Re: new option for a very low cost WiFi client for IoT

[an007_rld]

Hi,
ESP8266 SDK reach V1.0.0 (Espressif willing to pay $ 200 to anyone who finds errors in SDK 1.0.0).
Check here (use google translate): [url=http://esp8266.ru/esp82

[Old_but_Alive]

looks interesting.

a link to your info would be appreciated

[LzEn]

I am always scared with chips that do not have a lot of documentations. Although the one you proposed seems very good but if you want something cheap for WiFi/BLE  and greatly documented with and open SDK for development, you might want to check Broadcom's WICED platforms.
The SDK incorporates many example codes, schematics, datasheets, technical documents, Free IDE ...

The WICED SDK release is structured as follows:
  Apps          : Example & Test Applications
  Doc           : API & Reference Documentation, Eval Board & Module Schematics
  Drivers       : Drivers for WICED evaluation boards
  Include       : WICED API, constants, and defaults
  Library       : Daemons, servers, protocols and peripheral libraries
  Resources     : Resources used by the WICED webserver including scripts, styles, images and HTML.
  Tools         : Build tools, compilers, debugger, programming tools etc.
  Wiced         : WICED core components (RTOS, Network Stack, Wi-Fi Driver, Security & Platform definitions)
  Wiced/WWD     : The WICED Wi-Fi Driver (equivalent to the Wiced directory in previous SDK-1.x releases)
  README.txt    : This file
  CHANGELOG.txt : A log of changes for each SDK revision

I have attached the full readme text for more information.

The chip itself is BCM43362, and you can find many modules based on it. Murata, USI, Inventeksys.

Hardware Platforms
 * BCM9WCD1EVAL1  : Bare WCD1 WICED evaluation board (generic module)
 * BCM943362WCD2  : Broadcom 43362-based WICED Module with STM32F103 MCU mounted on BCM9WCD1EVAL1
 * BCM943362WCD4  : Broadcom 43362-based WICED Module with STM32F205 MCU mounted on BCM9WCD1EVAL1
 * BCM943362WCD6  : Broadcom 43362-based WICED Module with STM32F415 MCU mounted on BCM9WCD1EVAL1
 * BCM943362WCD8  : Broadcom 43362-based WICED Module with ATSAM4S16B MCU mounted on BCM9WCD1EVAL1
 * BCM9WCDUSI09   : Broadcom 43362-based WICED Module with STM32F205 MCU (includes WM-N-BM-09 WICED SiP) mounted on BCM9WCD1EVAL1
 * BCMUSI11       : USI 43362-based WICED+ Module (STM32F205 MCU, 8Mbit serial flash) mounted on BCM9WCD1MFI1
 * BCM9WCDPLUS114 : WICED+ Eval Board (includes BCM43362+STM32F205 WICED+ Module and BCM20702 Bluetooth module)
 * TWRK60D100M    : Freescale Kinetis K60 connected to a USI WM-N-BM-09 WICED SiP via SPI

The cheapest one I have found is on this link, I didn't buy it because the price sounds too good to be true. But if anyone tries it, let us know. I contacted USI myself and they gave me a price < 10$ for their module.
The chip itself is very cheap but it comes in a BGA package.

[an007_rld]

Please see the attached file.

What I find new is the module (with print antenna) can be buy in China for 15.20 yuan ($2.44).
see here: http://detail.tmall.com/item.htm?spm=a230r.1.14.9.6y0llT&id=40602285796&ad_id=&am_id=&cm_id=140105335569ed55e27b&pm_id=

At this point this is only one solution that ensure WiFi connectivity at this price...

There are other solution under $10 that can run OpenWRT (I can give some suggestions) but definitely a WiFi controlled lamp or switch will not need that...

Hope helps,
-an

[vfd]

There appears to be quite a lot of interest around the ESP8266 . There is some effort to get GCC working at http://www.esp8266.com/viewtopic.php?f=9&t=19&sid=2bcffde786cfebbd3a1cfa422e7ffdbb and if successful, the part can be the controller also instead of just a serially accessible module. There are some boards that breakout all the pins too such as http://www.aliexpress.com/snapshot/6226433326.html , I have some of both the normal modules and the full breakouts on order to see what I can achieve.

[bigtoy]

Wow, that's pretty cheap.

I've recently done a little proof-of-concept project using a Texas Instruments CC3200. The thing works well. It contains a Cortex-M4 which you get to program, plus a second processor which you don't get to program - that second processor contains the WiFi radio, protocol stacks, etc. So it's a one-chip (kinda) solution for WiFi. I say "kinda" because it requires a serial flash chip. $10 for a module, less than that for just the chip. So not expensive, but the ESP chip is obviously cheaper (although also not in the same league I suspect).

[motox]

I ordered a couple of ESP8266 modules two weeks ago from Ali.
The price is decreasing week after week. I paid 3.65€ per unit and they were more expensive on the week before. Now they are at 3.55 € single module, or 3.07 €/per module for a lot of two.

[puta]

Found a quick youtube video;

https://www.youtube.com/watch?v=cdBC79jvHkw

[towlerg]

For anyone not able/willing to use Ali, these modules are on eBay for £3.98 free post.

[motox]

Instructions of how to configure server/client TCP sockets with ESP8266:
https://drive.google.com/folderview?id=0BwK3EhAfht8uWTdBdG55NEFCakE&usp=sharing

[bobcat1]

Hi All

From my point of view it is problematic to develop a product complicated as WIFI ,without support from the manufacture
It is only good to hobbies or people who have time to play with this type of IC.
I will not touch it until I have full support from manufacture even if the cost go down to less then a 1 $
The 3100 & 3200 IC from TI cost more but have full support 

All the best

Bobi

[MisterX]

There is also an english translation of the datasheet here : https://nurdspace.nl/ESP8266

[an007_rld]

Well Bobi,

It seems the space become very crowded and at some point TI need to align the price... It will be a direct competition on that space and the estimated market is huge.
Anyway, there is another entry on the market: MT7681 from MediaTek.

Here is their 3'rd June announcement: http://www.mediatek.com/en/news-events/mediatek-news/mediatek-launches-socs-aimed-at-powering-smart-home-appliances/

A few low cost modules developed based on this chip (below you can find 2 of them); again the price is under TI offer:
http://www.alibaba.com/product-detail/802-11b-g-n-2-4GHz_60002469468.html and http://www.alibaba.com/product-detail/Hot-Selling-GWF-KM22-MT7681-Serial_1947601622.html

I will update the post as soon as I have more informations; connected home is a host topic today...

Regarding SDK and documentation, maybe this can be solved if you contact directly the company and probably sign an NDA with them...

Hope helps,
Regards,
-an

[biko4710]

In my option if I want to use a chip for IoT it should be ready with some kind of scripting language.
I don't want to implement another microcontroller to control the module.

wakeup every hour and do:
- get time via ntp
- get temperature via i2c
- store data on ftp
- goto sleep

An implementation of openWRT would be fine but it's too much.

[bobcat1]

Hi All

Please remember - if you plan a product with this kind of IC and plan to sell it in the US or Europe You will need to certify it (FCC) and the process cost lot of money
Other wise you can sell it!!!!.

Bobi

[Gallymimu]

any "module" sold without FCC compliance isn't really that useful as anything more than a toy.  Most major regions require some type of intentional radiator compliance.  Of course if you never get caught.....

[an007_rld]

As long as the chip was intend for IoT, obviously the product manufacturer will have to pass FCC as so as UL, CE, etc.. The chip manufacturer have in mind this as long as they target a high volume sells.

-an

[MisterX]

There is now a new website entirely dedicated to the ESP8266  http://www.esp8266.com/

[gan_canny]

A heads up on these 8266 chips.
They come in two versions and to make things a bit more of a challenge the boards are unmarked. Both versions are 3.3v.
The older version has a 57600 baud UART and may reply "ready" to some commands. The CH_P/D pin on the chip must be pulled high.
The older model may be doing the pull up on the pcb.
The newer model has a 115200 baud UART and CH_P/D needs an external connection to pull it up to 3.3v.
The newer ESP8266 repeats the command (EX AT note commands always end in CR LF) if it receives it correctly and appends OK CR LF
When CH_P/D is configured correctly the pcb will have a solid red LED and the second blue led will flash on briefly as the pcb powers up or just after the CH_P/D pin is brought high. The UART is a TTL like interface (only at 3.3v ) The ESP8266 needs some time (1 sec) after power up to prepare itself for use.
//////////////   ESP8266 WiFi interface ////////////////////////
///              baud 115200 8 bit N  1 stop ( standard RS232 at3.3v)
///          component side
//            +-----------------------+
//            |                       |
//            |       antenna         |
///           |                       |
///           | 8266         eeprom   |
///           |                       |
///           +-----------------------+
//            | Gnd GPIO02 GPIO00 Rx  |
//            | Tx  CH_P/D RST   3.3v |
//            +-----------------------+

GPIO pins and RST maybe left unconnected...There is a software RST command AT+RST as alternative to the  hardware RST pin.

[gan_canny]

An update
I have this working with the ESP8266 as client , but it required a fairly complicated FSM in the 3.3v UART interrupt service routine for a PIC 24 chip
The issue is first it is asynchronous and second there is no simple way to know when a response from the ESP8266 is finally complete.
The ESP8266 uses CR+LF as a terminator  a typical response is to repeat the command plus CR+LF then the response CR+LF
however some responses  EX. AT+STATUS continue with additional data then a final CR+LF.
Some commands EX AT+RST have many responses even sometimes  "ready" other times "OK" and even a dump of check sums
This approach makes for a simple PC terminal program interface but having a PIC act as client required a FSM in the UART isr and a timer interrupt
to close out responses that didn't complete within a time limit.
The ESP8266 got connected easily to my wifi access point and captured the IP address granted by the AP
The PC server side interplays well with the ESP8266 as they obtain each others MAC address via ARP then there is the simple 3 way TCP handshake
SYN SYN+ACK ACK this also works well and the ESP8266 easily links up. Data moves to the server via AT+SEND and the server can send asynchronously data to
the ESP8266. In a strict client server this asynchronous data transfer can be managed since the client is master and should not get any unsolicited responses from the server.
I'm no fan of the AT+command approach the NEMA sentence approach Ex GPS sentences with their start and end tokens and crc's is IMHO superior.
The TCP packets are often going to have small payloads like CAM packets since they may be controlling several leds or relays maybe a few as several bytes which is the approach I took.
Some will want a web page interface with POST GET to transfer data to from the ESP8266 but that's even more data stream parsing code for whatever is on the other side of the ESP8266.

[bigtoy]

As an aside, did anyone read the ALiExpress listing the OP posted. Some of what's on that page is hilarious. For example:

    In the absence of more cattle X chip out now is the most cattle X.

    Yue Xin advantages:
        5 ground gas, Xin Yue contacted each employee, are very enthusiastic and gas.

Better wear a gas mask when visiting their office!  :-)

[gan_canny]

I.m unsure as to why Chinese when mechanically translated into English often comes out goofy. In English very few words are tonal ...read is an example... same spelling but can be pronounced in two ways depending on context. It maybe that many words in Chinese are tonal and context is required to get a good translation. Anyway the ESP8266 is effective and by trial and error its responses can be chronicled. The sure thing is that a microprocessor doesn't understand Chinese or English  it just gets  a string of ascii characters... like the ESP8266 will spit out "no fun" at times which the microprocessor sees as an error since it wasn't the correct response like "OK"... a reset and a repeat of the command gets the "OK". I wish they had used a SPI interface for the PCB's the ESP8266 chip has SPI capabilities .....but the UART works even on unsolicited responses since they are delimited by +IPD,n:ddddddd where n is the number of chars after ":" ( represented here as d's )

Posted on: October 05, 2014, 08:25:50 20:25 - Automerged
I suspect there is a bug with the ESP8266 . With command AT+SEND=n,CR LF the ESP8266 response is '>' then a string of data is sent
when I trace my code I see the string written to the ESP8266 as having the exact chars to be sent....it appears the ESP8266 prepends LF to the string
and chops off the last char.....this is what appears in the TCP packet...the work around is to increase the length to be sent by one and to strip off the LF at the server. This isn't proof positive but the data going into the ESP8266 looks correct...it may be my code but it would have to be that the ESP8266 has a LF
sitting in its buffer that gets prepended and sent with the correct data. Debugging as with all asynchronous interfaces is a challenge.
Anyway unless this is confirmed by others the probability is that I have a loose LF somewhere.
Solution
The documentation always shows <0x0d><0x0a> as a delimiter however with AT+SEND=    it only needs one of these if two (<0x0d> or <0x0a> with both the extra
<0x0d> or <0x0a gets prepended to the data to be sent.

[Old_but_Alive]

I am still awaiting my modules to arrive from China, so I cant test directly.

I suggest you check the excellent site http://www.esp8266.com/

for any similar problems.

It sounds like a too simple bug to be a chip problem though, so its probably your code

[adelgado]

Take a look at this site, it includes examples of use for this module.

http://rayshobby.net/?p=9734

Best Regards.

[Cain]

Anyone know where to get the latest firmware?

I have updated to version 0.9.2.2 (is it the latest???) but experience several wierd things with some AT commands not working correctly.... For example when server is in MUX=0 mode it still allows several clients to connect when only one should be allowed and eventually module hangs with "busy..." message...