A Photography forum. PhotoBanter.com

If this is your first visit, be sure to check out the FAQ by clicking the link above. You may have to register before you can post: click the register link above to proceed. To start viewing messages, select the forum that you want to visit from the selection below.

Go Back   Home » PhotoBanter.com forum » Digital Photography » Digital Photography
Site Map Home Register Authors List Search Today's Posts Mark Forums Read Web Partners

Nikon D70 and lens selection (DX vs. others)



 
 
Thread Tools Display Modes
  #1  
Old June 23rd 04, 10:23 PM
Bob
external usenet poster
 
Posts: n/a
Default Nikon D70 and lens selection (DX vs. others)

On Tue, 22 Jun 2004 20:55:05 -0500, "Be_On_Neon" wrote:

larger sensor = more noise. don't get hung up on sensor size.


I think you have that backwards...

  #2  
Old June 24th 04, 04:22 AM
Bob
external usenet poster
 
Posts: n/a
Default Nikon D70 and lens selection (DX vs. others)

On Wed, 23 Jun 2004 18:33:54 +0000 (UTC), Paul Wylie
wrote:

Bob wrote:
Interesting.... I guess it's not like a Pentium chip, you can't have any
errors... I understand that CPUs wire out the defects, and have tons of
repeated circuits... Maybe someone needs to make a different style
sensor that can have faults that can be replaced.


I don't know where you're getting that information,


I'm not sure where I read it, and I don't know if it was specifically for
Pentium's, but I did read about CPU technology where the chip has redundant
sections and the ones that work are connected by the wiring machine attaching
the welded leads. It was no doubt from a magazine at work, we get some kind of
technology news. I'm on vacation so I can't go look for it!

but most general
purpose CPUs are not fault-tolerant. There's a lot of research in that
direction (HP was a big leader for a while), but the off-the-shelf
Pentiums from Intel are CPUs that made it through the defect-detection
gauntlet. Intel and other manufacturers have also been known to speed
rate their parts in QA testing and then sell those parts according to the
speeds they'll reliably achieve.

Intel's economies of scale come from using large wafers, the smallest die
sizes they can achieve, and high volume. They still throw out a lot of
chips, though.

--Paul
** Note "removemunged" in email address and remove to reply. **


  #3  
Old June 24th 04, 04:28 AM
Bob
external usenet poster
 
Posts: n/a
Default Nikon D70 and lens selection (DX vs. others)

On Wed, 23 Jun 2004 23:49:35 -0400, paul wrote:



In memories you can build in redundancy, but how you would do that in an
optical sensor I would not know.


I guess you could design a pattern where not all of the 'pixels' are used, or
more then the required amount are available, and just use good ones. Of course
that would mean a 10m chip would be a 6m chip or something, and I guess people
would rather take chances with 10m!!

And I think you would not except a not working pixel in your sensor.
well maybe that is not true...


I returned my first D70 because it had a permanent 'spot'. I don't know exactly
what it was.

Question does anybody know how many not working pixels are allowed
in a camera? I assumed none and I hope they are all working in mine... I
guess they can hide them with some software which would allow for higher
yields and lower prices, anyone?


I had a hot pixel in a Canon movie camera - totally ruined the video. It looked
like a blue firefly flying around with the anti-shake turned on!

Paul


  #4  
Old June 24th 04, 04:49 AM
paul
external usenet poster
 
Posts: n/a
Default Nikon D70 and lens selection (DX vs. others)

On Tue, 22 Jun 2004 20:04:11 -0400, Bob wrote:

On Tue, 22 Jun 2004 21:58:47 -0400, paul wrote:

On Tue, 22 Jun 2004 01:11:13 -0400, Bob wrote:


Although they are cheaper and lighter, you don't really want to invest too
many DX lens, unless you believe we'll be stuck in 1.5x crop factor. You
can't use them on film cameras. If there are breakthrough in sensor size
development, the DX lens would be worthless.


Isn't someone already making a larger sensor? I'm sure I heard about one that is
about 95% film size... it's only a matter of time anyway...


As long as the sensors will be made out of silicon wafers the film sized
sensors will be expensive. A 24*36mm piece of silicon will always be much
more costly than the a 19*30mm.

A quick calculation with the following assumptions:
defects/cm2 = 0.3
cost 8" wafer = $1500 (I do not know the price of a processed wafer, but
it would not be to far away from this)

24*36mm = $750 note: 2 good dies per wafer!!!
19*30mm = $167 note: 9 good dies per wafer

if the process improves the cost will go down
say for a good fab the defects per cm2 is 0.2 the cost price (without
package) would go down:

24*36mm = $250 note: 6 good dies per wafer!!!
19*30mm = $88 note:17 good dies per wafer

I do not know the details of the processes used for the sensors, but the I
do know silicon fabs. And the number of defects will be closer to 0.3/cm2
than to 0.2/cm2

In short the sensor for with factor 1.5 will be between 3 and 5 times
cheaper than a full size sensor!

Paul


Interesting.... I guess it's not like a Pentium chip, you can't have any
errors... I understand that CPUs wire out the defects, and have tons of repeated
circuits... Maybe someone needs to make a different style sensor that can have
faults that can be replaced.


In memories you can build in redundancy, but how you would do that in an
optical sensor I would not know.

And I think you would not except a not working pixel in your sensor.
well maybe that is not true...
Question does anybody know how many not working pixels are allowed
in a camera? I assumed none and I hope they are all working in mine... I
guess they can hide them with some software which would allow for higher
yields and lower prices, anyone?

Paul
  #5  
Old June 24th 04, 04:52 AM
paul
external usenet poster
 
Posts: n/a
Default Nikon D70 and lens selection (DX vs. others)

On Wed, 23 Jun 2004 18:33:54 +0000, Paul Wylie wrote:

Bob wrote:
Interesting.... I guess it's not like a Pentium chip, you can't have any
errors... I understand that CPUs wire out the defects, and have tons of
repeated circuits... Maybe someone needs to make a different style
sensor that can have faults that can be replaced.


I don't know where you're getting that information, but most general
purpose CPUs are not fault-tolerant. There's a lot of research in that
direction (HP was a big leader for a while), but the off-the-shelf
Pentiums from Intel are CPUs that made it through the defect-detection
gauntlet. Intel and other manufacturers have also been known to speed
rate their parts in QA testing and then sell those parts according to the
speeds they'll reliably achieve.

Intel's economies of scale come from using large wafers, the smallest die
sizes they can achieve, and high volume. They still throw out a lot of
chips, though.

correct and they use the newest technology to get more transistors per
square mm. But your pixel is limited in minimum size there it has to
capture a minimum amount of light.

Paul
  #6  
Old June 24th 04, 05:56 AM
Paul Howland
external usenet poster
 
Posts: n/a
Default Nikon D70 and lens selection (DX vs. others)

Bob wrote:
On Wed, 23 Jun 2004 18:33:54 +0000 (UTC), Paul Wylie
wrote:
Bob wrote:

Interesting.... I guess it's not like a Pentium chip, you can't have any
errors... I understand that CPUs wire out the defects, and have tons of
repeated circuits... Maybe someone needs to make a different style
sensor that can have faults that can be replaced.


I don't know where you're getting that information,



I'm not sure where I read it, and I don't know if it was specifically for
Pentium's, but I did read about CPU technology where the chip has redundant
sections and the ones that work are connected by the wiring machine attaching
the welded leads. It was no doubt from a magazine at work, we get some kind of
technology news. I'm on vacation so I can't go look for it!


I think this is all a red herring. With processors you can get away
with redundancy, backup circuits, etc, because it doesn't matter too
much *where* on the die the actual processing occurs. I can't see that
any of this discussion applies to a CCD however. If a pixel in the
middle of my image is defective, it doesn't help me too much by
switching to a redundant pixel on the outside of my sensor! In other
words, the position of the circuits in a CCD is critical, so you can't
use redundancy.
  #7  
Old June 25th 04, 09:45 AM
external usenet poster
 
Posts: n/a
Default Nikon D70 and lens selection (DX vs. others)

paul wrote:
On Tue, 22 Jun 2004 01:11:13 -0400, Bob wrote:



Although they are cheaper and lighter, you don't really want to invest too
many DX lens, unless you believe we'll be stuck in 1.5x crop factor. You
can't use them on film cameras. If there are breakthrough in sensor size
development, the DX lens would be worthless.


Isn't someone already making a larger sensor? I'm sure I heard about one that is
about 95% film size... it's only a matter of time anyway...


As long as the sensors will be made out of silicon wafers the film sized
sensors will be expensive. A 24*36mm piece of silicon will always be much
more costly than the a 19*30mm.


A quick calculation with the following assumptions:


You seem to be assuming that a defect will render a sensor useless: it
might, or it might not. If a defect takes out a whole line, then the
sensor is toast, of course. But if a defect only spans a pixel or
two, the sensor is still usable.

Andrew.
  #8  
Old June 25th 04, 08:14 PM
David J Taylor
external usenet poster
 
Posts: n/a
Default Nikon D70 and lens selection (DX vs. others)

"paul" wrote in message
news []
Yes you are correct. Any idea how many defective pixels are allowed by
different manufacturers? And how do I recognize them in my camera?

Paul


Paul,

The defective pixels are usually hidden by interpolation in the firmware.
I was slightly horrified to find that my Nikon Coolpix 5700 had at least
512 defective pixels (according to its internal table) but visibly the
image is perfect.

A zero-defect chio might be required for some high-end broadcasting
applications, and will have a commensurate cost!

Cheers,
David


  #9  
Old June 25th 04, 08:22 PM
Phil Wheeler
external usenet poster
 
Posts: n/a
Default Nikon D70 and lens selection (DX vs. others)



paul wrote:


Yes you are correct. Any idea how many defective pixels are allowed by
different manufacturers? And how do I recognize them in my camera?



There is a free program called DeadPixelTest that will analyze images
and tell you how many pixels are hot or dead, based on luminance levels
you can set.

http://www.starzen.com/imaging/deadpixeltest.htm

My 300D shows none at 2 seconds, 1 at 4 seconds and 13 or so at 30
seconds. So what it is seeing is noise, basically.

My original 300D had a hot pixel at 1/500 sec exposure -- truly a focal
plan defect -- and I had it replaced.

Phil

  #10  
Old June 26th 04, 01:26 AM
Roger Halstead
external usenet poster
 
Posts: n/a
Default Nikon D70 and lens selection (DX vs. others)

On Thu, 24 Jun 2004 06:56:39 +0200, Paul Howland
wrote:

Bob wrote:
On Wed, 23 Jun 2004 18:33:54 +0000 (UTC), Paul Wylie
wrote:
Bob wrote:

snip

I think this is all a red herring. With processors you can get away
with redundancy, backup circuits, etc, because it doesn't matter too
much *where* on the die the actual processing occurs. I can't see that


But the CPU layout is very complex and does have a geography
That doesn't leave much flexibility for fault tolerance and even then
it would be at the manufacturing level. With the simple CCD pixels
that fail out-in-the-field can often be mapped out.

any of this discussion applies to a CCD however. If a pixel in the
middle of my image is defective, it doesn't help me too much by


If a pixel in the middle of your image is bad, and most likely there
are a few bad (hot or dead) pixels in most any CCD the firm ware will
probably average it out with adjacent pixels and you'll never know
it's bad.

switching to a redundant pixel on the outside of my sensor! In other
words, the position of the circuits in a CCD is critical, so you can't
use redundancy.


No, you just block that one out, something you can't do in a CPU.
Also take into account even the most powerful CPUs for PCs are
considerably smaller than the CCDs. My 2.8 Gig Athlon XP Plus is just
over one cm square. Those CPUs also have many times the number of
discrete items compared to a CCD.

Soooo, although the number of transistors, diodes, resistors,
capacitors, and inductors in a CPU is probably an order of magnitude
or two greater than the CCD, which reduces the yield, the small size
and greater number of devices produced on a wafer increases the yield.

All yield figures I have are so old as to no longer be valid. Back
then we hadn't even been able to produce 6 inch wafers let alone the
10 inch which I think is pretty much standard now.

The larger wafers and improved manufacturing techniques have greatly
increased the yields so the cost of devices of a specific size is
going down.

When it comes to the industry moving to larger sensors There are
three questions to ask.

1. Do we need larger sensors?
2. Is there any thing to be gained by going larger?
a. If so, is it worthwhile?
3. Will the consumer demand larger sensors even though most don't
need them and only a tiny percent would ever make use of their
capabilities.

Roger Halstead (K8RI & ARRL life member)
(N833R, S# CD-2 Worlds oldest Debonair)
www.rogerhalstead.com
 




Thread Tools
Display Modes

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

vB code is On
Smilies are On
[IMG] code is Off
HTML code is Off
Forum Jump


All times are GMT +1. The time now is 12:08 AM.


Powered by vBulletin® Version 3.6.4
Copyright ©2000 - 2024, Jelsoft Enterprises Ltd.
Copyright ©2004-2024 PhotoBanter.com.
The comments are property of their posters.