What did I learn by throwing a bottle into my water tank?

Samridh Tuladhar
6 min readMar 30, 2022

Alright, time for some janky engineering!

Follow along!

If you’ve ever stayed in a house, then there is this little annoying problem — you don’t know how much water is left in your overhead tank! It’s always annoying to find out that you’ve run out of water.

The solution is simple — just climb up, peek into the water tank, and then decide beforehand if you need to fill it up.

Well, climbing up on a water tank, which is on the top floor of your house and typically located at the edges, isn’t exactly “safe”. But by now, you must have realized I don’t care much about safety, it’s more about being lazy. 🤣

So, let’s build something to solve this problem.

Hey! Hold on a second. I see a lot of ads on Facebook for water pump automation, that uses a device to measure the water level, and it decides to turn the pump on or off automatically. Why should I bother reading further?

  1. It is expensive. Costs NRs. 7k just for the device, not to mention you will have to run wires from the top to the bottom of your house and other installation charges that you will always be salty about.
  2. This is more about the lessons I learned. What I made is pretty jank anyway. 🤣

So, around mid-Jan 2022, I thought, let’s solve this problem.

I started to design even more complicated systems myself. After all, it has to survive rain, harsh sunlight, dust, and whatnot. And I did what I should have done at the beginning — googled a solution.

Let’s skip my over-complicated designs, and jump straight to the solution that I found:

If pictures can speak a thousand words, then a gif can certainly do better.

I honestly felt a little stupid after I saw this gif. 🤣 God knows why I tried to use my brain so much.

But wait, Samridh! You don’t even have half the skills required to manufacture, deploy or operate this.

And you are right. Of course, I don’t — I’m frugal as hell. 🤣

So, let’s go full jank, and here is what I’ve “built” ( if you can even call it that ):

Certainly cheaper than anything else in the market, don’t you think?.

That’s right! Just two water bottles, tied to each other with some nylon thread. Why nylon? Well, that was the find thing I could find.

But, principal-wise, it’s very similar to the gif. The gif has a 2 point leverage system, whereas mine will only have a single bend, which implies that when the water level is low, the readout bottle will be at the top of the tank because the probe bottle will be at the bottom of the tank.

This is how you become an “engineer” — by using fancy terms like “probe” and “readout”. 😝

Alright, let’s quickly set this up

  1. Fill up the larger bottle with water up to 70% capacity.
  2. Toss the larger bottle inside the tank, keeping the smaller one outside.
  3. Fill up the smaller bottle until it starts to pull the larger bottle.
  4. Remove just enough water so that the larger bottle pulls back again.

Overall, this is how it would look:

The readout bottle is nearly at the bottom, which implies that water is almost full in the tank.

For the sake of completeness, this is what it looks like inside the tank:

The probe bottle is at the near top of the tank, which matches the earlier observation

I know the water is very dirty, we have a whole filtration system, whose article has been in my drafts for the past 6 months. 🤭

Please don’t tell me it works?

It works! But it isn’t without flaws. The most obvious one is friction in the system, and it slows down the readout system. When the tank cover is closed, this is what it looks like:

The tank cover has a crack in it, where I can slip by the thread and some extra equipment.

The tank’s cap and the fact that the thread is resting on the tank body are causing friction in the system. The only way I can foresee solving this is to use proper ball-bearing anchor points, but there is no way to set it up without damaging the tank.😞

The readout bottle needs some water in it, otherwise, a simple gust of wind can swing it wildly, causing tangling and other problems. But it works, it’s a little bit slow, and tugging on the readout bottle helps to “re-calibrate” it. 🤣

The shape of the probe bottle is a cylinder and not a sphere, so the bottle can balance itself both vertically and horizontally while it’s floating inside, this impacts the reading a little, decreasing the precision of the instrument, it’s not too much, but it’s very noticeable.😞

But seriously, what’s to learn from this junk?

  1. Before you start a project, always check on the internet if anyone has already solved the problem. I have cut out the over-complicated designs I had wasted several days thinking about, whereas this google search only took me about 5 minutes.
  2. Simpler systems are better — the fact that there is no electrical system, and it’s just gravity and buoyancy doing the work means that I don’t have to protect it from the sun, rain, wind, etc. This system is super easy to maintain, replace and operate!
  3. Shapes — put more effort into it! It changed the precision of the product!
  4. Thread quality needs to be researched, I’ve noticed that non-tangling headphones use a flatter wire…, that shape seems relatively immune to tangling.
    However, in this case, the flatter thread system would further increase the friction. Not to mention that I have no way to manufacture such a material.
    Additionally, the tangling occurs only during maintenance and does not cause issues during operations.
  5. While sending information from one location to another, the correctness decreases as the distance increases. In this project, had I increased the length of the thread, or added more anchor points, the readout would have been more inaccurate.
    This sounds very obvious, but I didn’t consider this a serious problem until very late into the testing, thinking I’ll figure something out along the way.
    I tried to use other alternate readout mechanisms, such as using sound, electrical signals, wifi, pressure, but regardless of the mechanism, it would get inaccurate by the time the information is observed 3 floors below where I typically stay.

Main takeaway

The major lesson to learn from this project is that if you want to have a monitoring system, then use something digital. And, by digital, I do not mean using fancy chips & computers, it is about using definite-discrete values instead of continuous analog values.

Say in this project, instead of this — I created a system, in which a bulb would light up a different color at 25%, 50%, 75% & 100% capacity. This information would be easier to transmit across distances, as there are only 4 values that can be read out vs. a bulb that shows a continuous range of colors from 0 to 100%.

For the sake of argument only — let’s consider that I use a series of mirrors to keep reflecting the light from the readout bulb to my room. The 4 colors would be significantly distinct to me, compared to the continuous range of colors, which could be misinterpreted because when light travels through such a large distance it gets refracted, dimmed, and several other phenomena along the way.

Let’s also consider that I run a bunch of wires from the top of my house to my room, the same issue will occur. I will be able to see the discrete 4 colors better than the continuous range because of the resistance in the wires which will affect the final readout.

This article got a lot of introspective and “science-y” by just throwing a bottle into a tank. 🤣

I YOLO-ed more of this project than you think while I was working on it. It was only during the time I was reflecting on it, I got these thoughts that I felt I should share with you all. Hopefully, the next project that I’m going to post will have more cool moving stuff. 🤭

Samridh Tuladhar (@tsamridh86)



Samridh Tuladhar

A computer engineer, with a passion for cheap, affordable & environmentally friendly automation and utter disdain for paperwork and waiting.