Making a DSU from scratch for an orphan DCS100

[NikonWeb]

We are close folks!

I see you posted this on Youtube recently: https://www.youtube.com/watch?v=mRyBnOVRm3A

Will you start selling home made DSUs soon?

Jarle

[pieroc91]

Hey! thanks for posting it here, i tried but there was some issue with the forum and i couldn't do it.
That was indeed an advance and finally could transfer valid data but just a few pixels, the interface proved to be very slow and overflowed too easily.
The project is not dead and it will never be, is super slow yes but once i manage to store those bits that DCS is ready to shoot.
Right now i'm out of the country once again but when things stabilize i'll be giving it some progress to the DCS.
Thanks to everyone that are following the project

[pieroc91]

It is known that digital imaging testing must be done with the classic picture of Lena Söderberg https://es.wikipedia.org/wiki/Lenna
I don't have one of those so i draw my best on a CF card case.


Then i put myself to code and to wire the cheapest and chiniest logic analyzer i could get.


And lol and behold!! the first recognizable image from the modern recreation of a DSU.
With all of you... Lena:


Yes, it is a mess.
The image looks like is melting due to a very wrong timing of the CCD,
It is skewed due to the analyzer missing some bytes on the stream because it just decides to ignore my clock signal,
Also because of missing bytes the color information must be discarded,
Exposure is quite a hard thing since i can only get a reading if i expose it at the same time i'm reading it so i have to expose it faster than the sweep is, that means, using a flash.

But hey...
i got to make a picture of my mom's cat


Stay tuned

[pieroc91]

Now this is a Lena picture

[pieroc91]

The work contiunes and the advances keep on going, the blooming on the left got corrected, adding a 0.1ms delay between the vertical swipe and the reading got that fixed, there's still some ghost on the left side and some vertical lines are completely missing but i guess that all that is just fine tuning the timings. (cat picture for extra internet points).

[pieroc91]

And the glitches are gone, those wierd artifacts i'm quite sure are induced by the non timed read of my logic analyzer, improving that should improve the circumstances.
Pictured are my NC2000E and half of my D1X.

[pieroc91]

oh yeah, a little bit of fiddling got me to a picture that looks like an actual picture, now the D1X can be seen completely, and i know i'm reading the whole CCD which lets me know that my CCD is in good working order. i'll try to fiddle with the file on a hex editor to see if i can realign each line so i can apply a demosaic filter and see colors.

[pieroc91]

Things are moving fast, a lot of tweaking and a sunday code cleaning resulted on the DCS behaving as it should, now the shutter and readout times are the way the should and it takes pictures with the press of the shutter button with the correct speed and image size.
That means a DCS cat picture can be achieved.

[pieroc91]

fine tuning a little bit the reading and now i have some color, due to jitter i also have color banding but hey... even the DCS appreciates the red TrackPoint of a ThinkPad.

[pieroc91]

further tinkering adds support for more the one ISO setting, it should have 100, 200, 400 and 800, i could only get two working, the other two resulted in complete white pictures, it might have some problem on the preamp. On a side note i noticed the sensor to be almost not responsive to blue light, maybe the pigments went dead?