Velocity Reviews - Computer Hardware Reviews

Velocity Reviews > Newsgroups > Computing > Digital Photography > Higher bit conversion in DSLRs unlikely

Reply
Thread Tools

Higher bit conversion in DSLRs unlikely

 
 
Roger N. Clark (change username to rnclark)
Guest
Posts: n/a
 
      04-09-2006
bmoag wrote:

> Power usage is the most easily solved problem as far as the processing
> cirucitry itself is concerned. Decreasing power consumption, and therefore
> heat, is the current mantra for microprocessor design. Heat from excessive
> power requirements is what killed the Pentium 4 for Intel and drove what
> little is left of Apple away from the PowerPC.
> When you go 16 bit color you generate very large files. If current dSLRs
> were to go to 16bit color there is no way to write files of that size, even
> with the fastest Compact Flash cards, that would allow more than single
> frame shooting.
> This too is surmountable.
> Is 16 bit color all that desirable?
> As wonderful as 16 bit color seems we see and print in an 8 bit (and
> actually less, considering the limited color gamuts of viewing and printing
> processes and individual variations in color vision) world.
> Even many sophisticated digital photographers do not really understand, or
> prefer to ignore, the compromises involved in processing images in 16 bit
> color when they are down converted for real world 8 bit use by arbitrary
> device driver algorithms. That supposedly smoother 16 bit histogram is
> irrelevant to the 8 bit driver in the imaging/printing device that is
> incapable of rendering that gradation.
> A bigger advance than 16 bit color would be sensors and processing circuitry
> capable of exposure latitude greater than the current pitiful fractions of
> an f stop.


Like Scott said, we are already writing 12-bits to cards,
so 16 isn't that much more.

Good digital cameras have huge dynamic range, much more than film, and
with more bits to digitize the low end, dynamic range would be better.
Dynamic range is limited by how many photons you can capture
and what the read noise is to get the photoelectrons off
the sensor. Big pixel cameras like the 1D Mark II can store
as many as 80,000 electrons and have read noise of about 4
electrons. 80000/4 = 20,000 = 14.3 stops. See:
http://www.clarkvision.com/imagedetail/evaluation-1d2

For dynamic range info on sensors, see Table 3 at:
http://www.clarkvision.com/imagedeta...ignal.to.noise

14-bits would be a big improvement on many cameras, 16-bits
a modest improvement over 14.

Roger
 
Reply With Quote
 
 
 
 
Kennedy McEwen
Guest
Posts: n/a
 
      04-10-2006
In article <(E-Mail Removed) .com>, Rich
<(E-Mail Removed)> writes
>According to one engineer. Going from 12 to 14 or 16 bit A/D
>conversion is unlikely because
>the 14 and 16 bit converters (although they have provide higher
>resolution) have sampling rates
>that are too slow and their power consumption is too high, sometimes
>6-15x higher than
>12 bit converters. I guess it would mean a battery pack like some of
>those medium format
>backs have?
>

In 1980 I was using what was then a state of the art 20MHz 8-bit
convertor, a TRW part TDC-1007. Power consumption was around 5W and the
only reason it was supplied in a 64pin 1inch wide package was so that
they could get an adequate heat sink onto it! I still have one - the
biggest ceramic packaged silicon chip I have ever seen.

By 1990 ADCs with that performance at less than 1W dissipation in small
packages were commonplace and the state of the art was the Analogue
Devices AD9005, I still have one of those, stamped "Prototype". That
was a 20MHz 12-bit convertor with a power dissipation of 3W.

By 2000 the monolithic low power equivalent of that AD device was again
commonplace and we were designing circuits with 16 or more such chips in
the same space of the old AD9005. By then, state of the art was 16-bit
devices such as the AD9446, 80MHz 16-bit convertor consuming around 2.5W
and a more recent version of this will go even faster.

This should give you some idea of the trend - 4 bits per decade with
improved power and speed. So anyone that tells you that a 14 or 16bit
device won't be available now or in the near future with equal or better
power consumption of a 6 year old 12-bit device (using the Y2K
introduction date of the D30 dSLR as reference) doesn't have much
experience as an engineer.
--
Kennedy
Yes, Socrates himself is particularly missed;
A lovely little thinker, but a bugger when he's ****ed.
Python Philosophers (replace 'nospam' with 'kennedym' when replying)
 
Reply With Quote
 
 
 
 
Raphael Bustin
Guest
Posts: n/a
 
      04-10-2006
On Mon, 10 Apr 2006 10:37:03 +0100, Kennedy McEwen
<(E-Mail Removed)> wrote:

>In article <(E-Mail Removed) .com>, Rich
><(E-Mail Removed)> writes
>>According to one engineer. Going from 12 to 14 or 16 bit A/D
>>conversion is unlikely because
>>the 14 and 16 bit converters (although they have provide higher
>>resolution) have sampling rates
>>that are too slow and their power consumption is too high, sometimes
>>6-15x higher than
>>12 bit converters. I guess it would mean a battery pack like some of
>>those medium format
>>backs have?
>>

>In 1980 I was using what was then a state of the art 20MHz 8-bit
>convertor, a TRW part TDC-1007. Power consumption was around 5W and the
>only reason it was supplied in a 64pin 1inch wide package was so that
>they could get an adequate heat sink onto it! I still have one - the
>biggest ceramic packaged silicon chip I have ever seen.
>
>By 1990 ADCs with that performance at less than 1W dissipation in small
>packages were commonplace and the state of the art was the Analogue
>Devices AD9005, I still have one of those, stamped "Prototype". That
>was a 20MHz 12-bit convertor with a power dissipation of 3W.
>
>By 2000 the monolithic low power equivalent of that AD device was again
>commonplace and we were designing circuits with 16 or more such chips in
>the same space of the old AD9005. By then, state of the art was 16-bit
>devices such as the AD9446, 80MHz 16-bit convertor consuming around 2.5W
>and a more recent version of this will go even faster.
>
>This should give you some idea of the trend - 4 bits per decade with
>improved power and speed. So anyone that tells you that a 14 or 16bit
>device won't be available now or in the near future with equal or better
>power consumption of a 6 year old 12-bit device (using the Y2K
>introduction date of the D30 dSLR as reference) doesn't have much
>experience as an engineer.



Does anyone remember Bernie Gordon (founder of Analogic) stating
categorically that there would never be a monolithic integrated 12-bit
A/D converter?

Heh.


rafe b
www.terrapinphoto.com
 
Reply With Quote
 
Rich
Guest
Posts: n/a
 
      04-10-2006
On Mon, 10 Apr 2006 10:37:03 +0100, Kennedy McEwen
<(E-Mail Removed)> wrote:

>In article <(E-Mail Removed) .com>, Rich
><(E-Mail Removed)> writes
>>According to one engineer. Going from 12 to 14 or 16 bit A/D
>>conversion is unlikely because
>>the 14 and 16 bit converters (although they have provide higher
>>resolution) have sampling rates
>>that are too slow and their power consumption is too high, sometimes
>>6-15x higher than
>>12 bit converters. I guess it would mean a battery pack like some of
>>those medium format
>>backs have?
>>

>In 1980 I was using what was then a state of the art 20MHz 8-bit
>convertor, a TRW part TDC-1007. Power consumption was around 5W and the
>only reason it was supplied in a 64pin 1inch wide package was so that
>they could get an adequate heat sink onto it! I still have one - the
>biggest ceramic packaged silicon chip I have ever seen.
>
>By 1990 ADCs with that performance at less than 1W dissipation in small
>packages were commonplace and the state of the art was the Analogue
>Devices AD9005, I still have one of those, stamped "Prototype". That
>was a 20MHz 12-bit convertor with a power dissipation of 3W.
>
>By 2000 the monolithic low power equivalent of that AD device was again
>commonplace and we were designing circuits with 16 or more such chips in
>the same space of the old AD9005. By then, state of the art was 16-bit
>devices such as the AD9446, 80MHz 16-bit convertor consuming around 2.5W
>and a more recent version of this will go even faster.
>
>This should give you some idea of the trend - 4 bits per decade with
>improved power and speed. So anyone that tells you that a 14 or 16bit
>device won't be available now or in the near future with equal or better
>power consumption of a 6 year old 12-bit device (using the Y2K
>introduction date of the D30 dSLR as reference) doesn't have much
>experience as an engineer.


So does anyone know what 12 bit converters are currently being used
in various digital cameras? At least that way, you'd know what the
designers are willing to work with. The cost of 16 bit converters
is not low, from Analog Devices site, about $50/ea. per 1000-4999.
That seems like alot for one component of a digital camera aside of
course from the sensor.
-Rich
 
Reply With Quote
 
Scott W
Guest
Posts: n/a
 
      04-11-2006

Rich wrote:
>
> So does anyone know what 12 bit converters are currently being used
> in various digital cameras? At least that way, you'd know what the
> designers are willing to work with. The cost of 16 bit converters
> is not low, from Analog Devices site, about $50/ea. per 1000-4999.
> That seems like alot for one component of a digital camera aside of
> course from the sensor.
> -Rich

My guess is most are using one chip to control the CCD and do the A/D
conversion.
This seems like the the kind of thing either National or TI might make.

Scott

 
Reply With Quote
 
Raphael Bustin
Guest
Posts: n/a
 
      04-11-2006
On Mon, 10 Apr 2006 19:13:06 -0400, Rich <(E-Mail Removed)> wrote:


>So does anyone know what 12 bit converters are currently being used
>in various digital cameras? At least that way, you'd know what the
>designers are willing to work with. The cost of 16 bit converters
>is not low, from Analog Devices site, about $50/ea. per 1000-4999.
>That seems like alot for one component of a digital camera aside of
>course from the sensor.
>-Rich



The company I work for makes a line of chips used in
consumer MFPs.

These chips have (currently) several million gates and
include a complete three-channel 16-bit analog front-end --
and cost around $10 in large quantities.

The AFE is about 1/100 of the functionality of the chip,
by the way. The chip controls all the functions in a
typical MFP, including scanner, printer, USB, user
interface elements, and all image processing.

<http://www.sigmatel.com/products/multi/mfp/stdc3000.asp>

See Wolfson for AFEs:

<http://www.wolfson.co.uk>


rafe b
www.terrapinphoto.com
 
Reply With Quote
 
Rich
Guest
Posts: n/a
 
      04-12-2006
On Tue, 11 Apr 2006 00:47:19 -0400, Raphael Bustin <(E-Mail Removed)>
wrote:

>On Mon, 10 Apr 2006 19:13:06 -0400, Rich <(E-Mail Removed)> wrote:
>
>
>>So does anyone know what 12 bit converters are currently being used
>>in various digital cameras? At least that way, you'd know what the
>>designers are willing to work with. The cost of 16 bit converters
>>is not low, from Analog Devices site, about $50/ea. per 1000-4999.
>>That seems like alot for one component of a digital camera aside of
>>course from the sensor.
>>-Rich

>
>
>The company I work for makes a line of chips used in
>consumer MFPs.
>
>These chips have (currently) several million gates and
>include a complete three-channel 16-bit analog front-end --
>and cost around $10 in large quantities.
>
>The AFE is about 1/100 of the functionality of the chip,
>by the way. The chip controls all the functions in a
>typical MFP, including scanner, printer, USB, user
>interface elements, and all image processing.
>
><http://www.sigmatel.com/products/multi/mfp/stdc3000.asp>


I don't know what the differences are between those multicontrol chips
and the AD chips, but the price seems pretty cheap.
-Rich
 
Reply With Quote
 
 
 
Reply

Thread Tools

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

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are On
Pingbacks are On
Refbacks are Off


Similar Threads
Thread Thread Starter Forum Replies Last Post
You're unlikely to remember this... Pennywise@DerryMaine.Gov Computer Support 66 05-14-2007 03:59 PM
Likely/Unlikely and standards Richard G. Riley C Programming 4 02-24-2006 02:32 PM
Accessing higher security level from higher security level nderose@gmail.com Cisco 0 07-11-2005 10:20 PM
64 bit - Windows Liberty 64bit, Windows Limited Edition 64 Bit,Microsoft SQL Server 2000 Developer Edition 64 Bit, IBM DB2 64 bit - new! Ionizer Computer Support 1 01-01-2004 07:27 PM
Unlikely differences between design time and run time. Tzury Bar Yochay ASP .Net 3 11-29-2003 08:56 AM



Advertisments