Go Back   Steve's Digicams Forums > Digital Cameras (Point and Shoot) > Hybrid Still/Movie/MP3 Digicams

Reply
 
Thread Tools Search this Thread
Old Dec 19, 2009, 4:30 PM   #1
Junior Member
 
Join Date: Dec 2009
Posts: 7
Default Inside the zi8

I haven't noticed that anyone has taken a zi8 apart, so heres some documentation of my attempt to do so.

The Kodak is a bit tricky to disassemble compared to an Aiptek, but not nearly as bad as a DXG 595V. First you take off the battery cover and remove the battery. Then, undo every screw you can see. The lens cover slides off the end of the body. Run your fingernail around the edge of the case to pop a few catches near the power switch and the audio jack and the case separates. Pull the flat cable out of the board on the back cover and set the cover aside. Be careful not to lose the hinge pin for the USB connector, it will fall out.



If you can unpeel the copper shield or if you have some adhesive shielding tape to replace it, unpeel the copper shield, unplug the lcd, and set it aside.



Tilt out the lcd panel if it is still there and then undo the screws on the sheet metal cover/bracket and tilt it out also.



Undo one last screw on the daughter board and pull it straight up and out. On the daughter board is the image sensor, the power switch, and the macro switch. The main circuit board won't budge despite there being no visible fasteners holding it in. It might be bonded in place.





The image sensor appears to be the same Aptina 1/2.5" cmos used in the Aiptek A-HD+, the DXG 595v, and who knows how many other 5 megapixel cameras. Unlike those other cameras, Kodak has provided a cutout in the circuit board behind the sensor and filled the cavity with heat sink grease to conduct heat from the sensor into the large sheet metal bracket between the sensor board and the lcd screen. A cooler sensor means less noise. The window in front of the lens appears to have an anti-reflection coating on both surfaces, which would increase light transmittance by 8% compared to the uncoated window in front of the Aiptek lens. The uncoated lens cover of the Aiptek, however, in combination with the stepped cone behind it might be an effective contrast reducing filter to bring out shadow detail. Unlike the Aiptek IR cut filter which is easy to remove, the Kodak IR cut filter is glued into the lens and can not be removed without breaking it, which I did. If anyone knows an inexpensive source of IR cut filters, I'd be grateful. Sunex wants $17 which is about 10 times more than it should cost.
wbamb833392 is offline   Reply With Quote
Sponsored Links
Old Dec 19, 2009, 8:34 PM   #2
Senior Member
 
Join Date: Oct 2006
Posts: 584
Default

Great work. Fascinating they try to cool the sensor...
Does it look like the lens is 4 element or 2-element?

I took apart one of my DOD F200, it has a 4-element lens, much smaller aperture than the Zi8, with a smaller sensor - looks like a 1/2.33, and is 16x9 in shape. The IR filter is glued over the top of the sensor, not part of the lens. The good news is that the slight difference in resolution between my two units disappeared after I fitted stronger springs on the macro mechanism of this weaker one, and then put the camera back together
.

Last edited by Trevmar; Dec 19, 2009 at 8:41 PM.
Trevmar is offline   Reply With Quote
Old Dec 21, 2009, 4:14 PM   #3
Junior Member
 
Join Date: Dec 2009
Posts: 7
Default

Quote:
Originally Posted by Trevmar View Post
Great work. Fascinating they try to cool the sensor...
Does it look like the lens is 4 element or 2-element?
.
I can't tell how many elements the lens has, and I'd rather not ruin it finding out since I'd like to reuse it. The rear element is strongly aspherical, that much I can see. I don't understand why the IR filter coating is not applied directly to the cmos sensor cover glass. Doing so would simplify production and increase light transmission by a percent or two.
wbamb833392 is offline   Reply With Quote
Old Dec 21, 2009, 4:16 PM   #4
Moderator
 
fishycomics's Avatar
 
Join Date: Jul 2005
Location: NYC NY
Posts: 9,623
Default

a great job, very happy to see it opened and what it looks like, glad it was not super-hard, good luck.
__________________
These are Pc low Budget Economy Hybrid camcorders Great for the Daytime Outdoor shots
Light on the Object video made. No light No Video
fishycomics is offline   Reply With Quote
Old Dec 23, 2009, 12:24 AM   #5
Senior Member
 
Wayne12's Avatar
 
Join Date: Jun 2006
Posts: 1,071
Default

Sensor cooling can make a great difference to dark noise, the heat sink probably acts to reduce the mount of noise after it warms up.

It was one of the things I suggested back in the DIY camera projects, I even had a custom atmospherically sealed refrigerated design for one of my camera housing designs (but that adds heaps of power consumption).

It might also be worth noting that each time the manufacturer shrinks their sensor to a smaller chip design process, there is opportunity to lower the amount of noise, and increase performance.

Wbamb, would you know what ambarella processing chip it uses, or if you measured the size of the sensor or could see it's model number? Thanks.


Wayne.
Wayne12 is offline   Reply With Quote
Old Dec 26, 2009, 9:07 PM   #6
Junior Member
 
Join Date: Dec 2009
Posts: 7
Default

Quote:
Originally Posted by Wayne12 View Post
Sensor cooling can make a great difference to dark noise, the heat sink probably acts to reduce the mount of noise after it warms up.

It was one of the things I suggested back in the DIY camera projects, I even had a custom atmospherically sealed refrigerated design for one of my camera housing designs (but that adds heaps of power consumption).

It might also be worth noting that each time the manufacturer shrinks their sensor to a smaller chip design process, there is opportunity to lower the amount of noise, and increase performance.

Wbamb, would you know what ambarella processing chip it uses, or if you measured the size of the sensor or could see it's model number? Thanks.


Wayne.
There are no markings visible on the sensor chip, but it looks identical to the chip in the DXG 595V which according to the firmware is an Aptina MT9P401. The data sheet for the sensor gives the dynamic range as an incredible 70.1dB, but the SNR as a pretty poor 38.1. Neither of the first two cameras I disassembled had any provision for cooling the sensor at all, not even a hole in the circuit board, which surprised me since heat causes noise and the sensor does become quite warm in use. I'm going to add a sensor chiller to this camera to see how much the noise can be reduced. Thermoelectric chillers are large and wasteful, so I'll use a thermos of ice instead. Ice is pretty easy to come by. I'd like to be able to make some pro quality specialty cameras by hot rodding inexpensive cameras with extra cooling and better lenses. So far, trying professional video lenses and security camera lenses on the DXG cam has been very encouraging.

-Eric
wbamb833392 is offline   Reply With Quote
Old Dec 28, 2009, 12:31 AM   #7
Senior Member
 
Wayne12's Avatar
 
Join Date: Jun 2006
Posts: 1,071
Default

Quote:
Originally Posted by wbamb833392 View Post
There are no markings visible on the sensor chip, but it looks identical to the chip in the DXG 595V which according to the firmware is an Aptina MT9P401. The data sheet for the sensor gives the dynamic range as an incredible 70.1dB, but the SNR as a pretty poor 38.1.
Aptina has a few that size, and Kodak is supposed to manufacture their own (they have a hybrid bayer scheme that uses a transparent pixel in place of one of the greens to increase the low light performance). So, you could measure that the actual sensor array (not just the window) was the same size?

The signal to noise ratio is rather low. Until they get to at least 48-54 db Snr, I do not take them too seriously for consumer. This is the amount you would expect to get a descent 8-bit image in optimum light. For professionals, we would look for much more for low light, and an additional 6db for every extra luminance bit).

You did not notice the version of the ambarella camera control chip used?

Quote:
Neither of the first two cameras I disassembled had any provision for cooling the sensor at all, not even a hole in the circuit board, which surprised me since heat causes noise and the sensor does become quite warm in use. I'm going to add a sensor chiller to this camera to see how much the noise can be reduced. Thermoelectric chillers are large and wasteful, so I'll use a thermos of ice instead. Ice is pretty easy to come by. I'd like to be able to make some pro quality specialty cameras by hot rodding inexpensive cameras with extra cooling and better lenses. So far, trying professional video lenses and security camera lenses on the DXG cam has been very encouraging.

-Eric
Be careful, the cooling will attract condensation onto the sensor, and then anything attached in front of that, that can also get into he electronics. So a system to get around this is needed. There are coatings to stop this (however it is likely to get on something and accumulate), and insulating plastic etc, but optical properties of these is another thing.

Did you notice my post about adding filters to lens?

About low light. When I read things about coatings and ND filters being a certain percentage light loss, I wonder of what, through the whole brightness range or maybe just at a certain level (like a mid-day scene). I noticed when walking on a moon lit night, I get maybe twice the apparent brightness looking over my glasses as through them, even though they have high performance filters (like 95-99% throughput). This means that the actually loss of light could be many times more than the apparent 50%.
Wayne12 is offline   Reply With Quote
Old Dec 29, 2009, 3:54 PM   #8
Junior Member
 
Join Date: Dec 2009
Posts: 7
Default

Quote:
Originally Posted by Wayne12 View Post
Aptina has a few that size, and Kodak is supposed to manufacture their own (they have a hybrid bayer scheme that uses a transparent pixel in place of one of the greens to increase the low light performance). So, you could measure that the actual sensor array (not just the window) was the same size?

The signal to noise ratio is rather low. Until they get to at least 48-54 db Snr, I do not take them too seriously for consumer. This is the amount you would expect to get a descent 8-bit image in optimum light. For professionals, we would look for much more for low light, and an additional 6db for every extra luminance bit).

You did not notice the version of the ambarella camera control chip used?



Be careful, the cooling will attract condensation onto the sensor, and then anything attached in front of that, that can also get into he electronics. So a system to get around this is needed. There are coatings to stop this (however it is likely to get on something and accumulate), and insulating plastic etc, but optical properties of these is another thing.

Did you notice my post about adding filters to lens?

About low light. When I read things about coatings and ND filters being a certain percentage light loss, I wonder of what, through the whole brightness range or maybe just at a certain level (like a mid-day scene). I noticed when walking on a moon lit night, I get maybe twice the apparent brightness looking over my glasses as through them, even though they have high performance filters (like 95-99% throughput). This means that the actually loss of light could be many times more than the apparent 50%.
I have not been able to remove the main board from the zi8 to get a look at the processing chips. The sensor is identical in dimensions, shape, color, sensor area, and active area offset as the chip in the DXG camera. Whether it is a custom sensor with white pixels is anyones guess, but based on the better low light performance I would say it's a good bet.

Here in the dry desert condensation is not a problem, but if cooling works I plan to add some dessicant to the chamber between the lens and the sensor which is sealed anyway to exclude light. When changing lenses, a squirt of refrigerant gas from a common computer duster can purge the ambient air from the chamber and remove dust at the same time.

Filters are compared against no filter at all, and the result applies across all brightnesses. Optical coatings increase light transmission by about 7%, but a more important point for eyeglasses is that they reduce the reflection of the image of the eye back to itself which would otherwise reduce contrast greatly. However, the performance of your eyeglasses might not be because of coating, but the interaction with the eye itself. The most light sensitive part of the retina is the periphery, which is why so many people have the frustrating experience seeing a dim start out of the corner of the eye but seeing nothing when looking directly at it. The combined optics of your glasses and your eye might be blocking some light from the periphery of the retina giving you the impression that the scene is darker even though anything you look at in particular will be the same brightness.

-Eric
wbamb833392 is offline   Reply With Quote
Old Dec 30, 2009, 12:10 PM   #9
Senior Member
 
Wayne12's Avatar
 
Join Date: Jun 2006
Posts: 1,071
Default

Thanks Eric.

Quote:
Originally Posted by wbamb833392 View Post
Whether it is a custom sensor with white pixels is anyones guess, but based on the better low light performance I would say it's a good bet.
I wonder if a light source and a microscope would do it? it does not matter anyway, it looks like it has the goods.

Thanks for the verification on the sensor size and comments on filters. All glasses generally decrease light transmission, filters increase the light coming through, but I still find it darker than real life. It is likely the filers have there own minor hit, but small compared to the increased amount of light transmission. You can hold glasses over bright paper to see the result.
Wayne12 is offline   Reply With Quote
Old Jan 6, 2010, 10:54 PM   #10
Junior Member
 
Join Date: Dec 2009
Posts: 7
Default

Update: I finally managed to get the main circuit board out of the camera. It was held in by the large copper heatsink and heatsink compound on the main processing chip. The chip is an Ambarella A530C.

-Eric
wbamb833392 is offline   Reply With Quote
 
Reply


Thread Tools Search this Thread
Search this Thread:

Advanced Search

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 Off
Pingbacks are Off
Refbacks are Off



All times are GMT -5. The time now is 5:49 AM.