Home        Store        Docs        Blog

Building the world’s second consumer deep sea drop cam

Dear @rjehangir and @WorkshopScience, my son Pau and I are building a low cost deep-sea drop cam, inspired by the drop cams developed by National Geographic, tested to go down to 6,000m.

The specs for our camera are: up to 500m of depth, cost less than $500. In our design, our camera is a GoPro 4 Hero Session, activated by a WiFi-enabled Arduino microcontroller. We also designed and tested a burn-wire release mechanism, but in order to keep the whole solution cheap, we may have to sacrify sophistication and keep things as simple as possible. I think we will go with a galvanic timed release (need more research though). And still in design, lights will be mounted outside the enclosure that hosts the cam.

I have two questions for you:

  • We are prototyping our drop cam as an acrylic globe, made out of two domes 4", and in your website, we see that your dome end caps 4” are tested up to 500m. Our question is: do you think we can can attach two dome end caps and get a tight enclosure for our purpose? If yes, we just want to confirm that we will need the two dome end caps with their rings.
  • @WorkshopScience -We just discovered your drop cam, which has two dome end caps and a part that connects both caps, and it has some cable penetrators that allow for controlling or powering different elements. We assume this part was designed and 3D printed by you. Is this design open source? If yes, where could we get the 3D files?

That’s all for now,

Pau and Rafael Anta (Washington DC)

PD. Our target is to have the Paucam tested and ready for deployment in July 2019. First drops in Menorca Island!

Hello Pau and Rafael,

Your project sounds great, some aspects of it are quite similar to ours so I think I’ll be able help you out with a few things. The initially plan for the dropcam was to use a galvanic timed release like the one you are planning to use but after a bit of experimentation, I found that there are a few things that make it a bit difficult to use.

-In my design, the dropcam is supposed to work for a few hours hour per drop so finding galvanic releases that were so rapidly-corroding was difficult and they were also relatively expensive (around 10$ apiece). We got ours from https://www.underseareleases.com/info.htm

-The timed releases are also very heavily affected by temperature, there is a chart on the website linked above that shows release times at different conditions. As the dropcam descends, the temperature of the water will decrease so it can be pretty tricky to find out exactly when it will fully corrode. This is a problem since you won’t know when to start looking for it once it returns to the surface. Another consideration is that rapid galvanic releases need to be thin so can be mechanically weak.

The top camera housing flange you mention is actually made out of aluminium, a normal 3D printed part would not be strong enough for these purposes. The dome end caps aren’t designed to be attached to each other so I don’t think you will be able to get much pressure resistance if you do manage to attach them. Yes, I can send you the design file, could you email me at support@workshopscience.com for it. After some extensive testing, we found that the above-mentioned flange with a pair of normal 4” dome end caps could survive well past 500m. The weak points were the edges of the dome end caps, the flange on its own could probably easily go past 1000m.

I’d love to see your dropcam when its finished and please let me know if there is anything else I can help you with!

Kind regards,

Dear Ivan, thanks a lot for your quick answer! Sure, we will contact you to your email regarding the camera housing flange.

Let us share with you a few more details about our thinking for the release and recovery of the dropcam: in this trade-off between functionality vs cost, we still think we can have a very simple enclosure (two dome end caps, no holes for cables). In this case, we can’t control a burn wire mechanism for timed release. However, we could use the galvanic timed release and in order to compensate that its timing is not accurate, we could include a GPS (we like the Ultimate GPS by Adafruit) together with the Arduino MKR GSM 1400. With this solution, when the dropcam releases and gets to the surface, the Arduino could broadcast the lat-long of the GPS, and we could go find it by boat. What do you think?

More soon!

Hi Paucam,

I’m thrilled that you’re interested in doing this. I have a few notes:

  • The dome needs to seal against something that has an O-ring. That means that two domes back to back won’t seal, unfortunately. We built a special aluminum ring to attach them for this project.

  • I think the GPS and GSM tracking could work but I would definitely test it well first. Antennas very close to the water surface often have trouble getting good range. We looked into some low frequency tracking radios that we could find by pointing an antenna around, but we didn’t find anything suitable.

  • Also, note that our biggest challenge was keeping the camera cool enough! It often overheated in the dome enclosure until we added a small fan to circulate the air. If you pack a bunch of stuff in there you could run into a similar problem.


Hello Pau,

Thanks for reaching out to the email, I’ve sent you some more details. As Rustom already said, overheating made our Gear360 shut off after around 20 minutes but after installing a mini fan, it could run for an hour. This effect will probably be just as severe with the GoPro session however I haven’t personally tried it. Regarding locating the dropcam, another idea we had that Rustom mentioned was to use a system similar to that of the National Geographic dropcam: an RF radio beacon installed in the housing that would be pickup up by a directional antenna which you would hold and point in different directions to “hear” where the dropcam is. The only issue with that is that its difficult to find a beacon that would be powerful yet small enough but since then, I’ve seen a few new “Lora RF” modules that look quite promising so it may be worth to check them out. The benefit of this is that you aren’t relying on cellular connection that might be unavailable but you also have to bear in mind that the dropcam could drift quite far away while it is ascending/descending so you would need several kilometers of range to be on the safe side. This of course depends on the rates of ascent and descent that you get. The dropcam you saw in the video doesn’t actually have any GPS/radio locators, its just recovered by sight.

Kind regards,

Hi Ivan, many thanks to you again and also to Rustom. I spent a few days with one of the Nat Geo’s drop cams, which are amazing, and learned how to find it using the directional antenna.

I take note of the heating issue. We have several sensors inside the enclosure (including a motion sensor to start/stop recording) and will do several tests to measure temperature inside the drop cam to confirm whether we need the fan.

Remeber we put ourselves the challenge to do this drop cam below $500, so this makes us think every detail. We would really appreciate if you could share the file of your 3D object with us so we could rethink our design. And if we end up using it, we will give you the credit for that part (we will make our design open source).

Now, let’s move to the lighting of the drop cam for a second! In our design, we would like to install 2 lumen subsea lights. We think it is enough, but we saw four in your design, and we imagine it is because your cam is 360, is that right? And regarding batteries for the lights, what type of batteries do you use?

Thanks you!


Dear Rustom, thanks a lot for your feedback. This is what we realy needed! We will review or design and perhaps, we move away from the idea of two domes attached (just like a globe) to one dome and one aluminium tube. And we take note of every detail you mention.

Thank you!

Hello Rafa,

Sure, I’ll send the model as a reply to your email right now. Yes, there are 4 lights in our design because of the 360 camera however it would probably also work quite well with 3 since they have a large illumination angle. If you are using a single lens camera, 2 should be enough but you need to make sure you mount them in suitable location to avoid backscatter (there is a comparison of this in the video I made). You can use a range of voltages to power the Lumen lights but I used a 11.1V 3s 5000Mah Li-Po. In our design, this battery also powered the internal circuitry and the burn-wire.

Kind regards,

Ivan, thank you very much for sharing your insights and the flange model. One additional question: what type of material do you use for the burn wire? We have been testing with nichrome 80 30 AWG but perhaps you use a better material.

Will keep you updated on our progress!

Pau and Rafa

Hi Ivan, I hope you are doing well. We haven been progressing with the development of our drop cam. You can see attached the enclosure (the dome will face down). Now we are building the burn wire and are looking for parts to implement it. We like your red lever (we took a screenshot of your video), and we wonder where could we could buy something like this.

Last question: what’s the metal of the screws? Is that aluminium?

Thanks a lot for sharing your insights!


1 Like

Paucam or Ivan, has there been any investigation into electromagnetic releases? I’m hoping to build a dropcam along these lines in the near future, and something that concerns me with the burn-wire releas is that if the battery dies (or otherwise fails) before release, then the device is stuck. I know this can ultimately be solved with a galvanic release like was mentioned in Ivan’s most recent video - but I was just curious is electromagnetic release have been explored yet.

Dear Alex, our release mechanism will be burn wire, and will use a galvanic time release as a backup, but we are not very concerned with the risk of our battery running out of power. If you are really concerned, you could have a dedicated battery just for this.

Besides, I don’t know how to do an electromagnetic release. Instead, we do have a magnetic switch to turn the dropcam on.

Still building the dropcam. Here you can see two pictures.

1 Like

Regarding magnetic release, there’s this thing https://nicadrone.com/, and there’s also the possibility of using a beefy servo to move a permanent magnet.

1 Like

Hello Pau!

Apologies for the delayed reply, it seems like I must have missed the notification for your previous message from a while back. Anyway, to answer your question, for the burnwire, we’ve also been using nichrome wire and found that 32AWG worked quite well although we’ve recently had an issue with one of these not releasing for some reason. I’m not entirely sure why that happened and am still looking into it so please properly test your design before deploying it as our isn’t perfect yet. Thanks for sharing the photos! I made the red lever and the base to accommodate it myself so it isn’t available to purchase at the moment. I can however send you the 3D files for it in case you want to print it yourself. Both the 6 face screws and the 2 thumb screws are stainless steel (I think that the BlueRobotics face screws are made from a special non-corroding type of stainless steel). Looking forward to seeing your project progress and please let me know if you have any more questions!

Kind regards,

1 Like

Is this something that’s tethered, or are you looking at fuse/magnet methods of deploying some sort of flotation device to bring it back to the surface? Also, do you need to maintain orientation? Happen to have any links so I can read a bit more?

Okay, ignore me, I found the NatGeo video. That’s a fantastic project.

1 Like

Dear @rjehangir and @WorkshopScience, we are pretty advanced with everything and now, we are testing our burn wire. We have a doubt about the best option for the electronic circuit and I would like to know your opinion. In the fist picture, you can see the diagram of our circuit (pretty simple, we just activate the relay with a digital pin from the Arduino): my%20burnwire
And in the next diagram, you can see a circuit that I found surfing the Internet, and this is a burnwire used to release a GPS tied to a weather ballon:

The second diagram is more complex, it adds a diode and a transistor, but I am not sure it is critical. Any thoughts?

1 Like

Hello Pau,

We actually ran into this decision decision while building the current dropcam! From what I tested, pretty much all of the high-current relays required a relatively large activation current so the Arduino’s 40mA pin current limit would not be sufficient to trigger it. That was where more complex, second circuit came in: it allowed us to first activate a transistor directly from the Arduino which then in turn triggered the high-current relay. In essence, the transistor was working as a sort of intermediary relay to provide enough current to trigger the main relay. Right now I am working on redesigning our circuit to remove the need for relays entirely and to just use a few MOSETs (solid state components that are very similar to transistors) in order to briefly supply the high current necessary. This would make the footprint of this circuit a lot smaller and in theory also make the operation more reliable as there would be no moving parts. The only concern with such a setup is the over-heating of the MOSFETs due to the high current but since the wire burns almost instantaneously (under 50ms), this hasn’t been as much of a problem even when using only one MOSFET. Hopefully I’ll have some news to share with you about this new design in the next few days but to answer your question, out of the two designs you included, you would use to use second in order activate most high-current relays (of course, reed relays would not work as they are intended for low current usage but I am guessing that the first diagram is just a sample one you got from the internet).

Kind regards,

Hi Ivan, many thanks for your answer. We have been learning the hard way. The same day I posted this question, we decided to work with the second circuit, as you mention in your answer.

And here comes our (inadvertent) problem: in our dropcam we are using an Arduino MKR 1010 (WiFi), which can provide a maximum of 3.3V (I didn’t measure mA), but we did all the testing with an Arduino Uno, which can provide 5V. We built the circuit and it worked perfectly, but when we used it with the Arduino MKR, it doesn’t work. It’s 3.3V can’t activate the coil of the reed relay (requires 5V).

We just purchased a 3V Relay, will play with is as soon as we get it.

We look forward to seeing your update with the burnwire!

Pau and Rafa

What is the target amount of time from when the burn wire circuit is triggered, to when the camera is released?