<HOME
<Photography
<Electronics

Deep water camera DIY

There is a lake in Orivesi, Finland, called lake Äväntäjärvi, where I've been fishing with my father many, many times. It's not a big lake by any means, but curiously, there is an almost 50 meters deep abyss in it! In Finland, that's a really significant depth for a lake by any standards, not to mention a lake only this size! You know I wanted to sink a camera into it, to see what nightmare monsters lurk down there, and the Soviet nuclear submarine!
Rather than trying to make a watertight pressure vessel for a camera, I reasoned that the camera should withstand the pressure just fine, as long as it stays dry. And as long as I try not to maintain any pressure differential between the camera and the lake, there'd be no reason for any water to ingress.

I already had a bunch of cheap MD80 "MiniDV" cameras from eBay, which had flown onboard a couple of rockets. They were already modified for external control, so a microcontroller could mimic pressing their POWER and REC buttons, and read their two status LEDs. I thought I'd destroy them for kicks.
Here's my first attempt, which did succeed in destroying one. I took a piece of clear acrylic (this photo is looking at the acrylic side), taped the camera, 4×AA battery holder and a powerful LED to it, and taped a rubber film on the other side (I used a section of pink exercise resistance band). Here's another photo from the other side. I tried to leave as much air inside as possible by including bunches of crumpled tissue paper—the air should just compress as the pressure increases, and the inside pressure should be the same as outside, right?
After its deep dive, the thing surfaced all waterlogged, with the camera successfully destroyed. The "Gorilla" tape holding the rubber film to the acrylic had separated somewhere. The LED was still shining, and the camera still had power, but it was apparently not recording anymore, and was totally unresponsive. No video at all could be retrieved from its micro-SD card, not with my skills anyway.

Next I thought of a hot water bottle! With a couple of aluminum project boxes with rubber gaskets, and some threaded tubes and nuts (they are used in some lighting fixtures) and silicone caulk to seal everything, I envisioned a space station like assembly of modules, connected by tubes. One module would contain the batteries (four CR123 lithiums in series, which fit in a 2×18650 holder, and should provide 1 A of current for a whole hour, while maintaining at least 9 V voltage), LEDs and controller. A separate module would contain the camera (just to separate it somewhat from the light source), looking out through a hole covered with clear acrylic. The hot water bottle would provide a compressible air reservoir.
I calculated the total interior volume of the boxes and connecting tubes, which came out to about 200 ml. At a depth of 50 m, the pressure is some 5 atm more than at the surface, so I needed some 5×200 ml = 1000 ml of additional air to compress. The hot water bottle holds 1500 ml when not inflated to stretching, which gives ample safety margin as well. Some 2 kg of weight will be needed to sink it. A bit more for safety margin. 2.5 kg looks to be a standard size for weight lifting weights—perfect!

A linear regulator provides power for the camera and its controller, and a simple single-transistor constant current source powers two 3 W LEDs in series at 0.5 A (yes, that's below their max specs, but I wanted to be lenient on the batteries). The regulator and transistor are mounted on the project box lid, which acts as a heat sink. One LED is fitted with an additional lens to spread out its beam somewhat. I just could not fit the LEDs with their reflectors inside the box, so I mounted them outside, also using the same lid as a heat sink, and made an enclosure for them out of acrylic. This enclosure increases the volume of the system, but I think the hot water bottle should still just provide enough air to compress. Not much safety margin left, though.

A PIC16F690 microcontroller controls the illumination and the camera (which was already modified for external control). The PIC is needlessly big and overpowered for this purpose, but I just used what I happened to have in shelf. The camera was originally powered by a single internal LiPo cell, so it likes to run on 3.something V, now provided by the regulator. Happily, the PIC is also fine with that voltage.
I use a reed switch (actually a reed relay, with its coil not connected—again, what I had in shelf) to magnetically control the PIC from outside the box. I power up the thing with a magnet nearby, then close the lid, pump up the hot water bottle, and when everything's ready, I remove the magnet to tell the PIC it's time to start shooting. That way I can prepare everything comfortably indoors, and not have to expose anything critical to the weather. After the dive, I replace the magnet to tell the PIC to shut down the camera cleanly.
Just in case something goes wrong during the dive and the camera stops recording, the PIC monitors the camera's status LEDs and attempts to restart it, power cycling if necessary. The PIC also stops and restarts the camera every now and then, so it can write and close the current video file and start another one. You see, I am expecting this thing to leak regardless, and to destroy the camera. I just hope the SD card will contain something that I can recover, hopefully from the deep end of the dive.

Here's the complete contraption attached to a length of aluminum profile. It seems stupid to put the buoyant hot water bottle at the bottom end, and I was initially about to put it at the top without thinking, but then I realized that the squeezable hot water bottle will determine the internal pressure of the entire system. Placing it at the bottom where the hydrostatic pressure is greatest maximizes that pressure. The other modules will thus have higher internal pressure than the hydrostatic pressure they experience on the outside! So if (or when) anything leaks, it should just be air leaking out, rather than water leaking in! There's a Dunlop-type bicycle tire valve in one of the boxes, so I can easily fill the hot water bottle with a bicycle pump before sending it to the deep. I can also over-pressurize it just a tad, to provide some more safety margin (which I lost because of the LED compartment issue mentioned above). This valve certainly must be higher than the hot water bottle, so it won't admit water in!
I cut notches at the ends of a 50 cm piece of wood, and wrapped an ample amount of cheap Chinese "100 kg test" fishing line from eBay on it. (I've tested the stuff with an actual pull tester, and it only stood up to some 50 kg, but still good enough for this use.) One full turn thus equals one meter down. I already used this with my earlier attempt, and it at least worked just fine. It was a somehow eerie feeling finally hitting bottom after unwinding almost 50 m of line!

But what do you know, the stupid thing actually worked! I dropped it into the depths of Äväntä on 2022-10-15, and over half an hour it actually recorded video as it should, all through the dive! After surfacing, the camera and battery compartment were both absolutely dry, the camera was still in working condition, and the SD card contained video shot at nearly 50 m depth! Nothing very interesting, however—I now realize the camera and light source should have been angled downward, instead of looking straight out. Now I mostly caught brief glimpses of mud kicked up from the bottom, and where the actual bottom is visible, it's impossible to say even whether it's rock or gravel or silt. But regardless, these images were captured at a depth of almost 50 meters, using nothing more than a $20 eBay camera and a hot water bottle! How's that for DIY!!!???

Here's some of the very best frames from maximum depth and from the steep incline along the way up. Ignore the time stamps, they are bunk, but cannot be disabled in the cheap camera. In the first frame, something bright was moving at the lower edge of the picture—could that actually be a fish or a shrimp, or is it just a dead leaf the camera kicked up from the mud?

So there you have it: murk, dark, mud, and compression artefacts. If you're really interested, here's two of the best pieces of video (the first one has the "fish" or whatever): video1.avi (19 seconds), video2.avi (53 seconds). Actually, the eye can perhaps make out more detail (not just artefacts) from the moving image than from individual still frames, although the video is darker (it's straight out of the camera, whereas I adjusted the still frames a bit). Most of the sound you hear (when it's not destroyed by the camera's brain dead auto gain) is transmitted through the taut fishing line, as it scrapes against the side of the boat. Otherwise I'd guess it's pretty quiet down there, and being sealed in its container, the camera doesn't hear much anyway. At least while the camera was still in the boat, our voices were muffled beyond legibility.

Now that I've seen that the system works, I'll definitely angle the thing downward, and go for another dive next spring! That time I'll find the Soviet submarine for sure. Also (note to self!) make sure to trim any unnecessary ends of line away, so they don't cross the camera's field of view all the time. A small float on the line, some way up from the contraption, would also help keep the line out of the camera's way.

Yes, I know diving enclosures are commercially available for Gopro cameras and whatnot. I know they are rated for at least 40 m depth or whatever. But this project was all about DIY. But not only that, those commercial enclosures are pressure vessels which will break or leak when taken down deep enough. As long as there's enough air available to compress (add more hot water bottles if necessary), there's no limit to how deep this thing could go! Of course, the camera will experience the full hydrostatic pressure of that depth, but (unless something in the optics is sealed airtight) I don't see why a solid state camera shouldn't survive simple pressure just fine! Not getting wet is the issue, which can be resolved either by a sturdy pressure vessel, or as I did, by equalizing the pressures inside and out.

Antti J. Niskanen <uuki@iki.fi>

	There is a lake in Orivesi, Finland, called lake Äväntäjärvi, where I've been fishing with my father many, many times. It's not a big lake by any means, but curiously, there is an almost 50 meters deep abyss in it! In Finland, that's a really significant depth for a lake by any standards, not to mention a lake only this size! You know I wanted to sink a camera into it, to see what nightmare monsters lurk down there, and the Soviet nuclear submarine! Rather than trying to make a watertight pressure vessel for a camera, I reasoned that the camera should withstand the pressure just fine, as long as it stays dry. And as long as I try not to maintain any pressure differential between the camera and the lake, there'd be no reason for any water to ingress.
	I already had a bunch of cheap MD80 "MiniDV" cameras from eBay, which had flown onboard a couple of rockets. They were already modified for external control, so a microcontroller could mimic pressing their POWER and REC buttons, and read their two status LEDs. I thought I'd destroy them for kicks. Here's my first attempt, which did succeed in destroying one. I took a piece of clear acrylic (this photo is looking at the acrylic side), taped the camera, 4×AA battery holder and a powerful LED to it, and taped a rubber film on the other side (I used a section of pink exercise resistance band). Here's another photo from the other side. I tried to leave as much air inside as possible by including bunches of crumpled tissue paper—the air should just compress as the pressure increases, and the inside pressure should be the same as outside, right? After its deep dive, the thing surfaced all waterlogged, with the camera successfully destroyed. The "Gorilla" tape holding the rubber film to the acrylic had separated somewhere. The LED was still shining, and the camera still had power, but it was apparently not recording anymore, and was totally unresponsive. No video at all could be retrieved from its micro-SD card, not with my skills anyway.
	Next I thought of a hot water bottle! With a couple of aluminum project boxes with rubber gaskets, and some threaded tubes and nuts (they are used in some lighting fixtures) and silicone caulk to seal everything, I envisioned a space station like assembly of modules, connected by tubes. One module would contain the batteries (four CR123 lithiums in series, which fit in a 2×18650 holder, and should provide 1 A of current for a whole hour, while maintaining at least 9 V voltage), LEDs and controller. A separate module would contain the camera (just to separate it somewhat from the light source), looking out through a hole covered with clear acrylic. The hot water bottle would provide a compressible air reservoir. I calculated the total interior volume of the boxes and connecting tubes, which came out to about 200 ml. At a depth of 50 m, the pressure is some 5 atm more than at the surface, so I needed some 5×200 ml = 1000 ml of additional air to compress. The hot water bottle holds 1500 ml when not inflated to stretching, which gives ample safety margin as well. Some 2 kg of weight will be needed to sink it. A bit more for safety margin. 2.5 kg looks to be a standard size for weight lifting weights—perfect!
	A linear regulator provides power for the camera and its controller, and a simple single-transistor constant current source powers two 3 W LEDs in series at 0.5 A (yes, that's below their max specs, but I wanted to be lenient on the batteries). The regulator and transistor are mounted on the project box lid, which acts as a heat sink. One LED is fitted with an additional lens to spread out its beam somewhat. I just could not fit the LEDs with their reflectors inside the box, so I mounted them outside, also using the same lid as a heat sink, and made an enclosure for them out of acrylic. This enclosure increases the volume of the system, but I think the hot water bottle should still just provide enough air to compress. Not much safety margin left, though.
	A PIC16F690 microcontroller controls the illumination and the camera (which was already modified for external control). The PIC is needlessly big and overpowered for this purpose, but I just used what I happened to have in shelf. The camera was originally powered by a single internal LiPo cell, so it likes to run on 3.something V, now provided by the regulator. Happily, the PIC is also fine with that voltage. I use a reed switch (actually a reed relay, with its coil not connected—again, what I had in shelf) to magnetically control the PIC from outside the box. I power up the thing with a magnet nearby, then close the lid, pump up the hot water bottle, and when everything's ready, I remove the magnet to tell the PIC it's time to start shooting. That way I can prepare everything comfortably indoors, and not have to expose anything critical to the weather. After the dive, I replace the magnet to tell the PIC to shut down the camera cleanly. Just in case something goes wrong during the dive and the camera stops recording, the PIC monitors the camera's status LEDs and attempts to restart it, power cycling if necessary. The PIC also stops and restarts the camera every now and then, so it can write and close the current video file and start another one. You see, I am expecting this thing to leak regardless, and to destroy the camera. I just hope the SD card will contain something that I can recover, hopefully from the deep end of the dive.
	Here's the complete contraption attached to a length of aluminum profile. It seems stupid to put the buoyant hot water bottle at the bottom end, and I was initially about to put it at the top without thinking, but then I realized that the squeezable hot water bottle will determine the internal pressure of the entire system. Placing it at the bottom where the hydrostatic pressure is greatest maximizes that pressure. The other modules will thus have higher internal pressure than the hydrostatic pressure they experience on the outside! So if (or when) anything leaks, it should just be air leaking out, rather than water leaking in! There's a Dunlop-type bicycle tire valve in one of the boxes, so I can easily fill the hot water bottle with a bicycle pump before sending it to the deep. I can also over-pressurize it just a tad, to provide some more safety margin (which I lost because of the LED compartment issue mentioned above). This valve certainly must be higher than the hot water bottle, so it won't admit water in! I cut notches at the ends of a 50 cm piece of wood, and wrapped an ample amount of cheap Chinese "100 kg test" fishing line from eBay on it. (I've tested the stuff with an actual pull tester, and it only stood up to some 50 kg, but still good enough for this use.) One full turn thus equals one meter down. I already used this with my earlier attempt, and it at least worked just fine. It was a somehow eerie feeling finally hitting bottom after unwinding almost 50 m of line!