Another step forward on low cost sophisticated WiFi MCU boards?

Status
Not open for further replies.

stevech

Well-known member
This post is in hopes that discussion might influence a future product from PJRC - and shows the trend to interpreted scripting languages to do much more per line of code than C/C++.
This post is not intended at all to divert attention from today's Teensy 3.x - but rather, to discuss the future.
Info here may help those microprocessor fans who are interested in staying abreast of what's happening in the marketplace for makers (avocational uses) and for low volume products...
We're talking $12-$30 @1ea, and not the Windows IoT or Arduino Zero and kin.

And using languages that may be familiar to those working in servers and PCs, but not experienced with MCU/firmware and near-bare-metal.
Most or all of the interpreters, and the I/O libraries and "kernel" for these types of boards is written in C or C++.

  1. We've seen the low cost MCU board with a built-in Python interpreter + I/O library (STM32 MCU). Micropython.
  2. And the Particle.io Photon MCU board with Javascript and an on-line IDE that is very slick for many "maker" type of users. Lots of I/O libraries, nice event-driven system. STM32F MCU.
  3. Newly discovered is a lower cost yet MCU board. STM32F MCU. https://www.kickstarter.com/projects/857552561/wifi-mcu-stm32f4-wifibroadcom-lua
The third one listed above is in kickstarter status it says, for shipments in Oct. 2015 (next month as this is written). if they reach the funding goal. From ShenZhen.
Here's the kickstarter: https://www.kickstarter.com/projects/857552561/wifi-mcu-stm32f4-wifibroadcom-lua
And this: http://www.wifimcu.com/en.html

Confusing, but the vendor linked to below is taking orders at just a bit more cost than the two-fer price on the kickstarter. This seller says they've been a webstore for years.
http://www.smartarduino.com/wifimcu-wireless-wifi-development-board-using-lua-from_p94744.html says 300+ in stock. How can that be, given the kickstarter?

Developers go by the moniker "DOIT; see http://doit.am/ (they begain with an ESP8266 - now moved to Broadcom as did Particle.io)

Well whatever/whenever, this price point board is amazing, if true. Too low to be realistic? Doesn't read like the suspect sellers of boards built from known-reject chips (or worse).
  • Broadcom WiFi module
  • STM32F411EC MCU (512K flash, 128K RAM, floating point, lots of on-chip peripherals). 2MB SPI flash chip on PCB.
  • USB
  • SWD flash and debug option
  • Standard firmware has a script interpreter for the Lua language. (OMG, does my brain have room for Lua as well as Python, Javascript, C, C++ ?)
  • Says full open source. Bootloader and all system software, I/O libraries, and the interpreter - sources on github.

Some time browsing these web sites... "WiFiMCU" seems to have comprehensible API documentation. And the source is open.

Well, October should show what $17 for two brings, on the kickstarteraward. The smartarduino.com price is about 10% higher (supposedly).
 
Last edited:
Did you notice their website is also selling a counterfeit Teensy++ 2.0 at $35.69. They've also got lots of counterfeit Arduino boards, labeled as Made In Italy and Arduino.cc... which is kind of a contradiction, given Arduino's recent legal turmoil.

Who knows, maybe they will ship a top-quality Python experience, but I'd be amazed if they do anything more than send you to Damien's website for all software and tech support.

FWIW, I do have MicroPython and Espruino Javascript on my radar for next year, after we've got a Teensy++ 3.x board that's powerful enough to run those languages well. I don't know exactly how it's all going to end up, but I can tell you 2 things. #1: When I'm directly involved, I'll be porting libraries and Teensyduino's drivers, not settling for scripting but you have to bring all the tough low-level coding. #2: PJRC isn't going to match or even come close to the prices of these Chinese counterfeiters.
 
Indeed - if there were a factory made Teensy with WiFi and scripting, that could be big. Lua footprint: "Lua interpreter built with all standard Lua libraries takes 242K and the Lua library takes 414K."

3. Newly discovered is a lower cost yet MCU board

The third one listed above is in kickstarter status it says, for shipments in Oct. 2015 (next month as this is written).

Confusing, . . .

Confusing indeed - almost like it already exists to ship that fast - that distributor "with stock" better sign up and buy some more or it will not kick start out very well. It is not 'discovered' enough yet - they only want $3,000 and have 10 days left ... and are sitting at $1,337 pledged.

Hoping the ESP8266 has life in it - I'm about a month out from getting my KStart Oak/Acorn units. They have agreed/signed onto Particle.io as a supported platform and BLYNK too. I'm happy the T_3.2 released with spare power to run the $7 ACORN unit. It is a somehow customized ESP8266 (4MB flash shipping) board and the OAK is that soldered to a daughter board with breadboard pins and USB/LDO regulated power.
 
Did you notice their website is also selling a counterfeit Teensy++ 2.0 at $35.69. They've also got lots of counterfeit Arduino boards, labeled as Made In Italy and Arduino.cc... which is kind of a contradiction, given Arduino's recent legal turmoil.

Who knows, maybe they will ship a top-quality Python experience, but I'd be amazed if they do anything more than send you to Damien's website for all software and tech support.

FWIW, I do have MicroPython and Espruino Javascript on my radar for next year, after we've got a Teensy++ 3.x board that's powerful enough to run those languages well. I don't know exactly how it's all going to end up, but I can tell you 2 things. #1: When I'm directly involved, I'll be porting libraries and Teensyduino's drivers, not settling for scripting but you have to bring all the tough low-level coding. #2: PJRC isn't going to match or even come close to the prices of these Chinese counterfeiters.

You're referring to the distributor "smartarduino" ? So we can put that seller on the ignore list.

Setting them aside, the Kickstarter is the main interest.
This board has Lua for scripting, rather than Javascript or Python.
Teensy-next to support one of these... needs lots of RAM - more like 128KB or greater- that's a main complaint of users. The CCM RAM in most ARMs is problematic in how to ensure that an app cannot mistakenly try to use this named region for DMA.

Indeed, these products do all the speed critical I/O and some compute intensive things in the C code that implements the scripting language's API.
 
Last edited:
Google and I found lots..

Th EMW3165 seems to an OEM module that is a big step up from the ESP8266. It seems to retail at $8 or less vs. the $3-4 for the ESP8266. MXChip seems to be the module's designer. The module's RF design is from Broadcom and the MCU is an STM32F411 with 512KB flash, 128KB RAM, floating point, etc. Broadcom provides under NDA the reference design, software for the 802.11 MAC, and an IDE called WICED.
This module, as I read, runs the 802.15.4 layer 2.5 MAC as they call it. And it, like Photon and others, do time sharing of the CPU with user app code in a manner that lets the MAC get all the CPU time it needs to support 802.11b/g/n MAC processing. IP/TCP et al are on the module too.

Nice blurb on using an SWD programmer/debugger with the EMW3165 module.
http://www.seeedstudio.com/recipe/344-programming-emw3165-with-broadcom-wiced-and-gcc.html

That module brings in many packagers selling boards employing that module and, e.g., a buck/boost power supply, USB plug, pins for GPIO, et al. They likely adapt Broadcom's reference design and maybe Broadcom's 802.11 MAC and operating kernel. Then some pacakger-vendors get real FCC certs for use in the US regulatory region. Maybe not so much certified for the EU and the infamous 2.4GHz regulations of Japan and France - not sure.

So I think the kickstarter for Oct. is for one such board that employs the module. Not sure why there'd be a kickstarter if others have already gotten a board with the module to market.

But these boards are amazing... if one chooses a good board manufacturer. The modules will all be identical, I assume.
 
Last edited:
As you mentioned in the original post, the reason for the kickstarter is strange, since the modules seem to already exist.

The price _is_ good. They are slightly one-upping the specs for the Particle Photon (Cortex M4F vs Cortex M3). The Broadcom WiFi they are using seems to be the same BCM43362 as the Photon. And $8 beats the $12 cost of the Photon (comparing to the Photon P1 module with the built-in antenna). Although we are talking about small quantities here. I'm sure the prices would be a lot different if you were to purchase enough.

On the other hand, with a company like Particle you are getting a lot more documentation, and software support. And more help from other users.

I have a little bit of experience with MXChip. I bought two of the EMW3162 modules (the predecessor to the EMW3165) and one of the dev boards back in December.
It took months to get more than the barest documentation for the EMW3162. The EMW3165 seems to be slightly better supported, but not a lot.

And the ESP8266 has a new rev 'E' out now, that supposedly fixes all of the stability issues we've seen in the past. If those issues are really fixed, perhaps that is the way to go. I have a few of the rev E's, but haven't had the chance to stress them much.
 
Indeed, prudent. But what do you think of the Kickstarter for this and the people behind it? (irrespective of that one distributor)?

Well, I'm having a hard time separating the "WifiMCU" Kickstarter and "SmartArduino" in my mind. Why? Well, after clicking just a few links, I ended up here:

https://github.com/SmartArduino/WiFiMCU

The github user publishing all their software is named "SmartArduino". That tiny fact makes this whole thing pretty tough to evaluate irrespective of that one (apparently disreputable) distributor.

At first I thought "wow, 106 commits already", so they must be actively developing software, right? Well, if you look at the commit log, nearly all those commits are merely editing the readme file or incrementing the version number!

The first major commit did bring in a *lot* of source code. Nearly all of it appears to be the files distributed by MXCHIP, who makes that module. Their "IDE" appears to be a terminal emulator talking to the Lua intepreter. That alone doesn't necessarily mean they're not actively trying to build value on top of other open source code. But the lack of software platform development details on Kickstarter, lack of any entity other than a disreputable Chinese clone vendor, and the only visible development activity on Github appearing to be merely the readme file which served as the material for the Kickstarter page all add up to a picture that doesn't inspire much confidence.

The extremely low prices also aren't a good sign.... in terms of a revenue stream to fund substantial software development.

But I could be wrong. I've been wrong about plenty of things before. Maybe this thing will eventually turn into a viable, well maintained platform. Who knows, maybe they'll someday displace Arduino?! But that seems like quite a long shot, given all the not-so-positive signs.

They do have a forum set up, it seems. Take a look and see what you think?

http://bbs.doit.am/forum.php?mod=forumdisplay&fid=52

At this point, my main thought is I don't wish to spend even one more minute looking at this. I have far better things to do... like implementing MTP protocol.....

But if you feel it's worthwhile, or might become worth your time, go ahead and back the Kickstarter. Worst case, you'll lose $20 and never get anything. A more likely case is a $20 unsupported paperweight, which you'd probably still be able to use if you go to the trouble to get the support package from MXCHIP and run with the normal STM toolchain. Well, actually, the most likely scenario seems like they won't reach their meager $3k funding goal.... but if you throw in $20, that'll at least help.
 
Last edited:
My experience with tech stuff from kickstarter is there is only one item that I've bought that was ever near their expected delivery date, namely the Teensy 3.0. I used to think the Teensy was the only kickstarter project that I had ever invested in that actually shipped on time, but Paul S. has said that he was a few days late. The delay time for the others ranges from a month to almost a year.
 
Indeed, we had trouble with the USB connector "floating" and becoming mis-aligned during soldering on that first Teensy 3.0 batch. Many boards required rework to straighten & resolder the connector. What was supposed to be plenty of boards to ship all the September rewards turned out to be only a trickle, and a lot of time spent inspecting boards and working with the contract manufacturer.

Most rewards still shipped on time, but I believe we shipped a couple hundred of the ones promised for September on October 1st. Many others shipped in the last few days of September, rather than the middle of the month as I'd originally planned, so they arrived in the first week of October if in the USA. While still technically within the September promise if we shipped before Oct 1, it fell short of my goal.

All the rewards promised for October did ship within the first couple weeks. Then it took us 3 months of painful work to ship the last hundred rewards, which had imcomplete or wrong info or missing payment for international shipping. Whenever Robin and I talk of maybe using Kickstarter again, we remember how painful and how long that dragged on, after we were "done". If we ever use Kickstarter again (a pretty big if), we'll give people maybe 2 weeks after rewards ship to submit their info, before we just refund and cancel.
 
When backing (buying) a dev board which attempts to launch a new platform on Kickstarter, I believe the larger risk is whether the author actually invests in continued development after the rewards ship.

For example, while we had Teensy 3.0 on Kickstarter, two others launched. Both raised more money than Teensy 3.0, probably because I'm too conservative in promising features.

https://www.kickstarter.com/projects/kuy/galago-make-things-better

https://www.kickstarter.com/projects/fairduino/smartduino-open-system-by-former-arduinos-manufact

I believe it's safe to say both of these were abandoned after rewards finally shipped. Ultimately very few, if any, backers actually ended up with a useful product.
 
Last edited:
And the ESP8266 has a new rev 'E' out now, that supposedly fixes all of the stability issues we've seen in the past. If those issues are really fixed, perhaps that is the way to go. I have a few of the rev E's, but haven't had the chance to stress them much.
I think the EMW3165 uses SPI rather than a UART. To the end user, other than (often unnecessary) speed, the API hides this. Price sells. But for hobby use, I'd rather pay the cost of a Coke on the Champs-Élysées more and get quality and documentation to reduce frustration.
 
Aty this point, my main thought is I don't wish to spend even one more minute looking at this. I have far better things to do... like implementing MTP protocol.....
What with Photon (Javascript), MicroPython, and this MXChip board (Lua), all being a different concept that Arduino's C/C++, it's good for us to know the Maker-world trends. We're looking outside the box.

Presumably these board makers' plan is to enlarge their potential market by using languages that are simpler than C++ and like Teensyduino, come with good reusable code modules for a more coherent system.
 
Last edited:
I see after another 36 hours the p#1 KStart is not yet half of the $3k funded with 80 backers and 9 days left.

They refer/point to http://doit.am/ (Doctors of Intelligence & Technology) and that page top has a SHOP banner link that goes right to the clone seller smartDuino - just like the GitHub link Paul followed.

I clicked Paul's p#12 KStart link to Galago his one year after update starts:: "year filled with challenges I never would have expected before setting out to complete this project. In fact, had I known about even some of those problems, I likely wouldn’t have started!"

I wish I had seen Teensy sooner! Glad I didn't buy into too many other things I'm not ever going too use - or that aren't usable.

It took a year(?) for ESP8266 to get where it is - and that is piggybacking on established Arduino IDE. If this EMW3165 does as well with it's added horsepower it will take similar infrastructure to become usable - that won't come from an opportunistic clone seller. As pointed out - maybe the maker world will turn to it - but it is a different beast. As different as Raspberry Pi. Dropping an interpreter onto it will suck the (deterministic) power out of it like dropping a GUI on your OS - or putting Linux on your rPi. And while making it easier with a nice language interface - it could also obscure the power Teensy preserves and multiplies from the 8 bit AVR world - and except for making IOT specific NEST type control systems - that hardly need an FPU - the bare metal will be far away.

I'm still excited to get my ESP8266 units - to use as a $7 coprocessor WiFi board to my Teensy. Maybe the EMW3165 can do much of that alone - and it does have impressive radio range spec that would fit my long driveways better. But I'm not looking to be a pioneer - 'face down in the dirt with arrows in my back'.
 
Presumably these board makers' plan is to ..... come with good reusable code modules for a more coherent system.

Damien George is doing this on MicroPython, Gordon Williams is doing this on Espruino, and Bogdan Marinescu & others are actively working on eLua.

But there's little indication, at least I can see, that Jimmy Wu, Ping Yang, Liuji Shang, Honggang Li & Chenfu Yi of WiFiMCU & SmartArduino intend to put any significant effort into any software development. There is plenty of anecdotal evidence, especially the extremely low price and github commit log lacking any code development, which suggests their business model is merely shipping a breakout board which leverages existing software.
 
I clicked Paul's p#12 KStart link to Galago his one year after update starts:: "year filled with challenges I never would have expected before setting out to complete this project. In fact, had I known about even some of those problems, I likely wouldn’t have started!"

While I rarely talk of these little problems, I can tell you those sorts of things happen all the time. Just yesterday and today we're dealing with some unknown change with one of the chips which appears to be affecting soldering and our assembly vendor's automated inspection. It's unclear (so far) if first pass or final yield will be affected.... in fact, much about it is still unclear today. That's how these things go. They suck up lots of time behind-the-scenes, in highly urgent troubleshooting.

Last week was a problem with those little 320x240 displays failing. That one turned out to be some non-conductive film on the header pins (perhaps a flux chemical used when they soldered?) which cleans up pretty easily with isopropyl alcohol. But before that was discovered, it appeared we had 3 large batches with high fallout, which casts a lot of doubt on the ones which did pass testing... of course, while orders are pending to ship to customers.

Over the years I've seen many, many people try to enter this market. Some stay, but most give up. Manufacturing quality products *and* consistently improving software requires a lot of dedication and discipline. It's totally awesome on a creative level, but work-wise it's really tough. Just cloning existing products and reselling stuff is far easier and more profitable.
 
I'm not celebrating the possible demise of this unit (except to the degree they are or associate with IP thieves) - but this doesn't look good for the project.

What's to prevent someone from taking an existing, shippable open source project and adding a bit of low cost "value added" such as a breakout board or some simple gizmo, and putting it up as a Kickstarter?

The core of the one URL'd above is suspect in this regard?

People on eBay buy from eBay sellers then resell the same item at a higher price. Somehow they avoid double shipping, i.e., they get the original seller to drop ship after the resale (?).
 
Last edited:
Status
Not open for further replies.
Back
Top