The ESP32 Beta Units Arrive

A little more than a year ago, the ESP8266 WiFi module showed up uneventfully in Seeed Studio’s store. Since then, the documentation has been translated to English, a proper development environment for this chip was created, and everybody is using this cheap but powerful chip for the latest Internet of Things things.

The company behind the ESP8266, Espressif, is not one to rest on their laurels, and for several months they’ve been working on the next generation of powerful WiFi-enabled tiny, cheap systems. They have their silicon, and already 200 lucky people have their hands on the very first test units of the ESP32, the next generation of Espressif’s WiFi chips. The teardowns have begun, and [LadyAda] streamed her initial experiments with the chip to the Intertubes (available below). [Martin] is also one of the guys who received these early beta chips, and he was kind enough to post his thoughts on Espressif’s newest chip.

A little bit of information on the ESP32 has dribbled out, and [LadyAda] and [Martin]’s demo unit confirm all we’ve suspected. There are two Tensilica L108 processors running at up to 160MHz, a lot of peripherals including ADCs, DACs, I2C, SPI, I2S, and PWM, more RAM, AES and SSL for security, and Bluetooth Low Energy. WiFi has also been upgraded, and the ESP32 will support speeds up to 150 Mbps.

The beta units that were handed out to developers are actually Espressif’s in-house development units labelled ESP31B, and the form factor of the current board isn’t confirmed as what will ship. These beta units are boards with castellated pads with a 1.27mm pitch; this isn’t exactly a user-friendly package, and people have already destroyed beta boards.

While the features are great, Espressif has said the ESP32 is not a replacement for the ESP8266. They’re different markets, and if you just wanted to add WiFi to a project, there’s no reason not to choose the ESP8266.

With this new chip comes a lot of new possibilities. The folks at Espressif are doing a lot of experimenting with their internal beta version (ESP31). [Sprite_tm] is working for Espressif, too, so you know there’s going to be some very cool demo projects, like a Sega Master System emulator, with video. This is just a port of the SMSPlus emulator to the ESP31, with the LCD being driven at 15fps over SPI. All the code for this is available over on Github.

With cool new hardware comes the obvious question: when will it be released? That question is hard to answer, but the best guess is, “soonish”. This chip is coming though, and when it’s released it will put a lot of processing power, WiFi, and Bluetooth into a lot of projects. It’s already shortlisted as the best new chip of 2016, but other than that, there’s not much more information.

40 thoughts on “The ESP32 Beta Units Arrive

  1. Here is somthing I never understood. When looking at this data sheet for the esp8266 I was surprised by how little information was actually available on it compared to just about any other ic manufacturer data sheet for such a part. Does the company really not want to encourage development or am I missing something?

    This is less of a complaint and more of a genuine question.

    1. I’m guessing it’s because it’s an ASIC of their own design and they don’t need to/want to tell any more than they ave (the ESP8266 is some ARM core so could be stuck with anything and the new one uses a synthesisable/configurable core from Cadence)

      1. Yea but every other manufacturer includes information not because they aren’t scared that someone will copy but because without that information you can’t confidently design the ic into a place where you haven’t already seen it used or resort to making assumptions that have to be tested wasting more of your time. There’s a reason other companies include that info I just find it ridiculous that this company doesn’t.

        I guess it could be that there’s a much more detailed NDA version of the data sheet though.

        1. I think that datasheet was translated, so it’s possible that only essential parts were done? I’m completely guessing. But I noticed the same thing… completely different from pretty much any other silicon manufacture.

        2. They probably haven’t spent the effort to characterize their chips unlike most of other “western” chip vendors that are in the industrial application. Those additional info cost time & money, so unless they have large customer(s) that need the info, your typical low cost Chinese chip vendors aren’t likely to provide that on their own.

          If there are *guaranteed* parameters on a datasheet, then there would also need to be additional tests at the chip fab/assembly to make sure that the parts meet those parameters, Once again these affect yields, (test time) and cost.

          1. Despite that level of available documentation, the ESP8266 has proven to be a capable little beast. And it’s been an interesting and useful experiment to “open-source” the creation of more useful SDKs and frameworks the way they have. The ESP8266 is starting to show up in commercial products, so there must be some confidence in them.

    2. The ESP8266 wasn’t originally designed as a development platform, it started out life as a way of adding WiFi cheaply to low-end Android tablets and you used their firmware and drivers. I don’t think they’ve ever put the resources in to properly document it since.

      1. I think you’re right about the chip’s origin. I’m grateful that they chose to expose the programmability to the public last year. Everyone benefits I think.

        I found that the quality of documentation got better with every SDK release.

    3. Some of us still remember the days when radios and TV sets, even computers, came with a big manual with complete schematic and part list. Decades of fucked up copyright laws and consumerism gone out of control brought what we have today: black boxes we are forbidden to look into, let alone attempt to repair. The worst aspect is most people are so brainwashed to consider this normal.

      1. Preach…
        Seriously. There should be a market niche for if not user servicable, repairable/upgradeable devices. Go into your local store bung it down and you pay to get it serviced and parts upgraded then bing there you go. Good for another two years.

        Instead nope. sealed boxes you can’t open or even look at too hard without offending some corporate douchbag. Christ I know we want things cheap, but i”m tired of things that break in two years. I want longevity in my devices. I mean look at the apple II, the game boy, and so on. There are plenty of examples twenty or even thirty years later still running (albeit with replaced buttons or a patched this or that.)

      1. Thing is, this thing is supposed to be a “step up” from the ESP8266. It is meant to be more expensive. So, to me it doesn’t sound unrealistic if they stay more expensive than the ESP8266 modules.

          1. @qwerty
            IMO the best feature of this board is using STM32, it’s looks like Node MCU with ARM – and these cores are more popular, better documented than Tensilica used in ESP, have mature toolchains and plenty of libraries… Weird that they didn’t get founded on that Kickstarter, is paying $5 extra per dev board to get ARM instead of some weirdo really such a bad deal?

          1. Before the EMW3162, there was the MT7681, same or slightly better spec as ESP8266, same starting price point ($6), but not enough volumes to drive price down (and no AIThinkers to take care of that market). There are a lot of very cool ARM-based cheap SoCs from Mediatek also that somehow did not make it big in the hobbyist market. I think larger firms like MediaTek are not interested in the hobbyist market, too costly to run a low-volumes/low-margin business for them. Espressif is small enough to find an interest and they have a go-to-market route with AIThinkers. Internal/domestic chinese market is very active too when you look at the chinese forums and sites.

  2. I want to know more on its Bluetooth LE capabilities, as to serve as an iBeacon or just as range meter to devices allowing proximity triggers on IoT is something many people here waits.

    AlsoI’d like to see 5ghz version of both ESP32 and ESP8266, in wifi congested buildings a 5Ghs link means freedom for IoT deployment.

    1. Yup. The ESP8266, ready to roll on an AIThinker board for $3 or less, is astounding value. They’re more powerful than many wifi-less microcontrollers that sell for more.

      The ESP32 will be a big brother in terms of capability. At anything under $10 for a module with antenna, I’ll be all over it as well.

      Espressif assured me I was one of the lucky 200 betatesters… but I haven’t recieved a ESP32 yet. Maybe it will be waiting for me when I get home. Please, Santa???

  3. This thing is a complete fail until the Chinese begin selling copies at <$5 free shipped. Also, if you watch the video, she says the regulator is getting warm. Not a good sign for low power designs. A next gen ESP8266 should be made more efficient at the same or lower cost, not the opposite. An "advanced" power hungry wifi/iot chip with some extra IO lines isn't really that useful. Also watching this video is painful. The girl, who supposedly is an MIT grad, starts by not even lining up the pins and jumps right into soldering. 0 F's given. It works, but wow, really not professional. And forget about the linux/software battle. Just skip to the end.

    1. Eh, this is a recording of a live event, I’m inclined to give her some slack. Although I also cringed when she soldered the first 2 pins without aligning the thing… altgough I heard it’s harder to do things like that when you have a microscope/camera in the way.

      The ESP31 as it is now is mostly showing off the functionality and power of the product; a fair few of the low-power modules either aren’t in the SDK or are not available in this demo firmware, so I’m not surprised the regulator generates some heat.

      1. yup, it’s much more difficult to align pins when you’re doing it at arms length and at an angle so you cant look straight down for alignment. there’s a slight delay so you can’t use the monitor for visual feedback.

        since you’re spritemods and have a cool site /resource – post up a video of you soldering a 0.5mm QFN under a HDMI/USB microscope and we’ll post that up as an example for folks to check out!

  4. So programming the esp8266 started with the lua, but then a standard toolchain was released…

    What will the development environment look like for the esp32? Will there be (closed) binary blobs for the tcp/ip stack, free-rtos?

    I’ve been looking for low-cost network chip for transmitting/streaming audio.. any ideas if this will work for that type of application? Or is it best to implement a tcp/ip stack and just have the mac/phy networking done off-chip (ex. enc28j60)?

Leave a Reply

Please be kind and respectful to help make the comments section excellent. (Comment Policy)

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