rcholas
Offline
•
Upload & Sell: Off
|
 p.25 #14 · A7/A7r - performance with WA RF lenses | |
theSuede wrote:
The first layer (closest to the sensor) sits quite easily detachable to the sensor mount frame. It takes between 3-5 minutes to free the plate from the frame with a sharp knife and some previous experience. It is in no way hard-bonded to ANY part of the sensor material, only to the mount frame. There is a thin layer of optical gel between the plate and the sensor surface proper in some camera models, depending on manufacture.
It's quite obvious we are not talking about the same thing.
- What I mean by "sensor" is the silicon sensor chip proper.
- What I mean by "sensor package" is the thin, purple-colored ceramic carrier in which the silicon chip is placed, and that is sealed with a transparent cover plate on the top side, and has electrical contacts on the lower side that are soldered to a printed circuit board.
I'd be, er, impressed if you were able to remove that transparent cover plate from the ceramic carrier using a mere knife, without causing significant damage to the ceramic and the cover glass.
Not to mention that you'd be exposing the fragile silicon surface, which had been placed in the ceramic carrier in a clean room, and carefully sealed with the transparent cover glass in a controlled atmosphere to prevent any external contaminants from reaching the silicon, the bonding pads and the bonding wires.
So, I'm afraid you're confusing the "sensor package" with the silicon chip itself.
I have had a D800 sensor package in my hands at a trade show, and happen to have been told by a Nikon engineer that the cover glass, bonded to the ceramic carrier, was made of birefringent material, hence my original post in this thread.
Unfortunately, pictures of the D800's sensor package seem scarce on the Internet.
There are, however, some substitutes that will, hopefully, illustrate my point.
An IR filter + AA filter optical stack typically consists of:
- an IR absorption filter
- an IR hot mirror coating
- the first birefringent layer, for the AA filter's X separation
- one wave plate to restore the polarizations
- the second birefringent layer, for the AA filter's Y separation
How are these layers typically organized ?
The "Canon 5D Mark II" white book has, on page 21, this informative picture.
We can see, from the lens's side, that there are three "groups":
- The first birefringent layer (pink), that also has a piezoelectric dust removal vibration function on the Canon 5D Mark II.
- A wave plate (blue) to which an IR absorption filter (green) has been bonded
- The second birefringent layer (orange), which, on the Canon 5D2, also seems bonded to the silicon chip's ceramic carrier, just like the Nikon D800.
As for Nikon, iFixit has nice hi-res pictures of a teardown of the D600.
- D600 ceramic chip package, containing the 24MP silicon chip, likely sealed with a birefringent crystal (the transparent cover).
- D600 "sensor mount frame", containing, from the lens' side:
- a transparent plate with a dust removal piezoelectric driver. This plate is highly likely to be the first birefringent layer. The thin blue-colored layer that seems to be applied to that transparent plate's back might be an IR hot mirror coating, or a bonded wave plate.
- a flexible, black-colored peripheral dust seal
- a blue-ish plate, consisting possibly of an IR hot mirror + IR absorption filter, and perhaps a bonded wave plate, if said plate wasn't already bonded to the first transparent plate.
- green PCB on which is soldered the ceramic sensor carrier with its bonded birefringent crystal cover
The "step 30" picture of iFixit's Nikon D600 teardown shows that the "sensor mount frame" seems to contain only two optical plates — i.e. the transparent first layer with its dust removal piezo driver, and a smaller, blue-ish plate.
Thus, considering the number of optical plates on the teardown picture, the Nikon D600's IR+AA optical stack seems very similar in its layout to the Canon 5D2's, and there's little reason to believe that the Nikon D800's IR+AA optical stack has a much different structure either.
Considering the typical elements in an IR+AA filter stack, it is thus highly likely that on the Nikon D600, just like on the Nikon D800 and Canon 5D2, the indispensable second birefringent plate is, in fact, the transparent cover glass that's bonded to the ceramic chip carrier.
It's thus highly likely that the "first layer (closest to the sensor)" that you mention as removable from the "sensor mount frame" using a "sharp knife" is not, as you seem to be imagining, a birefringent layer, but rather, the blue-ish plate (IR filter ?) visible on iFixit's "step 30" picture.
The flangeback distance is the distance between the lens' bayonet mount flange and the imaging plane (film or sensor). The optical path length of the flangeback distance is a constant, that must be rigorously adhered to by a camera body if it wants to achieve proper focusing with a Nikon lens.
If you want to argue that on the Nikon D800, the ceramic carrier's covering glass is not a birefringent layer, but just a simple piece of glass, you'll have to explain why Nikon would waste some of the tightly metered space between the imaging plane (the silicon chip) and the lens mount with a piece of glass that has no optical role.
Space in a full-frame DSLR is at a premium as every millimeter allocated to the optical stack behind the focal plane shutter must come at the expense of the space allocated to the reflex mirror, and advancing a reflex full-frame mirror towards the lens is not really an option as the mirror might collide with the rearmost glass element of some lenses.
theSuede wrote:
The "best" solution would have been to use the absorption filter for sensor base coverage, and a simple optical flat with the hot/cold mirror treatments on each side for the top dust-off plate. This would necessitate a ~0.4mm offset of the sensor surface proper in the sensor cradle (or indeed by moving the entire sensor assembly by shims or any other mechanical solution). The decrease in number of surface interfaces and the decrease in total stack thickness would decrease both the astigmatism delta and the SA delta to less than half of what it is today, and it would also decrease the corner smearing due to less reflection/scatter....Show more →
I'd be interested to hear why combining a simple optical flat and a forward or backward offset of the sensor, say, by 0.4mm, would be sufficient to improve the "astigmatism and SA delta".
May I remind you that, with non-telecentric lenses, an optical flat induces — among other aberrations — a degree of "field curvature", which is an aberration that you seem to confuse with "SA" (spherical aberration), despite them being totally different concepts.
Correcting field curvature would, by definition, require curving the image field — i.e. curving the sensor chip, a rather difficult endeavor — instead of simply linearly translating a flat sensor...
Edited on Oct 21, 2013 at 10:56 AM · View previous versions
|
(or any other recent Nikon camera either for that matter).
