Growing food: I just ordered a MicroGrow Kit from Hamama


Hi folks,

For the last month, I’ve been wanting to start actually growing food indoors, but it’s been hard sorting through all the options. Building a food computer is out of my budget, and it seems more complicated than what’s necessary to accomplish my main goal–grow fresh salad! I’ve been researching ready-made systems from companies like HydroFarm or General Hydroponics and DIY options like the Kratky method. But, until I stumbled across Hamama’s website tonight, I hadn’t found anything that seemed like a good fit for getting started cheaply in the very small indoor space I have to work with.

Anyhow, this is what I got:

I intend to get a small Sunblaster T5 fluorescent grow lamp to go with it.

My plan is to use the MicroGrow kit as a test subject for collecting data on how plants are growing–mainly by taking timestamped photos and logging my observations–as opposed to building a climate control chamber or logging environmental sensor data.

If anybody cares, I just found out about Hamama tonight by randomly exploring links on Twitter. But, it turns out that it’s Camille Richwoman’s new company that she started with Daniel Goodman in San Francisco. From what I read, they both worked on OpenAg last year. Also, nobody asked me to write this. I don’t have any connection to Hamama other than buying one of their kits tonight because I think it’s going to be a good fit for my goals.

If anything potentially interesting comes of my MicroGrow Kit experiment, I’ll post more about it here.

Data Model Exercise: Cornell Hydroponic Lettuce
Choosing Between PFCv2.1 and MVP

Great for you wsnook. And your point is ?


@Pabu01 I can’t tell if you’re just trolling or if you actually want an answer. I’d prefer to take the positive view, so I’ll assume your intentions are good and that you’re genuinely curious and puzzled.


I didn’t tell the whole story when I said I want to eat salad–that’s true, but there’s more. According to your account profile, you haven’t spent much time reading posts here on the forums, so I’m not surprised that you don’t understand my perspective. I’ll try to explain.

I believe the discussions and software development on the OpenAg forums have the potential to become more productive and beneficial as people gain experience actually growing food. The baseline level of knowledge in the OpenAg community is unbalanced relative to Caleb’s stated vision. I see this in what I read on the forum, and I feel it as my weak botany background limits my progress in writing software. I want the situation to improve, so I’m working to educate myself. Just like with my other guides and howto posts, my intent in this post is to offer a suggestion for other people who want to improve their knowledge.

The Long Version

Part 1: Building a stronger community

From what I see, this community is strong on computing and engineering experience and weak on indoor growing experience–some people are knowledgable, but that’s the exception rather than the rule. By working to improve that balance, we can do better and more relevant work. If you read through the forum history, you can see that people with lots of growing experience sometimes show up here wanting to talk about hydroponics, but it seems like they tend to get frustrated and leave. That’s unfortunate–our loss.

People seem to hear Caleb talk and come away with the idea that food computers will save the world, but what he’s actually saying is much more nuanced. The v2 personal food computer is just a DIY climate control chamber with an automated nutrient dosing system. Both of those technologies have been around commercially for a long time. Sure, DIY is fun, but funded researchers can just go buy the real thing.

The vision Caleb talks about is large and inspiring: As a community, how can we address the problem of aging farmers and the loss of agricultural expertise as they die? How can we compensate for the loss of fertile farmland due to pollution and climate change? How can we improve the nutritional content and flavor of food? How can we reduce food waste? How can we encourage young people of various cultural backgrounds to see farming as a viable, interesting career choice? There are a lot of opportunities there.

MIT has the money and labs to do formal research with expensive equipment, and it makes sense for them to do that. But, the idea of a community project to build a crowdsourced distributed research platform with food computers all over the world is problematic. Sure, it would help researchers who want to analyze the dataset. But, where would people get the money to build those food computers? Is it even ethical to encourage people to spend their money in that way? Personally, that second question troubles me. On the other hand, if we reduced the cost of participation by an order of magnitude, then a crowdsourced research network might be reasonable.

The big opportunity for making relevant, meaningful contributions on the community side of OpenAg–that’s us–is to help make hydroponics and indoor farming more accessible. The vast majority of people on the planet don’t have thousands of spare dollars for spending on hobby projects, so accessible inherently means cheap. But, cheap equipment isn’t helpful without knowledge of how to use it–accessible also means having educational curriculum. The Raspberry Pi Foundation is an excellent example of doing this well in the context of computer education.

I see growing microgreens as comparable to learning to write a Hello World program on a Raspberry Pi, but building a v2 food computer is more like writing your own Lisp interpreter. Both are valid things to do, but one is much more accessible and relevant for beginners.

Part 2: My software project

At a more technical and personal level. I’m in the early stages of a software project that I hope will further these goals:

  1. Help people learn the basics of hydroponics and indoor farming so they are better prepared to contribute to community projects.
  2. Explore ways of building image based datasets that could be used for measuring phenotype expression with computer vision and AI algorithms.

To work on the software, I just need to be able to grow plants simply under semi-repeatable conditions with a camera pointed at them and some basic environmental sensing. Microgreens seem like a good fit for a cheap plant to watch with my camera, and I can eat them too. If you care about the conceptual motivation behind my project, you can learn more by reading this, this, and this.


WOW! Thnx for the TL;DR :slight_smile:
Reading my post afterwards might give an impression of trolling. Sorry about that. My point was I didn’t get why would you post it here, when it doesn’t have anything in common with this project (apart from one person that used to work on PFC).
Your explenation with giving a historical context makes lots of sense. Thanks for that.

Still, I don’t think that both projects are comparable. I don’t believe you can learn too much about hydro from just soaking a piece of “cloth” with seeds into water and then waiting. I love small greens. I used to make them as a boy and they are delicious to add into yoghurt. Have to restart that!
Still the complexity of closed environment seems beyond that.
I’m a beginner here as well.


Thanks for your guides. You rock!


Yeah, I agree. “Just add water” won’t take people very far on its own, but it is a good first step. The main benefit is that it’s an easy way to experience success that can inspire us to learn more. Our brains are built to chase little rewards–that’s why games are so fun. I’d like to see people get the idea that studying hydro and botany is fun and exciting–it can be.



Although I think the PFC at its current price point can be useful as well, I agree that it would be nice to have a lower cost alternative to the PFC, seeing as I can make a $200 prototype aquaponics system, and it costs a few hundred more to add sensors/actuators, it should be possible to build a cheaper PFC.
I still like the idea of a closed environment, since many of the kickstarter hydroponics setups seem to be just a glorified bowl with a light on top.

I have a question: why did you decided to start a new software project instead of going with openag_brain, is it because you prefer golang over python, or were there some other architectural concerns you had?


Thanks for asking–I’ve thought about this a lot. I took a long, careful look at the PFC2 design, and I came away with the impression that its highest and best purpose is being a photogenic demo. It seems to do a good job of that. People see it, they realize it’s possible to grow food in a climate controlled chamber, and that catches their attention. I suppose it could also be a good Maker project for people who want to practice fabrication skills. Beyond that, I’m not sure what it’s good for.

What I’m doing with golang and simplehydro is playing and having fun while teaching myself about hydroponics and botany. Philosophically, I like creating simple, efficient, inexpensive things that are built from a place of deep understanding. Before I try to build a device for growing plants, I need to understand how to grow plants.

My vision for simplehydro is to create software that teaches people to be competent indoor farmers. I want to be the first student. If you’ve read The Diamond Age, educating the Mouse Army is my inspiration for how it might work. Conversational AI is probably too lofty of a goal, but I can do lots of interesting things short of that. In particular, I’m interested in extracting the key info from long, dry academic sources and using it to create games and training sims. Another big thing I want to do is have a good place for recording notes about stuff I grow. For now I’m thinking that will look like photos, temperature logs, and text. I’m envisioning that implemented as a golang app that I can control with a web browser from a phone, laptop, or whatever.

Games and sims might be good for getting people ready to grow without immediately killing all their plants. But, building higher levels of skill would involve applications and practice. Calculators and system design guides could be useful for helping people do DIY hydro. Computer vision or maybe a text or web based conversational agent could be useful for diagnosing and remediating nutritional problems, diseases, and bug infestations.

I guess openag_brain ended up being too narrowly focused to hold my attention. I don’t see what problem it solves. I’d rather focus on helping people learn the foundational skills to go make DIY hydro systems that fit their local needs.

[edit: On python vs golang, I love python. At the moment, though, I’m enjoying learning go–I’m attracted by its efficiency and built in support for concurrency. It seems ideally suited for squeezing a lot of performance out of a Pi Zero.]

[another edit: I see some stuff here on the forum about educators using food computers in the classroom. My impression is that those folks are creating curriculum materials on their own, or that they’re getting curriculum from elsewhere. To me, having open curriculum is the most interesting part.]


That is an amazing vision. It’s good to see where you’re coming from now! (I haven’t heard of The Diamond Age, but it’s now on my list)

It’s actually pretty close to what OpenAg is trying to do, (I’m attempting to create a tool to assist in PFC design) but we can probably do a better job communicating it…

I think some of the goals in the PFC include exactly what you mentioned:

  • CV diagnosis of disease, measurement of plant health
  • Open Data(Climate Data, Recipes, etc.)
  • Learning how to grow plants (related to ↑)
  • Building a “virtual PFC” model to train on.

One of the things I disagree with, is why people would need to DIY their own hydroponics systems. I suppose there is merit to making custom hydroponics systems for various needs, but I didn’t build my oven or refrigerator, and I’d rather not have to bother with designing and making my own. I assume I’d be pretty happy with a set of “generic” systems.

I’ll ask the people doing the education pilot if they plan on releasing curricula as open courseware or something, there might be some legal things or whatnot that I’m not aware of, but it will be nice to default to open.

The current openag_brain might be tuned to the PFC 2.0 hardware, but it’s designed to be more general so it makes me a bit sad that you’ve decided to go with your own project, but I’m looking forward to your progress and will be watching! :slight_smile:


@spaghet I look forward to seeing how things progress with the PFC and openag_brain design work. It would be great if it turns into something affordable that achieves the goals you pointed out.


@wsnook I’m glad I found someone doing same thing I’ve been doing for the past 2 months!

I’m growing butter crunch lettuce using the kratky method.

I’ll post a link to my google doc that has pictures from germination to today (I do have a months lag… life got in the way…)

I do intend to add a built in camera so I don’t loose data of the growing process.


I have found that although I don’t have the full PFC I am getting pretty much the same data, I have a clip on fan to maintain the airflow and humidity, it’s in a spot that has the temperature needed, the lights work on a timer.

I’m basically using half of the PFC features at a cheaper price.


Thanks for the info, I’ll try the raspberry pi camera and see what I get.

I believe that the PFC idea is great, just it needs some modifications to it. I would suggest that they start with a beginner version.

PFC Stage 1:

  1. Use lettuce -(or any similar greens) quick germination and 30- 45 days to harvest

  2. Find any spot that has a steady temperature and humidity (any thermostat I got from Accurite on Amazon for less then $10) and determine that it’s stable.

  3. A clip on fan

  4. LED shop light on a timer

  5. Use the Kratky grow method (no maintenance needed once the pH, conductivity and water temperature are stable. No need for any air pumps as the roots will create an air space for oxogyn once the water lever goes down. (I used a poler spring 3 gallon bottle and cut off the spout)

  6. Use the raspberry pi with the camera for documentation (you can add a IP relay to control the lights as well)

  7. Sit back and enjoy your results!

Stage 2:

  1. Add to the raspberry pi a pH, conductivity and temperature module to learn what’s actually taking place inside the reservoir.

@Caleb let me know what you think of my idea.


@wsnook @spaghet @Pabu01 @Webb.Peter @jake @adrianlu

Great dialog, very interesting to read and Im glad to have critical viewpoints expressed. I completely agree that an MVP PFC is a sorely needed "mod" of the PFC. I am seeing two very clear, and differentiated, paths of development on the PFC starting to organically (pun intended) take shape.

  1. One dev is trending towards a precise piece of scientific equipment to push the boundaries of our understandings of plants and their complex environmental, physiologic and morphologic interactions. This kit will be expensive by virtue of quality, reliable and precise tooling required and the skill level to assemble will be high (chillers = expensive, LED lights that put off 400s mol / m2 in correct spectrum w/tune-ability = expensive, the more sensors = the more expensive).

2.The other is a dev kit specific for folks just getting started (students, makers, etc) with a moderate degree of skill required and moderate cost (sub 300 USD…maybe even sub 100 USD?).

It is my dream that we (community) approach dev on the PFC like a custom desktop. In a custom desktop the standardization of hardware/software interfaces is KEY. You can get an empty tower on the cheap, throw in just enough processing, memory, graphics etc to get you started for an affordable price.THEN over time as you develop the need for additional power, capabilities, you can upgrade (better CPU, more RAM, dual graphics cards, water cooling for an over clocked processor, etc) and you have a catalog to choose from, user reviews, software capabilities that allow for expansion, etc.

This should allow for many communities to enjoy the PFC and for it to start developing user interfaces and BOM’s that are mass customization.

Of course there are alternatives:

If you want to purchase a basic functioning system, non networked, non data gathering, not DIY, not opensource BUT frankly that grows really well on the cheap - your best option is honestly AeroGarden by miracle grow. This little bot is well designed (*remember miracle grow is Scotts and Scotts now owns general hydroponics the biggest player in the space), consumables are widely available, it can grow a variety of crops (tomatoes, peppers, greens) and they have a wide variety of bots to fit your budget (some now with apps). There are of course many others in this category - grove, urban cultivator, etc etc. Frankly a new one is added every day it seems and kickstarter is overflowing with them, take your pick and get started growing (buyer beware, but I have nothing but love for anyone wanting to grow).

If you want to build a hydro system, YOUTUBE IT :slight_smile: literally 100’s of video’s like this (the 5 dollar hydro kit) are available for your perusing pleasure. You can also go to just about any hydroponics store (at least in the US) and the folks there will help you get set up (beware though the parts costs are often 3-10x amazon or hardware store prices) but they have all the nifty components you’ll need organized in an easy to find way. ***If you just want to watch things grow to first true leaf stage here is a tip - buy some expanded poly ethylene matting, throw seeds on it (greens, beans, peas whatever), water them a bit and BOOM :slight_smile: you’ll have sprouts. Its almost impossible to kill sprouts, you dont even need light or fertilizer (its built into the seed) to do it. I encourage everyone to do this, its really fun.

Now the question I had in my mind…I didnt want an off the shelf proprietary product (just not my style) AND I didnt want to build one off / unique systems because it would never be very optimized / modular enough to expand with my interest (*disclaimer I started by hacking the off the shelf ones, then building my own from youtube video’s / unsavory :wink: chat room) plus I wanted HELP! Lots of smarter people than me in different fields (chemical engineering/fertilizer, horticulturalists, electrical engineers, data scientists, plant biochemists etc etc). This is where the idea of an open source / open standards bots (at that time called CityFARM, now morphed into OpenAg + PFC + Food Server) began.

I specifically made it a priority to launch this community and contribute what we had been working on. None of it is perfect :slight_smile: none of it is finished / polished / documented in a way Im happy with yet but the project is off to the races and frankly taking on a life of its own outside of our little lab. Not to mention funded to continue at MIT for at least the next 3 years (which was a monumental undertaking all on its own, trust me. Try going to a financial sponsor saying,“Give me money so I can give away IP for free” lol)


  • To work with / adapt / create standards that keep us working together.

  • To personally contribute meaningfully to the community project with whatever knowledge and tools you have. (WW"MIT"D was never the idea and “demo or die” should be our attitude)

  • Not to fork / disband and become either of the two alternatives mentioned above :slight_smile:

That being said…

Lets have the first truly community derived device be the MVP PFC what do you all think? I am game if you are and will bring in some help from the lab

We will have to agree on core components (ie. should we stick with rpi and arduino? or maybe move on?), core sensors (frankly you can do a lot with a little on sensing), core acuations (nix heating/cooling? go back to flourescent?), frame material (its aluminum right now…but it could be wood that we finish well…or even plastic or foam?). This should be a sprint and build challenge (*not a prolonged conceptual forum conversation). Then we run the different machines and share our results. What do you all say?

As a starting place I would argue (just my opinion as an individual) to try and keep the same dimensions as the V2.0 PFC and design language (structure + skin + “motherboard”). That way what we learn on the MVP PFC can benefit those already starting to build V2.0. Should we set a BOM cost of 300 USD? 200 USD? 100 USD? Check out what we did for schools with the foam farm activity a while ago for 500 USD

Feedback? thoughts? Questions? …btw I wish there was some community cowbell I could ring @gordonb so we could get the most feedback possible.

Building a food computer on developing communities
Growing Veggies packed w/ Flavor & Nutrition
MVP - Community Development
Choosing Between PFCv2.1 and MVP

@Caleb Awesome post! Congratulations on the funding.

To the extent my little side project can eventually feed back into the mainline flow of where stuff is headed, I’d like for that to happen. I’ve been quiet about it lately because I’m busy writing code and don’t have results to show yet. I’m documenting my progress on GitHub and Twitter if anybody’s curious.

To your point on standardization, it seems like the best standards emerge as documentation of consensus following periods when people try lots of stuff to figure out what works. That’s what was going on in the Homebrew Computer Club days.

You could make a fair argument that, because lots of people got Altair BASIC for free, BASIC became a de facto standard. Publications like Dr. Dobb’s Journal helped people share game code, the market expectation formed that new computers should include BASIC, “Micro Soft” changed its licensing practices accordingly, and the rest is history.

Have you thought about starting a regular email newsletter? That could be your cowbell in addition to an educational resource. Your description of experimenting with hydroponics would make a great article. It would be great to hear more about the science of what your group is doing in the Media Lab too.


I was beginning to explore this in another topic, but will move over here.

Defining goals is critical, and I think you are starting in that direction: building a system with a growth path compatible with the PFC. I like the analogy to the PC and its core compatibility. If you want a hand-held calculation - that is a different product (go look for something on you-tube).

A critial question is whether there are any serious research projects that could be taken on with something less than the full blown PFC (nutrients?)? What is the minimal set-up needed to give valuable data? Hardware and software interfaces are very important, but DATA is the most important key. I can provide data without hardware, and hardware/software standards without data standards will just give you a mess.

I can envision a three step growth path to the Food Computer:

  1. Absolute minimum: Coffee jar or bus tub, air pump, timer and light. There need to be two or three standard light recommendations, that could be reused in the next build. GE Bright Stiks1 (100w equiv) in a reflector are working for me (and are cheap (~$7), or possibly jump up to a CREE COB1.
    This could be enhanced with an Arduino to occasionally take some sensor readings (temp/humidity, LUX). The Arduino could run basic sketches, with the results recorded on paper or spreadsheet. This sets up the sensor/automation growth path.
    Data needs some standards, and a way to manually enter/upload it.
    Some parts could move to the next phase. Price about $30 (without the Arduino & sensors)
  1. Next level would be to make an enclosure (Foam Farm?).I, like the idea of a standard size, but standard material does not seem to be relevant. This is when the automation core becomes significant, the RaspberryPi and software infrastructure. A temperature/humidity sensor will get you started, along with fan (circulation and ventilation) controls to regulate temperature. About everything else can be a future addition (fertilizer & pH can be done manually).
    Price about $200 to $300 (assuming all new parts)
    Stay with RaspberryPi and Arduino for now. Focus on standardizing data and data interfaces; create facades to hide the OS, ROS, etc. and you can move to something else in the future.
  1. Final level would probably be the PFC - better enclosure and adding pH and dosing controls.

Again, it is back to defining goals, and particularly goals for data & research. Unless you magically are going to pull all phenotypic data from cameras (underwater cameras for roots?), you will need some manual data inputs; which will require definitions, standards and protocols. This will be harder than the hardware issues.


I believe it would be much better for the project to agree on things like data formats, and standard connectors. It does not make sense to standardize on specific things like computers or sensors as this limits what can be used to make different versions of PFC. For example, I’m currently considering using Beaglebone Blue and personally, I don’t like ROS, Docker or even CouchDB as I believe in a minimalistic approach. However, just because I want to use a different software/hardware stack does not mean that I could not contribute something, even it’s data, or plans and recipes.

Standardized connectors and protocols will allow us to use different modules, sensors, and boards without making huge modifications to the PFC.


This is where things get difficult (and fun).
Standards are not monolithic; just look at the stack of standards around a USB printer, there is the physical plug standard, TCP/IP standard for communication over USB, the standard for driver handling. There should not be ‘THE OpenAg Standard’, but a collection of layered standards.

  • The bottom foundation layer should be the data standard (definitions and semantics)
  • Next is data representation (Json, XML, csv, …).
  • Then we can get into exchange protocols & APIs for exchange
  • Only after this should we get into (multiple) software standards (Raspbian, ROS, database, …)
  • The least significant standard is the physical hardware (RaspberryPi v3, Arduino Duo, PFC box size).

We need to remember: software is portable and APIs can be implemented in multiple languages. We are already a stack of Linux, Python, C++. The real issue is open standards and DATA.


I agree and love both of the directions that Will @wsnook and Howard @webbhm are proposing.

I have been working at the OpenAg lab now for 4 weeks with @caleb. My main task is to do exactly as Howard suggests: define the data plan. A second fun task is to help define MVP system. I like everything that Will is doing, but would love it to be part of the plan to share / create the same data as the PFC.

I would like to work together as a community to reach a common goal for Data and the MVP.

I see two discussions we need to have with the goal of coming up with a plan and definition for Data and MVP. Are new categories / topics in this forum the way to start? Or perhaps we can also make two new sections on our wiki as a more permanent place to write up a design / definition / documentation?



Technicals (especially with me talking) get boring very quickly. I like to
keep this public so anyone can join and give input, but it should
definitely be a topic under ‘Data’ so anyone not interested in data can
ignore us., I would suggest a new category for the MVP, as I can envision
many different topic threads under it.
Any conclusions (definitions, formats, …) should be put on the wiki.
I am not sure if the MVP should end up in the wiki. That depends upon how
much it is a sponsored project of OpenAg -vs- a community activity that
operates on the fringes of the community - either way, no decision needs to
be made on that at the moment.



Wise words. That sort of discussion would be better off in its own thread.

[edit: @rbaynes, I saw your DM and sent you a reply. I’m definitely interested in finding ways for people with an official PFC build to meaningfully exchange data with folks using other equipment. My bias is in favor of iterating on working prototypes as the main way of moving forward–I think that’s what Caleb was suggesting.]