Question about DLP

Archived from groups: alt.tv.tech.hdtv (More info?)

"Nicholas" <none@nowhere.com> wrote in message
news:S8Aed.9251$Lk3.8570@newssvr12.news.prodigy.com...
> Has anyone ever had the color wheel assembly fail on their DLP television?
>
> Anyone know what the MTBF is supposed to be?
>
> Thanks,
> Nicholas.
>
Samsung had a few early color wheel assm fail because of their bearing
design. They make a horrible racket that sounds like bad bearings in a
computer Hard Drive. Even with the racket they still worked properly. I've
seen samsung DLP sets with over 3,000 hours and no color wheel failure, or
lamp failure for that matter. Still too early to say anything else as the
3000 hour sets were the longest runs I've seen. The DLP sets have only been
out a couple of years.

I'm naturally distrustful of anything mechanical and expected more failures
than I have seen. As every month passes and I get fewer and fewer calls on
older DLP sets I think they might have a very long MTBF.

Archived from groups: alt.tv.tech.hdtv (More info?)

Terry Joyce wrote:

> Nicholas wrote:
>
>> Has anyone ever had the color wheel assembly fail on their DLP
>> television?
>>
>> Anyone know what the MTBF is supposed to be?
>>
>> Thanks,
>> Nicholas.
>>
> Mean Time Between Failure
Sorry, Reread and saw the "the"

Archived from groups: alt.tv.tech.hdtv (More info?)

dg wrote:

> Sometimes I have to wonder about the MTBF listed on consumer products. For
> example, the Western Digital Raptor hard drives list the MTBF as 1.2 million
> hours. Correct me if I am wrong, but 1,200,000 / 24 /365 is 136.98 years!!
> How can they possibly come up with such an incredible number? That is an
> absolutely UNREAL figure. That would mean that for every drive that is DOA,
> there is another drive out there that lives to be 273 years old, without
> failure.
>
> Kind of makes their 5 year warranty sound pathetic (but is actually the
> longest warranty in the industry).
>
> In conclusion, MTBF is for all practical purposes, not a good figure to base
> any thoughts on. I don't know why they would even claim such a number.
>

MBTF is really a way of expressing failure rate. It actually has meaning
for companies that have 1000s of drives (and there are many of these).
The figure quoted is for drives under perfect conditions that have never
been mishandled or driven outside the specified operating parameters.
Most DOAs never leave the factory, so very few are factored into that
figure.

In addition, the definition of "failure" is very strict. If a drive
recovers from a fault after being power cycled, then, for some
manufacturers, it has not "failed" for the purposes of this calculation.
The failure of multiple attempts to write the same data in the same
place is required before the drive is considered to have failed. This
allows them to remap the bad sectors on the fly so they don't have do
deal with warranty claims. Sometimes rezeroing the drive is required
after some number of retries, too. There are similar specifications for
read retries.

At 1,200,000 hours you would expect at least 7.3 drives per thousand to
fail in any one year. The industry (at least the part I have worked in)
believes that that 1% of all drives in service will fail in any year.
This takes into account that not all handling is perfect and drives are
sometimes abused for some period of time. That is 10 of every thousand.

Similarly, 14.6% of all devices with a MBTF figure of 60000 hours are
expected to fail every year.

Matthew

Archived from groups: alt.tv.tech.hdtv (More info?)

dg (dan_gus@hotmail.com) wrote:

: In conclusion, MTBF is for all practical purposes, not a good figure to base
: any thoughts on. I don't know why they would even claim such a number.

You aren't looking at it correctly. The MTBF doesn't add up per device, it
is used to determine the failure rate for a group of them.

If one drive has a MTBF of 1,200,000 hours, if you used 10 of them in a
similar fashion, you should get a failure in 120,000 hours instead.

Think of it as if you owned a large company and needed to buy 1000 of them
for the employee workstations, this cuts the MTBF to just 1200 hours or one
failure every 41 days.

If you were a retailer like Sears or Best Buy needed to purchase 3 or 4
thousand of them, that would be close to one going bad every couple weeks.

In other words, don't look at MTBF on a single unit, figure it out on a
group of them.

-bruce
bje@ripco.com

Archived from groups: alt.tv.tech.hdtv (More info?)

The more I think about it, the more I wonder. According to the info you
have given me (and I thank you, not trying to piss on your answer), if I
bought 1000 drives I could expect 1 failure every 1200 hours, if I bought 10
drives I could expect 1 failure every 120,000 hours. Continuing that math,
if I bought 1 drive I could expect a failure every 1,200,000 hours-right
back to where we started. At some point between buying 1,200,000 drives and
1 drive, the line is crossed between "realistic" and "unreal". Where is
that line?

What you said kind of makes sense, but still it seems like there should be
an asterisk next to the MTBF listed on device specs that says exactly how
many devices they are basing that number on-because as we have shown here,
one can interpret the numbers with the same method but using a different
sample size and come up VERY different results.

Still not confident in MTBF ratings on products.
--Dan

"Bruce Esquibel" <bje@ripco.com> wrote in message
news:clr36p$eh$1@e250.ripco.com...
> You aren't looking at it correctly. The MTBF doesn't add up per device, it
> is used to determine the failure rate for a group of them.
>
> If one drive has a MTBF of 1,200,000 hours, if you used 10 of them in a
> similar fashion, you should get a failure in 120,000 hours instead.
>
> Think of it as if you owned a large company and needed to buy 1000 of them
> for the employee workstations, this cuts the MTBF to just 1200 hours or
one
> failure every 41 days.
>
> If you were a retailer like Sears or Best Buy needed to purchase 3 or 4
> thousand of them, that would be close to one going bad every couple weeks.
>
> In other words, don't look at MTBF on a single unit, figure it out on a
> group of them.
>
> -bruce
> bje@ripco.com
>

Archived from groups: alt.tv.tech.hdtv (More info?)

"Matthew L. Martin" <nothere@notnow.never> wrote in message
news:10o2lnk2uo0250d@corp.supernews.com...
> The point is that the value of the MTBF figure is very dependent on the
> sample size being measured. That is what is causing your quandary. A
> sample size of one is insufficient for making a prediction of MTBF. A
> sample size of 10,000 is more than enough to get two digits of accuracy
> in an MTBF figure. You have observed that in your post above, but missed
> the bit on this being a statistical projection, not a statement of
> product life.

Thank you for helping me understand MTBF.

Thanks,
--Dan

Archived from groups: alt.tv.tech.hdtv (More info?)

But it does make SOME sense. If Dell is building 100,000 computers they can
estimate repair costs by looking at the MTBF of the PC components and pretty
accurately estimate their rate of replacement and then multiply by how much
it costs to replace each part. But for me, just an end user who wants a
quality product, the MTBF is totally useless.

--Dan

<graph this> wrote in message
news:82t4o0ds91tjlbqllna5fdgu739pjef7h6@4ax.com...
> MTBF is just like the six sigma dog and pony show your Quality
> department puts on for the customers.
>
> Looks so impressive.
>
>

Archived from groups: alt.tv.tech.hdtv (More info?)

I have a Samsung HLN5065W in which the color wheel failed after 13 months.
Samsung graciously agreed to repair out of warranty. Also, I had a
horizontal sync issue (double images when switching channels). Samsung
replaced main board and audio / video board.

"Nicholas" <none@nowhere.com> wrote in message
news:S8Aed.9251$Lk3.8570@newssvr12.news.prodigy.com...
> Has anyone ever had the color wheel assembly fail on their DLP television?
>
> Anyone know what the MTBF is supposed to be?
>
> Thanks,
> Nicholas.
>