Climate Computer - Brain Box


#1

Hey Guys,

Its been a long time coming but its almost here. I have been trying to take the idea behind the food computer and apply it to bigger growing systems.

The Challenges
The food computer measures over two million datapoints per plant using sensors and machine vision. The design so far is focused on optimizing data capture for scientific use. We want to create a similar computer that focuses on automating growing envrionments for commercial and residential use. We sacrifice data accuracy and control in favor of farming as many plants as possible in an economical way.

The Requirments
In order to do so, the new computer would have to be safe to use, able to control a small growing environment independent of the growing system used (aero, hydro, or acquaponics), scalable, and easy to use and setup.

The Design
That is how the climate computer came to be. The CC is basically a modified version of Jake Rye’s brain box. I added a power supply (for the DC outlets) and AC outlets and we are working on a weatherproof case.

Here are some pictures of the prototype.



The food computer inspired me to make it easy for people to automate any indoor growing system. So, far our modified “brain box” has all the sensors used in the food computer, a 300-watt power supply, 10 DC outlets and 4 AC outlets controlled by relays, a raspberry pi, and an Arduino used to read the sensors. The power supply size, sensors, and number of DC and AC outlets are customizable.

I do not know anything about coding so I hired a programmer to write a webapp I could use to control the computer from my phone. There are three main programs you can assign to each relay: cycle timing (15 seconds off, 2 min on), scheduler (on 6 am, off 6:30 am), and sensor control (on when temp is over 30C, off when is under 25C). Here is the code’s github. I hope it helps.

The next step is to create a circuit board to get rid of all the wires and improve the webapp. We will be selling our climate computer as a diy kit you could built in less than one hour.

What do you guys think? We could built an openag version if enough of you are interested.That would make building a food computer much faster. Let m know if you want me to post the BOM too. I would greatly appreciate any inputs on the design or the software.

We are going to start a beta-version campaign for our climate computer as soon as the commercial version is ready, let me know if you would be interested to be part of it. The price for the beta version computer is going to be around $350.

Thanks!

Juan


My shot at a "Food SuperComputer"
My shot at a "Food SuperComputer"
Grow your own algae protein: Spirulina Photobioreactor
#2

Hello, Juan, it’s great to see this!

  1. What are you plans/timeline for commercialization?
  2. Are you interested in working together with other members of the forums to bring this product to reality?
  3. Do you plan to sell an entire Food Computer, or just the “brain” as a kit?
  4. How do you intend to keep the cost so low while still incorporating pH/EC/Co2 sensors?
  5. Is the web-app self-hosted, or will you provide a subscription service and host via Google/Amazon?

#3

1.- We want to start selling the computer as a kit to indoor residential growers and people using the food computer first. A first version kit could be ready in 9 months. Eventually (in 2 years) we want to sell wireless sensors, a wireless camera, and wireless relays all as separate components connected through a cloud. This way we could scale the system to farm size.

2.- Yeah i would be very interested! No one in my team has programming skills and that is slowing us down, we appreciate any help. If you are interested in joining the team at some point just let me know. All the hardware and software we are making will be posted here anyways. Even if you do not want to participate directly we could really use your feedback.

3.- Just the “brain” as a kit for now. If there is a lot of demand for the food computer we may built one around our “brain” and sell that.

4.- The cost includes a Co2 sensor but does not incorporate pH and EC, they add around $340 dls to the computer. We are selling the two probes as separate items thought, they are already incorporated in the software.

5.- The web-app would be hosted in a cloud platform, either Google or AWS. We are still looking into it though, the pi could potentially act as a server. On the long run we want an edge architecture instead of beig cloud-based. Any ideas about this? Like I say, we are missing technical talent in our team.


#4

Hi Juan,
Looks really cool. I have been researching ways to add a computer to my current hydroponic setup and this looks closer to what I am looking for.

I have a web dev background and would like to help out with the web interface. Do you have an API for the web server and the food computer figured out?


#5

Hey Brad,
You are exactly the type of people we are building this for, am glad you are liking it.

Thanks for the help I greatly appreciate it! To be honest, I don’t know but thanks for the question, let me ask my programmer and get back to you. The program should be working but I have not had time to test it yet, I am going to do that these coming weekends and keep you guys posted. I included all the code my programmer wrote on the github link above, you can check it out if you want. Let me know if you run into trouble.


#6

I talked to my programmer, yes the API is included in the code and it is separate from the rest of it. For now, we are using the pi as the server, that will change in the future though.


#7

@Juan1 I’d like to get your thoughts regarding the data model @webbhm is crafting. We too have been looking at using MySQL, which appears to be no longer supported: Sensor Data Modeling

I am working now to develop a commercialized kit version of the $300 Food Computer for use in schools. Have you seen the brain that @iancollmceachern is currently selling for the MVP? @bdavis you may be interested in this as well:
image
https://www.openagriculturesupply.com/product-page/open-agriculture-supply-mvp-food-computer-kit

Please let me know your thoughts on how we can work together and not duplicate efforts. If you can post a BOM/PCB design that would be beneficial as well. I’m especially interested in the sensors you plan to use going forward and the methodology behind choosing those.


#8

Peter, let me take a look at the data modeling with my programmer and I will get back to you. I am familiar with your work its great by the way! Yeah I saw that brain, I thought it was really nice how they put it together, Where did you get the case @iancollmceachern? And how are you controlling the outlets?

Here is the BOM Peter, I do not have a PCB yet but we are going to develop one for sure. If you know of any design we could use let me know!

Right now our main goal is to have an MVP, so I used the same sensors as on Jake’s brain computer. You can see them in the BOM. They are all open source and easily available. Ideally, we would want to standardized how sensors connect to the computer and create a big sensor library so you can connect other sensors such as a water level or light sensors. We will probably be changing them in the future for chip-based sensors like the ones made by Adafuit to reduce costs too.

The Computer price has gone up a bit since I last updated it, Amazon has been bringing up prices like crazy.


#9

Guys I was hoping I could get a bit of your input, I would greatly appreciate it!

Our strip has 2 8 channel relays to control 16 different outlets, but we can have 12, 8, or 4 channel relays too.I added a power supply and DC outlets (10 orange pins on the side of the box) to resemble a bit more the brain in the FC. Outlets can be any combination of AC (wall power sockets) and DC, and we can plug power supplies of different sizes up to 1050 watts. We are thinking of selling three versions so far: the open Ag which is 650W power supply and 16 DC outlets, another version with 300W power supply 12 AC and 4 DC output, and the AC version with 8 AC outputs (cheapest one). These models are work in progress, what do you guys think? Is there another combination you would prefer instead of these?


#10

I’m fairly certain it was 3D printed. The outlets are controlled via the same software stack @webbhm has created: https://github.com/webbhm/OpenAg-MVP-II

I believe that @thiemehennis has been developing one. I’m still waiting to hear about the status of the OpenAg brain PCB from @Poitrast. It would be great if we could all be using the same thing…

We appear to have very similar designs. One difference is that we’re using a SI7021 temperature/humidity sensor (less than $8). We have plugins for the same Co2/Water Temp sensors (@webbhm’s GitHub). He has also got water level monitoring, and top-off dosing working on his MVP as well as basic computer vision to detect plant area. I believe @ferguman will be incorporating pH/EC sensors into our stack as well.

@TechBrainstorm @catbarpH @Beskhue @pspeth are also working towards similar goals I believe. I hope we can all work together on a cloud solution or at least agree on the same data model so that we can share/exchange our experiment results.

Previously I had been leaning towards the 8 AC relay option. Then in a sense all we’re creating is a “smart” power strip though and I feel like that’s been done:

Sorry, I don’t have more concrete answers. It’s hard to get anyone to agree on standards at this point. I think everyone has their own idea of what this looks like. One of our goal’s with the MVP was to pull together as many people as we could and then include them in the development of the software and data model standardization. I can’t emphasize enough how important it is that we focus on the phenotypical variables, as well as the sensor variables. Who cares what the temperature was in your box if we don’t know how it affected the plant?

I am working on a commercial venture with @drewthomas89 to put food computers in schools. It is critical that this device is safe, UL listing is important to us, especially when dealing with AC power. We are actively looking for partners and areas of mutual interest where we can work with other teams to develop our “MARSfarm” prototype and bring it to market.


#11

We are currently working on final assembly of the first batch of prototypes for our AgBrainBox. @Webb.Peter, @webbhm and all the rest of the St. Louis folks will get our first prototypes along with a handful for our internal team. We hope to ship them out early next week.

We also just made them for sale on our website here for $425 (without the AC power sensor option) https://www.openagriculturesupply.com/product-page/agbrainbox-alpha-all-in-one-power-strip-based-food-computer Please note that while the hardware is functional, the software for these devices is still being architected, written, debuged and released by the community as a larger group so we can’t fully support the software on these boxes yet. They do come with a new 32gb SD card with NOOBs pre-installed!

Thank you @Webb.Peter and @Juan1 for the great discussion. I think that this is analogous to the Operating System/Computer Hardware paradigm. All of our different boxes have the same “chipset or processor” (raspberry Pi) and the same internal peripherals (relays, i2c sensor bus, etc.). As long as the high level product architecture and also the software architecture allow for this we will all be fine. Just as Lenovo, Dell and HP all make computers that can run various flavors of linux, windows, etc. so can we. To do so I think we need to agree on specific pinout blocks for the RPI header, Say DIO pins 0-7 are earmarked for relays, the i2c bus is already there, and a few other items that slip my mind at the moment. If we set standards for these high level basics all of our boxes should work with all of our software, as long as we allow for a little setup/configuration, etc. Perhaps we can agree on a .txt file header that just sets which pins are what function that way us box builders can just follow that standard? I am getting ahead of the process here, we need to hammer out what this looks like as a group but I think you all get the idea.

Answering some questions from above:
The case for our AgBrainBox is fully custom designed and made by me at Open Agriculture Supply in California. It is made of laser cut acrylic panels and 3d printed parts with the goal of eventually mass producing robust, economical and very nice looking metal enclosures which will eventually be UL listed for them in the very near future once we have the design a little more nailed down.


#12

indeed, have a look at our RPI extension shield - to be found here WeVolver: https://www.wevolver.com/border.labs/astroplant/blob/master/documentation.md

I see the 3D file is not there yet but we have it (the image files are there). Let me know if you need it. We’ll update our the coming weeks when the developers are back from their holidays.


#13

I really like your idea of start creating a standard. Here is the pins I used for the relays
GPIOs_array=[11,12,13,15,16.18,19,21,22,23,24,26,31,32,33,35]
numbers start in relay 1 and go to relay 16. I connect the arduino nano and the pi through a usb cable.

On another note, what do you guys think of using solid state relays instead of electromagnetic ones? Because of the nature of hydroponic systems, which have to turn on and off every minute almost, electromagnetic relays may fail too fast no?

@Webb.Peter if we only do AC systems we will look more like an environmental controller.

There are some startups in the market with patented versions of cloud based wireless environmental computers, but all of them are very new. The technology is just catching on.

By the way have you guys heard about Libelium? Its an open source IoT platform with hardware already included. It is really amazing, specially for developers. You should check it out!


#14

@Juan1 Thank you! Thank you too for the pinouts, our standard has begun!

My experience with SS relays is that they are more costly, power hungry, especially when you are switching higher loads. I think that mechanical relays have been refined to the point where they have a very long life. It is typically spec’d so we can do some quick simple calculations and see if they are suitable - I’ll take a look! Great suggestion!

Libelium looks very interesting, but I can’t find where I can actually buy some hardware, dev kit, etc? Is it available in small quantities to the hobbiest community?


#15

Yeah they sell dev kits but you need to dig a bit on their page into something called cooking hacks. it’s like an Adafruit kit but more advanced. Here is the link.

https://www.cooking-hacks.com/shop/kits/kits-by-categories

I think we should standardize sensors too so they can be compatible with the software. I was thinking of using chip sensors from adafruit to reduce costs. What do you think?


#16

Ian,

I see you are using RJ-11 jacks to connect the I2C sensors, what do you think about them? Are they weatherproof? I have been thinking we should standardize the sensor connectors too, maybe something this? What do you think?


#17

@Juan1 Indeed, I think they’re “meh”. They are nice because they are ubiquitous, but not nice because they’re just not that great of a connector (especially for this application). We were thinking about different connectors but arrived at the RJ-11’s because they’re easy to find, there was some documentation from some other stuff using them for I2C I found for precedent, and we wanted to go quickly. I like the ones you link to, they lock nicely and are waterproof, a huge plus. They are also rated for lots of power, so it would be nice to find something smaller if we can. I wonder if there is a connector from the sprinkler/security system world that may work for us. I have also thought about ps/2 or svideo connectors but they aren’t latching, water resistant or terribly durable.


#18

Yeah you make good points. If I find something less bulky Ill let you know. What do you think about 4 pins though?

Crazy idea but if you separate the arduino + sensors in their own case, from the pi and relays, and power the sensors with the arduino you could connect everything by connecting the arduino to the pi, no? That way we would only need 1 usb connector.