New Part Day: The ESP8266 Killer

Around this time last year we first heard of the ESP8266 WiFi module. It’s still a great little module, providing WiFi connectivity for all those Internet of Things things at a price point of just $5. It’s an attractive price for a great module with a huge community pumping out a lot of projects for the platform.

Now there’s a new kid on the block. It’s called the EMW3165, and like the ESP it provides WiFi connectivity for a bunch of wireless projects. It’s much, much more capable with an STM32F4 ARM Coretex M4 microcontroller, a ‘self hosted’ networking library, more RAM, more Flash, and more GPIOs. How much, you’re probably asking yourself. It’s a dollar more than the ESP8266.

The datasheet for the module goes over all the gritty details. While this chip has 3.6V I/Os, there are some 5V tolerant pins – a boon for the Arduino crowd. It’s also surprisingly low power for something that connects to an 802.11n network. The real bonus here is the STM32F4 core – that’s a very, very powerful microcontroller, and if you want a 2-component WiFi webcam build, this is the part you should use. There will be a lot of interesting builds using this part. It’s also passed FCC certification. Very cool.

183 thoughts on “New Part Day: The ESP8266 Killer

        1. to be fair, the folks behind ESP8266 have provided decent English documentation for the chip, since it was released: and there is a strong community at as well. As an electronics engineer, and probably the folks at Hackaday as well (where i first read about this chip), I got excited about ESP8266, at a time when it was barely documented (in Chinese or English), because it was obviously a game changer; ESP8266 has great specifications and such a simple BOM that for first time, you think about integrating this chip directly into your design, without using a module or a degree in RF black magic, and have WiFi functionality with a 160MHz CPU, ADC, low power options, great form factor, and great WiFi range for probably less than $2. i can’t see how EMW3165 comes close in any of the measures. In the book of electronics, David always slays Goliath, not the other way round.

        1. Exactly. Even if the FCC cert is fake it does not matter. Because this is a sub-assembly, you would need to have your design certified anyway (if mass produced for sale), so the FCC cert doesn’t really matter.
          Although using a pre-certified sub-assembly would speed the certification process.

          1. No, that’s completely FALSE where an FCC-certified module is concerned. But it is pretty much true for CE certification (there are a few exceptions), though that will change soon as the EU snuggles up to the FCC rules.

            An FCC-certified module may be used in a system WITHOUT FURTHER CERTIFICATION if it is used according to the limitations and guidelines listed in the module certification. That’s a simple fact.

            The “real” problem is lazy certifications done by device or module makers. Certification testing is not done *by* the FCC: The FCC only approves or shoots down whatever is submitted. The FCC mandates that the testing must be done by a certified lab, but does not specify the maximum testing to be done, just the minimum.

            I’ve had to reject attractive modules because their tests were designed by idiots (which the FCC approved with joy because they’re the easiest to check), and could only be reproduced using laboratory equipment, and never as part of a high-volume system.

            A good certification test will define the complete operating envelope within which the module will meet and maintain FCC certification. What range of VCC? How much noise on VCC? Floating ground OK? Antenna load limitations? SWR limitations? It takes a whopping big test to cover all this terrain, and such tests are EXPENSIVE.

            Some (most?) module vendors are simply too lazy or too cheap to do a “real world” certification. Which then puts the burden on the module user (the system designer), since there is no way to economically manufacture a system where each and every individual box will match the lab conditions of the original (cheap) module certification testing. So someone dead-set on using a module with a lousy cert must then do a full cert of their own proving the final system will meet FCC requirements over the full range of manufacturing variations.

            So, if you see a module that has a 10-page FCC certification, just walk on by, no matter how good the price looks. A “decent” FCC module certification package will be at least 30 pages long, and often 50-100.

            Now, one neat trick I’ve seen system vendors do (and have done myself) is stick a module inside another module, then do their own module cert that covers their own system manufacturing tolerances, which then lets them use that module in a wider range of their own systems (and keeps the module from their competitors). This may be the only reason for “short” module certs to exist: Such modules aren’t really intended to be used on their own.

            The next time you open up a router, check to see if there’s a PCB between the radio module and the main board: That’s a sure sign that a “modularized module” was used.

    1. There are a lot of “ESP8266 killers” out there already – for example the CC3200, with arguably better support, better documentation, more professional distribution channels etc. But people keep ranting and raving about the ESP8266 in the hacker community just because of price, and that won’t change as long as the price of alternatives is higher. No matter how crap it is, if your price is the lowest you win a lot of attention, and even though the ESP8266 is quite usable and good for many things, many people just don’t put a price on their time and effort and don’t understand “you get what you pay for” usually applies.

      1. If the CC3200 is like the CC3000, it is simply a big fiasco, full of bugs, such hangs after few hours of TCP stress tests. And TI never followed up on those bugs because they don’t know how to fix them since their design was outsourced in India. No, Thanks, TI doesn’t have better support …

        1. Yeah i don’t know what is going on with TI recently but i have wasted so much time with some of their components like TMP006 i.e. and though I am doing something wrong? but when i bought THEIR test board i realised it was total crap! numbers all over the place, unusable…
          Bought another chip from another manufacturer and it just….works (like a charm)!
          Gee TI had me scre@#d too many times, i wonder how did they had guts to release such a peace of crap?

      2. CC3200 it’s barely used by Energia IDE, has almost un-existant support in arduino community, almost un-existant shields or usable board for a microcontroller enthusiast.
        I have an energia based board with this chip and I get many disconnections, also note that it’s power it’s much less.

        Texas support it’s not the best around, I mean the disastrous experience with CC3100 that has several hardware bugs and texas never helped to solve.
        It’s also costly.
        The ESP8266 it’s another story, it can works as standalone module, it cost nothing and has a strong user support.
        There’s no comparison!

      3. CC3200? Come on! It’s a total fiasco!
        The CC3000 series it’s simply a disaster! About the Texas support? Not exist!
        Look at what they do with CC3100 (one of the biggest fail I ever seen around, hardware bugs and Texas refuse to admit until was too obvious).
        And the price? Bah….

          1. The real issue is volume: “ePacket” delivery means they open a sea or air container, and don’t close it until it is full of mail, after which the entire container is addressed to a US Post Office Distribution Center and put on a ship (big/heavy packages) or aircraft (small/light packages).

            The air/sea containerized traffic between China and the US dwarfs that between any other two countries, and the economies of repetition and scale apply. Big time. Other nations may be paying to send partially-filled containers, instead of waiting weeks or months for them to fill completely.

            I really don’t know how the US can afford their end of the deal unless the US imposes draconian restrictions on the content of the containers (such as pre-sorting and bundling, like it does for national junk mail rates). I suspect it may also have something to do with international postal regulations, about which I haven’t a clue.

          1. arduino on the esp8266 is just lousy, I could see if it just wrote to the user half of flash, as in write in arduino compile to lua and upload to esp8266. but as it is its just junk.

        1. No good alternatives. For *years*, the arm-gcc toolchain was just left as an exercise to the reader. Now, someone, I think ARM themselves, maintains current builds on launchpad.

          Atmel Studio is a giant download and is fairly oddly designed internally. Every chip vendor has some reskin of Eclipse, and most are pretty hamfisted. Energy Micro, Freescale have their own. Then there are the third parties, Embedded Artists and others. And then there’s IAR with their own whole deal.

          I think SEGGER is doing their own no-bullshit barebones IDE, but it’s Windows only. Then there’s CoIDE, but I dunno how well that is maintained.

          Even then, you have to deal with Makefiles, which are terrible. Alternatives of Gradle, Maven, SCONS, Ant, etc are all so much worse, in terms of clarity and accessibility.

          And then if you want a particular vendors implementation on the CMSIS, you have to slog through their site and find it, get it set up, etc.

          This is why Microchip persists, despite having fairly old MCU cores. Their MPLAB X IDE works very well out-of-the-box, based on NetBeans which is pretty clean and straightforward, and is a free and easy download.

          Arudino IDE is horrible, but it works. But it’s horrible. But it just barely works.

          At some point, someone made an Arduino makefile, and some command lino ino-build tools. I don’t think either of them ended up getting maintained at all, though.

          1. I actually prefer a GCC + makefile setup to using IAR/Keil or other IDEs. I can use the ATOM IDE (cause I’m a hipster wannabe) and even nemiver (Linux) as a gdb frontend. I have a lot of respect for Eclipse but I do not enjoy using it.

            I used to be big on the Microchip product line and they definitely get points in my book for the MPLABX IDE. But the facts are that ARM micros are much more popular, better designed and exhibit lower power. I’m also not a big fan of intentionally crippling the free version of their gnu (fsf based) mips-gcc compiler (for the pic32 parts) to make money. That along with ridiculously buggy hardware has kept me away.

          2. The toolchain hasn’t really been the issue for a while now at least, as long as you’re OK picking between a dozen different options… The real problem is the lack of any kind of consistency in the header files, linker scripts, and startup file (crt.o). Even from the same vendor you can find multiple competing and incompatible sets of files out there for a single chip. This shouldn’t be hard, but I don’t think ARM per se cares that much because what really matters to them are the vendor’s IDEs.

            I’d love to be able to take the time to gather and refactor all of these things (for all the vendors’ chips), and arrange them so building a baseline project (e.g. with sane defaults for RAM/flash setup) is no more complicated than building for an AVR: arm-gcc -mmcu stm32f415 main.c. Time, however, is the issue.

    1. I’m hoping more for a breakout board. Maybe something akin to the ram slots you get in laptops, leading to some nice friendly headers or even some easier to solder pads for a through hole connection

  1. Jsut the facts, please. No ‘killer’ drama.
    Although it has ‘better’ specs than the ESP8266, it does not look good enough to run Linux. For your suggested wifi webcam, you can probably do that easier with a $8 wifi router stick running OpenWRT.

    1. The fact that it doesn’t run Linux is an advantage. It doesn’t take 20 minutes to boot, doesn’t require you to be a system administrator of all your devices, upgrading them each time they find a new SSL hole, and generally, “just works”.

      1. Ahh, more drama…
        Embedded linux boot times like OpenWRT are on the order of seconds, not minutes.
        I agree Linux is overkill for some things, and in those cases I’ll use the ESP8266. I paid 246c a few months ago, and as others have pointed out here, they are as cheap as 225c now.

    2. there are a lot off better versions of stuff that don’t run linux. i bought a very good vacuum cleaner,better than most, I am very satisfied, and… surprise: it doesnt run linux. and it has all of TWO buttons, instead of a ssh interface, even though i could probably patch in an arduino with a network shield to interface two buttons to the ssh, or even dick around with kernel patches to support external buttons….

      neither does it run windows, which means i dont get to dick around with licensing/pirating issues, or drivers.

      guess what? it does the job it was designed for, it sucks the dirt, all without running linux.

      if this chip can talk to other chips and connect to wifi, screw linux. linux is just one way to get stuff done, and one shouldnt care HOW the job gets done, provided IT DOES. besides, a wifi camera is only one suggested project idea, and i have a brain full of others, who would probably be ill suited to run on a linux platform.

      I like linux a lot for various reasons, and i hate linux a lot for others. my feelings are equaly distributed for any operating system i tried out.

      but please dont bring the argument that it sucks for not using whatever construct you might be more familiar with, ok? generally.. forget any kind of golden standards, and admit that you are a lazy fuck that doesnt want to learn something new. it is ok, we are all lazy fucks who dont want to reinvent the wheel. it should be socially acceptable.

  2. I’m not sure about the “2-component WiFi webcam”. This uses a STM32F411, which does not have the DCMI camera interface. It seems like the F411 could be replaced with another F4 with more I/O (like those with DCMI) and then you _would_ have a very capable webcam system.

    1. I was wondering about that. I don’t know much about cameras or DCMI, but it seems like meeting the time constraints for bit-banged camera interfacing would be difficult (impossible?), especially while simultaneously running a WiFi stack on the same MCU. Though one of the real gurus like [Sprite_tm] might have a legitimate shot at it. 128KB RAM also tends to restrict it to cameras with built-in JPG.

  3. Forget price.
    This thing has a fully opensource stack! Much more IP features (IPv6 for example)
    And FreeRTOS support.

    This is a easy winner. No more crazy esp binary blobs or missing documentation.

      1. Definitely the first thing I noted, digging through the spec-sheet- there’s really no information on what wifi module is used, or what it’s programming interface is (besides that it’s connected to the ST via SDIO): big red flags that this could be anything.

      2. I’m not sure that is true. But I haven’t really worked enough with that stack yet.
        I see some references to .bin and .a files. But I’m not sure if they are used.

        At least I’m pretty sure the IP stack is in source. If some low level WiFi stuff is a blob, well. I could live with that.

          1. Ha! Can’t say I’m surprised, though. No one else should be either.

            WiFi chipsets typically have no built-in hardware checks that ensure you’ve set the TX frequency within an acceptable range. “Acceptable” meaning one allowed by the FCC (or your particular regulatory agency), rather than the limits of the hardware. With that range possibly differing by region, or being changed over time, enforcing or limiting that in hardware could lead to multiple hardware revisions. And that’s expensive.

            Instead, limits are enforced in software. At very least, the part of the software that handles frequency setting is in a proprietary BLOB, so that the check cannot easily be bypassed. And the details of the low-level hardware registers protected by NDA, so that one cannot easily write a replacement for the BLOB.

            So you want a cheap, programmable transceiver, that’s either FCC pre-certified, or can be built into a certifiable product? The price of that is a BLOB.

  4. So basically a knockoff Sparkcore / Photon with a different STM32 micro.

    That pricing sounds bogus. There’s no way this can be anywhere near price competitive with the ESP8266. The STM32F411RE is $4.60 qty 2.5k at Digikey. Even if they can halve the price with volume and negotiation it’s as much as a complete wholesale ESP8266 module before factoring in the wireless chip, the rest of the BOM, and mfg / test / distribution.

      1. Technically speaking, there is no reason why this should be more expensive than esp8266. The specs sucks. It doesn’t have more ram or higher cpu speed nor better power consumption. 260mA on receive? That’s totally crazy. Compare with 80mA on esp8266 and my esp8266 module works all the way down to 2.2V. The technology behind esp8266 is light years more advanced than emw3166.
        There is very little documentation put out on emw3166, compared to esp8266. The thing about these big companies is they do not give a sheet about the community because they don’t make money on it. they released this “killer” through an unknown module house like Mxchip. I bet they will discontinue this module a year from now, because the specs are just not competitive enough. Esp8266 has certainly raised the bar.

    1. “Until we can program it with Arduino IDE and it has a DIP package”

      What is this, 1990? I guess you’ll be expecting 5V logic levels too? The world has moved on, and Arduino should be moving on too.

      1. Most of its pins are 5V tolerant.

        Arduino IDE sucks. Downloading the Arduino IDE and having a complete compiler chain without having to do 37 hours of pulling the pieces together and compiling and dealing with versionitis is something I can live with.

        Getting an LED blinking and serial port talking and timers going with babytalk code; then swapping it out for something without so much useless abstraction when I want it faster: that’s also survivable.

        1. Have been anal about RF all my life. And so take a look at this 100 page FCC report for Espressif module. I wonder how did they squeez all the RF circuits into yhis tiny chip. Their specs are at least 6-10dB better than the EMW3165 module. This is what it takes to build a real product. I will definitely choose Espressif modules over the Broadcom EMW3165.

        1. The newest one, directly from Espressif, is genuinely FCC certified. The pricing is similar (actually a bit lower IIRC). It’s called the “WROOM” module. If you’ve worked in-depth with the ESP8266 modules, it also has quite a few useful pins that many of the configurations leave out because, well, I guess whoever pinned the old modules out was in a rush and didn’t bother figuring out which pins were actually useful (TOUT is a good example – it’s the one ADC pin on the ESP8266, and most of the variants don’t have it…).

          Of course, I also trust that the antenna and front end are matched better than the generic “chinese hobbyist” (usually by “AI” or whatever) ESP8266 modules, given that it’s actually FCC certified.

          With that said, having a FCC certified module doesn’t automatically preclude you from /all/ FCC testing…

          1. Cheapest is USD $4.40 / module @ Aliexpress. Espressif is selling it directly on their Taobao site at 20 Yuan (USD $3.22) / module. I think that’s really only for domestic (China) shipments, so add the bulk shipping rate and additional discounts to get an estimate for volume pricing (if you care about FCC certification, then the only pricing you should care about is volume pricing, I’d imagine)

          2. What’s better than just FCC certification, Espressif is now shipping their modules with Japanese TELEC certification. I’m seeing few local distributors for about $6-7, So living in Japan I can design with $3 modules from eBay, and chances are if I ever make something into Kickstarter, I would only have to quickly check my product against regulations for unintentional radiators, slap the $6 WROOM-02 onto it, and ship right away. Local regulations here don’t require the whole product to be certified as a radio device if the radio module is separate and has valid certification.

            What revolutionary product to actually design in my backyard is an entirely different difficulty though.

          1. Wrong. You need your design to pass verification but can reuse the certification. Quite different and cheaper. But you can get certification from a China lab for $1500 these days anyway.

          2. The need to pass verification require time and money. The certification only applies to the component that was certified by the FCC Just because the sub assemblies used in the design are FCC certified, does not mean your completed design is. There are no FCC test firm accrediting bodies or telecommunications certifications bodies in China but there is OCA, BSMI and NCC. If they give their certification, a validation is still required under Section 2.902 47 CFR. If a foreign manufacturer wants FCC certification, they can send samples to the FCC for certification.

          3. Any Chinese lab certification would require verification from the FCC. As Mark said, certification of a sub-component does not give certification to the entire design. The certification on the sub-component means nothing if the complete design does not pass verification.

          4. That’s the whole point of modular certification, that the finished device can simply reuse the fccid of the module. If you’re not adding more clock sources you probably will pass verification pretty easily. There are plenty of 2.948 listed labs in China and elsewhere in Asia that can perform certification tests.

          5. The entities mentioned in Section 2.948 47 CFR outside the US cannot give FCC certification. They are “telecommunications certifications bodies” or “test firm accrediting bodies”. If they are outside the US they can only give their own certification (OCA, BSMI, NCC or other foreign cert). That certification then requires further verification from the FCC, which simplifies obtaining FCC certification. As Mark mentioned above, if a foreign manufacturer wants FCC certification, they can send samples to the FCC for certification.
            See the following to learn more about the process-

          6. Right so all you have to do is pay a filing fee to a tcb after you get the test report from the lab. I don’t think anyone sends simple things like WiFi modules to the FCC for testing these days. Look through the FCC database there are tons of foreign companies with test reports from foreign labs.

    1. If it is metal, then it would probably block the signal.

      No reason why you can’t swap it out for a non-metal one though. Get a big enough one that you don’t have to have the module actually inside the junction box.

    1. Because you think that hackaday has an advertising contract with MXCHIP? I love to hear about new parts like these, IoT is a growing space and anything that is better than the ESP (and close in price) is very interesting in my book.

      1. Agreed, it’s an interesting looking part and this HaD post is the first I’ve seen of it.

        Maybe buZz needs to go find some other site that better matches his ‘bro’ needs.

      2. do some research on the actual part, i read many people have more problems with it, then a esp for example.
        and it also seems it has got some hardware bugs ? but not sure about that.

    2. Why don’t you just immediately skip any articles titled “New Part Day” then, rather than complain? Because there’s no surprise, every one will be written similar to this… here’s a new part, here’s what it does. It’s not an ad, I highly doubt anyone’s paid for it, there was no bias in comparing it’s specs. And it’s something almost everyone in the community is excited to hear. It’s one of my favorite regular features.

      All the contest BS is pretty advertorial, in the same sense that TV networks advertise their own shitty shows in the comercial break for their own shows, but, that is what it is. Same as you, I can easily identify them and just skip those articles.

      1. nah, let him nap. It will easily take anohter year until this device is down in cost, availability, ease of use, sample code, etc. pp. The ESP has a strong cummunity with excelent Examples in Code and Hardware (ESPToy for example). It will take several Month until we even see this new component integrated into the culture… and I’m not yet convinced that it will get the same love as the ESP. So, yeah, let him nap some more.

  5. I’m amazed at how much time & effort ppl are investing on custom projects that involve tens (if not 100s) of hours of soldering, programming, tinkering and building, using equipment costing 100’s (of not 1000’s) og dollars, and eagerly spending a _lot_ of extra time on chips with custom architectures and bad/non existent documentation.

    Just to save a few dollars in BOM.

          1. Well, since I know I’m not saving any money (quite to the contrary, time is also money, quite a lot of money), I tend to concentrate more on the learning stuff part :P

          2. Are you sure you’re not just looking for an excuse for all those hours wasted on cheap crappy HW?

            Besides, there are plenty of other problems to solve, no need to go chasing low BOM on a hobby projects where the real expenses are time investment, equipment and plain failure to make anything work at all.

  6. Here is the FCC application:

    So, it is FCC indeed, unlike the cheap ESP8266 modules with FCC engraved on the shield but without ID.

    However, it is only 13.5 dBm output power vs. 19.5 dBm for the ESP8266, so forget about controlling a quad with it, you won’t go past 50~70 m. No serious figures regarding power consumption yet.

    But there is no way a module with 2 chips (including an STM32F4), 2 crystals and an SPI Flash will be cheaper than the single ESP8266 chip with a single crystal and SPI Flash. Given that the STM23F4 is already ~$4-$5 in medium qty, this module won’t go below $5 soon, whereas you can get ESP8266 at half this price today. Of course, not FCC-compliant…

    1. I believe the ESP-12 module is FCC certified and has an ID, it’s just not printed on its can/shield. The reason is that FCC ID’s are usually printed or engraved on the external enclosure of a part that an end-user would see. Since there is almost no chance in most designs that end-users would see the ESP-12, there’s no need to mark it. Instead, the regulations say you should print its FCC ID on the outside of the product you make.

  7. The part about “hooking it to an Arduino” is what caught in my throat.
    An STMF4 is a much more powerful processor than what is found in an Arduino.
    It is like dropping a Lambourghini V12 engine into a VW Golf,
    make that 2 V12’s…

  8. This is basically a slightly more powerful version of the Spark photon for less than half the price. I like the fact that it uses an STM32F411…I’m already familiar with that part and reference manuals/software frameworks (STM32Cube) already exist for that part. Two concerns that I have are:
    – I see no guides/tutorials/ proper documentation on using the MICO/MXCHIP framework

    – The README also says that it supports the IAR Workbench IDE. Providing GCC/Gnu Make support is critical to the openness and accessibility of this part.

    Please correct me if I’m wrong as I’d very much like to be wrong about these points.

    1. mbed is a good starting platform, and I’ve also had good experience with LPCExpresso from Embedded Artist. Good selection of boards and solid developemtn platform.

    2. The mbed NXP LPC1768 is good (as mentioned above) (3.3V IO), or Teensy 3.1 (5V tolerant IO).
      If you are adventurous and need a ton of IO you could try the Arduino Due ($18 out of China and 3.3V IO).

  9. Personally I’m very skeptical of these built-in modules with soc and radio all built-in. I know this is old school but I very much prefer using an AT command based WIFI to serial module that takes care of the WIFI stuff along with my favourite microcontroller. This way I don’t have to learn how to use a gazillion different frameworks/SDKs for every flavour of the month WIFI+ Micro SOC.

    The ‘HiFlying LPT-100 / USR-IOT USR-WIFI232T’ modules are a great example. They are small, low power, have relatively decent documentation, can be configured via web interface or AT command sets, cost $10, and can be added to any micro of your choice.

    The HLK-RM04 is another good choice if power consumption is not a concern

    1. A $9 hell yeah why not use the CHIP for WiFi heavy processing some non-real time IO stuff. You can also use many programming languages; python, node, go e.t.c not just c/c++. You also have access to tonnes of pre-made apps and libraries. The Wt toolkit for example is a great framework for building IOT apps.

      If you need real-time add a cheap cortex-m0 like the stm32f0 which still comes with plenty of peripherals, and IO’s in many packages and form factors.

      I’ve recently become less skeptical of the $9 CHIP due to free-electrons announcing that they’ll be involved in the project. I have a lot of respect for the guys at Free Electrons

  10. I cant see how this is going to exite the community the way the esp8266 did.
    There was no documentation or support so it was a challenge taken up by the community.

    Now we have a very cheap solution that can be used in different ways by different people.
    And now the community is driving the manufacturers to get more out there to fill demand.
    You cant plan for that sort of take up. But they wished they could.
    Which is why this is being released.
    It wont replace the esp8266.
    We have to wait for the next piece of obscure hardware to come along and capture the communities attention.

    1. my problem with the ESP (I have a handful of them and have had very mixed success) is that its great for hacking, for low-cost and for doing one-offs or POCs. but would you really design something that needs to have consistent and trustable hardware that is buyable for years? do you think this module or its api/interfaces will be stable for long enough to justify designing things that use the ESP board?

      I feel uneasy betting on it. it came, another will too. it will then be old and busted and something else will be teh new hotness ;)

      we should demand that things we support are going to be around for long enough to justify the effort to learn them and design with/for them.

      and so, I’ll continue to play around with modules like these, but I will not commit to designing a longer term product that depends on these kinds of things. if the vendor can commit to long term stability and support, that will get me seriously interested.

      1. Can any IC manufacturer guarantee they will produce a particular part forever? They only produce parts as long as the demand is there, and any mass produced product you use that part in will only be in production while there is demand for it.

    1. I have some doubts about Atmel and STM in the area of WiFi compared to companies like Marvell, Qualcomm and Espressif. If you have read their specs you would understand.

    you could see FCC number on the module.
    why always say esp8266 not pass fcc
    it means the module with esp8266 could gain fcc cert
    then if the buyer care it, just choose the factory follow the game rule

    1. I personally use an ST-LINK/V2 (ISOL) mainly because it has that isolation (working with industrial machines and different power nets here). The non isolated version is not bad either, but a little bit expensive for what it is.
      Definately the best, cheapest and easyest way to get some form of USB->JTAG/SWD for Cortexies is to just buy one of the STM32 Discovery or Nucleo EvalBoards. I do really like the 429I-DISCOVERY for it’s small graphic touch display and quite large 64MBit RAM, but if you only want the ST-LINK portion of the board, there are of course cheaper boards, i think they start at around 6-7$…an they all have the same ST-LINK as the stand-alone thingie.
      Of course, if you need deeper debugging (ETM, CodeCoverage,…) there is almost no way around spending ~1K$ on the debugger. (There are some places where you can rent an ULINK PRO on a monthly basis too. We just had such a case where we had rented 10 PRO debuggers running in parallel to increase the chances to find a nasty spurious bug.)

  12. I think we can celebrate a new part without shitting on something else, no?

    Last fall I ordered a Spark Core & Photon, and a handful of ESP8266s. Initially the ESPs seemed like a joke, but I persevered, and evaluated different frameworks, and started getting good results with custom firmware. I’m not laughing now, I’m planning a short run of something using ESP8266s.

    To me, yes we will each have our favourite parts, but wifi iot connectivity is going to be a commodity soon; and most parts will have more or less the same range of capabilities, so given that all ‘surviving’ wifi modules are reliable and certified, the choice will come down to availability, price… and ease of development. Having a big base of working examples is a serious advantage.

  13. I think a lot of people are forgetting something very important when it comes to the price, while yes you can the cheaper ESP01-03 for about $3 off ebay, however if you’re like me and would rather have any of the slightly better warrants for flexablity, then ones such as the ESP-12 cost about the same as buying this new EMW3165 so I know which one I’d rather have if I’m paying just as much for it… now you could argue about it not having much support/community but if everyone’s response to a new product is “I like the old one better and it’s slightly cheaper” then of course its not going to have a community, $8 seems completely fair for what you’re getting.

    1. Esp-12 can be bought for less than $2.50 from aliexpress. To be fair, we should wait and see what price the mx module becomes available for after a while. The esp modules weren’t always this cheap at the start.

      Having both options is great. For really low cost stuff that doesn’t need a heap of io I’ll use the esp8266. When I need more io (spi/hardware i2c/adc/dac) I’ll spend more on the mx. Adding these extra features to the esp8266 via external chips adds a lot of $ to bom cost.

        1. The EMW3165 FCC results are terrible. I think this module is not suited for mass production. What a shame that it’s using a Broadcom part. This EMW3165 part is managing only a measly 13dBm, whereas ESP8266 does more than 20dBm. This has gotta be the worst killer ASSassin i have ever seen, cus it’s freaking dead on arrival. Kaputt!
          I bet that this MXCHIP module house is a run by a bunch of pure marketers without any idea what hardware or engineering is. Too bad, Broadcom is also going the way of Intel when under Otellini, except that Intel is alive today. Broadcom is now being sold to Avago, sub-con their product marketing to China, and wants to claim to be a killer with this big ASS module, while startups like Espressif run circles around them. What a joke.

        1. Yes ! I’ve already saved this datasheet. I’m looking more about software, tools, libraries, tcp/udp stack, etc.
          It sure that since it is based on STM32F411, there is GCC, and even STM32Duino, but are the libs and stacks compatible ? Is there any things one the Broadcom chip side relevant to know ?

  14. Since somebody mentioned, let’s keep this discussion pure technical. This 3165 thing can connect to external antenna, so you can wire your antenna to anywhere. But the signal needs serious testing around the spinning fan. So I guess, the best solution, which is irrelevant to any wifi module, is to put the wifi module in the wall switch and use some tricks, D2A, or pulse generator to relay the control.

  15. ESP8266 killer or not, I’m going to give this module a try. I’ve been following the comments on this page for a while and it appears that no one actually has a module to play with just yet. I am sure a lot of people just ordered them from Seeed after this post went live, so hopefully in a couple weeks there will be a lot of hacking taking place.

    I went ahead and set up a forum at to hopefully help nourish community development on this module once it makes it way to the hands of hackers. For now I’m just collecting links to useful resources, but once my modules arrive I’ll be posting my successes and failures. Anyone and everyone welcome to join in :-)

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.