation.

Well, if you don't know what the tolerances are it is meaningless.

I believe (from other posts I've read) the vertical axis is the average value across the entire surface of the lens and the horizontal is the highest single value recorded for that example.

Again, what is the manufacturing/design tolerance?

Quarter wavelenght?

You wont fiind that infromation on the website and that is why I think the work is suspect.


What about camera failure rates or electronic failures in general? It's well known that about 3% or more of cameras fail within short order due to manufacturing issues. This has nothing to do with the consumer or the care given to the camera. These cameras would fail in anyone's hands. And these cameras more likely than not passed all of the manufacturers' tests.

If this is the case with complete camera failures, it stands to reason that there would also be a fraction of cameras or lenses that are optically duds.

Also, manufacturer/design tolerances are often applied to individual components within the camera or lens. Having hundreds or thousands of components in a camera or lens, it's entirely possible that some cameras would get a certain combination of components, that while all of them individually are within spec, cause the camera as a whole to perform poorly. It's just the luck of the draw.

A manufacturer could make the tolerances so tight that it would be virtually impossible for a camera to be a poor performer, but this is often prohibitively expensive. So, usually the specs are loose enough that a certain percentage of the completed product will be duds.

BTW, I rarely give thumbs up but I'm going to give it to jtan163 in this case because he's clearly more on the ball and the unwashed masses are giving out the likes to the wrong side, as usual. I honestly wish they would do away with the thumbs up system.

Post (hide subjects) Posted by
Keyboard shortcuts:
FForum PPrevious NNext WNext unread UUpvote SSubscribe RReply QQuote BBookmark MMy threads
Color scheme? Blue / Yellow