This video explores the process of designing a Water Tank Monitor System from scratch. We explore the process in-depth, from both a technical and philosophical perspective. Facebook: / leos.bag.of.tricks
This is pure gold. Especially your analysis on how "competitive analysis driven development" can cause a products feature set to diverge from what people actually want.
Jeremy clarkson kind of outlined this in his interviews with Jennifer Saunders and Johnny Vaughn on top gear, except it more is designing a product to what a focus group think it wants rather than what people actually want or need.
Loved the video. I think it demonstrates how product development should actually be done: hands-on in the field, and not from behind the desk of a marketing department. Well done!
Super stuff. A friend of mine developed a very similar audible acoustic wave based system in the 1990s, but the target was heating fuel tanks for rural properties. It was a great success and was still in production a few years ago. It used a radio link at very low baud and transmitted once an hour or so. It was definitely in the more premium price bracket though.
I've got some(3) water tanks, and I really don't see the point of such device *unless* it can be connected to something. So I think it would be nice to have 2-3 relay outputs (tank empty, tank full, tank halfway). Maybe at the display side? so the main unit could be kept minimal (2 connections), and an additional unit (relay output unit) could be created and connected to the led bar (the data is already there). Please do some livestreaming, chit-chat videos also (less effort from your side), I'm really hooked:) Every video is pure gold. I've created with your prototyping technique a furnace controller board:)
I am a long time served professional design engineer and find your retoric very refreshing to hear. I agree whole heartedly. Having spent the time developing this stuff....how can anyone else get there hands on it or use it?
I really like the solution you came up with, I build one about the same time you were , but used a waterproof US sensor. First tried to place the sensor in a pipe but that didn’t work then I settled on mounting the sensor in a box that I mounted on top of the tank. Here in New Zealand everyone in a rural area have to use a water tank for household water and most commonly they are 25,000 or 30,000L The advantage is that on most of them the top is a dome so that gives you the 40cm clearance that US sensors need to work properly. After installation the system worked well but the next morning it didn’t, so I removed the sensor box from the tank and found that condensation had formed a drop on the sensor after wiping it off it worked well again but the next morning the same problem so I mounted a 12V lightbulb inside the box and run it on about 9V so it will never blow, this solved the condensation problem. It has been working very reliable now for a year and a half. Probably due to my age a am a firm believer in cable connections so I power the sensor box with 9V for the light and then make 5V with a regulator for the arduino nano, this arduino sends the water level in % out of the serial port into a RS485 interface so I can have a display up to several 100 meters away and also have multiple displays on the same line. Currently I have a simple 7 segment led in the pumpshed and a graphic display in the house as it using low speed RS485 you can connect it with cheap CAT5 cable. I really like your led solution that looks really cool.
I use a 30 kPa pressure sensor at the bottom of the tank connected to the analog input of an ESP8266 that sends the data using MQTT to my Node-red based home automation system. Works quite well and also shuts off the water pump so it doesn't run dry. Also not very expensive, the sensor being the most around 1000 baht from Lazada.
Leo - Love all your videos - so much thoughtful engineering and useful tips. I was thinking: 1) those vent holes look big enough to be access ports for mosquitoes. 2) I wouldn't want to add an external tube on the outside of an existing water tank for fear of seal failure and leakage.
1) In my (tropical south India) experience mosquitos usually don't breed in overhead water tanks. Maybe it is too high for them, maybe it gets too hot in the sun, I don't know. 2) Up there near the tank there is very little water pressure, so if the rest of the house plumbing holds up, not something to worry about.
Greetings from Argentina! Very nice design. I love all the analog part, it is generally lacking on most other creators (of any sort) and reduces the cost greatly (besides being super cool). The only weak point I see on the design is bugs. I think you should add some sort of filter to prevent anything water-seeking entering the tube. I've seen live frogs on supposedly "closed" tanks. PS. The DIY prorotyping tool is genious, I use a similar method but was never happy about the cutting method. Gonna try it soon
Awesome explanation, as usual, thanks for sharing! The analog processing section is very intriguing, I wonder though if it'd be possible to perform it in the/a microcontroller itself, in software
I dunno about that, my knocker actually broke off a long time ago and ive never put another on plus i lock the screen door on the outside so i cant hear anybody knock most of the time.,
Very nice a reliable solution to a practical problem that helps everyday life, excellent. I like the adaptive threshold circuit, that removes the headaches of the double, triple bounce and sound speed temperature variation. The wavelength in the pipe has a different than in free space does that cause a prpblem?
Worked for quite large companies. Learned one thing.. Decision makers are no-brainer finance (department) people, not engineers. All they know how to please their managers, not necessarily product / service work.
An excellent video, very cleanly communicated and edited. It's a cool product and I especially like the fact that it can be installed parallel to the tank. You accurately read my mind since the whole time I was wondering why the clicking was so frequent! Would it be possible to use lower frequencies, or would the convenient planar effects you mentioned go away? Also, if you're already using an enclosed pipe for measurement, could other measuring techniques be utilized? For example, a force sensor at the bottom?
That is a great first step, but how reliable are speaker, microphone and flood sense wires long term? The speaker forms a superb waterproof surface towards the container, but the back side is practically open, it's covered with felt leading to the chamber behind the membrane, where delicate wires live and will probably get corroded eventually. And while IP rating of the microphone guarantees that it's OK to submerge for an hour or two, i'm not convinced that the crimp on the rear PCB side of the microphone capsule, while perfect at first, doesn't become a capillary channel after a few temperature cycles, as the PCB jacks the crimp open by thermal expansion. Inside is an impedance sensitive JFET like K596 that is easily disturbed by even elevated ambient moisture. I also don't even know what they did to protect the front and how durable it is long-term. The rear of these components is in a waterproof chamber, but how water- and vapour-proof is it really? There's sense wires sticking right through the plate, there's connectors on top, there's the mounting holes for the microphone and speaker themselves... And once moisture gets in, can it ever get out? It's likely locked in there forever. I also remember someone testing all kinds of flowerpot reminders, all kinds of sensors and things conductive and capacitive that you stick into a pot of soil and which are supposed to tell you when the moisture level is too low and you have to water your plants, and none of them survived a few months. This makes me suspicious of flood sense wires. Not to say that they can't be done, just to say that special measures are needed for sure. Outdoor stuff is rough. I have an RF thermometer here that i can't get to work reliably, it always fails in the winter, i reflowed every joint with leaded, and i waterproofed it best i can, lacquered the electronics etc, and it's still wonky! And there's barely even a proper winter where i live, it snows about once every 4 years! The single most robust component i can think of is a piezo disc transducer, and i can vaguely imagine it being used for both generation of the pulse and sensing. These sit at the bottom of a humidifier and last a good few years. Should that be viable, you might just laquer coat the finished PCB and not have to worry about moisture in the PCB chamber.
Perhaps instead of soldering the full tank probes in, could you use Mill-Max pins and basically press fit them in? Put the two connectors on their own PCB with pins that go to the main PCB. Install the connectors into the top shell, put the main PCB down over the connector PCB and solder, install the bottom shell, then insert the probes from the bottom into the Mill-Max pins. I know you don't like radios, but an alternative model could use the same basic main PCB and LoRa radios. Low power, incredible range, and low cost. You could run the transmitter side off a Tadiran TL-5930 Lithium Thionyl Chloride battery for 10 years, easy. The unit the house could remain LEDs, or you could go to a smaller display at some point. One major feature you have to your advantage is that you only need unidirectional communication.
Great vid! Could you maybe make the case halves thread together? Have a circular, through-hole connector centrally mounted on the PCB (pointing up). If your tolerances were right, the lid could compress an o-ring on the perimeter of the base and one on the connector body as it bottomed out. Also, maybe 2 stainless PCB mount screws that are long enough to function as probes?
Have you thought about surge protection on wire inputs? I didn't see any in the schematic. Long ago we had a TTY current loop on a wire between buildings. After a thunderstorm, the serial port on the minicomputer fried, even without a nearby strike. Fortunately they repaired the computer by replacing a few chips. A door bell solenoid is tough, but not transistors. Now wired sensors always remind me of that lesson from my job in high school. Using a separate tube makes a lot of sense, and the full level probe a good idea. Does tube diameter matter? Can a cheap 1" pvc pipe work? Now I wonder if you could use a resonant circuit, a water trombone, with a frequency meter for more reliability and accuracy. I've been looking into a capacitive water sensor with 2 concentric pipes, but haven't had time to experiment. Maybe acoustic in a tube is better, especially for longer distances.
You are exactly right- it needs an extensive protection network on the power and data connections. Step one is validate the overall concept before investing the time to develop and test the protection scheme, it's not a simple matter to test this stuff!
I wonder why a classic gas tank gauge wouldn't work. Maybe has to do with corrosion or the design of the tank, as the lever in a gas tank gauge is designed for longer tanks, not taller tanks, but I'm guessing the values could be converted.
Bruh why not use a float and potentiometer? Maybe a donut-shaped 3D printed part that slides up and down a rod with the water level, and the rod has a carbon track embedded in it that the float drags a contact along. Then the rod can just be cut to length and the resistance value for empty calibrated. The track could even be inside the rod and coupled to the float with a magnet, if it's not a good idea to have it exposed to the water. Or, the float could pull on a rope that's directed out of the tank with a pulley and counterweight or spring, so the electronics are out of the tank entirely. Or the pulley itself is what's measuring the level and it's an encoder. Just throwing some ideas out there c:
I have seen a related solution as a water level sensor in a humidifier. It was a longish cylindrical protrusion in the plate, in which a reed switch was placed. On the protrusion, would be placed a plastic float with a magnet inside it, and it would be suspended by the water. For this design, this would make a robust flood indicator that can be waterproof and can replace the sense wires currently foreseen. Just needs some kind of cap or circlip that would prevent the float from sliding off the cylinder. The reed switch can be placed such, that when the float rests on the limiter, the float activates the switch, and when the water rises above the float resting position, the connection is interrupted. However the float exposed an issue with the mechanism, that contaminants and dissolved calcium salts from the water would accumulate on the cylindrical track and the magnet, and eventually the float is stuck in one spot. Of course, this spot will be where the usual water level was, so when the float failed to descend as water was depleted, the humidifier overheated and burned out, and required replacement of the piezo plate and power transistor. If the device was used as an emergency indicator that only sometimes (and preferably never) gets submerged, maybe that issue can be avoided. You also absolutely don't want to build a potentiometer exposed to water, for the same reason. My first thought following up from yours was to place an oppositely oriented magnet and have it drive the potentiometer inside the cylinder; however constructing a meter long potentiometer is hard, so it would be much simpler to make a tape of regularly spaced Hall switches similar to LED tape, and use that to read the float magnet. The device in question is Fakir Ultronic LB10, which had been around since at least the late 90s if not earlier, so there can't possibly be any active patents left on it, if there were any to begin with.
Very cool video, I'm a fan of your work... Personally I would have approached the problem in a "silent" way using light instead of sound (see the I.C. VL53L1X). As a curiosity, how much energy does the sensor side of the device uses? (in case you want to go wireless with LoRa or similar...)
Great tutorial, I think using a cheap time of flight sensor would be a lot less work and cheaper, no need for an external pipe or the flood sense wires. Maybe use an esp32 to make it wifi enabled
I have a small boat with 2 fuel tanks, 2 water tanks, and 1 wastewater tank, and zero working tank guages. Your solution might find a market in the boating world. Wires would be no problem, and you'd have 12 or 24 volts DC available. I think you'd need to move your vents down into the tank. Boat tanks have vents.
very nice. do you sell it? I've been looking for something like this.. suggestion: the face of the printed part that holds the speaker and mic should not be "flat", but shaped like an anechoic chamber : this should reduce rebounds
So what happens during a wind storm and creates the bottle effect like when you blow over it making the bottle a flute.. or what happens if the tank constantly has a drip on the outside or any external noises that happen to hit inside that signal window? Because inside a tank when a noise is made the sound is amplified and distorts which might create a false read... betting you tested all of this long ago but just wondering...
Great video, really helpful! What kind of external connectors did you use? I've used GX16 plugs for a couple recent projects but yours looked much higher quality.
at first they look like waterproof christmas lights connectors they make with 2, 3 and 4 pins... but they really aren't i think lights used some specific connector as inspiration
As a fellow electronic designer myself, I really enjoy watching your videos. Thank you very much for sharing your work with us. My 2c : what about an "amphenol like" circular connector screwed on the top lid that would plug directly in soldered contacts on the main board when you assemble the whole thing. No wires.
Hi Leo. Your analogue section can potentially be replaced with a PGA460, have you looked at that chip? Then you can use a single transducer (like a bog standard 40kHz proximity sensor). I did not look at the total cost of your solution so I do not know if that would make sense from that perspective. Let me know if you need any help with that.
Good product explanation, but too complex. I would use one amp & use ADC to do the signal processing in digital. Also, I live in Pakistan, & EVERY house has two tanks. Some simple & reliable solutions are available. My favorite one just uses a pressure transducer, a microcontroller & a seven segment display showing 0-100%. Designed by my friend. Very elegant. He sells it for about 15$. It can be installed anywhere in the house & connects to the plumbing via an air tube.
Super. Very good project. Very well explained. You seems to have thought to everything. But what about the humidity getting in your speaker and microphone? It would be a shame to waste all that beautiful knowledge of making that board if it dies because of the humidity. I have used those ultrasonic sensors (similar process) and tried to waterproof them but the humidity always get to them. I really really like your design, but... afraid that there isn’t much we can do to protect the speakers and mic. Have you though of something in that regard?
Hello Leo! I am thinking about using HC-SR04 with a pipe. Have you ever tried such basic sensor in this configuration? In my case we are speaking about underground water tank. Sometimes it gets overflowed. I am affraid that such event will be a disaster for electronics. (mic and speaker mostly). Perhaps this can be avoided by turning upper part of the case into a cup? You can put the electronics into this cup so air pressure can prevent water from turning the electronics into a piece of junk. Connectors would become troublesome however this is easiest way of flood prevention. Polymer coating is a must have to protect from the vapour ofc. Best Regards!
couldn't you get by with a single sample-and-hold peak detector? since you only need the threshold on the second pluse, you could use the first pulse to "prime" the S&H and then measure the delay with the second one
i didn't get how it's calibrated exactly? does it try to switch like 1000 adressable leds on and then measure current to figure out how many is actually left after you did cut extra ones off. then turns that into maximum depth by adding distances between leds up to millimeters? kind of makes sense for me
Simpler than all that. It lights up the LED's in direct relation to the distance from the water level. The LED's you cut off would not light up anyway, if the tank is 6 feet tall, the lowest water level is 6 feet down.
Another excellent video ! Are the speaker and microphone waterproof ? Could you please show a bit more about your smd prototyping technique next time ? Looks nice. Thanks for the quality content.
Just remembered you made a great video about soldering : ru-vid.com/video/%D0%B2%D0%B8%D0%B4%D0%B5%D0%BE-vq968AFgPhg.html My bad, I'll watch it again 20 time.
I am happy to have found your channel. It seems that a mind like yours is in extinction, paradoxically due to the flood of cheap Arduino modules, which limits the detailed research of the problem to find the most appropriate and reliable solution to the problem. Modular is great, but in this way, creativity and technical skills are killed in the long run, and to be a good product designer you need both (the "Arduino generation" is already creatively dead). I believe that your channel will gain in value and views over time, although it will be increasingly difficult for most people to understand what this guy is doing😄. Keep going, I will be your follower for lifetime
I agree with your sentiment, but it's a bit harsh to say "the 'Arduino generation' is already creatively dead" since having access to easy learning/prototyping tools can enable those with creativity to actualize their goals. Of course there are a lot of people who will utilize the tools to make braindead ideas, but that's existed forever. Look at the existing water measurement tools showcased in the video as exhibit A.
@@ollie-d Lack of creativity is a general problem in a world of abundance not only of the Arduino generation. Creativity is strengthened when you have few things available, not when you have too much of everything already done. How would you make this water level measurement tool even more simple, cheap, reliable, precise and robust than the one from the video with Arduino, without the use of an ultrasonic or laser sensor? Describe your solution and I'll describe mine😉
I understand the appeal of non contact methods of measurement, but wouldn't a donut float with a small tube in center (with conductor loop inside) and a magnet glued to to the float be the best low tech solution for sensor development? Then the sensing mechanism is just a dumb magnet float which sends an analog signal along the conductor inside a tube. Process this analog signal. The position of magnet is then calculated like an LVDT sensor moving its magnetic field with the surface of the liquid in the tank.
@@leosbagoftricks3732 www.omega.co.uk/technical-learning/linear-variable-displacement-transducers.html I realize the idea I presented is not as simple as I made it out to be. I said magnet when I guess I meant a "core" and way more complicated wiring setup with primary and secondary coils were needed to produce these sensors that what I suggested. I haven't use these type of sensors in over 15 years back at the university, so I forgot how they actually work. They are "variable transformers" with coils that register the "movement" of a core with an oscillating supplied voltage. The readouts are V/mV/mm or V/mV/in. Full disclosure, I am not an EE, but I cant help but think there is a conceivable and known geometry in the universe we can engineer that couldn't "sense" position like this device and still be made from "dumb" and "cheap" components like a magnet and wiring configuration somehow. I was just thinking that copying this idea has merit from an engineering a robust sensor. It would only be a float and some geometry of wiring (hopefully easily controlled in a manufacturing environment). The problem is of how to place alternating primary and secondary coils vertically along many different tank depths without needing different manufactured products and have the core on the float is hard to solve. Obviously this idea is not even ready for prototyping, more thinking is required to feel confident to waste materials on this idea.
Consumer behavior ultimately dictates weather a product will be successful or not too! A completely separate animal from good engineering. I would think most water tank designs probably come with an autofill solenoid valve will supply water from a city municipality and then cut off with a couple of float switches installed at the correct depths before tank overflow or being drained completely. 99% of the time the consumer would know they have water in their tank with no monitoring. The switch and valve mechanism doesn't cost a whole lot and requires almost no knowledge of microcontrollers or more sophisticated instrumentation. Then how many more scenarios are we engineering for? There is probably a segment of the consumer market that loves data. The "I want to know everything that's going on in my tank" people. There's definitely industrial clients that want to know everything going on in their tanks. But my guess is both of our ideas at continuous sensing in water tanks wouldn't sell that well in the consumer market when a simple float switches installed at the correct depths is super cheap and set up an auto-filling valve. This makes both of our ideas obsolete or selling to niche crowd. I must admit I am a part of that crowd, I love sensors and smart home crap. Its funny that I get so mad at all the truly crappy smart home products that its sole development by a companies marketing team (not engineering team) was meant for me to buy it cause I am curious and then it doesn't work.
@@mikenorfleet2235 Water tank designs in many countries involve a pump which pulls water from a well and puts it in an overhead tank. Overflow sensors are your ears, when you hear water hitting your yard you go and turn the pump off. At that point you know exactly how much water is in the tank.
I would put a third prong in - for a 'High-High' reading - basic all shits hit the fan Sirens going off call the emergency services signal. Seriously, You don't need accurate sensing so close to full. Use three prongs - one set indicates tank full (High) and another set for error (pump has not been switched off when 'High' level has been reached) so over full hence 'High High'. How about a 'Lo' & @ 'Low Low' for pumped output applications to protect the pump/motor from running when dry. Use stainless with a crimp on it. Noise of ping is too load (even at 15sec intervals). Coloured LED down side - power drain unless you incorporate a solar charge and battery arrangement and might be hard to see in strong sunlight (unless high power of course) Vent holes to big/many possible ingress to crawlies and the like, or use a course foam filter to screen and protect probe connections from moisture through condensation. You can get a vent specifically for this purpose.
Solution for you wire-probe problem: 1) Use stainless-teel for the probes 2) Sole the soldering problem by electroplating copper onto the last 5 cm of the probes. Attach a DC supply to a clip lead attached to some bare pc board or other source of copper and attach the other side of the supply to the stainless-steel rod. I’ve had good luck using water/vinegar for an electrolyte. This has worked well for me for years, for example when I want to solder a SS rod to a PL-259 or SO-239 to make an antenna. I’m sure it will work for you too. -Greg, WW5GG
While I love your videos I think you're guilty of the same sins you're criticising. Your system is complex with more points of failure than I have fingers to count. You want a reliable system? Piece of wire, a float, a pulley and a counterweight. That'll last 40 years and anybody can install 🙂