I have 2 20-gallon fish tanks that struggle to stay cool enough for my fish’s liking during the summertime. A while back, I installed 2 80mm computer fans with a small 12-volt wall transformer to keep the tanks cool. This solution worked well, but it ended up overcooling the tanks, causing the heater to kick on frequently. While this was generally fine, it led to significant water evaporation, as expected.
Recently, I purchased some small OLED screens for my microprocessor projects and thought it would be fun to build a fan controller for the fish tanks. There are many interesting features I’d like to incorporate, even if they aren’t necessary for this project, as it would be a great learning experience. I’m sure there are better ways to approach this, but this is my adventure into this project.
08/15/2023
Finally got some time yesterday to really start this fan controller project. I got the screen running with a text library working, an encoder wheel dialed in and debounced, and two thermostats mostly calibrated. The thermostats are accurate at about 70 degrees and go higher as the temperature increases. I just need to get the power supply I am going to use, take some new points, and input that function. For now, it’s fine. I also got a photoresistor working. All sensors are polling at the correct times and storing their data into global variables.
And something I’ve never done before: this IC has one kilobyte of EEPROM. I’m using it for savable settings like the polling intervals and set temperatures, so they’re not hard-coded. I also made it so that when it boots up, if you’re pressing the button, it will load and save the default settings. This was originally because I don’t have the menu structure built up yet to change variables and save them, and I needed to test things I was still designing. However, I think I’m going to leave that feature because it seems like a great idea to be able to reset to defaults, especially since I don’t plan to do a lot of input verification. This way, you could break it, and being able to reset the defaults sounds like a great idea.
I also made the EEPROM easier to use. Instead of writing one byte at a time, I check what size variable is being stored, break that variable into bytes, and store those bytes separately. When retrieving the information, it does the opposite: it checks what variable type you’re storing it into, pulls that many bytes, and strings them together for that variable. This way, I can store any size variable without having to think about it too hard, which is awesome.
08/15/2023
All right, I think I’m done for the night. Instead of turning the fans on and off with a relay, I decided to use a MOSFET instead. I can switch this on quite quickly. Using one of these would allow me to control the fans with a PWM type controller. Each millisecond, I’ll turn the pin on, then turn it back off between millisecond one and two based on an integer for the duty cycle.
At 255, it will turn off right at the very end of that first millisecond and immediately turn back on, meaning the pin was on for the entire millisecond. At 127, it will turn off halfway between 1:00 and 2:00, so the pin is off half the time. This gives me a pretty good PWM signal at one kHz. If you look at my cheap oscilloscope, you can see it rising from 0 to 255 over the course of a full second, starting at a 0% duty cycle and going to 100% duty cycle to verify it works. I think this will work out great.
Instead of having thermostat hysteresis, I’ll be able to have it ramped. Actually, I think I should still have some hysteresis so it’s not toggling on and off when it’s at the minimum temperature to start the fan.
08/20/2023
I got in about a half hour of work on this. I added little fan logos with each temperature that animate when the fans are on. I also made a bar graph showing the PWM signal going to the fan, ranging from 0 to 255. This will indicate how hard the controller is trying to cool down the tank. I also set up input events for the individual page, so now when I turn the knob or click it, it’s receiving events, although I’m not doing anything with them yet.
In the blank space on the top right corner, I’m planning to add some small pictures for a menu. I want one for a graph of the last 24 hours and another for the main text menu structure. Turning the knob left or right will change these pictures, and clicking the knob will go to the selected page. This leaves the bottom right empty space for something else.
I’ve got about 50 pixels left to right, so I’m thinking of maybe doing a summary graph of the last 24 hours with no text indicators, just to use that space. However, if I come up with something different, I might use it for that instead. I initially tried putting the temperatures on the same line horizontally instead of vertically, but I couldn’t fit both three-digit numbers without it looking confusing and leaving a lot of top and bottom space. I think that’s enough for tonight.
08/21/2023
I made some logos for the menu selection on the main screen.
08/24/2023
I’ve got some of the menu written, and this is what it will look like. The menu screen will be able to run a function, change a variable, set a checkbox, and change the menu (submenus). I’ll work on scrolling the list when you hit the bottom and top tomorrow. In the video, you can see a little bit of a delay. I’m not sure what’s causing that, but I ran out of time to investigate.
Here’s the menu structure that I have so far:
- Main
- Polling Timers Menu
- Back
- [Empty slots for future options]
- Screen Auto Off
- Back
- [ ] Use Photo Sensor
- Photo Threshold
- Auto Off Timer
- Temperatures
- Back
- Tank 1
- Back
- Fan Low Temp
- Fan High Temp
- Hysteresis
- Tank 2
- Back
- Fan Low Temp
- Fan High Temp
- Hysteresis
- Adjust Temp
- Back
- Tank 1
- Back
- Read Temp
- Actual Temp
- Read Temp
- Actual Temp
- Read Temp
- Actual Temp
- Read Temp
- Actual Temp
- Tank 2
- Back
- Read Temp
- Actual Temp
- Read Temp
- Actual Temp
- Read Temp
- Actual Temp
- Read Temp
- Actual Temp
- Manual Set
- Back
- [ ] Fan 1 Manual
- Fan 1 Speed
- [ ] Fan 2 Manual
- Fan 2 Speed
- Info Debug
08/25/2023
I also got the 4809 chips I mentioned needing to swap out because the 328 I was using just doesn’t have enough memory. I’m hoping to switch over to that tonight. In the meantime, I’ve been using an oversized controller that is way too big for this project to continue writing the interface. This is the PIC I plan on actually using. It only comes in an SMD, so I think I will only ever use it on a development board and not the IC standalone.
09/03/2023
If I can share my progress
09/10/2023
I was able to get in about two hours of work on this project and managed to complete quite a bit. The only things I have left are graphing, which will be a significant task, and the last items on the menu structure: a calibration temperature entry and an info screen that shows all the raw input and other details.
09/17/2023
So I’m starting on the graphing, and I need a really small font so I can include a lot of numbers for the legend without taking up much space. Super, super small… I think I can make a font using only a 3x4 pixel array!
I think this will work
09/22/2023
I was kind of unhappy with the readability of my super tiny fonts. It’s hard to make a three, eight, or two with only a 3x4 grid of pixels—most of the numbers, LOL. So, I caved and added one more line of pixels, making it 3x5. It doesn’t take up much more room, but the letters are a lot clearer.
09/23/2023
I was kind of proud of this one and wanted to share. My tiny font does not have the alphabet, but I wanted to write the words “tank” and “fan” on this minimum/maximum screen. So, instead of making the entire alphabet or even just the letters I needed, I made one single character that looks like the word “tank” and one single character that looks like the word “fan.” LOL, so only two characters.
09/24/2023
I decided to go with KiCad for the PCB designer. It takes a long time to design in this, but it’s so dense with so many options that I think it will be good to learn. Here is the wiring diagram for my fan project. It’s mostly a breakout board for the ATMega4809 dev board rather than a circuit design, but that’s okay.
Sadly, I ran out of time to etch this bad boy. I needed three jumpers to make this single-sided board work. One of them is because the dev board has two ground pins, but I connected them together outside the dev board to ensure I’m not pulling a lot of power from one ground to the next. To achieve that, I needed a jumper. The program says I have all the connections correct, so I’m going to trust it and test it. It would take me a while to look over each connection to make sure it is correct. Well, so far so good!
09/30/2024
Well, I got the circuit board design done, and I finished turning my 3D printer into an agitator that even warms up the bed to the temperature the acid likes to etch at. I think I got the frequency set well so that it agitates effectively for the amount of acid I’m likely to use.
10/07/2023
A couple of trials, but I’m zeroing it in. I’m realizing now that I had originally made my traces unusually thin, so I think I’m going to redo the layout using slightly thicker traces. Then, I think this will be ready to etch.
10/08/2023
When I first designed the board, I measured the maximum amperage it would pull with absolutely everything on. Using that information and the thickness of the copper clad, I determined the size of the traces to use. However, looking back, I realized that was the minimum size I could make them, and that particular number was really small. So, I went back and redesigned the board with traces that are half a millimeter thick, which is about twice what I originally did. This will give me a little bit of error room for when I etch the board.
While experimenting with the laser, I discovered that going slow with low power creates much finer cuts. This will take about half an hour to completely etch my board, but as you can see, the detail is really nice. I think I might be able to etch this one in acid and have it work, hopefully. After etching, I’m going to try to cut a hole through one of the two boards I was already playing with that are failures because I didn’t know what I was doing, LOL. It’s nice to finally see this coming together.
Also, to brag a little bit, I think the magnets I put down on the sheet metal to line up a new board every time is a beautiful idea, LOL.
The acid etching worked well, and then I used the laser again to uncover the solder pads. This is when I realized I might have made them a bit too small, LOL. I checked some of the longest traces for continuity, and it looks like it worked well.
10/09/2023
I got to do a little bit of cosmetic work. The bottom is green, and the top has a white silk screen with black writing. When I revealed the pads over the green, I was misaligned because I had taken all of my alignment stuff off to do some cutting for a cosplay thing for Jenny. When I do it again, it will stay lined up. I also realized that I forgot to mirror the entire design for the bottom of the board. That’s okay, though; this one’s just another experiment.
The top of the silk screen looks messy and different because I was zeroing in on how much power to use and how fast to move. Near the bottom of the silk screen is what I decided to stick with.
This weekend, I’ll try to combine all these things together properly and try to drill it. We’ll see if anything fails with that.
10/15/2023
This one here, I think, is ready to solder and find out where I went wrong, LOL.
It looks much less like a circuit board now and more like a breakout board for the dev board, though
Well, there goes 13 bucks… Right where the highlighted trace goes around that data pin. It’s not supposed to connect to that pin, but I think it does.
So the first wire I set up was the power. I designed it to run off 12 volts. To switch the 12 volts using the MOSFETs, the 12 volts also runs to the dev board, which has a voltage regulator that provides 5 volts and 3.3 volts, which is convenient. However, right where that 12-volt line swoops around, there’s an IO pin directly connected to the processor. When I plugged it in, I accidentally shot 12 volts directly into it and let the blue smoke out of it.
10/16/2023
My damage ?? So now I’ve got my desoldering tool and I’m going to desolder everything, reprint another board, and give it another go.
I did price having these done professionally. The best option I found is buying a minimum of a five-board run, which would only cost about $6 for all five boards but would take about two weeks for delivery. Still, I’m going to try to do this one myself, though.
10/17/2023
Okay, here’s a second attempt at a board I will solder on. I’ve redone a lot on the board. Instead of trying to stick with industry-standard size pads and things, I am stepping back a little bit and trying to compensate for the quality that I can achieve. So, the traces are much thicker, the pads are much larger, and the spaces in between everything are much larger. I have to admit, I love the way this looks and how it turned out.
10/20/2023
So much easier with the proper tool, LOL. I was trying to do these with a pair of needle nose pliers the last few days and was failing miserably.
Got the first iteration of the box done. I didn't print any interference Parts I'm just going to print the whole thing and see if I need to make adjustments
It appears I had only one pinout error, which I successfully resolved in the cable linkage.
It's working! Bahahaa
10/21/2023
How many months did this project take? Well at least it's finally done
