OV5647 vs 400 grit, or: how to remove the IR filter

[M33P]

Once upon a time, a great man said:

according to Omnivision the IR filter will be almost/completely impossible to remove.

I'll shut up now and let the pictures do the talking.

IR01.jpg (62.64 KiB) Viewed 16110 times

IR02.jpg (45.36 KiB) Viewed 16110 times

IR03.jpg (34.13 KiB) Viewed 16110 times

...

[M33P]

...

IR04.jpg (41.42 KiB) Viewed 16111 times

IR05.jpg (39.05 KiB) Viewed 16111 times

IR06.jpg (41.5 KiB) Viewed 16111 times

...

[M33P]

IR07.jpg (49.09 KiB) Viewed 16109 times

This last picture was taken at 100x magnification via afocal mount on my microscope. It illustrates not only pretty gold bond wires but small bits of crud that are an inevitable consequence of removing the IR filter in this manner. Contamination that will degrade image quality unless removed.

Now all I need to do is figure out some sort of surfactant and solvent to remove them...

[nicknml]

That didn't take long

[recantha2]

Respect! Now put it back together again :p

[Maxion]

Now *that* is nicely done!

Now you just got to find a way to mount that lens again at it's focal length!

There are many solutions available for cleaning sensors. A blast of air should work fine (Don't use a compressor as it can spew tiny oil droplets and if you use a can of compressed air *Be careful* as the propellant can spew out as well, best is a manual "rocket" blower).

There are also wet cleaning solutions available, either ready made pads or just the solution itself and modified q-tips that don't lint.

The sensor itself though is very prone to scratching, so don't use any pressure while cleaning it.

[jbeale]

Very nice work getting the sensor completely apart. Have you tested to see if it still works at all?

Yes, there is a large body of experience- and products- built up over the past 10 years with cleaning DSLR sensors.
Here is just one example: http://www.amazon.com/Digital-Survival- ... B000PNGM18
However, with a DSLR you are touching an anti-alias filter/IR filter, not the bare sensor.

In this case, apparently the bare sensor is exposed and if you are using a mechanical wiping tool, the bond wires might be in harm's way. Using a liquid (eg. Eclipse ) and a squeeze-type air puffer might be my choice.

People warn about using compressed air (or flurocarbons) in cans, where liquid droplets may spray onto your sensor and cause trouble- perhaps I've just been lucky but I've gotten away with that on my DSLRs.

[poing]

Is the IR filter stiff or can you bend it easily? Would it be possible to glue a stick to it and pull it out through the lens mount?

[M33P]

Very nice work getting the sensor completely apart. Have you tested to see if it still works at all?

It worked fine until this morning when my ham-fisted attempts at wiping using the gentlest touch squashed two of the bond wires together.

Then I really did kill it by attempting to separate them using a needle and a microscope - which promptly pulled one of them off the chip. Just my luck it wasn't a redundant power or ground connection

This sensor is *very* naked once you get the filter off. I had planned on putting a microscope cover slide on top of it after cleaning and gluing that down on the remaining plastic surround. Suffice to say that filter removal is strictly for those with a desire for destruction.

Next I will try to profile the absorption characteristics of the filter itself (far easier with it removed) - you can buy IR LEDs with 950nm-800nm wavelengths for about 20p each. Maybe the cutoff isn't so bad.

[scorp]

I did the same on Friday. Assembled it back and adjusted focus. Made some pictures in pretty dark room and pictures were brighter than with IR filter. But I used lamp to illuminate the room rather than IR LEDs, didn't have them in my hands at that time. Then I noticed one small spot and decided to remove it later leaving camera open on my table. My cat ruined it next day so it has few wires snapped like on your one. Will try to fix but I suspect I need to order new one.

[meltwater]

Much respect for going for it, and trying. Great to see the close up pictures and construction.

[jamesh]

I believe the phrase "Almost impossible to remove without damage" may be appropriate here!

Godo work nevertheless.

[scorp]

It is not that difficult at all even without damage. My damage was my silly cat(or rather silly me) and doing it in dusty room

[peepo]

regarding dust removal, on a PS3 eye camera an artist's paint brush does the trick.

seems strange but works well, can't say why.

it may be the sensor has some glass or similar plate on top, don't know.

even seems to remove finger prints.

[jbeale]

I did the same on Friday. Assembled it back and adjusted focus. Made some pictures in pretty dark room and pictures were brighter than with IR filter. But I used lamp to illuminate the room rather than IR LEDs, didn't have them in my hands at that time. Then I noticed one small spot and decided to remove it later leaving camera open on my table. My cat ruined it next day so it has few wires snapped like on your one. Will try to fix but I suspect I need to order new one.

Did you use the same method, sanding down the lens holder? Or perhaps something different, if you were able to reassemble it afterwards?

A pointed paint brush apparently works well for dust removal due to static attraction between the brush and the small particles. There is even a brush-type tool sold for this purpose which spins around before use to build up more static charge. I didn't know about the fingerprints, maybe it is the same principle as a microfiber cloth (oils wicked in by capillary action?)

[scorp]

I did the same on Friday. Assembled it back and adjusted focus. Made some pictures in pretty dark room and pictures were brighter than with IR filter. But I used lamp to illuminate the room rather than IR LEDs, didn't have them in my hands at that time. Then I noticed one small spot and decided to remove it later leaving camera open on my table. My cat ruined it next day so it has few wires snapped like on your one. Will try to fix but I suspect I need to order new one.

Did you use the same method, sanding down the lens holder? Or perhaps something different, if you were able to reassemble it afterwards?

When I'm back home today I'll put some pictures of my parts.
What I did, I used scalpel to detach black plastic case from flexi cable. Once they apart I had access to the filter but it was not loose inside so I had to unscrew lens to push it from lens' side out of the case. Then the case nicely sits back. I didn't have any damage to the wires doing that. Just make sure you don't put scalpel too dip and don't bend sensor metal bed too much. Fit scalpel in-between glue and case in flexi-glue-case sandwich to leave more glue on flexi rather than on the case.
I suspect that small drop of super glue will be enough to have case and flexi(with sensor) back together.
Just noticed that CPC is out of stock so my second attempt won't be soon.

[jbeale]

What I did, I used scalpel to detach black plastic case from flexi cable. Once they apart I had access to the filter but it was not loose inside so I had to unscrew lens to push it from lens' side out of the case. Then the case nicely sits back. I didn't have any damage to the wires doing that. Just make sure you don't put scalpel too dip and don't bend sensor metal bed too much. Fit scalpel in-between glue and case in flexi-glue-case sandwich to leave more glue on flexi rather than on the case.
I suspect that small drop of super glue will be enough to have case and flexi(with sensor) back together.

Very interesting, thanks for your notes! I will look forward to photos also. Your technique sounds easier and cleaner than M33P's sandpaper approach. With the lens holder removed intact, that also makes it easier to experiment with different lenses. I wonder if it is possible to using 3D-printing to make a suitable holder for a similar lens or a wider-angle lens.

My experience with superglue and optics is all bad. The cyanoacrylates all seem to outgas and deposit an opaque whitish residue onto any nearby surface, especially within an enclosed space like the lens holder assembly would be. I gather the original assembly used epoxy, and that seems likely to work well enough. If you use a few tiny droplets of 5-minute epoxy it is possible to take apart later if have to, especially if you get it warm first which will make the epoxy soft.

[M33P]

When I'm back home today I'll put some pictures of my parts.
What I did, I used scalpel to detach black plastic case from flexi cable. Once they apart I had access to the filter but it was not loose inside so I had to unscrew lens to push it from lens' side out of the case. Then the case nicely sits back. I didn't have any damage to the wires doing that. Just make sure you don't put scalpel too dip and don't bend sensor metal bed too much. Fit scalpel in-between glue and case in flexi-glue-case sandwich to leave more glue on flexi rather than on the case.
I suspect that small drop of super glue will be enough to have case and flexi(with sensor) back together.
Just noticed that CPC is out of stock so my second attempt won't be soon.

I did consider this method at first but I deemed the risk of bending the metal backing (and thus cracking the chip) to be too great. I did attempt to find a seam on which to abrade rather than cut the epoxy but the bond is quite tight on 3 of 4 sides. The side with the ribbon offers a slightly bigger bead of epoxy.

The filter is glued to the plastic case, you can see the remaining epoxy in one of my photos so either way you have to remove the front lens to pop it out.

By practicing on the now-deceased chip I can use a (thoroughly cleaned) sable-hair needlepoint brush and damp lint-free tissue to remove pretty much all contaminants.

This is still pretty high-risk to even attempt on a working chip - I did not expect my sensor to survive the disassembly as it was a bit of a tear-down inspection.

[poing]

Very interesting, thanks for your notes! I will look forward to photos also. Your technique sounds easier and cleaner than M33P's sandpaper approach. With the lens holder removed intact, that also makes it easier to experiment with different lenses. I wonder if it is possible to using 3D-printing to make a suitable holder for a similar lens or a wider-angle lens.

Curiously both those lenses have a built-in IR filter

[sharix]

Very nice to see you succeed!

[scorp]

Here it is. Sorry for the quality, hands were shaking.
As you can see, when connector at the bottom, there are two caps on the right just next to the edge. It is easy to damage them so I would suggest trying to cut only on the left/top/bottom. Top and bottom caps are not as close as right one.

[scorp]

Curiously both those lenses have a built-in IR filter

First one has IR filter optional

[mathewL]

Great to see someone trying this before me. This is awesome!

before I go in with a scalpel or sand paper, I'd like to know if poing's suggestion might work-- could it be pulled out from the lens end with a dab of glue? does it look like the filter is flexible at all, or is it glass? how good is the mount to its frame?

[jbeale]

If it's glued around the perimeter to the frame, then the IR filter flexibility probably doesn't matter. I think you will not be able to overcome the epoxy edge bond (shear stress) by pulling with a center bond in tension, even if your epoxy adhesion is very good.

[scorp]

Great to see someone trying this before me. This is awesome!

before I go in with a scalpel or sand paper, I'd like to know if poing's suggestion might work-- could it be pulled out from the lens end with a dab of glue? does it look like the filter is flexible at all, or is it glass? how good is the mount to its frame?

Filter is a glass. It is wider than lens hole and difficult to detach from the case even by hooking on the edge from inside. I had to push it hard from lens side..