Onocoy is a GNSS DePIN project with a "bring your own" hardware invitation. You buy a base station, set it up, and start providing its correction stream to Onocoy.

The "bring your own" model has a couple advantages over projects that require you to use their hardware.

First, you can easily use your own data stream by connecting directly with the receiver. Second, you can really dive into the details of what you're doing to see where you can improve. Third, using your own station requires you to learn a little bit more about the world of PNT (Position, Navigation, Timing), which is just flat out fascinating.

Of course, for us regular (non-GNSS nerds), a "bring your own" station is a little intimidating; it ain't quite plug 'n play. Let's go through the broad strokes you'll need to do, then get down into those details to make sure you maximize your rewards.

First, you have to buy a base station. The good people over at SparkFun were gracious enough to send me over the best of what they have, which is a SparkFun RTK Mosiac X5.

Here it is sitting on my workbench, merrily collecting signals and issuing out an RTK correction stream. Yes, I've covered up the location of my top secret hidden lair in San Diego.

I installed it in my office with the antenna up on the roof. I used PoE (Power over Ethernet) because I avoid WiFi connections wherever I can. I've got a Ubiquiti router and PoE switch, so injecting power and finding a port isn't a problem.

Still, for both antenna and PoE I had to cut some holes in my wall and run cable through the walls and up to the roof. I think that's fun work, but I also don't have to do it every day.

The more stable your antenna on the roof, the better. Ideally the antenna mount is bolted on with zero (millimeter or less) wiggle. I've got mine on a heavy duty non-penetrating roof mount. While it's not perfection, it's pretty darn stable.

Ok, that's the physical deployment side. Let's jump into the geek-side and plug our faces into the computer.

To get the bare minimum setup for the Mosiac, just plug the thing in like I have (connected to power and ethernet and the antenna), then navigate on your network to whatever local IP you've set up. In my case, it's 192.168.50.46. Obviously that'll be different for you. You'll see when you go there the "Septentrio" branding; the Mosiac X5 is their most compact and high precision GNSS receiver.

Once you're on the local IP for your Mosiac it'll walk you through basic setup. SparkFun has a guide that makes it dead simple, here.

You'll also need an Onocoy account in order to take your shiny new RTK stream from the Mosiac and feed it into Onocoy. You can set up over at console.onocoy.com. Once you're there and signed up, open the Reference Stations tab and hit "Connect Server".

Onocoy has excellent documentation on setup and it will change over time. One thing I've learned with writing these rough guides is that it's far better for you if you refer to the project's set up docs than mine, at least for basic setup stuff.

Use the Onocoy Installation Guide to connect your station's stream to the Onocoy network.

Once you've finished setting up your Onocoy & Mosiac, you should see something like this on the Mosiac side:

That shows that you're receiving signals from satellites, and feeding out a data stream to Onocoy. Cool, right?

After 24 hours of having it up, your Onocoy console should look something like this. I mostly pay attention to the Satellites and Signals boxes, and spent a bunch of time comparing with local stations.

The general conclusion I came to after looking at my station vs all the other local ones (including ones clearly set up by professionals) is that the SparkFun setup I've got is as good as it gets out of the box; there's nothing better on the market for what we're doing.

The only real issue is that I've added a station in San Diego that either wasn't desperately needed or is more useful than others but is in a deployment area that's already saturated.

If you're thinking of deploying a station and your main motivation is the Onocoy tokens, use the Estimator tool available on the Onocoy Explorer to get an idea of what you'll earn.

That's it for now! Poke around on both the Septentrio & Onocoy sites to learn more. Next up? How to use the SparkFun RTK Torch!

For a little under $8,000 you can have your own very capable private AI at home that you can feed data to and get answers from.

It's not hard to do, it'll put you out at the cutting edge, and I think this offers significant advantages to those who follow through with it. This guide talks about what the machine consists of, how to put it together, what you might use it for, and what is still fantasy.

This is written for us semi-normal people interested in AI. We aren't going to build our own LLM from scratch and we may not code for a living. We've used ChatGPT or Perplexity and think it's neat, and we can copy/paste commands into a CLI. We may have never built our own PC but that's fine; I hadn't either.

We're willing to read a bit and muck about, but in general we don't have time to develop deep domain expertise; we just want something that mostly works and puts us out at the forefront of the AI revolution.

You can also spend less if you want. My budget was $8k so that's what I used. $3k is probably as low as you can go and get reasonable performance, but I'm sure some wizard out there will let me know in the comments that if I just re-ionize toothpaste and flame-char duct tape to exactly 417.314 fahrenheit then smelt my own iron, I could do the whole thing for $5.

I didn't do that, just spent the money, enjoyed building an excellent thing, and called it "Monstra".

The GPUs are a little too close together to run for long term comfort, but for relatively short tasks the thing is awesome.

Yes, you can run an LLM on your home computer right now; it'll just be a lot slower than a purpose-built monster. "Monster" is relative; for a home setup this is on the "overbuilt" side. Compared to top of the line massive racks of A100s or H100s, this thing is puny.

Ok, so what do you get out of building your own personal AI machine?

• Privacy in your queries. Maybe you don't want OpenAI to know about the product you're building that you need help coding.
• Related to this, you get the ability to use an AI trained on data that you might not want to share, like tax records or shipping information for every sale your company has ever made.
• You get flexibility in setting it up exactly how you want it; how much "horsepower" you use, how creative you want your responses to be, and specific things to focus on.
• You get to experiment on the cutting edge, and find out how much you can actually do and how much is still you fantasizing about having your own drysine queen.
• You can also participate in renting out spare capacity.

I'll start with what I've been experimenting with, then we'll get into the details of the build and how to set it all up. This is not (by a long shot) all you could do, these are just some of the problems I'd like to solve for myself.

Training Plans (race prep, overall fitness, etc)​

A few years ago, I hired a private high-end coach to help me build a training plan for a hike 'n fly paragliding race I was preparing for. This is a former Olympian who ran strength and conditioning at Red Bull. I'll admit it; it was probably a lot more coaching horsepower than I needed.

He charges $800 a month for what I wanted, which is a customized plan with unlimited calls to an elite coach and alterations to continually make the changes required of any training plan. We spoke every day, it was awesome, and he was worth every penny. Still, $800 is $800.

Now, that's on the expensive side. If you want excellent coaching you could hire the folks over at Evoke Endurance, who charge $350/month for a similar service.

If you had a private AI, you could take PDFs of all the books that Scott Johnston over at Evoke has written, transcribe all the Evoke (and Uphill Athlete) podcasts, add in whatever other content then you want, then feed it to your AI and let it help you develop your own custom training plan.

If you wanted to be a little entrepreneurial, you could open that up as a service for others. It won't be quite as easy as I'm making it sound, as you'll need a RAG and nerdy vectorDB skills, but with AnythingLLM you can probably pull this off for yourself with some elbow grease.

Let me be clear here: As of July 2024, the AI won't be as good as a professional coach. In fact, when you start it may be worse than a bad coach. Still, I'm *guessing* that'll change by December of 2024 (the ol' "6 months in AI is a new century" rule). That's a guess and I'm likely to be wrong.

LiveChat Agent​

Since 2009, my wife and I have run Paleo Treats which has both an online store and a brick and mortar shop. Since 2014 we've used LiveChat to chat with folks who visit the website. If you've ever been on the PT site and opened up a chat, you were talking to me or my wife. That's cool, but...we turn off the chat when we sleep, and throughout the day when we have other things to do.

LiveChat offers an AI service called ChatBot that starts at $52/month. With our HomeAI setup, we can replace that with a custom agent and have more control (and learn way more about setting those up) to give super customized chat experiences to folks who want to buy gluten, grain, and dairy free desserts online.

Taxes​

Having been in crypto since 2014 or so, it's always a discussion with our tax preparer every year; "What are these token-things? Why are you still doing them? There's not any great guidance on this..."

Because taxes are about money and money is so personal, I'm not sure I want to be uploading all my financial docs to every tax site promising "We do it with AI!" out there.

I'd much rather feed my private AI the US tax code (which you can find here, Title 26), give it my last 10 years of returns, then work through the current year with the overwatch of an AI knowing that eventually, as long as I put in my data correctly, I'll be able to do my taxes far more efficiently with my own personal consultant that knows every aspect of my finances.

Again, this won't be easy or perfect (or maybe even that helpful) right now, but if I start working on this now, I'm guessing by the next time I need to submit taxes it'll be a button push, a few questions, and perhaps the best tax prep experience I've ever had.

Our current tax preparer charges about $6,000 per year to integrate 3 businesses, personal taxes and my various crypto adventures.

Inventory & Operations​

Again, as part of the Paleo Treats business, we're always ordering ingredients and consumables, doing production runs, then packing and shipping out items. I've built a series of Python scripts to help keep track of costs, but the ability to start to integrate all of our Operations data into an LLM and ask "How many packages have we been shipping to Sugarland Texas this month, and how does that compare to last year" or "Take a look at this giant csv of every package we've shipped over the last 10 years and tell me which city gets the most packages" etc etc starts to make finding opportunities much more efficient.

Yes, I know you can do that with Excel or Google sheets. I think the ability to talk to a personal and secure AI starts to change the game for small businesses who aren't necessarily data wizards but still have legitimate questions about operations opportunities.

Funzies - Grinding a Vanity Solana Address​

I stumbled on a post by Nick Frostbutter on how to "grind" a Solana vanity address using a built-in command in the Solana CLI (Command Line Interface, which is where nerds go to talk to computers.). That particular set of instructions uses your CPU. With the machine I'd built, the CPU was pretty powerful (gear list below), but the whole reason I got into this was I wanted the GPU capability for LLMs, so...why not use the GPUs?

My buddy nosmaster found this Github repo on how to use GPUs to grind, and after a full weekend of fiddling about with various settings, I managed to grind out a gking... Solana address in about 45 minutes. I know, I know, it's good for almost nothing and you don't get a mnemonic phrase (the "twelve words"), just a barely decipherable private key, but it was a fun use of Monstra.

The Build, Part 1: Understand What Everything Does​

Before we get to a parts list, I thought it'd be useful to have an overview of what everything does. It was all a mystery to me when I started, and just knowing what each thing does has been helpful in grasping the overall concept.

Every computer we build for AI will have 8 constituent parts:

CPU - Central Processing Unit: This is the "brain" of the thing.

GPU - Graphical Processing Unit: Fundamentally, this is what gives the AI its horsepower.

RAM - Random Access Memory - Think of this as the working memory of the device.

Storage - This is the long term memory. Not as fast as RAM, but way more of it.

Motherboard - The coordinator of all the other parts, and the physical structure they all plug into.

Power Supply - Making sure everything stays on without flickering.

Case - Big enough and airy enough to keep everything proteced and cool.

Cooler - Fans to make sure the CPU specifically is cooled off.

Additional Cooling Fans - Drives air through the whole case to keep all parts cool.

Other nerd tools - Thermal paste, screwdriver, cable management, etc.

The GPUs are critical for making the LLMs actually work (at a reasonable speed). The CPU manages overall workflow, and the RAM supports large datasets. Everything else supports that. Of course, you can scale up (or down) from here.

Ok, so with the basic stuff out of the way, what should you buy and how does it go together?

The Build, Part 2: Buy & Assemble Your HomeAI​

I'll start by saying that I'm impatient and Amazon is easy, so I bought everything there. I'm sure if you hunt around and have some patience, you can save yourself some money.

All up it ran just under $8k. Of course, if you want someone to build it for you, Gigabyte is offering similar builds starting at $11k. You can also check out a Google sheet I put together for different "size" builds from $3k - $8k. The "V3" on that list is Monstra.

I ordered everything on a Monday and, being Amazon, it was all there by Thursday including hiccups.

Assembly was more or less straightforward, though I had significant help from Vortal, nosmaster, and Rob who are all part of the mysterious advanced tech exploration group called AppleFuckers over on the GK Discord.

Each part comes with plenty of guides and manuals, and from my limited experience, if you just follow the Motherboard manual it'll walk you through everything you need to do.

Still, if this is your first build, I'd find a computer nerd and have 'em help you with it; many things are not obvious, and having an experienced guide to walk you through updating the BIOS (if that's even needed) is super helpful.

What To Expect The First Week​

First off, know that you're an experimenter. Nothing will work out of the box; everything requires the ability to use the CLI to do basic install stuff.

Still, if you can search the internet, copy/paste, and use ChatGPT to get you going, you'll be up and running your own private AI in just a few days.

It won't literally take that long, as installing these things once you know how is about a ten minute job, BUT...learning how to do it, what works and what doesn't, and what you should and shouldn't pay attention to takes time.

Here's what I loaded up and used the first week:

• Ollama (with a bunch of different models)
• OpenWebUI
• AnythingLLM
• Stable Diffusion

Each of those is its own "flavor" of AI, or in the case of OpenWebUI and AnythingLLM, a way to interact with an LLM.

About that RAG​

If you want to add in your own content (like the US Tax Code or your shipping info or the contents of a book), you'll use a RAG, or Retrieval Augmented Generation. I'll start off by saying this: The *concept* of a RAG is easy; add in the customized thing you want your AI to reference when answering your questions. The *application* of getting the thing to work is the sticky bit.

AnythingLLM makes it pretty easy to add in documents, though it's not ultra clear how to assign them to the specific model/thread you're working on. OpenWebUI is good for a few docs, but not so good with a ton of docs, and doesn't really give citations (the way, say, Perplexity does).

The use of RAGs is still well ahead of "easy". The data you feed the AI has to be clean and well structured, and frankly, most of us humans are walking around with such a mess in our heads regarding how we organize and understand our own data it can be a mind-bender to translate that to your machine.

Still, it's been VERY useful for my own thinking to move through the steps of introducing data to any of the LLMs I've worked with.

Next Steps​

Rather than giving you a step by step of exactly what to do with each LLM, I'd encourage you to explore this AI jungle on your own. Know that other folks (like me!) are out in the jungle with ya and we're all learning how to interact with our own private AI in our own way.

So far, of all the things I wanted to do, none have come out the way I want. I'd pretty much expected that. Still, I'm getting exactly what I wanted in the form of hands-on experience working with and around a personal AI and seeing both the opportunities and the limitations.

I found the following two YouTube videos helpful when I was starting out:

Host All Your AI Locally

How To Use Your Local AI (TechnoTim vid)

Self Hosted AI That's Actually Useful

If you're experimenting with this as well I'd love to hear about it! If you buy a bunch of parts and need help, jump in the GK Discord server, there's usually someone awake and willing to point you in the right direction.

See ya on the other(AI) side!

Archived Comments​

Sheree - 2/5/2025​

Love this guide. I’m interested in spinning up a private instance as well. I’m working on the experience side for an enterprise, would be fun to have one at home to experiment with. Thank you for a thoughtful and very entertaining guide

Chris - 10/16/2025​

Hiya, thanks for this write up, they are harder than they look- or were ? It has been more than a year, how has the project gone? Do you still use it? Have advancements in the field unlocked significant new utility? Maybe there is a Part 2 please post a link if so.

Nik - 10/16/2025​

Hey Chris, I have learned a ton since I built Monstra. I work using Cursor every day, so it's much easier to interact with AI/LLMs in general than it was a year ago. Monstra has made it very easy and convenient to run local LLMs on data I want to keep private, and it's been amazingly useful to have as I FAAFO. Still, building/rolling your own only has utility if you like exploring/learning. If you just want to GSD, running the OpenAI API (or any others) is probably a more efficient use of cash & time unless you're going to run a shitload of jobs. Does that help?

"What the heck is that thing?" It's something I hear via the built-in mic all the time as I've been driving my new Frodobot around Normal Heights.

A Frodobot is a small, remote controlled robot designed to "crowdsource a massive global real-world robotics dataset" according to founder Michael Cho.

For most of us, it's just a fun little RC robot to drive around and wonder at the tech allowing you to control a robot via an internet browser from anywhere in the world. It connects via its own little built in cell phone, so as long as you've got cell service where it is, you can drive one.

It's kind of like the nice version of being one of those drone pilots stationed in Nevada, except instead of executing airstrikes you're delivering desserts.

You could also be exploring with your buddies; every Frodobot has 2 cameras (front & rear), a mic, and speakers. You could be in Belfast, your buddy could be in Kinshasa, and both of you could be driving robots together in my hometown of San Diego.

Now, that's not quite reality yet; Frodobots are still super new, and there are only two options to drive one. First, you sign up to drive one in the Frodobots fleet, which for the most part are all on an industrial building's rooftop in China that's set up with a maze to navigate through.

The other alternative is to buy your own Frodobot, which is what I did. In fact, I bought 2. They advertised $199 each, which was a little less than what I actually paid, plus $236 in shipping.

They shipped via DHL from Hong Kong, arriving just 4 days after I'd ordered them; super fast!

Now, you might ask, "Nik, why the heck did you buy a Frodobot, never mind two of 'em!?"

First off, I love geeky little things that are new. Second, I thought they'd be fun to use at my other company, Paleo Treats, as part of a local marketing strategy. Third, I think there's some small income potential here to rent out the bots to people from around the world. Fourth, I'm betting that Frodobots is going to be a DePIN, and I've seen how important it is to start early on ordering hardware for DePINs.

The first thing doesn't need a ton of explaining.

The second thing is pretty straightforward; we've been using our little Frodobots to make mini-deliveries of our desserts to other businesses. It's a fun thing we can do to to collaborate with other businesses. Nothing crazy, just delivering stuff via robot. You can check those out over on the Paleo Treats IG.

The third thing or fourth things are why you might be here.

The question to answer there is:

How do I make money with remote controlled robots?

It's pretty simple: Frodobots is setting up a marketplace where you can rent out time on your Frodobots to anyone else in the world. A user will log in to something like driver.frodobots.com, pre-pay for time and probably some insurance payment, then they can select your bot, and as long as you have it on and charged up, they can drive it around.

Now, I'll get a little hand-wavy here and just guess at a couple of reasons that driving a Frodobot might be worth paying for. None of these have been vetted by GK Legal, they're just ideas.

The Maze​

You could set up a local "maze" course in the parking lot (or warehouse) of your business and offer prizes for who completes it the fastest, or, if you have more than one Frodobot, who completes it first out of a group. The prizes might be money, or a coupon to shop at the business, or more time on the Frodobots marketplace.

This seems the most realistic, as Frodobots is already doing this in terms of renting out time for money and then awarding special roles as prizes.

Local Gifting​

We think it might be fun to offer up the chance for someone to deliver a gift to their friend. Let's say you live in Alabama, but your bestie lives down the street from a business with a Frodobot. You might buy something at the business, have 'em put it in the Frodobot basket, then drive it down to your friend's place as a special delivery. Since there's a camera, mic, and speaker, you get the fun experience of interacting with your friend through the robot. Sure, you could call 'em, but this is something new: "I delivered your gift myself, with a robot I controlled." Not everyone can say that at a dinner party.

Local Challenges​

Given local street layouts, it might be fun to do a larger version of the maze, where you captain a Frodobot around to local markers, "tag" them, and come back, all under a certain time. If you're an enterprising type, you could print up QR placards that you affix to the entryway to a business at Frodobot camera height. You could then automate the whole thing, and award prizes based on when the QR was scanned.

Casual Walk​

This was actually one of the first ideas I had for this. My nephew lives overseas and my parents live in other states. I thought it would be cool to hand over control of a Frodobot to them before I go on my morning walk with the dogs. In a way, I get to take them along on my walk, get to chat with them and share the same things I'm seeing; when I feed the crows with peanuts, what my little doggies are getting up to, and just have a slightly different experience than they might have if I just Facetimed 'em and they had no choice in what they got to look at.

All those are just ideas of what you can do with a remote controlled robot. Let's dig in a bit to the tech next, then we'll finish up with the "What if it becomes a DePIN?" question.

The Tech​

Ok, so what's in 'em? The components are in two separate areas; the sensor package, which is up in the "head" (where the camera, speaker, and mic are) and then the locomotion package, which is down in the body.

I asked ChatGPT to help me go through what I saw up in the head:

On Main Board (labeled TC-RV2216 Core)​

• Rockchip RV1126: A high-performance AI processor from Rockchip, typically used in smart cameras, AIoT, and other intelligent devices.
• Rockchip RK809-2 : This is likely a power management integrated circuit (PMIC) used to manage power requirements of the device, working in conjunction with the main processor.
• SEC 110 K4F8E30 : This appears to be a memory chip from Samsung Electronics (SEC), possibly RAM or flash storage.
• SEC 149 C041 : Another memory chip from Samsung, probably an eMMC flash storage used for the operating system and data storage.
• YXC 24: This might be a crystal oscillator used for timing purposes, with "YXC" being the manufacturer and the numbers representing part code or frequency.
• SP485EL: A low power half-duplex transceiver chip used for serial communication, likely RS-485.

Over by Modem (top right)​

• PAM 8610: A digital audio amplifier chip, used for driving speakers and providing sound output.
• Fn-Link 6223a-ST : A Wi-Fi module providing wireless connectivity.

Other components​

• Slkor LM2596S-12: A voltage regulator chip, likely used to convert higher voltage down to 12V.
• GG2302: This could refer to a specific part or chip on the board, but without more context, it's unclear what its function is.
• Buttons for Power On, Reset, Update
• USB OTG: Stands for USB On-The-Go, a USB port that allows the device to switch between acting as a USB host or USB device, enhancing connectivity options.

Then, down in the body, the following.

As you'll see, GPT wasn't sure what all of 'em are, but for some of the nerds who'll read this they may make sense, so I'm leaving them in.

• Espressif ESP32-S3-WROOM-1 : This is a Wi-Fi + Bluetooth module from Espressif. The ESP32-S3 is a series known for its high performance and versatile application range, including IoT, wearable electronics, and smart home devices.
• ADB210 YRZ #713: This could be a part number or a specific chip, but without more context, it's difficult to determine its exact function. It might be a part of a larger integrated circuit or a specific identifier for a custom component.
• SS14: This is a Schottky diode, known for its low voltage drop and high efficiency, commonly used in power rectification and as a protection device in circuits.
• 4701, 2202, 1003: These are likely resistor or capacitor codes, indicating their values. For example, 4701 could mean a 470-ohm resistor, and 1003 might be a 100 nF capacitor. These components are fundamental in electronic circuits for controlling voltages and currents.
• SS36: Another Schottky diode, similar to the SS14 but likely with different voltage or current handling characteristics.
• BAXY NO58: This appears to be a specific part number, which could relate to a custom component or a specific version of a standard part. The exact function would depend on the device's design.
• T4: This could represent a transistor or a specific code for a component. Transistors are used for switching and amplifying electronic signals.
• IRFR7440: This is a power MOSFET, used for switching and amplifying electronic signals in high-power applications.
• THP4 M51: Likely a specific part number or identifier for a component, needing more context to determine its function.
• HIP4081AIBZ : A high-frequency, high-voltage full-bridge N-channel FET driver IC, used in motor control circuits and other high-power applications.
• WCH CH240C : This might be a USB to serial chip, often used for connecting USB devices to serial interfaces.
• MAX3485: This is a low-power, half-duplex RS-485/RS-422 transceiver, used for serial communication in industrial environments.
• JM32R LM259: This seems to be part of a product code, possibly related to a power management chip like the LM2596 (voltage regulator), but the exact part is unclear.

Ok, so what are the takeaways from that? First, that the team at Frodobots is making little to no money on selling hardware; at $199 this is a pretty darn good price.

Second, the presence of the RV1126 tells me that we'll see obstacle avoidance as a feature in the future, and I won't be surprised to see some autonomous driving capabilities like "Hey Frodobot, can you go down and pick up a loaf of bread at the corner store" at some point in the future, where the route gets mapped out, sent to the Frodobot and it drives the route on its own.

Third, the Fn-Link in the head (WiFi connectivity) means we may see not only local programming possibilities wirelessly (though you can probably just plug in to the USB port as well), but there may be some crossover with other DePINs regarding mapping out WiFi networks.

That brings us to the last of the reasons I bought this:

What if Frodobots becomes a DePIN?​

A "DePIN" stands for Decentralized Physical Infrastructure Network. It's a new type of business model where people (or businesses) buy hardware that provides some kind of service. The service can be wireless connectivity, or weather data, or dashcam imagery. Whatever it is, providing the service is rewarded with tokens. Those tokens can usually be traded in cryptocurrency markets.

A good example is Helium Mobile, where you can buy a Helium Hotspot, provide coverage for people using a Helium phone plan, and be rewarded in tokens for the amount of data that flows through your Hotspot.

In the case of Frodobots, you've got to ask a few questions: What is that they want? What is the most efficient way to get it? How would you build an incentive structure to achieve that?

As founder Michael Cho has said, the goal of Frodobots is to "crowdsource a massive global real-world robotics dataset". The DePIN play there seems pretty straightforward:

Provide a way for people to earn tokens by generating a real-world robotics dataset, and make that ability globally accessible, whether you're the robot owner, driver, or fleet manager.

Of course, any proposed business has to have some value, so how might be figure out the value of this market? According to a market report from Marketsandmarkets.com, "The global AI in Computer Vision market size is [...] projected to reach $45.7 billion by 2028, growing at a CAGR of 21.5% from 2023 to 2028."

Of course, that's just AI in computer vision, which is only a part of what Frodobots can do. As [Forbes has said](http://Data is the lifeblood of modern artificial intelligence.), "Data is the lifeblood of modern artificial intelligence", and the real world data these robots can provide at a large fleet level may be staggering.

Again, from the Forbes article:

Getting the right data is both the most important and the most challenging part of building powerful AI. Collecting quality data from the real world is complicated, expensive and time-consuming.

[-Forbes](http://Getting the right data is both the most important and the most challenging part of building powerful AI. Collecting quality data from the real world is complicated, expensive and time-consuming.), Synthetic Data Is About To Transform Artificial Intelligence

Frodobots may offer an alternative, or better yet, a supplement to synthetic data that isn't complicated, expensive or time-consuming.

The value of real-world robotics datasets lies in their ability to reduce time, cost, and resources required for developing, training, and validating robotics systems and AI models. You don't have to look very far to see super smart people diving deep into figuring out how to build foundation models for robots to interact in the real world.

Basically, if you have access to a large data set from the real world, and especially if that dataset involves the interactions between humans and machines, that is potentially a very valuable thing.

Contrary Research unpacks this idea in the Scale AI Report that we *used* to not have enough data.

A long-running issue with building artificial intelligence and machine learning applications was a lack of well-organized data required to build models.

-Peggy Wang & Sachin Maini

Companies like Frodobots may help to start alleviate that data shortage, helping to continue building our models and understanding of the world in new ways.

All of that adds up to why I bought a couple of Frodobots. Frodobots represents this fascinating intersection of technology, business, and community engagement, and I love being a participant at that glorious intersection.

If you buy one and want a code for $50 off, use: 8HSR5H

If you're looking for a mobile plan for yoru Frodobot, the best I found for unlimited data was a Helium Mobile plan at $20/month and $5 off the first month if you use that link or code KAS6KPL

Rock on!

Archived Comments​

Oro - 4/11/2024​

Thanks your thoughts on Frodobots! I discovered them here and will likely be ordering one or two using my Discord discount. Here is a Reddit AMA from the co-founder that has more information about them https://www.reddit.com/r/frodobots/s/4fMb7SztC5

Nik - 4/11/2024​

Yeah, Michael did a great job over on Reddit with those, thanks for linking 'em!

Oro - 4/11/2024​

NP! Nik, what IoT 4G carrier are you using for your FrodoBots Zeros? I'm considering Soracom. Do you have any experience with them?

Nik - 4/11/2024​

I'm using a pSIM from Helium (hellohelium.com). $20/month unlimited. $5 off first month with code KAS6KPL Mint Mobile has $15/month for the first 3 months but limited to (I think) 15 GB.

Oro - 4/13/2024​

Thanks, Nik! Mint Mobile has an Unlimited Plan for $15/mo for 3 months rn. The same plan/price can be extended too: “Once the 3 months are up, you can renew with a 3, 6 or 12-month plan to keep the savings going.” https://www.mintmobile.com/plans/ I’m going to get two lines for two Zeros. I’ll update once everything arrives. ? ?

Nik - 4/14/2024​

Oh nice, I didn't see the unlimited plan option when I signed up. Nice find!

Oro - 4/24/2024​

Hey Nik! I got my Zeros and The Tech looks the same. There are some cosmetic differences though: my Zeros have a small 3-D printed cap over the wires connecting the head electronics to the body electronics instead of the full stainless panel that yours have. I like the look or yours better; not sure why it was changed (cost? repairability? both?). I activated one of my Zeros and the Mint Mobile Unlimited plan I used is only $15/month for 3 months; it doubles to $30/month if renewed for 12 months. I'll likely be moving to another carrier in 3 months.

Nik - 4/24/2024​

Right on. I've thought a little about how to improve that side panel, hadn't thought of replacing it with a 3D printed cap though. I was thinking more of hammering out a little channel for the wires (though it's clearly not necesssary.

Oro - 5/1/2024​

Hey, Nik. Do you happen to have pictures of the back of the circuitboard in the head? I'm running my Zeros in a low bandwidth area, and I'm thinking of adding physical switches to disable one or both cameras so that not as much data is sent to the Agora API. The two HD video streams are not necessary when I'm driving using line of sight and they slow down the responsiveness of the Zero. Hopefully in a future version of the firmware turning off the cameras will be an option without using physical switches.

Nik - 5/1/2024​

Yep, here ya go:

In the early spring of 2023, I began investigating building my own VO2 mask as an amateur athlete and professional tech enthusiast. I stumbled upon directions to build one, which you can view here.

According to the original designer, “The output includes continuous output of VO2, calories consumed, volumes of expired gas, as well as performing such functions as Basal Metabolic Rate.”

The device looked like it would work, so I ordered the parts, printed them in the GK 3D printer and put it all together. I was dissatisfied with the results.

The ventilation tube threads were weak, and there was no VCO2 option or pressure option (for correcting for altitude.) It was a little bulky for some of the exercises an athlete would do in the normal course of working out (pull ups, almost anything with a bar in the anterior plane), and from my experience was more an exercise in what could be than a great working mask.

However, it was a superb start. It allowed for calibration against local air, and the battery life was good for at least 3 hours with readings every 5 seconds, about 2,160 readings.

The original designer had gotten it to work well enough to output accurate data as measured against a lab unit, so I knew it was possible.

“We tested the device in a physiology lab against a $60,000 machine and found it gave the about the same results.”

Rabbitcreek, Instructables

They had also integrated it with Zwift, a popular biking app, and to an iOS app called Sensirion.

The thing was (is) cool, but I thought there's plenty of room for creative improvement.

Before we go deeper, I'll give a little background for readers new to the GK blog. I'm active in the outdoors and tech-savvy enough to follow instructions, add some problem solving, and either deploy or improve geeky projects. Along with a group of like-minded geeks, I've worked in the Arduino and Platformio worlds to build projects tracking WiFi and Bluetooth users and developed a soil moisture sensor housing

We’ve also developed a radio bridge that moved data from rtl_433 to a public LoRaWAN.

On the physical side, I've run long distances, lifted heavy weights, flown paragliders, and participated in a bunch of sports, mostly endurance related. I've paddled the Catalina Classic (32 mile open ocean paddle), was an athlete in the very first CrossFit Games, run Leadville, and a million years ago represented the USA in international competition as a Naval Pentathlon athlete (came in 5th on both of the swims.) I’ve used commercial products like Oura rings, wearable force plates, stride analyzers, and heart rate monitors.

I've also been a soldier, with all the usual experiences there; long marches, heavy backpacks, jumping out of planes, diving underwater, lots of guns and explosives. Basically, most of the things in the brochure.

All of that is to say I thought of myself as an excellent user for a VO2 mask; someone who could build it, use it, and improve it. I looked around at the options, and they all face two big problems:

First, they're super expensive. The VO2 Master is probably the closest thing out there to what I'd want. It's a very cool piece of kit but at $6k a unit it ain't going mainstream, and as of this writing it didn't have the ability to measure VCO2, which is incredibly important if you want to get the maximum value out of the uncomfort of putting a mask on your face while you work out.

We'll take a very short diversion here into what VO2 and VCO2 are, and why they matter.

First, the obvious thing: VO2. It's stands for Volume of Oxygen, and a measure of the oxygen in your exhaled breath. If you can do that accurately enough, and you know how much oxygen is in the air around you (from calibrating before you put the mask on) then you know how much oxygen you're using.

Very simply, the more oxygen you can consume, the more metabolically fit you are.

VO2 measurements for the most part are only used for elite and pretty good athletes to show off at parties, mostly because it's so damn difficult to get readings frequently enough to inform your training. You go to a lab, strap on a mask attached to a hose attached to a "met cart" (metabolic cart), and then a technician makes you work harder and harder on a bike, treadmill, or if you're fancy, a rower until you decide you've had enough.

The tech goes through your readings with you, explains what they mean, and typically suggests what you might do to improve. It's very cool, but it's a cumbersome metric to obtain.

I'm NOT saying it's not a legitimate marker, I'm just saying that because it's so difficult to get readings, it's not really that practical to care about your VO2.

In a perfect world, you'd want to see VO2 (and VCO2, more on that in a minute) as easily as you see your heart rate. That way you can tell if the workouts ("interventions" in lab-speak) are working.

Ok, so what about VCO2? That's the Volume of Carbon Dioxide you exhale. If you know how much carbon dioxide you produce AND how much oxygen you consume, you can math a little and figure out how many calories you're burning, and if those calories are coming from burning fat or carbs.

Without going into why someone might want to know that, I'll just drop this: A device that can show you VO2 and VCO2 accurately, in real time, in the field (meaning you can wear the thing anywhere, not just be hooked up with a plastic octopus on your face in a lab) doesn't currently exist on the market.

DIY versions like the one I built do exist, but they ain't ready for prime time. That brings us back to where we are now: Building a VO2/VCO2 mask ready for the market.

That's what I'm working on now, and it is very likely that's why you're reading this; you're trying to figure out what the heck Gristle King Inc is and why I asked you for help with building a mask. Now you know.

My goal is to build a mask that

• retails for less than $600
• accurately reads VO2 and VCO2 in real time
• transmits short and long distances
• has a display mechanism for that data (app, watch, tablet, etc)
• is comfortable enough to wear that an athlete serious about monitoring it can tolerate it for hours
• easy enough to use that you don't need much more than the technical ability to put together an Ikea cabinet.

That's the deal. If you can help, I'd love to hear from you.

Archived Comments​

Eric Chelini - 2/27/2024​

Very cool, Nic. Would love to check it out. How can I help? Since I am writing you, I am interested in the DePIN space and Web3. I also have a background in overseas fundraising.

Nik - 2/27/2024​

Awesome, thanks Eric! Good to know re. the fundraising side, will reach out if that seems like a good option. Let's see what happens with this grant.

Fraser Thurlow - 5/29/2024​

Hi Nik. I've recently finished my PhD at the Australian Catholic University. As part of this, I frequently used the COSMED portable metabolic systems, as well as laboratory systems, to determine a range of measures in exercising athletes. Considering that VO2 measurement is the gold standard for exercising testing and prescription, I've always thought there is a great need for a more compact, user friendly system that's available at a reasonable price for the everyday athlete. It's great to see the design you've made so far! I don't have expertise in product design, but I do have a lot of knowledge and experience regarding the use of these systems and how they can be best applied to optimise physical performance. I also have access to university facilities should product testing be required. I'd love to chat further :)

Nik - 6/8/2024​

Awesome, thanks Fraser! I ended up not getting the gov grant I applied for, so the project is in a holding pattern for now (though everything is open source so anyone can continue work on it). If I start it back up I'll def reach out to you! Can always email me at my first name then at, then gristleking.com. :)

frnkr - 10/19/2025​

Hi Nik, what's the status of this project?

Nik - 10/19/2025​

Haven't touched it in a while. I didn't get the grant I was going after, and as fun as this was to build I wasn't willing to put up the kind of R&D blood & treasure it would take.

DavidG - 10/23/2025​

Hi Nik, Nice article. I recently tested the zonex which is a from a French startup but currently measures only VCO2 and VE with the aim of giving you your ventilatory thresholds. There are other devices coming onto the market now as well. So maybe getting a bit crowded. I'm hoping to build the original project myself at some point.

Nik - 10/24/2025​

Right on David. I've gotten a couple of emails about this article in the last month; seems like people are getting fired up about VO2 now?

The Rabbit r1 launched to great fanfare on January 9th, but nobody I knew really understood what to use it for. It was one of those "Holy shit, that thing is rad, but..what does it do again?" moments.

What is the r1? For my regular readers, I'll start with this: It has nothing to do with DePINs or crypto (yet).

For everyone else...let's start with what it looks like. It's a little orange box about a thumb-width thick and maybe as long and wide as your index finger. Basically, a little smaller than your palm (unless you have plumber hands and can rip a phone book in half, in which case it's half the size of your palm.)

photo credit: Rabbit

It's got a scroll wheel and a screen, a camera that can roll its own eye up into a cover (so it can't watch you all the time) and a place for a SIM card so it can connect to the webz.

Now, far more interestingly, what does the r1 do? It uses fancy AI (technically a LAM, or Large Action Model) to watch what people do on their screens, listen to what they say while they do it, and then encode that into repeatable actions.

What the fuck does that mean?

Basically, it's like a very smart 10 year child with perfect recall who does exactly what you tell it to. It can watch and mimic what you do on a computer, sometimes with uncanny accuracy. Whether you're editing videos and could use an "auto mask" assistant, or you're always booking trips online and get tired of the details, you let the Rabbit watch what you do a few times, then, it just...does it for you.

Technically, a Rabbit is an agent, a "software entity capable of performing tasks on its own." Those of us with a love of sci-fi will remember the hotel agent in the book Altered Carbon; we ain't there yet, but we're getting closer.

The key points here are that a Rabbit will:

A) watch and listen to you

and

B) you can tell it what you want to do

No programming necessary. It doesn't have to integrate with an app any more than you do, it just listens, learns, and does. That's it.

Oh, and one more thing. The L in LAM. It doesn't just listen to and learn from you. It does that for everybody.

So if someone who is a movie editing wizard decides to walk their Rabbit through their process, well, not only can they reuse that process, but they can *sell* that process to you.

[Rabbit] is confident it can make business-sustaining money through licensing its platform.

Kyle Wiggers, Tech Crunch

This means the the experience-buying market is about to get even more frictionless. Upwork used to be the hot place to drop a tech gig you wanted done now, but that's still pretty messy. There's sifting through people, hiring one (or a few), testing 'em out, then seeing if you like working with them.

Of course, if what you want is an imaginative engineer to solve a current problem in a novel way, well, Rabbits prolly ain't for you. If, however, you'd like to just buy any kind of online executory experience that might have taken someone years to master and they now do as casually as Stephen Curry makes free throws, well, there may be a Rabbit for that soon.

So, where does that leave you, someone who owns or has ordered a Rabbit and is wondering what you'll do with it?

One thing we'll probably see is an explosion of "Master class" style offerings, except instead of Neil Gaiman teaching story-telling, you'll be be able to buy the story-telling Rabbit who was taught by Neil, then write your own Gaiman-esque stories.

If you were to consider preparing for your Rabbit, you might want a "What to do while you wait" pamphlet.

In broad strokes, it's reasonably safe to start thinking of any Rabbit action or sequence as requiring three things:

• Intent through spoken word
• Screen recording
• Desired outcome

An example of this might be:

Intent: Check across 30 Discords for the general sentiment around DePIN in crypto
Screen recording: Going through the Discord servers I'm in, searching through the latest posts, giving each server a rough "sentiment score" on DePIN in crypto
Desired outcome: Overall sentiment score for the idea of "DePIN in crypto", maybe on a scale of 1 to 10.

That's not by a long shot the best or only or most complex, it's just one idea based on some of the work I already do.

Obviously we'll see Twitter flooded with the "My boss thinks I'm a genius, but I've just got a Rabbit in my pocket" ads. I'm sitting down with my buddies to think up of all the ways we might use ours to help YOU! They may steer crops, make movies, manage social media details, or book that trip for St Cuthbert's walk.

In the meantime, what can you think up that your Rabbit might do for not just you, but all of us?

• booking a flight
• pentesting & wardriving (say no more!)
• growing crops using sensor networks
• writing new songs
• setting up complex network rules (firewalls, device tunnels, etc)
• visualizing all sensors in a space
• visualizing all DePIN miners in a given location

Batch 1 ships starting March 31st, 2024. To the future!

Curious about the Rabbit? Buy yours here (no, not an affiliate link) --> https://www.rabbit.tech/

Archived Comments​

Warren Bowman - 1/12/2024​

Nik, I love what you do. I was just wondering what the hell this rabbit thing was. Thanks for explaining it. If I may pick one small nit: editing is a subjective craft. Each piece has it’s own look and feel. As a career film editor, I doubt that this device can compare performances on a subjective level, intuit what the writer intended, realize the director’s vision, and give the producer what they want. Am I missing something?

Nik - 1/12/2024​

Right on Warren. As a brand new film editor, I'm on the other side; a ton of stuff that is probably super basic for you is time-consuming for me; drawing a mask around an object, moving that object through frames to simulate motion (think the "superhero landing effect"). That's just a small example, but expanding it out, let's say you had a Rabbit watch you edit, oh, 2 or 3 short films. As you edit, you tell it what you're doing, then you package that up and sell it as a "editing assistant" on the Rabbit marketplace. I buy the WB-Edit-Assist for my RAbbit, and then it's like I've got you in the room with me for when I want to do all the small tech tweaks that are natural to you. Does that make sense?

Fernando Cassia - 1/14/2024​

Hardware specs? Is it wifi or LTE enabled or both? How much bandwidth does it use and what degree of control do I have on bandwidth usage?

Nik - 1/14/2024​

2.3 GHz MediaTek MT6765 Octa-core (Helio P35) processor 4 GB memory 128 GB storage 8 MP, 3264x2448 photo 24 fps, 1080p video Bluetooth 5.0 / Wi-Fi with 2.4GHz + 5GHz / 4G LTE All specs on their page: https://www.rabbit.tech/

You - 1/14/2024​

People need to be realistic about what $200 can get them lol. ChatGPT is $20 a month and is barely agentic, and somehow idiots expect they can get a smart agent for $200 with hardware?

Joe Rizo - 1/15/2024​

I am most interested in the rabbit teach, which i'm sure is what is creating the biggest stir in the developer community. i just pre-ordered the rabbit r1, but would like to get started now on creating the action agent i need. when do we get access? i dont see anything for developers on the site.

Oro - 4/29/2024​

Thanks for this Nik! Now i have a better understanding of an r1. So an r1 and it’s LAM are like an automated code writer for Selenium WebDriver or Puppeteer/headless Chrome? I used to write test scripts similar to the web example on the Research page https // rabbit tech/research using them. BTW, did you receive an r1 yet? What are your first impressions? Also, what do think about the Ai Pin? Does it have DePIN potential? https // humane com/aipin

Nik - 4/29/2024​

Right on! I just got my R1 today. I'd say it's a fun gadget that's not yet delivering on the hopes that got built up. For me that's totally fine; it was $199 and after so much experience with crypto hardware I'm just psyched they delivered on time. :) I don't have any experience with the AI pin. I don't think either of 'em will go the DePIN route, though I heard rumblings at one point they might do a token.

Oro - 5/1/2024​

Thanks, Nik! Interesting article on r1: "The Rabbit R1 is probably running Android and is certainly powered by an Android app under the hood." https //www.androidauthority.com/ rabbit-r1-is-an-android-app-3438805/ Hmmm.

Shane - 5/16/2024​

Does rabit1 have an affiliate program

Nik - 6/8/2024​

I'm not sure. If they do, I'm not on it yet.

Randy - 6/18/2024​

Seeing the demo on January 9th I thought cool, and only $200. Let's give it a try. I've been an early adopter (for the most part), for a long time ... My Rabbit R1 arrived yesterday. Not sure what I can do with this yet but its been fun playing with it. Recording a meeting and having it summarized could be useful, but I haven't tried it yet. Visual object identification is cool and hopefully will get better over time.

Nik - 6/18/2024​

From what I've seen with mine, you won't be able to do much yet, but it's fun to experience.

Ok, so ya caught the Wingbits bug and are willing to play the game just to see how good you can get? It's fun to mess around with the Tar1090 tool (which I've covered in a previous post) and see the pretty pictures, but...what about them numbers?

Pretty clearly here we're seeing that Setup 2 is better than Setup 1. For right now, let's skip the "why" and ask, "How do we KNOW this?"

I know, I know, you're super smart and you can just look at the pictures and see that #2 is reaching out further.

Still, you might be a little geeky, and you might enjoy digging into the graphs and learning about the system so that you know for a fact not only how to move from #1 to #2, but how to measure it along the way to make sure you're making the right moves.

So, how do we do evaluate our setup, which include the dongle, any filters or amplifiers, cable, and antenna? One way is with graphs1090.

Once you have Wingbits installed, you get to graphs1090 by going to http://192.168.xx.xxx/graphs1090/ You'll want to replace the xx.xxx with your local IP.

That will bring up a page that looks like this.

Officially, Wingbits calculates your rewards with this equation:

user rewards = uptime % * data provided in hex % * hex rewards * 1 Wing

The rough translation of that is "I want you to be a reliable source and cover as much ground as accurately as possible."

For this post, we'll focus on the second term, data provided in hex %.

Let's start with the big picture:

For every hex out there, there are X number of available messages to collect each day. Your goal is to have a setup that collects as many of those messages as possible and submits them to Wingbits.

@acci via Discord

We have (at least) two issues when understanding how those messages count: How many messages we're getting out of the total, and the clarity of the messages.

Think of it this way: If you receive an envelope in pristine condition, open it up and have a clearly legible letter inside, you can understand the message. If, on the other hand, you get a torn-in-half envelope with water-stained and faded writing on the letter inside that you can barely read, or the thing is on fire when you get it, well...you got the message, but it's not really useable.

You want to get lots of envelopes that are clearly readable.

There's one more thing when it comes to testing your setup: Some of us have an unfair advantage when it comes to testing.

I live in San Diego, which has an international airport and is a tourist destination. We almost always have aircraft in range. I've got two rigs outdoors and up high. During most weekdays each can easily see over 500 message/second. You'll see the "messages/second" down in the ADSB-Maxima box, and it's a good measure of just how busy things can get.

My buddy r4wk, on the other hand, lives way out back of beyond and routinely gets less than 50 messages a second. Unless he decides to pack up and live closer to a major air traffic hub, it'll just take him longer to gather enough data to make a good decision.

So, how much data do you need? It depends on what you're trying to know, and how sure of the data you want to be. You can get a rough idea of how many aircraft you're tracking by looking at the ADS-B Aircraft Seen / Tracked graph, but keep in mind that's not the total, just a per second amount over the course of the time period you're looking at.

If you want to know exactly how many planes you see in a given time period, you can install and run the Unique Aircraft Tracker which you can find on my Github. By default its set for 15 minutes, but you can change it to any length of time.

gk@wingbits:~ $ python3 unique_aircraft_tracker.py
Unique aircraft in the last 15 minutes: 106

If you're geographically lucky and can "see" a ton of plans, your testing period can generally be shorter.

If you're like my buddy r4wk, you'll need more time to gather the same amount of data. You'll need to figure out what works for you, in your situation.

Ok, so with "how many planes/messages do you need" addressed, what else should you look at?

We'll start with using graphs1090 to tune your setup a bit. I'd recommend you do this in the early evening, after a full day of messages. Set the timescale to 24 hours (at a minimum), then look at the section marked "Messages > -3 dBFS". This is the first ROUGH step of gain tuning, there is a deeper dive into why dBFS isn't the only metric to use over here.

In general, you want that to be less than 10%.

The "dB" is probably familiar to those of you who've deployed Helium Hotspots. This is your RSSI, or Received Signal Strength Indicator. The FS refers to the dB being measured Relative to Full Scale, where Full Scale is 0. You don't want to reach Full Scale. :) Think of Full Scale (in our envelope example before) as your envelope being on fire. Very hard to read a burning envelope.

Now, you might think you just want to stay hell away from those red lines. How would you do that? By turning down your gain. Think of (and yes radio nerds, I know this isn't exactly right) gain as volume. As you crank the volume, at some point it gets distorted and becomes difficult to hear what's going on.

How do you tune your gain? First, there's a way to definitely NOT do it; DON'T use the the Automatic Gain Optimization script. It won't get you where you want to go fast enough, and it may overshoot in either direction.

You can see your current gain in the graphs1090 section labled (oddly), "Misc". Here, you can see my gain is set to 38.6

Changing gain is easy (and let's be honest, it's fun to fiddle with this!). Open up a Terminal to your Pi or whatever you're using and type this, where "xx.x" is the gain you want.

sudo readsb-gain xx.x

Your options are limited to the following. If you set something else, it'll snap to the nearest number in the following list.

0.0 0.9 1.4 2.7 3.7 7.7 8.7 12.5 14.4 15.7 16.6 19.7 20.7 22.9 25.4 28.0 29.7 32.8 33.8 36.4 37.2 38.6 40.2 42.1 43.4 43.9 44.5 48.0 49.6 -10

You CAN get around this with an external gain tuner, but for now let's just use what you already have.

Once you've set your gain, you can wait a bit and watch the 1090graph tables, OR you can stay in the command line and execute this, to give you a real time gain read over a specific time period since readsb started.

grep -sh /run/{dump1090,dump1090-fa,readsb}/stats.json -e '' | jq '.total.local | ((.accepted | add), .strong_signals, .signal, .noise)' | xargs -n4 echo | awk '{printf "\nPercentage of strong messages: %.3f\nSignal: %.1f\nNoise: %.1f\n", $(2) * 100 / $(1), $(3), $(4)}'

That will return something like this:

Percentage of strong messages: 6.657
Signal: -10.0
Noise: -22.3

In that command, you can change the word total in .total.local to one of the following to read the last 1 minute, 5 minutes, or 15 minutes.

last1min
last5min
last15min

I'd suggest at least 15 minutes, but part of the joy of this whole thing is mucking about with it and deciding what works for you.

Now that you've set your gain, let's make sure you didn't set it too high (or too low).

We'll use the ADS-B Range graph to help us confirm that we're still getting enough range even as we turn down our gain.

You can also run the Unique Aircraft Tracker Python script from above, changing your time period to match what you're testing.

Here in San Diego on a weekday, I'd be pretty comfortable running 15 minute tests after I changed something on the rig. That could be the gain, or a different antenna, or a new filter, etc.

At this point (mid-November 2023) it's difficult to say what to aim for with Wingbits. It's not clear if we want max range, or a low amount of "too hot" messages, or the maximum number of aircraft. My *guess* is that you'll be best off striking a balance with all three. Remember, that's just a guess; the official goals haven't been released.

Going Further - ADS-B Analysis With Dirk Beer​

Ok, let's say you want even more number-y stuff. Where might you turn? Why, my good friend and fellow San Diegan Dirk Beer! Dirk has written an additional Python script to help assess your range.

It does this by making an assumption; the aircraft that just transmitted via ADS-B will do again. If you received the first transmission, you should receive the next. If you don't receive the next, it's not because the aircraft ceased to exist, it's that your rig no longer has the range to adequately hear it.

You can find Dirk's scripts along with his well written instructions over on his Github.

Running them will give you a graph like this:

You can use this to act as a check the range reported with my Python scripts, then continue to adjust your gain (or other aspects) accordingly. Dirk has been working a bunch on this lately, and you may find that you'll get most of the information you need just from his script.

Whatever you do, have fun with Wingbits, enjoy learning about new tools and ways to use 'em, and have a blast! Huge thanks to @acci, @grayhatguy, @dirkbeer, and @r4wk for the help with this post; this community is so rad!

Archived Comments​

A Rough Guide to Wingbits - Gristle King - A Guide to DePIN - 12/28/2023​

[…] Graphs1090 gives you some more stats to geek out on. If you want to go deep­er into it, check out this article. […]

For many years, the Gristle King blog was devoted to DePINs, or Decentralized Physical Infrastructure Networks.

As of October 2025, that is changing. I've had a lot of fun building in the world of crypto, but there is a wider world yet calling.

From sensors using LoRaWAN to building community to general marketing consulting to flat out hard work and the many and varied ways one can make their way in the world, I'm fascinated by the intersection of technology, hard physical work, community, and business.

If you like that combination, you'll dig what I'm doing here.

I've written about the basics of Wingbits in a recent post; if you haven't read that yet or you're new to the project, take about 7 minutes and read through it; it'll make the rest of this much easier to understand.

This is the first in a few of the "Advanced User Series" posts on Wingbits. I'll be making a few assumptions for these: First, you understand the basics of the project. Second, you want to do the best job possible. Third, you're not trying to do this the cheapest way possible. You can definitely do a shittier job more cheaply and easily than what I'm about to describe. If minimum effort is your game, this ain't your post. Fourth and finally, you don't mind mucking about (a little bit) with code, mostly copy/paste.

Let's start with setting you up to see how your rig is working. We'll do that using two tools that came with Wingbits and that are built by a Github user named Wiedehopf. Those are called Tar1090 and Graphs1090. This post will focus on Tar1090.

Tar1090 allows you to see what aircraft your rig is currently tracking, including their altitude (by color), the furthest point they were picked up, and if you want to dive into details, their airspeed and how far from you they are. It looks like this, and is hosted on the machine you've got Wingbits on.

To see this, once you've installed Wingbits on your Pi and have it up and running, go to http://192.168.xx.xxx/tar1090/ replacing the xx.xxx with your IP address.

Ok, so why do you care about seeing that (other than that it's pretty cool to watch all those planes zing around the sky)? Well, you might want to know what you're rig is covering so you can decide whether or not to buy or bid on a SkyHex. The SkyHex map is made up of large (Uber res3) hexes, and looks like this in my local area:

I live in the red circle, in San Diego. You can see that a couple of hexes around me have been bought at least once. Each SkyHex has 3 slots available. Owning a slot allows you to earn off signals received from that SkyHex. You don't have to be deployed in the SkyHex to earn from the signals captured from it.

The question here is: Should I just buy the San Diego hex I'm in, or should I buy a few of the ones around me? What could I actually earn from?

To figure that out, I'm going to overlay the SkyHex map on top of my tar1090. Now, in the instance below I used my massive graphic design skills to make it super easy to show you this. There isn't currently a way to overlay the SkyHex map on the tar1090, so I did it in Canva. I highlighted the edge of my range (using a bone-stock antenna) with the blue.

On the surface, I should probably buy all the hexes above me (I'm the black dot with the white ring) and the one to my southeast, but not the south or southwest one. But wait...the next thing you'll want to know is how much traffic is happening within the blue lines.

With a few tweaks, tar190 will allow you to see a heatmap of where most of your flights are tracked. This makes is easier to decide whether or not you want to buy surrounding hexes. After all, it's probably not ultra useful to buy a hex where you can only occasionally track aircraft.

How do we set up the heatmap? Get ready for some geekery!

First, open up a terminal window and log into your Wingbits instance. In my case, it's on a Raspberry Pi and I'm on a Mac, so I'm using (duh) Terminal. We’re going to add a directory and change ownership of it to let readsb store heatmap data. All this is detailed over in Wiedehopf's guide, but I found some of it hard to understand and put together, so @acci over on the Wingbits Discord helped walk me through the basics. Thanks @acci!

Start by pasting in the following, one line at a time:

sudo mkdir /var/globe_history
sudo chown readsb /var/globe_history

Once those are done, you’ll want to go into your readsb file and add some lines to it. To do that:

sudo nano /etc/default/readsb

Add in the heatmap command --heatmap-dir /var/globe_history --heatmap 30 to the bottom, like this:

JSON_OPTIONS="--json-location-accuracy 2 --range-outline-hours 24 --write-json=/run/readsb/ --heatmap-dir /var/globe_history --heatmap 30"

Once that's done we've got two more steps. First, we're going to restart the service.

sudo systemctl restart readsb

Then we're doing to go set up the URL so you can actually see the heatmap. It will be different than your normal tar1090 URL. Copy and paste the following address, changing out the xx to match your address, then bookmark it so you don't have to remember it each time.

http://192.168.xx.xxx/tar1090/?heatmap=200000&realHeat&heatRadius=3

It may take a few hours (or a few days, depending on your air traffic) to really fill in. You can change how it looks and what it displays using Wiedehopf's system, but the above "args" will give you something like this:

Obviously that gives you a much better idea of how much area you're consistently covering, and you can decide what hexes to buy based on that data. Feel free to play around changing out those args to suit what works best for you; you can change how long a timeframe you're looking at, how big the individual heatmap dots are, their opacity, and more.

Now that I've seen that, I have a better idea of what I might reasonably cover and what SkyHexes I might buy.

Another important tool you can use with the tar1090 map is use the Hey What's That instructions to get an idea of what your maximum theoretical range should be.

If you're hitting it or coming close, you'll know you don't need to do much more. If you're not, well, there's work to do!

Remember, this is a crypto DePIN project, which puts it as a high-risk enterprise. Don't put any money in you can't afford to lose! I've gotten the most joy participating in these projects by focusing on learning new things and doing the best job possible. Earning tokens is a great way to keep score, but fairly obviously not a reliable way to pay the rent.

Huge thanks to @acci over on Discord for helping walk me through the heatmap workflow! Any mistakes are mine, no warranty or guarantee is made on any of this, have fun learning & pursuing excellence with me!

Archived Comments​

Wingbits Optimization: Graphs1090 Plus - Gristle King - A Guide to DePIN - 11/19/2023​

[…] how good you can get? It’s fun to mess around with the Tar1090 tool (which I’ve cov­ered in a pre­vi­ous post) and see the pret­ty pic­tures, but…what about them […]

Warren Bowman - 12/23/2023​

This is great information and really helpful. Thank you so much! I just got my antenna, registered and active today, so tomorrow I will work on getting that heat map going. Thanks again.

A Rough Guide to Wingbits - Gristle King - A Guide to DePIN - 12/28/2023​

[…] I’ve writ­ten up a tuto­r­i­al on how to get a lit­tle more out of Tar1090 than just the basic install, if you want to do that, go here. […]

Tracy Johnson - 2/8/2024​

Thank you Gristle King for this writeup! Much appreciated.

William - 7/2/2025​

7/02/2025 On Tar top right side it says total aircraft. Is this a daily number or monthly number? On Tar top right side it says history. Is this aircraft history for month or year? Just trying to clarify a statement I made about this site.

I've been watching Wingbits for a while now. They're a new DePIN rewarding you for tracking commercial and private planes, and some parts of the project remind me of the early days of Helium, back when I wrote the original Rough Guide for Helium. Wingbits is a project that allows you to choose your hardware and that rewards you for providing "better" service. In the early days of Helium I gained a significant advantage by trying to provide the absolute best service over just sticking a Hotspot in a window. I *think* the same kind of thinking will be useful for Wingbits.

First, let's go through what Wingbits is. Wingbits rewards you for placing hardware (a small computer like a Raspberry Pi connected to a radio receiver and an antenna) and receiving automatic transmissions from most commercial aircraft regarding where they are and how fast they're going. If you want to fast forward and just buy the gear now because you're all excited, skip ahead here. Just remember to come back here and keep reading.

The transmissions you'll capture are called ADS-B, or Automatic Dependent Surveillance - Broadcast. They provide the GPS location, altitude, ground speed, and other information about an aircraft. This transmission is public, and can be received by anyone. It is more accurate than radar, and is required for most aircraft in the US.

Basically, ADS-B is a way for an aircraft to say, "Hey, I'm here, I'm going this fast, at this altitude; DON'T HIT ME!"

Like all DePINs, in return for you setting up the hardware and passing data, in this case ADS-B data through Wingbits, you receive tokens. Those tokens can be traded for other currency.

Ok, so why has Wingbits particularly caught my attention? Because unlike many other DePIN projects, you can build your own hardware, you can make that hardware better, and better hardware means potentially more tokens.

Now, NONE OF THIS IS FINANCIAL ADVICE, so don't go spending money just because you read about it on some guy's blog on the internet. Do your own research before jumping in.

For me, Wingbits represents three opportunities: First, the challenge of building something in the radio world. Second, the chance to learn about something new. Third, the ability to write about this and share it with folks. All of those are super fun for me. Take careful note that earning tokens is NOT in my top three reasons for doing this. Yes, it's fun and a way to keep score, but for me that's about where it ends.

Before we get to how to build a rad ADS-B set up for Wingbits, let's talk about why this particular DePIN exists.

ADS-B data has traditionally been gathered and supplied by volunteer hobbyists. These are the kinds of nerds who just like finding things out and sharing what they know. For years, these hobbyists have provided the data they collect to a company called ADS-B Exchange for free; the company made it easy to see all the other information provided, and most hobbyists were satisfied with that arrangement.

Then, in January of 2023, ADS-B Exchange sold to JetNet for a reported $20,000,000. All of a sudden, the hobbyists who'd been providing free info realized that what they'd been doing had value. They just didn't know how to capture it. Enter Wingbits, an attempt to capture the value provided by hobbyists and return some of it to them for their contributions, using the DePIN model.

Ok, so HOW do you make the best possible setup? Let's approach this from two angles: What the network "wants", and how best to provide that.

Originally, the Wingbits teams licensed out hexes. If you bought one, you could earn from it; if you didn't, you couldn't. As the project grew, they decided to change that. I think I've scrubbed out references to "buying hexes" below, but keep in mind that at the end of the day, the person who provides the "best" coverage in a hex is likely to earn the most tokens.

Let's start with the size you'll have to cover. In the Wingbits Litepaper (a rough sketch of what the project is about), the rewards system is set up around hexagons from the Uber H3 system. The hex sizes are res 3, or about 4,600 square miles of area and about 73 miles between parallel edges. In the below view you can see those hexes.

There are 41,162 res3 hexes in the world, most of which are in the ocean and probably can't be covered. That leaves us earth-dwellers with about 15k hexes to populate. Each hexagon has a fixed amount of Wings tokens assigned to it daily. As of 08June2024, that number is 24.

Rewards in this system are directly tied to the quantity and quality of uploaded telemetry data

Wingbits SkyHex Medium article

These rewards may change over time. Please refer to Wingbits documentation for the latest.

Ok, what does that mean for you? Wingbits wants you to RELIABLY capture LOTS of HIGH QUALITY data. The better you do at any of those, the more you'll be rewarded.

So, let's do a couple of examples.

Wings Tokens Earning Example 1​

You're out in Montana, all alone in your hex and you've deployed 1 antenna. Let's say you're trying to be awesome and keep this thing on all the time, for 100% uptime.

You'll earn: 100% uptime x 100% of data (you're the only antenna) x 24 Wings = 24 Wings/day.

Wings Tokens Earning Example 2​

For this one let's imagine you're also in the hex I am, here in San Diego. You're committed to excellence (like me) but you've got janky internet, so you're not up all the time. Still, you're in a great location, so you actually capture 70% of the data to my 30%.

You'll earn: 70% of data x 24 Wings = 16.8 Wings/day.

Wings Tokens Earning Example 3​

Worst case scenario here: You're in San Francisco. Aside from the excellent food and being in the heart of technological innovation, you've got some serious downsides; namely the number of people competing with you.

You're only capturing 15% of the unique data; the other antennas are just outcompeting you.

Available Wings: 15% x 24 = 3.6 Wings/day.

Important to Note: You Don't Need To Be Deployed In Your Hex​

You just need to provide coverage for the hex you bought. You could live outside of it, or on the edge of it way up on a mountain, not actually have you antenna deployed inside the physical hex, and still be providing excellent coverage.

Now, HOW you capture that data is where I think the fun part comes in. Remember, you want to RELIABLY capture LOTS of HIGH QUALITY data.

RELIABLE means you've got a strong and fast internet connection so your hardware doesn't go down. I know some of you are going to use the WiFi side of the Raspberry Pi, and to be honest, that'll probably work fine. I'll be using an ethernet cable. I'm aiming to win, yo.

LOTS of data will depend on where you are as far as your "sightlines" to aircraft and where you place your antenna. I know, I know, it's easy to put it inside near your window and call it good. Just remember, if there's one person like me anywhere near you, they'll be putting that antenna up high where it has clear lines of sight everywhere, and they'll crush you. This is a contest, after all.

The other thing to remember with LOTS of data is where YOU are vs where your antenna will cover. In many cases you''ll have a range of over 100 miles, so even if you're in the middle of a hex (and you buy the one you live in) you could reasonably buy surrounding hexes and pick up traffic there.

HIGH QUALITY data is where we start to venture into the black magic of radio. You'll want to be capture exactly the right signal, and you'll want that signal to be clean and strong. Getting that combination will require an antenna tuned to the ADS-B frequencies. There are 2, the main one is 1090 MHz and the other one, much less used, is 978 MHz. I'll roll with one of each just to see what happens, although my focus is on 1090.

Just for clarity on that, the 1090 is the worldwide standard. 978 is US only, and really for smaller craft. I'll go after 978 because it's fun and I'm in the US. Most folks should focus on 1090.

The next part of high quality data, once you get a signal on a tuned antenna, is filtering out the "trash" around it. For that, you can use any number of filters. I've linked one that is fine below. If you want to go batshit crazy and get the custom-made pure glory, DM me. It probably doesn't make a difference, but I like nice stuff, so I bought one.

Third is an LNA, or Low Noise Amplifier. This is built into the gear linked below, so as long as you get that you don't need to worry much about it.

Now, I'm POSITIVE some radio-nerd will read this and point out how I could read the ADS-B off the UFOs on the backside of the moon if I just did [insert your favorite radio hack here]. That's what makes this fun; there's all kinds of stuff you can do to level up your game. Remember, this is just the Rough Guide. It'll get ya 80% there, and probably 95% if you buy everything below.

Gear List​

You'll need at a minimum a small computer (Raspberry Pi), an SDR (radio receiver) and an antenna. Separate the Raspberry Pi from the SDR with a USB extension.

You may want to invest in nicer gear, although I'd caution against that unless you know what you're doing. I'll cover optimization in a later post; the first big step here is to use gear specific to 1090. I know, I know, you have some old Helium antennas laying around. It's true that they work, but it's like bringing a knife to a gun fight. You want to win, right?

If you want to support the Gristle King enterprise and say thanks for writing this blog, use the following Amazon referral links to get a basic set up.

[ninja_tables id="6525"]

If you want to spend more money for possibly better results, you can use this gear list:

[ninja_tables id="6685"]

Keep in mind that while the gear in the second table is "better" that may not help enough to offset the extra cost. I like doing things well, so I've basically used the second version, but it IS more expensive.

Build It​

Build instructions are straightforward. I used Simeononsecurity's DeFli dual guide setup the first time. Then I cut out the DeFLI setup part and focused on Wingbits for mine. If you don't mind me assuming you know a bit about the Pi, here's what to do:

Start by "Registering an antenna" on your Wingbits account.

The ID will be the station name, something like "super-secret-scorpion".

Next, we'll turn to your Raspberry Pi.

• Set up your Raspberry Pi, connecting your SDR Navflight stick to the Pi using the USB extension cable, then the SDR to your antenna. If you bought the filter, put that between the SDR stick and the Pi. It'll look like this, from the top: Antenna, filter, SDR stick, USB extension, Raspberry Pi.
• Flash an SD card with RaspOS Lite. Use the Gear settings to enable SSH, set the hostname, and timezone. Fancy.
• Insert the SD card in the Pi and power it up.
• SSH in and run the command found in your Wingbits dashboard, something like this:

curl -sL https://gitlab.com/wingbits/config/-/raw/master/download.sh | sudo bash

You'll probably find that the first time you run this code you'll get an error at the end, something like

vector is active. ?
readsb is inactive. Waiting 5 seconds...
readsb is still inactive.

Just reboot and install again. Yeah, I know, a little janky. Hopefully you won't have to do this, but as of November 5th 2023, that's what I had to do. By the way, I was able to get mine working on PoE with a splitter that was 5V and 2.5A.

You'll be asked for your "antenna ID" once that finishes. You got the antenna ID in the first step, remember? It was something like super-secret-scorpion. Make sure you include the dashes!

• Set up your location using YOUR coordinates, not the 33/-115 ones below. You can use Google Earth or latlong.net to find yours.

sudo readsb-set-location 33.67854 -115.12356

You CAN set autogain if you'd like, but I'd recommend against it.

Troubleshooting isn't particularly easy or clear, and I've found that most times if you just check your connections (Pi-Dongle-antenna) and then reboot a few times, it all works.

Weidehopfs Tar1090 and Graphs1090 are included, so you can see what's working by going to http://192.168.x.yy/tar1090 and http://192.168.x.yy/graphs1090 making sure to replace the .x.yy with whatever your appropriate details are.

The Tar1090 is a cool local (hosted on your Pi or whatever computer you use) interface to see what flights are being tracked. It's fun to watch. It looks like this:

I've written up a tutorial on how to get a little more out of Tar1090 than just the basic install, if you want to do that, go here.

Graphs1090 gives you some more stats to geek out on. If you want to go deeper into it, check out this article.

I expect within the next few months we'll significant improvements in both of these regarding testing antennas and setups and how it effects token earnings. Here you can see where I went offline for a few hours as I moved from a test rig inside on my desk up to the roof, with an accompanying bump in stats.

Ok, that should do it! As a teaser, I've heard that these things use 40-60 GB a month, so an off-grid setup might be something fun to explore, although as a long-lasting solution it's probably not the best move with current tokenomics if your goal is to maximize earnings.

In the meantime I'd strongly recommend joining the Wingbits Discord to learn more and stay up to date, please say "Hi" when you come in (I'm @gristleking in there). Huge thanks to @!simeononsecurity, @acci, and @gwosty on the Wingbits Discord for all their help with this. Now let's track some aircraft!

Extra Credit for the Nerds​

Airspace is broken up into several classes; you can think of them as areas where there is more or less control by "the tower". Class A is the most restricted, generally beginning at 18,000' above mean sea level and up to 60,000'. This is where commercial airlines, cargo operators like FedEx, and high performance aircraft fly.

ADS-B is also required in Class B airspace (all around your busy airports) and Class C airspace (less busy airports).

Archived Comments​

Wingbits For The Advanced User: Tar1090 - Gristle King - A Guide to DePIN - 11/8/2023​

[…] writ­ten about the basics of Wing­bits in a recent post; if you haven’t read that yet or you’re new to the project, take about 7 min­utes and read […]

Scott Mitchell - 11/11/2023​

So how am I paid for this? Is there a wallet that gets attached to the station name? Is there or will there be app like Helium that you will onboard the station in, and tie into a wallet that way? I currently feed ADS-B exchange, Flight Aware and Flight Radar 24 with 3 different units. 2 are home built Raspberry PI and 1 is bought from Flight Aware, so I don't think I can modily it.

Nik - 11/11/2023​

No tokens yet, so no pay. Eventually they'll sort that out, you're still what we might call "uncomfortably early." :) You should be able to add in Wingbits to one of the Pi units, but I'm not sure.

Warren Bowman - 11/15/2023​

I have spare helium miner sitting around. I understand it can be repurposed for Wingbits, would raspberry pi perform better, or is it more a question of antenna size/placement?

Nik - 11/16/2023​

Hi Warren, the performance will be much more about antenna placement than what kind of Pi you're running. Does that help?

Wingbits Optimization: Graphs1090 Plus - Gristle King - A Guide to DePIN - 11/19/2023​

[…] so ya caught the Wing­bits bug and are will­ing to play the game just to see how good you can get? It’s fun to mess around […]

Cambo - 11/22/2023​

Great post, thanks for the detailed intro! Are there any considerations for the antenna cable? I'm looking at a 25-30ft cable run from antenna to filter/stick to get the antenna high up on the roof. Heavy on Helium and this was a key factor for setting up the gear, e.g. LMR400+ for long cable runs to minimise signal loss. Can't see much of an impact for wingbits though as just receiving signal, no POC etc.

Nik - 11/22/2023​

Yeah, cable loss will effect how much your rig can "hear". I'm using LMR400 for anything over a foot of run.

Baron Hall - 11/22/2023​

Is there any work being done to shim Wingbits into an existing PiAware setups?

Nik - 11/22/2023​

I think so; def. check in on the Discord, lots of folks are running multiple projects on a Pi there. Cheers!

Baron Hall - 11/26/2023​

I found balena-ads-b which supports all the ADS-B feed sites, including Wingbits. It took me just a few hours to port my existing PiAware setup (since 2014) over to it and I am now feeding both from one install. I'd also recommend the FlightAware Pro Stick Plus as it includes the 1090 band pass filter for the same cost as your recommended SDR. Lastly, if you bother to also send the data to FlightAware, they upgrade your account to Enterprise level for free (a $99/month value), which is non trivial.

Pengo - 12/27/2023​

Thank you for the clear instructions. One question, do the services restart automatically in case of power loss?

Nik - 12/28/2023​

Mine usually do, but occasionally don't. I almost always check.

Derek - 1/13/2024​

Hi Nik, Thanks for this! Will utilize your affiliate links. What do you think of this: https://www.amazon.com/ADSBexchange-com-RTL2832U-Antenna-Software-Industrial/dp/B09F2ND4R6/ref=sr\_1\_2?keywords=flightaware+pro+stick+plus&qid=1705194966&sr=8-2 Re: Lastly, if you bother to also send the data to FlightAware, they upgrade your account to Enterprise level for free (a $99/month value), which is non trivial. Would you be able to do that ONLY with their SDR or is that something you would setup in the Wingbits software?

DePIN – Infraestructuras Descentralizadas – Planeta Selene - 7/8/2024​

[…] Wingbits es una startup innovadora con sede en Estocolmo que está revolucionando el seguimiento de vuelos mediante tecnología blockchain. La plataforma recompensa a la comunidad por configurar hardware que captura las transmisiones automáticas dependientes de vigilancia (ADS-B) de aeronaves comerciales. El ADS-B proporciona información detallada sobre la ubicación GPS, altitud, velocidad terrestre y otros datos de una aeronave, que es más precisa que el radar tradicional. La misión central de Wingbits es agregar datos de vuelo a nivel global para desarrollar modelos avanzados, como algoritmos de predicción de retrasos. 1thehub.io2depinhub.io3youtube.com4gristleking.com […]

Jim - 8/1/2024​

Can't find a working discord invite on site or on X? Any help?

Nik - 8/1/2024​

To Wingbits? Should be down at the bottom of their page; I just tried it, worked for me.

Jim - 8/1/2024​

Wingbits yes, thanks for replying. I manually copied the invite into discord now and it has worked. Before was relying on browser invite and it kept failing.

Peter Kennedy - 8/2/2024​

Would you be willing to build me a Wignbits station? I live in SD North County. I have my antenna installed and I have a Wingbits account. Please give me an estimate if possible.

Nik - 8/3/2024​

Hey Peter, you can totally build one yourself. All the parts are listed on this page. If you run into any problems you can ping the Wingbits Discord. What has you most worried about the build?

Peter Kennedy - 8/8/2024​

Hey Nik - You have more confidence in doing this than I do. I asked someone who regularly builds for Flightaware to build me a Wingbits. After a week he gave up, too hard for him. Everything with Raspberry Pi is Greek to me. Please let me know if you would and how much it costs. Sincerely, Peter

Nik - 8/8/2024​

Hi Peter, hit me on email with "Wingbits build" in the subject. I'll get some more info and then get a price to you. My hope is that I can convince you to build it yourself, but if not, I'll have an option for you. Email is my first name, then gristleking.com.

shawn - 11/24/2024​

What's your understanding of the latest rewards program from Wingbits for mainnet. Has it changed from the testnet rules? Latest from Wingbiits docs seem to conflict with your statement`You Don’t Need To Be Deployed In Your Hex`, they're FAQ mentions flagging stations that are located outside of their registered res6 hex - https://docs.wingbits.com/project/wingbits-faq My cursory understanding from browsing their docs on upcoming mainnet(starts Jan/2025) and using their HW, MGW310 or WB200, is that only one person can register a station in each res6, and the only requirement for registering is being the first to do it online and the station will have to be located in that res6 hex. Then each station can capture/report data from as many nearby re3 hexes around it as it can reach and get rewards that way. But that seems to imply only registered res6 stations will be allowed to participate and earn on mainnet. Maybe you can clarify/confirm, do you know if other stations can be deployed and live on mainnet and located in a res6 hex of which it is not the registered station but can still be able to earn rewards such as if it has a better antenna, etc? Wingbits says they will drop your res6 registration if your station reports outside the res6 hex, but it's not clear how that would be executed, as no one else can register for a res6 hex that is already registered. It seems confusing, maybe I am misunderstanding though, I want to get a station and be ready come Jan/2025, but not seeing an incentive as of now, what's the point for people to buy/build a station if the res6 hex you live in is already taken/registered? Seems like the product could have just followed a PoC proof of coverage model for station rewards instead to allow the network to dynamically use the best coverage overall rather than this registration process.

Nik - 11/24/2024​

Yeah, this sounds right (first to register), but I'm not sure.

ROBERT SCHUITEMAKER - 2/12/2025​

Seems Wingbits no longer does BYOD (bring your own device). They did for a while and then you needed to purchase a device to confirm your location but they no longer sell them. Pity as I already have a SDR feeding FlightRadar24 with ADSB data. I installed the Wingbits software and my site was marked as 'online' but GeoSigner was labelled N/A and Tar1090 didn't show flights. From what I read, and see on their map, you can share a res6 hex with another site, but not a res3 hex. I tried to claim my residence (street) but get the error "Hex already reserved or claimed", which sucks.

Nik - 2/12/2025​

Yeah, BYOD is long over. I'd reach out to the team to see if your station might quality, though I don't think it will.

So you've read the piece on the Helium Bridge and you want to use one to bring your favorite non-LoRaWAN sensor data onto the Helium Network? Rad!

Let's start off with the current limitations, just so you don't get all hot and bothered then realize what you wan't ain't doable yet. Right now, it works "out of the box" with 4 sensors:

• WS80 from Fine Offset (Ecowitt), a small weather station.
• WH45 from Fine Offset/Ecowitt, an air quality monitor
• Acurite 592TXR Indoor/Outdoor Temp/Humidity sensor
• Springfield PreciseTemp Soil Moisture

For the WS80 specifically we've published a specialized set of instructions so you can make some changes regarding how much it reports on Helium vs how quickly it collects data. You probably don't want to be sending data out as quickly as it's collected, every 1.7 seconds, as that would be a fairly large DC bill.

Keep in mind that this is just the beginning, and that doing this project will introduce you to (force you to?) level up your geek chops. This is not a small "build an Ikea table" project. If you've never done this before, plan on about 3 weekends to put everything together so you can reliably build a bunch of Bridges.

You'll be ordering PCBs and components, soldering them together, and programming what you've built. It's really cool, and we'll walk you through step by step so it's as easy as possible, but be ready to hang in there a bit!

If you want to use the Bridge with other protocols like BLE or ANT+, you'll need to take the hardware basics and do some tinkering, or wait a bit as we continue to tinker as well and add more options.

Find The Helium Bridge Github Here

You can find the a detailed set of instructions over on Github for how to build a Helium Bridge. In this post, I'll add in a few explanatory points and help guide you through some of the things that aren't intuitively obvious to us less-geeky types.

Let's think of doing this in three distinct stages:

• Assembling the hardware
• Programming the device
• Using it with specific sensors

Assembling the Hardware​

Ok, so step one is to order the PCB, or Printed Circuit Board that will be the base of the Bridge. To do that you'll go to the Github and download what's called a "Gerber File". You'll find that in the "pcb" folder, here.

You'll then upload the Gerber to a place that'll print 'em for you. We used OSH Park, which you can find here. When you go there, you'll seen the options to upload your files right on the main page.

It can be a little confusing going through the ordering process, and it seems to change every few months or so. The important parts are to make sure everything lines up correctly and to make sure you order what's called a "mask" when you order the board. Trust me, this'll make applying the solder paste much easier.

You see, we're NOT going to hand solder this board. You're going to use what's called a "Reflow Oven" to attach the various components. You can buy one of those, or, way more fun, you can build one! We built the Controleo3 from Whizoo using their kit and a toaster oven!

Once you have (or have access to) a reflow oven, you'll need the board and the components to put in and "cook it" as well as consumables like solder paste to make the electrical connections.

Everything you'll need to buy or have on hand other than the reflow oven is listed in the Githhub. Rather than double-listing 'em (and risking have separate lists in different places), I'd just go over there and put in your orders. It can take a few weeks to get everything in.

In big picture terms, the PCB is the structure to which the brains (the Lilygo computer), the radio (the CC1101), the power (the 18650 battery), the charging source (the solar panel) and the receive & transmit antennas to attach to. You could connect all those manually (that's how we started), but it can be a little tricky to do that, and using the PCB makes for a much cleaner final project.

Here's the printed PCB and a PCB with the "mask" taped over it and ready to have the solder paste applied. Note how the holes in the mask line up with the places where the solder paste will need to go on the board.

Using a card (usually supplied with the mask), smear on the solder paste once you have the mask taped on. Remember to tape the board to the mat first so it doesn't slide under the mask.

Once you've got your components placed, carefully remove the mask, then slide everything into your reflow oven and cook it!

Here's the PCB after we added solder paste and the CC1101 (the square green board). We got a little ahead of ourselves on this one and cooked it before we added the UFL connector (right below the green board). :)

Ok, so that's cool. You'll end up doing a little hand soldering, but nothing crazy.

While it may seem confusing to read about this, once you have all the components in front of you it will make far more sense, and you can basically "see" the signal flow.

Pay careful attention to the Github instructions, especially the parts about polarity on the battery. Oh, and don't lay this thing down on a metal surface with the battery in; it'll short! Ask me how I know...

Once you've got it all together, but BEFORE you put in the battery or connect the solar panel, you'll connect to the LilyGo (the thing with the red squiqqly line over it in the picture above) with a PC and program the thing.

Programming the Device​

I've always found programming any of these boards to be a little finicky, which has almost always been a result of my ignorance and ineptitude. Read the directions carefully, go slowly, and if you have any problems jump on over to the GK Discord and look for the Helium Bridge channel; we'd love to help out!

If you use Platformio (which is what we used and recommend) and you haven't before, it'll be much trickier to start, but once you're up and running with Platformio it's universally liked far more than other options. So, prepare for a(nother) steep learning curve. Rad, right?

The step-by-step version is over on Github. If you have questions please drop 'em here or jump into the Discord link above.

Using It With Sensors​

Ok, so as it currently stands you can use with sensors listed above. Some of them are on 433MHz (the soil moisture sensor and the temp/humidity sensor listed above), and some are on 915Mhz (the weather station and the air quality monitor above).

As thousands of sensors are out there, it's not practical to list every single option and configuration. You're heading off a little into the wilderness here as you add more and more sensors. If you don't find that fun, this may not be for you. :)

Keep in mind that the Bridge doesn't distinguish between what it hears; it's not like it's just "listening" for one device. If there is a weather station and an air quality monitor in range, it'll pick up and retransmit both sets of data. We're not sure of the limits here as far as how many devices an individual Bridge can handle. For now, if the signals overlap in being received, you'll lose a reading.

It's reasonable to think you'd easily handle 5-10 sensors per device, but that's not a guarantee.

The data is separated out in the decoder, so you can have a couple different sensors all flowing through one Bridge and getting separated out in the Decoder on whatever Console you're using.

One of the cool things Dirk did was write the decoder so you don't have to adapt it for Helium Console or the MeteoScientific Console; it'll just work. Neat, right?

Areas for Further Exploration​

• Level up MQTT - Have each sensor have its own named MQTT topic.
• Adding more sensors - Modify the rtl_433 esp libarary and add more sensor decoders to the javascript decoder
• In the WS80 version of the code, we have a bunch of features like "sleep to save power", reporting on battery status, and aggregating data. Integrating some of those functionalities into the LMICnode.cpp would be pretty rad.