The Digistump Oak; An ESP8266 On Kickstarter

When it was first released, the ESP8266 was a marvel; a complete WiFi solution for any project that cost about $5. A few weeks later, and people were hard at work putting code on the tiny little microcontroller in the ESP8266 and it was clear that this module would be the future of WiFi-enabled Things for the Internet.

Now it’s a Kickstarter Project. It’s called the Digistump Oak, and it’s exactly what anyone following the ESP8266 development scene would expect: WiFi, a few GPIOs, and cheap – just $13 for a shipped, fully functional dev board.

The guy behind the Oak, [Erik Kettenburg], has seen a lot of success with his crowdfunded dev boards. He created the Digispark, a tiny, USB-enabled development board that’s hardly larger than a USB plug itself. The Digispark Pro followed, getting even more extremely small AVR dev boards out in the wild.

The Digistump Oak moves away from the AVR platform and puts everything on an ESP8266. Actually, this isn’t exactly the ESP8266 you can buy from hundreds of unnamed Chinese retailers; while it still uses the ESP8266 chip, there’s a larger SPI Flash, and the Oak is FCC certified.

Yes, if you’re thinking about building a product with the ESP8266, you’ll want to watch [Erik]’s campaign closely. He’s doing the legwork to repackage the ESP into something the FCC can certify. Until someone else does it, it’s a license to print money.

The FCC-certified ESP8266 derived module, cleverly called the Acorn, will be available in large quantities, packaged in JEDEC trays sometime after the campaign is finished. It’s an interesting board, and we’re sure more than one teardown of the Acorn will hit YouTube when these things start shipping.

58 thoughts on “The Digistump Oak; An ESP8266 On Kickstarter

  1. Thanks for the coverage! We’re super excited to contribute to the ESP ecosystem – I did want to add one thing that this article doesn’t mention:

    The Oak is both about a new development board (and the Acorn module that powers it) and equally about the cloud we’ve built to support it – the RootCloud – which is free (and optional) with the Oak boards.

    The biggest feature this enables in Over the Air Updates – you can upload code to the Oak from the Arduino IDE or from our advanced (Arduino Compatible) Online IDE.

    You can also use Serial.print() style debugging wirelessly – it appears right in the IDE just like the Arduino Serial Terminal but without wires.

    The RootCloud is also a full IoT data cloud – implemented as fully API compatible with the Spark API – and adds some features on top of it, like data storage, retrieval, event persistence, and more. The RootCloud also allows you to build point and click dashboards (and share them) from the real time data from your devices.

    We’ll also be releasing a local hostable (Linux/Win/OSX/Raspi) version of the RootCloud API server for those who don’t want their data in the cloud.

    Our aim is to create a full Ecosystem for the Oak/Acorn – not just a development board. Of course, the Kickstarter has all the details!

    Erik Tylek Kettenburg

    1. Starting with the Digispark, I didnt see any code contributions going back to the Arduino IDE. Adafruit seems to have a much more cooperative relationship with the open source community. For example the USBtiny is supported by the official IDE.

      1. If open source community-ness is judged by formal support in the Arduino IDE then few have achieved that – that tends to be more politics than open source. Paul (of the Teensy, and an awesome community member) is a huge contributor to the Arduino IDE codebase and even his boards aren’t supported! I’m hoping 1.6.3+ brings more cooperation with the Board Manager (and hopefully that will be improved).

        1. I’ve just moved to Teensy and Paul is a real force. He had credits on 1.6.1 and 1.6.3 for fixes, but was surprised by 1.6.2 release. Funny thing is apparently a fix they took for 1.6.3 is something he tried to submit many moons ago, but wasn’t accepted at the time. So bitten again he made it work and offered it back and it was accepted as it solves a problem in library management (detecting duplicate libs). Paul has a thriving ecosystem and long standing interaction with the IDE and JAVA that runs it.

          That just says that few outside persons have the ability to do anything but keep up with the changes as they ship – let alone get system wide changes pushed back to the core.

          Also noteworthy is that the Teensy support package was near 30MB before 1.6.0 and with added support to 1.6.3 is now over 65MB. That is a lot of dedicated code that far exceeds what has ever shown up for even the DUE or any SAM variant.

          It seems code base in the IDE is really a nebulous metric . . . how many pre-lease nightly builds were there of 1.5.8 .vs. 1.6.0 >> 1.6.2 to get to 1.6.3 with substantial changes involved.

  2. “repackage the ESP into something the FCC can certify.” Unlikely to happen unless the firmware is locked down on the ESP8266 and it standalone not to be integrated into any other device.

    1. The ESP actually isn’t a hard chip to get ready for modular approval – we’re working with RF design veterans to ensure we can get approved.
      No need to lock down the firmware – there are plenty of chips – like those by TI and Nordic that have modular approval subject to certain guidelines or settings being followed and have firmwares that can be edited, additionally the firmware can just lock out the ability to increase the transmit power and the like.
      We’re pursuing modular approval (approval for the Acorn module) and we don’t see any major obstacles in doing so – nor is this our first time going through FCC (we’ve done it for commercial projects before). We’re not the only ones pursuing approval for this chip either.
      Of course our final design will be open source – so while our approval won’t be portable to clones/derivatives – it should provide a detailed look at what it takes to make a module that can be approved with this chip.

      1. What Adafruit did was build a breadboard to mount an existing “Me Too” ‘reference design’ ESP package – those are already on Tindie. Erik seems to have invested more already with added technical details and extensions on his KS page that you can find on the Adafruit product page that points to outside resources. And his hardware is actually evolved from the reference design.

  3. I’ve backed Digistump before and rushed to do it again – looking forward to it.

    “He’s doing the legwork to repackage the ESP into something the FCC can certify” … AGAIN

    FCC certification is already on some pre-packaged factory units. The OAKs will be improved custom assemblies needing their own testing – but variations on the same core radio hardware and user loadable software scheme should not prevent FCC certification as the radio unit and processor are proven with good layout when shielded properly. I suppose that ‘design & approval overhead’ was why the KickStarter goal was set where it was – and that was doubled in about the first two days.

    1. If you’re racing to the bottom, go cheaper? Adafruit use the ESP-12 module on that board. which is the part that is actually FCC certified. Not breadboard-able or have convenient reset buttons, but at 1/4 the cost.

      What you’re really paying for here is the cloud backend, as well as the “user friendliness”. It’s also as if the ESP8266’s are running a custom core in order to facilitate wireless code uploads too, but I could be wrong.

    2. Unless they use EMI filters (they don’t BTW) on all the pins out of these breakout boards, the long dangling wires etc of breadboard would be acting like antenna. So not sure what the FCC certification would mean anything other than the fact that it can be imported legally. Ultimate the user is still legally held responsible for the interference.

    3. They’re priced about the same – ours comes with the cloud service and OTA updates and super easy setup – no USB UART dongle required. Ours uses a custom module and has custom bootloader/firmware. I think they are too very different offerings – an easy to start dev board vs. a ESP breakout – plenty of room for both, I know I could use both!
      Of course we had no idea this was coming out either, but even if it had come out before we launched, we would have none the less, pretty different amounts of efforts to build something with these.

    1. These are IOT connected devices – if you want a processor and I/O for hardware – buy a Teensy. If you want WiFi connectivity for IOT and buying into the ESP family – the cloud/Network is and must be there (either WWW or local) so the OAK looks like a perfect system to get connected and get working. To get the best of both when needed – buy one of each and set up Rx/Tx between them so the Teensy is free to do what it does best and get OAK’s added IOT WiFi and a supported development environment with true Dual Core multitasking.

      Buying 3 minimal packaged ESP units to start from scratch and supply power and use the cheapest you can find with no extended I/O or fancy features . . . no thanks

    1. dictionary words cannot be trademarked, e.g. Acorn, Apple, Silicon, etc. What are you trademark are words like Intel, Espressif, XKCD, etc — words that were not in common usage, prior to this. You can also potentially lose your trademark, if your brand loses its original meaning and is used to refer to other products, e.g. Coke is used sometimes to refer to Pepsi as well, prompting Coca-Cola to use the phrase Coca-Cola instead.

      1. Yes they can. I’ve run into this problem before; I have a registered TM that is dictionary word + acronym. Another company had-and-lapsed dictionary word+number, they attempted to re-register just the dictionary word but couldn’t because mine was blocking it for the same class(es) & attempted to move to invalidate mine. USPTO had no problem with that but they waited too close to deadline and ended up just re-classifying the mark.

    1. NodeMCU is really cool – this is more Arduino style (vs. Lua) and more of a comprehensive package of IDE, OTA updates, data cloud, etc – using a compiled language (Wiring/C++) also means less RAM issues and faster execution – with support, and a strong community (though NodeMCU has a great community too!)
      I think NodeMCU and Oak are pretty different projects using the same chip, with some features in common – plenty of room for both and NodeMCU is great!

  4. The one meg of RAM is a bit better then 512 that comes with the standard ESP – 01, however like the ESP – 12 with an adapter board for breadboarding, it’s too wide so you only have one row of pins open on each side.
    Adapting a ESP-01 is a 10 minute job for anyone who can use a soldering iron.

    1. Technically, it’s increased ROM not ram. You can stick much larger SPI flash chips on then if you need more code space.

      Espressif support up to 4096KB in their current SDK.

      There’s a few ESP-12D or ESP-12E modules on aliexpress that have larger SPI flash chips. Shouldn’t affect the FCC certification.

      1. Oops that’s what happens when I type on Android phone rather than a proper keyboard. Ive seen the Winbond 16mb/2MB chils for around 30c on Aliexpress, so it is pretty cheap to upgrade them if you need the bigger flash.
        One thing I don’t see mentioned is how much flash is chewed up by their OTA update code.

        1. KS details for OAK say 300KB+ available to the user in the running system. OAK starting with a 1GB flash – then the system STACK is stored twice – once for runtime and once for Anti-Bricking reload capability. Since OAK is not an off the shelf low end package it seems it is using a faster flash allowing overclocking that may not be possible with the lowest cost proof of concept reference units repacked by others.

  5. Obviously this builds upon a lot of heroic work done by many others:

    – ESP-XX by AI thinker
    – gcc tool chain adaption
    – ESP8266 Arduino adaption by community
    – Spark core firmware

    He claims that the firmware is a “clean room” reimplementation of the spark firmware. The hardware itself is hardly worth mentioning – a new SPI flash and an adapter board for the “Acorn”, which is almost identical to many of the chinese boards.

    Of course it is a lot of work to repackage everything, but he is not shy to claim ownership for many things that seem to be in a a gray area. Some people may feel duped for all the free work they were contributing. Where is the line?

    1. I’d like to add, that the published API documentations contains sections that were taken verbatim from the spark API documentation.

      For example:

      “The STM32 microcontroller has internal pull-up resistors (resistors that connect to power internally) and pull-down resistors (resistors that connect to ground internally) that you can access. If you prefer to use these instead of external resistors, you can use these argument in pinMode().”

      STM32? … I don’t think so…

    2. Like many projects we built this on the shoulders of other projects and try to give credit where it is due, especially when we release sources, etc – however – I want to clarify a few points here:

      – ESP-XX by AI thinker: Why? We’re not using their module or layout? In fact we started this before the ESP-12 was out. We’ve played with their modules and they’ve been hit and miss (in fairness the misses were likely counterfeit) – yes they did show FCC cert. was possible, and they seem like a pretty cool company, but I’m not sure what I’m missing here….
      – gcc tool chain adaption: YES! Much more credit due here, I’ll work that into the project text this coming week!
      – ESP8266 Arduino adaption by community: No and yes – I’m sure we’ll benefit greatly from it, but so far our adaptation/core is our own, we’ve been at this for 6+ months – the other one was just released, no way we could have thrown it together since that release.
      – Spark core firmware: API YES! (and we used their Docs as the basis of ours and we put a big giant credit at the top of the docs saying so) and we try to give credit, Zach of Spark and I are also talking on how we can work closer together. Firmware: NO – it’s a very different implementation (very different chip after all) and uses a different protocol that we tweaked to be spark api compat later in the process.

      Anyone is welcome to say where more credit is due and I’m happy to make sure it is given! Digistump has been a solidly open source company for 3+ years, built largely (like many open source projects) on the work of others and hopefully contributing back to that knowledge base as well.

    3. I wanted to follow up here and say – “thanks for the important reminder” – while, as I said, I don’t feel credit is due for some of those things – I’ve added a “Thank You” section, which was on my todo list but should have been higher up – to the Kickstarter page to make sure it is known whose work this was built on

    4. Totally agree. He is great promoter.
      However, the project is not truly open, which part will be open sourced ? Doc ? Make me laugh.

      Why not show us some scheme, or code ?

  6. Haters gonna hate.

    Erik has a proven track record, as he’s already delivered on three previous successful Kickstarter campaigns. Personally, I’m happy to pay a few bucks extra (versus a raw ESP module) in order to get a product that’s ready to work out-of-the-box, with a dashboard/data cloud service that’s ready to use with it, that’s already got a bunch of library porting work done, etc.

    Yeah, I could buy some cheap modules, and spend a bunch of time hunting down half-assed documentation on the internet, and writing my own API wrappers to somebody else’s (or my own) data service. And in some ways, that sounds like fun. But I just don’t always have time for that. I’ve got a full-time job, a family, and a house to take care of. The little project time I can find has to be focussed and fast.

    1. I can’t agree more with this. I often feel “hackaday” should change its name to “hateaday”. It seems to be getting harder and harder to find useful posts in the comment section.

      I facepalm when people jump to the Adafruit breakout board as “new”. Breakout boards for the ESP 07/12 modules are not a new thing. Boards have been shared on Github/OshPark and for sale on Tindie for months previous to this. It’s good that a bigger player is offering a ESP breakout board but I do worry that the 3.3v regulator they used (250ma) is undersized. The ESP is quite power hungry.

      1. I’m almost certainly switching to a 1A regulator for the Oak (the prototypes actually have 1A regs on them already) – and while pulling more than 500ma through it would produce a lot of heat – I rather have heat vs part selection be the limit given how cheap 1A vs 500ma vs 250ma regulators are. Like all wifi radios – these can suck down some power (220ma worst case in my testing) – 250ma would mean no room to power anything else in those cases – or worse it could mean it was unreliable when transmitting at max distance. If you don’t give an ESP chip enough power it tends to stall and then reboot (WDT) – which can be very frustrating to deal with and debug.

      2. I can and do agree more with Dougal . . . I also find it noteworthy (if tinged with sadness for going unbacked) that Erik’s first Kickstarter campaign was about DASHBOARDS – Erik has been working 6 months behind the scenes creating an updated version of what he wanted as soon as he saw the opportunity to do it – and made time to finish his last KS Pro unit and transition out of his Day Job to make it happen.

        With this dedication and commitment I expect this to not only come off on Schedule and be all it says it will as far as robust and supported – but I’ll be surprised if there isn’t a phase II follow on to expand it. After all at $10 a piece shipped – I see more work than profit margin, especially with initial design and FCC certification. Adafruit has a barebones $7 ESP module and repackaged breadboard unit now at $10 – they can afford to make it in house and it is “just another product”.

        Adafruit makes some cool stuff, I’ll make an order someday soon – but not for their ESP units. Especially since they are borderline powered. I’m looking to use ESP8266 instead of nRF24 hardware – if you search on that hardware the surge power requiring an added cap to get reliability. ESP is a different device entirely – but risking under powering the radio is questionable.

  7. How would you compare this to the electric imp? It seems similar in form factor, arduino style development, Wi-Fi uploading, cloud IDE, among other features. This is half the price though. Pros and cons?

    1. Of course my comparison is biased (and I’m keeping it short) but I’ll say that I was super excited about the Imp when it came out – but less so once I found how closed it was – but at the time it came out it was amazing for the price point – I’m glad Spark and others broke down those price barriers with open source hardware not too long after-

      Oak Pros vs Imp:
      Oak is open source
      Oak is about 1/3 the price
      Oak has 10 i/o (11 counting ADC) vs Imps 6
      Oak is completely Arduino compatible
      Oak lets you can host your own cloud

      Oak Cons vs Imp:
      Imp has more pins with ADC and a slightly more accurate ADC
      Imp has lower power consumption is sleep modes
      Imp is available now
      Imp’s cloud is established and well tested already
      Imp has more versatile peripherals

    1. OAK is 3 days announced – with 6 months conceptual/Proto work behind it and 5 months development/production ahead. It is “open source” so it will be posted. It is building on other prior work for stated compatibility with similar systems – so any changes from that will be necessitated by hardware or functional improvements, that you can re-code as desired.

    2. Mike – by scheme what do you mean?

      The device to API protocol? The core? The compiler?

      It will all be posted and documented by the time we ship – but if someone is interested in more details about a specific part, let me know which part and I’ll clean up and post whatever I can find time for.

      Digistump is a very small team (I’ve personally responded to every support email for the last 3+ years) – so time often limits what we can clean up, document, and put up in the alpha stages (once a board is in production we have some breathing room to document the production release), but we’re bringing on some more staff to ease the load and get additional expertise so that I can focus more on that kind of thing.

  8. If [Brian Benchoff] or anyone else from Hack-a-day is reading – we’ll have a small number of alpha units available for public testing if Hack-a-day has any interest in getting their hands on one and giving this all a spin, please email us – we’d love to provide one!

  9. I am surprised by the amount of scepticism in the comments! Can we all just celebrate this wonderful addition to wifi microcontrollers? The Root IDE alone is enough to make me happy, nevermind the great price and functionality. I hear some tones envy in the in the criticism. For me personally, the OAK is the best thing to ever happen to the IoT. Great work Erik!

  10. Unfortunately, even though this project had a lot of promise, it’s ultimately been a failure. The hardware shipped way past the promised date and the final firmware still hasn’t been released almost a year past the project kickoff. I expect delays with Kickstarter projects I’ve supported but this one has been worse than normal and communication with customers was terrible.

  11. Unfortunately, even though this project had a lot of promise, it’s ultimately been a failure. The hardware shipped way past the promised date and the final firmware still hasn’t been released almost a year past the project kickoff. I expect delays with Kickstarter projects I’ve supported but this one has been worse than normal and communication with customers was terrible. This unfortunately has made hesitant to support other Kickstarter projects.

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.