Ok guys, here is a list what I need to control. I made groups with the kind of signal that I need to control them. Sorry but it is a long list but the project has lots of different aspects.
Digital Signals:
- PWM Signals:
I need to control light intensity by providing PWM signals to some LEDs. Most probably I will be using 4 different channels (warm white, cold white, red, moon light). Frequency will be ~1kHz. The cable from the main controller to the LED drivers will be ~2m-3m. Additonally the PWM signal needs to be upated roughly once per second to get nice and smooth intensity changes.
The speed of some pumps needs to be adjusted using PWM signals. The number of the signals will depend on how far I want to go with automation. Nearly all of the pumps are located next to each other, cable length to main controller appr. 2m. One for the water exchange will need a longer cable (~10m). Frequency of the signal is not that critical. A few 100Hz are ok for sure. Resolution of the signal is not that important. For the pumps used to inject fertilizer the timing needs to be accurate (duration the pump is operated needs to be controlled to less then 0.2s jitter, 0.1s would be preferred).
- Digital Control Signals:
Controlling Valves using a logic signal (0V off, 5V on) as well as switching some function on or off via relais (Filter, Heating, ...) should be an easy task. The signals need to be stable but they are not changed often. Most of them maybe 2 or 3 times a day some might change a few times per hour. Distances for these signals are no more then 5 meters.
- Digital Input Signals:
The flow meters provide a digital signal. A pulse equals a certain amount of water passing through the flow meter. They use a 3 wire connection for this (GND, VDD and Signal). Power consumption is low (I measured 3mA when operated from a 3V3 source). The signal is quiet nice (step flanks) even when using a few meters. The signal frequency can reach from 1Hz to 300Hz. It is necessary to count the pulses either to measure a water volume (for water exchange) or a flow through a tube (checking that the flow is as expected by counting the pulses per second). Currently I use interrupts the call a function that adds one to a variable. But this might not be the best solution.
Analog Signals:
- Temperature measurements using thermistors:
The thermistors I use have either 10kOhm or 50kOhm resistance. I use them in a voltage divider with a precision resistor and measure the voltage using a ADC. I can measure the thermistor characteristics at work by comparing them with a precision thermometer (0.1K accuracy traceable to national standards). Due to the high resistance of the thermistors long wires are no problem and temperatures can't change quickly (so I can use lots of averaging either on the analog signal side or in software).
- Glass electrodes for pH, redox measurement:
Tricky measurement as these electrodes generate a voltage which is proportional to the value of interest but their impedance is huge (typically 1GOhm to 100GOhm). The electrodes come with pre installed wires (and connectors) which are long enough to reach the main controller. Tricky part is to amplify the signal (fA input bias OpAmps) and to get rid of all the noise which couples into the signal by shielding and averaging. I need to do some more testing (and find the time for that) but I think I can solve that. After that stage of frst amplification everything else isn't complicated.
For those who wonder what these measurements are used for: pH measurement is used to dose the CO2 injected into the water. Redox value is a good indicator if something goes wrong in the aquarium. Normally you can see this by looking at the fish but if you are away for some days such a value is a good indicator that can be checked remotely.
- Water conductivity measurement:
Not really started yet. The conductivity of the water is typically measured using two electrodes wired in series with a known resistor (voltage divider). If the characteristics of the electrodes is known you can calculate the water conductivity from the measured voltage. Issue here is that polarity of the signal needs to be altered to avoid electrolysis (typical frequencies for this seem to be a few hundred Hz). There are lots of example circuits on the web and they look reasonable and usable. They also have no problem with wires of a few meters.
- Weight measurement:
Measure the weight of the CO2 bottle in order to known how much CO2 is left. Measurement is done using a weight sensor which output a (small) differential voltage which is proportional to the weight on the sensor. Using an instrumentation amplifier enables measuring and amplifying the signal. After that it's easy to handle.
User Interface:
Some buttons (debounced and connected to digital inputs) and a 20x4 character display (I2C interface) will be used for user interaction.
Communication:
As I am regularly away for some days I wanted to have some remote access. Plan is to connect the Teensy main controller to a Banana Pi or similar computer that runs a web server and some application that talks to the Teensy. Connection will use some UART port for this. I checked that the communication is working in principal. The two boards can be located in the same box, therefore cable length is not an issue here. Everything else will be done after the controller is working.
Possible Future Plans:
- Water Level Measurement: Ultrasonic range meter? Float Switches? Capacitive Sensor? ...?
- Detect if the aquarium is damaged: Glass break detectors don't seem to work as the water is changing and damping the signal used by these sensors? Water detection sensor on the ground?
- Temperature regulate the water for the water exchange: Tankless water heater?
- Simulate Rain/Thunderstorm? Ok, I start to get crazy.
Ok, thats the list. There are basically two questions. The first is an architectural one. Should I really controll all of these from a single centralized micro controller or should I use several controllers with one centralized main controller to coordinate them? The second question depends on the answer to the first one. If I use several controllers I need a communication interface in between them. Then the question is what would be a reasonable communication bus. If a single micro controller is used the question is how to connect the external peripherals and how to assure that the different functions don't affect each other?
As written above I started with the idea to use one single micro controller. But due to the trouble transmitting the signals and especially because I fear that the different functions will affect each other I was thinking about alternatives. But it is hard to decide this kind of questions if you don't have experience with communication buses and reliably transmitting signals over some distance.
Regarding PWM frequency and disturbing the fish. At seems that the effect vanishes somewhere between 200Hz and 500Hz, depending on the species (reports from people who use PWM dimmed LEDs). Therefore 1kHz should be fine.
Regarding the RGB LEDs. I thought about that but decided against it. I would need to many of them in order to achieve the desired light intensity. Some plants need a huge amount of light the grow nicely. I target something like 10000lm as a maximum value for the light intensity and I want to have small light sources to get some nice shadow effets. Currently I plan to use Cree CXA COBs. Mixing warm white and cold while to get a nice color spectrum and adding some deep red LEDs (still looking for some high power versions) as white LEDs miss that part of the spectrum which is important for photo synthesis.
Thank you for your help
Daniel
