The D50 FAQ -Version 1.03 - June 21 2006

The D50 and the search for the Holy Grail.

No, King Arthur was not searching for a D50! Rather, he was searching for something that is unobtainable to mere mortals: Perfection.

The perfect astro DSLR would have 100% quantum efficiency and no, internally generated, noise.

No, the D50 is not perfect, in fact it is far from it, but it is, in my opinion, a little bit closer to the perfect DSLR than any other at this time.

How does the D50 compare to other DSLRs?

Compared to the D70/S/200, the D50 has lower noise, especially amplifier glow and a higher dynamic range. Compared to the Canon 350XT/20D/30D the D50 exhibits lower noise, less amp glow (somewhat temperature dependent, however), has higher quantum efficiency, and a deeper full well at ISO200 and higher, for greater dynamic range, during long exposures.

To sum up, the D50 compared to the 350XT/20D/30D:

20-25% greater quantum efficiency (actually, I mean total ability to convert photons to electrons, per pixel).

higher dynamic range at ISO 200 or greater, during long exposures

lower noise, especially dark current

10% larger sensor (12.5% larger than the 350XT)

The D50 is superior to any other, half frame sensor size, DSLR for astro work, with the exception of it's suitability for automated imaging. Compared to full frame (FF) DSLRs the D50 also appears to have an edge( possibly) in sensitivity and (probably) noise reduction, and this must be balanced off as to whether the larger field of a FF DSLR is worth the extra cost. The D50 like the D70/S requires an IR remote signal to stop and start exposures longer than 30 seconds. I, typically, use the Nikon IR remote, ML-13, for this purpose, although I understand that a typical multi-function IR remote, can be programmed for the same function. A computer can also be programmed to issue the IR remote signal via standard serial or parallel cable that is linked to a IR transmitter. A computer, so equipped, can then act as an intervalometer (image sequence timer) to take a predetermined number of images of any duration, up the the firmware limitation, of 30 minutes.

What sensor does the D50 use, and where can I get the specifications?

The D50, probably, uses the Sony ICX413AQ, or some variant of it. You can download a PDF file with the technical specifications here:

http://ceres.informatik.fh-kl.de/pbw/lehre/20041/foto/resourcen/Datenblaetter/Sony/icx413np.pdf (2 page, short summary)

or here:

http://www.framos.de/pdf_sheets/ICX413AQ.pdf (very technical, 19 page document)

So, why isn't the D50 more popular for astro-imaging?

There is a serious problem with the D50/70/S/200 DSLRs that limits
their usability for astrophotography. When the camera completes an exposure of
greater than 1sec duration, it automatically runs a median filter on the image.
This median filtering occurs regardless of whether the in-camera noise
reduction is on or off. The exact purpose of the median filter is known only to Nikon but
it seems to be aimed at reducing the effects of hot pixels during long
exposures. Unfortunately, the median filter algorithm seems to confuse stars
with hot pixels, and removes both with equal enthusiasm! Needless to say, this
is a very undesirable state of affairs for anyone interested in doing
astronomical work with a Nikon DSLR.

Is there a fix for the median filter problem?

Ar first glance it would seem that Nikons are simply unusable for serious
astrophotography, but for one important discovery: 'Mode 3'. Mode 3 was
documented by Christian Buil, a French astronomer, who also created the name
"mode 3". He noted that if you turn noise reduction 'on', and take a long
exposure image, and turn off the camera during the noise reduction phase, 'Job
NR', with the on-off switch, that the camera will flush the buffer and save the
current in memory image to the memory card, WITH OUT running the median filter,
and the result is a true RAW image, with all the data intact.

'Mode 3' saved Nikon DSLRs from being completely driven out of the
astrophotography market, but it is quite a bit chore to have to turn the
camera on and off after each exposure to recover a true RAW image, with the
stars intact, especially when your friend is using a Canon, which can be
completely automated, while tethered to a PC. Most astrophotographers simply
choose the Canons due to their ease of use and their hitherto the D50, superior performance.

The D50's wide dynamic range enables it to take very long exposures while preserving information in both the faint and bright areas of the target object. So where another DSLR might be required to do multiple images of an object, both to preserve detail and to control noise, and then stack them, the D50 can probably take 20-30 minute exposures of the same object, and still not burn out the bright core areas, while noise will remain very low. This limits the amount of interaction needed to use the camera in 'Mode 3' as having to flick the on-off switch every 30minutes is not an unreasonably tedious chore. The resulting images will simply amaze you with the wealth of recorded information and the extreme lack of noise, including amp glow. However, for piggyback work, with a fast lens or for use with an scope that has an automated focuser, this is still not an acceptable solution. Appeals to Nikon to change the firmware, did not produce any results, so I began to try and find a way to automate the D50 while still using 'Mode 3'. After some trial and error I have developed a method to automate imaging in Mode 3, although it is a little complicated, it is not unworkable:

A Nikon DSLR, will also save a true raw image if the camera
detects a low voltage condition in the internal battery during the 'Job NR' phase of
the in camera noise reduction. Basically, as a safety feature, any image in the
buffer is saved to the memory card, without having the median filter applied.
There is a range of internal voltages which will cause the camera to abort a
'Job NR' and save the image to the memory card, unaltered.

I obtained a home built intervalometer that can trigger a IR remote signal to
start and stop an exposure with a D50/D70/S, and I wired a relay into the
circuit that can interrupt an external power feed to the camera. The process
goes like this:

Intervalometer opens the shutter

Intervalometer closes the shutter

camera begins 'Job NR'

Intervalometer shuts off external power to the camera.

camera reverts to internal power (deliberately supplied at a low voltage or via depleted internal battery)

camera senses low voltage and saves the image in the buffer to the memory card

no median filtering!

Intervalometer restores external power

Camera reverts to full operation.

Intervalometer then opens the shutter for the next exposure to begin a new sequence.

I have used the above technique in actual practice and it works well. I have been able to acquire 20 or more 10min images, in a continuous sequence, however, for greater reliability it will be necessary to use a regulated power supply to replace the depleted internal battery. I plan to do this by using a modified MS-D70 internal battery, pack wired to a suitable power supply.

Obviously, a PC that has an IR transmitter attached to a external port could handle the above sequence quite easily, and could also easily control a relay to interrupt the external power supply when required. I will be working on this in the near future.

All this to work around a fault (or feature, maybe, from Nikon's point of view) that could be easily fixed with a simple firmware change!

Does the median filter completely destroy the camera's ability to do astrophotography?

No. It should have relatively little impact when the image is oversampled, as the median filter is looking for single pixels, or when doing crowded star field imaging, where the loss of a few fainter stars will go unnoticed.

What other performance issues effect the D50?

The D50 does generate a fair number of hot pixels, some of which are random in nature and will not be removed by a simple dark frame subtraction. Higher temperatures seem to increase the hot pixel rate. However, hot pixels do not have much impact on the camera's overall performance, especially when you consider that you have 6 million pixels to work with, but it is annoying all the same, and can be a time consuming chore to remove, depending on your software.

There is some concern that the D50's compressed NEF (RAW) format, does not generate a true RAW file. There is some evidence that Nikon decided to compress the file in such as way that the full 4095 (12 bit) range of data value is not preserved. In practice, however, this does not seem to be an issue, that has any noticeable impact on the images. The advantage to the compressed format is a decreased need for storage space and faster download and processing times. In theory, Nikon could address this issue via a firmware change.

The D50 does not have a mirror lock up feature. In practice, I have been unable to detect any mirror slap issues with the D50, and careful monitoring of the autoguider guide star, has not revealed any detectable movement when the mirror/shutter is fired. The D50 seems to have a very soft mirror slap, probably because Nikon engineers thought that the typical D50 user is unlikely to want a high frame rate, where a more rapid mirror movement would be likely.

Is there any other performance advantages of a D50?

Oddly enough, Digital Photography Review ( http://www.dpreview.com/ ) found that a D50 had  a better USB 2.0 data transfer rate, than just about any other DSLR around, including the Canon 5D! A quick check (June 15 2006) t the above site, showed that the D200 for one was somewhat faster, but the D50 is still tops in the budget category.

Where can I find a technical analysis of the D50?

This article was written using data from my non-modified D50:

.

Why bother?

I went to this effort because the D50, is so superior to its contemporaries that it is worth the effort. It is a superb jewel of a camera, that when modified, by removing its IR blocking filter, actually compares quite well with a traditional cooled, colour ,CCD camera at a fraction of the price!

Stay tuned More to come!

Revisions
1.02 added urls, fixed a few typos.
1.03 fixed a few typos, added links to CCD specifications.