Update docs and split up the main readme (#200)

pull/213/head
Thorsten von Eicken 8 years ago committed by GitHub
parent 9515ea538f
commit 64fb2323c3
  1. 112
      BUILDING.md
  2. 192
      FLASHING.md
  3. 1
      Makefile
  4. 545
      README.adoc
  5. 163
      README.md
  6. 14
      RESTMQTT.md
  7. 36
      TROUBLESHOOTING.md
  8. 177
      UC-FLASHING.md
  9. 87
      WIFI-CONFIG.md

@ -0,0 +1,112 @@
Building esp-link
=================
Before you build esp-link, consider that you can download ready-made firmware images!
Just head over to the [release section](https://github.com/jeelabs/esp-link/releases)
and download the tgz archive.
If you decide to build your own, there are a number of options:
- On linux x86 download the ready-built toolchain and patched SDK like the automated build does
and compile the firmware
- On linux use a docker image with the toolchain and the SDK to compile the firmware
- On linux download and build the toolchain, download and patch the SDK, then compile the firmware
- On windows use a docker image with the toolchain and the SDK to compile the firmware
- On windows install mingw, python, java, and a slew of other tools and then build the
firmware
Once you have built the firmware you will want to flash it to your esp8266 module.
Assuming you already have esp-link running you can either go back to the initial flashing
via the serial port or you can use the over-the-air (i.e. Wifi) update method, which is faster
and more reliable (unless you have a non-booting version of esp-link).
The OTA flashing is described at the end of this page,
the serial flashing is described in [FLASHING.md](FLASHING.md).
### Automated builds
For every commit on github an automated build is made. This means that every branch, including
master, and every pull request always has an up-to-date build. These builds are made by Travis
using the instructions in `.travis.yml`, which basically consist of cloning the esp-link repo,
downloading the compiler toolchain, downloading the Espressif SDK, and running `make`.
If you're looking for how to build esp-link the travis instructions will always give you
accurate pointers to what to download.
### Docker (linux or windows)
The [esp-link docker image](https://hub.docker.com/r/jeelabs/esp-link/) contains all the
tools to build esp-link as well as the appropriate Espressif SDK. *It does not contain the
esp-link source code!*. You use the docker image just to build the firmware, you don't have
to do your editing in there. The steps are:
- clone the esp-link github repo
- checkout the branch or tag you want (for example the tag `v2.2.3` for that release)
- cd into the esp-link top directory
- run `make` in docker while mounting your esp-link directory into the container:
- linux: `docker run -v $PWD:/esp-link jeelabs/esp-link:latest`
- windows: `docker run -v c:\somepath\esp-link:/esp-link jeelabs/esp-link:latest`,
where `somepath` is the path to where you cloned esp-link, you probably end up with
something like `-v c:\Users\tve\source\esp-link:/esp-link`
- if you are not building esp-link `master` then read the release notes to see which version of
the Espressif SDK you need and use that as tag for the container image, such as
`jeelabs/esp-link:SDK2.0.0.p1`; you can see the list of available SDKs on
[dockerhub](https://hub.docker.com/r/jeelabs/esp-link/tags/)o
Sample steps to build esp-link v2.2.3 on a Win7 Pro x64 (these use the docker terminal, there
are multiple way to skin the proverbial cat...):
1) Install Docker Toolbox ( http://www.docker.com/products/docker-toolbox )
2) Install Git Desktop ( https://desktop.github.com/ )
3) Clone esp-link from Github master to local repository ( https://github.com/jeelabs/esp-link )
4) Open Docker Quickstart Terminal
5) cd to local esp-link git repository ( C:\Users\xxxxx\Documents\GitHub\esp-link )
6) Run "docker run -v $PWD:/esp-link jeelabs/esp-link" command in Docker Quickstart Terminal window
Note: there has been one report of messed-up timestamps on windows, the symptom is that `make`
complains about file modification times being in the future. This may be due to the different
way Windows and Linux handle time zones and daylight savings time. PLease report if you
encounter this or know a solution.
### Linux
The firmware has been built using the https://github.com/pfalcon/esp-open-sdk[esp-open-sdk]
on a Linux system. Create an esp8266 directory, install the esp-open-sdk into a sub-directory
using the *non-standalone* install (i.e., there should not be an sdk directory in the esp-open-sdk
dir when done installing, *if you use the standalone install you will get compilation errors*
with std types, such as `uint32_t`).
Download the Espressif "NONOS" SDK (use the version mentioned in the release notes) from their
http://bbs.espressif.com/viewforum.php?f=5[download forum] and also expand it into a
sub-directory. Often there are patches to apply, in that case you need to download the patches
from the same source and apply them.
You can simplify your life (and avoid the hour-long build time for esp-open-sdk) if you are
on an x86 box by downloading the packaged and built esp-open-sdk and the fully patches SDKfrom the
links used in the `.travis.yaml`.
Clone the esp-link repository into a third sub-directory and check out the tag you would like,
such as `git checkout v2.2.3`.
This way the relative paths in the Makefile will work.
If you choose a different directory structure look at the top of the Makefile for the
appropriate environment variables to define.
Do not use the source tarballs from the release page on github,
these will give you trouble compiling because the Makefile uses git to determine the esp-link
version being built.
In order to OTA-update the esp8266 you should `export ESP_HOSTNAME=...` with the hostname or
IP address of your module.
Now, build the code: `make` in the top-level of esp-link. If you want to se the commands being
issued, use `VERBOSE=1 make`.
A few notes from others (I can't fully verify these):
- You may need to install `zlib1g-dev` and `python-serial`
- Make sure you have the correct version of the SDK
- Make sure the paths at the beginning of the makefile are correct
- Make sure `esp-open-sdk/xtensa-lx106-elf/bin` is in the PATH set in the Makefile
### Windows
Please consider installing docker and using the docker image to save yourself grief getting all
the tools installed and working.
If you do want to compile "natively" on Windows it certainly is possible.
It is possible to build esp-link on Windows, but it requires a
[gaggle of software to be installed](WINDOWS.md)

@ -0,0 +1,192 @@
Flashing esp-link
=================
### Hardware configuration
This firmware is designed for any esp8266 module.
The recommended connections for an esp-01 module are:
- URXD: connect to TX of microcontroller
- UTXD: connect to RX of microcontroller
- GPIO0: connect to RESET of microcontroller
- GPIO2: optionally connect green LED to 3.3V (indicates wifi status)
The recommended connections for an esp-12 module are:
- URXD: connect to TX of microcontroller
- UTXD: connect to RX of microcontroller
- GPIO12: connect to RESET of microcontroller
- GPIO13: connect to ISP of LPC/ARM microcontroller or to GPIO0 of esp8266 being programmed
(not used with Arduino/AVR)
- GPIO0: optionally connect green "conn" LED to 3.3V (indicates wifi status)
- GPIO2: optionally connect yellow "ser" LED to 3.3V (indicates serial activity)
If your application has problems with the boot message that is output at ~74600 baud by the ROM
at boot time you can connect an esp-12 module as follows and choose the "swap_uart" pin assignment
in the esp-link web interface:
- GPIO13: connect to TX of microcontroller
- GPIO15: connect to RX of microcontroller
- GPIO1/UTXD: connect to RESET of microcontroller
- GPIO3/URXD: connect to ISP of LPC/ARM microcontroller or to GPIO0 of esp8266 being programmed
(not used with Arduino/AVR)
- GPIO0: optionally connect green "conn" LED to 3.3V (indicates wifi status)
- GPIO2: optionally connect yellow "ser" LED to 3.3V (indicates serial activity)
If you are using an FTDI connector, GPIO12 goes to DTR and GPIO13 goes to CTS (or vice-versa, I've
seen both used, sigh).
The GPIO pin assignments can be changed dynamically in the web UI and are saved in flash.
### Initial serial flashing
Download the latest [release](https://github.com/jeelabs/esp-link/releases) or use the
`user1.bin` and `user2.bin` files that are produced by the build process.
You will need to flash the bootloader, the firmware, blank wifi settings, and init data.
Detailed instructions are provided in the release notes.
_Important_: the firmware adapts to the size of the flash chip using information
stored in the boot sector (address 0). This is the standard way that the esp8266 SDK detects
the flash size. What this means is that you need to set this properly when you flash the bootloader.
If you use esptool.py you can do it using the -ff and -fs options. See the end of this page for
instructions on installing esptool.py.
The short version for the serial flashing is:
- flash `boot_v1.X.bin` from the official SDK or from the release tgz to `0x00000`
- flash `blank.bin` from the official SDK or from the tgz to `0x3FE000`
- flash `esp_init_data_default.bin` from the official SDK or from the tgz to `0x3FC000`
- flash `user1.bin` to `0x01000`
- be sure to use the commandline flags when flashing the bootloader to set the correct flash size
- some of the addresses vary with flash chip size
After the initial flashing if you want to update the firmware it is recommended to use the
over-the-air update described further down. If you want to update serially you only need to
reflash `user1.bin`.
### 32Mbit / 4Mbyte module
On Linux using esptool.py this turns into the following for a 32mbit=4MByte flash chip,
such as an esp-12 module typically has (_substitute the appropriate release number and bootloader
version number_):
```
curl -L https://github.com/jeelabs/esp-link/releases/download/v2.2.3/esp-link-v2.2.3.tgz | \
tar xzf -
cd esp-link-v2.2.3
esptool.py --port /dev/ttyUSB0 --baud 230400 write_flash -fs 32m -ff 80m \
0x00000 boot_v1.5.bin 0x1000 user1.bin \
0x3FC000 esp_init_data_default.bin 0x3FE000 blank.bin
```
I use a high baud rate as shown above because I'm impatient, but that's not required.
### 4Mbit / 512Kbyte module
```
curl -L https://github.com/jeelabs/esp-link/releases/download/v2.2.3/esp-link-v2.2.3.tgz | \
tar xzf -
cd esp-link-v2.2.3
esptool.py --port /dev/ttyUSB0 --baud 460800 write_flash -fs 4m -ff 40m \
0x00000 boot_v1.5.bin 0x1000 user1.bin \
0x7C000 esp_init_data_default.bin 0x7E000 blank.bin
```
The `-fs 4m -ff40m` options say 4Mbits and 40Mhz as opposed to 32Mbits at 80Mhz for the 4MByte
flash modules. Note the different address for esp_init_data_default.bin and blank.bin
(the SDK stores its wifi settings near the end of flash, so it changes with flash size).
For __8Mbit / 1MByte__ modules the addresses are 0xFC000 and 0xFE000.
## Updating the firmware over-the-air
This firmware supports over-the-air (OTA) flashing for modules with 1MByte or more flash,
so you do not have to deal with serial flashing again after the initial one!
The recommended way to flash is to use `make wiflash`
if you are also building the firmware and `./wiflash` if you are downloading firmware binaries.
The resulting commandlines are:
```
ESP_HOSTNAME=192.168.1.5 make wiflash
```
or assuming mDNS is working:
```
ESP_HOSTNAME=esp-link.local make wiflash
```
or using wiflash.sh:
```
./wiflash.sh <esp-hostname> user1.bin user2.bin
```
The flashing, restart, and re-associating with your wireless network takes about 15 seconds
and is fully automatic. The first 1MB of flash are divided into two 512KB partitions allowing for new
code to be uploaded into one partition while running from the other. This is the official
OTA upgrade method supported by the SDK, except that the firmware is POSTed to the module
using curl as opposed to having the module download it from a cloud server. On a module with
512KB flash there is only space for one partition and thus no way to do an OTA update.
If you need to clear the wifi settings you need to reflash the `blank.bin`
using the serial method.
The flash configuration and the OTA upgrade process is described in more detail
in [FLASH.md](FLASH.md).
## Installing esptool.py on Linux
On Linux use [esptool.py](https://github.com/themadinventor/esptool) to flash the esp8266.
If you're a little python challenged then the following install instructions might help:
- Install ez_setup with the following two commands (I believe this will do something
reasonable if you already have it):
wget https://bootstrap.pypa.io/ez_setup.py
python ez_setup.py
- Install esptool.py:
git clone https://github.com/themadinventor/esptool.git
cd esptool
python setup.py install
cd ..
esptool.py -h
## Installing esptool.py on Windows
Esptool is a pythin pgm that works just fine on windows. These instructions assume that git and
python are available from the commandline.
Start a command line, clone esptool, and run `python setup.py install` in esptool's
directory (this step needs to be done only once):
```
> git clone https://github.com/themadinventor/esptool.git
Cloning into 'esptool'...
remote: Counting objects: 268, done.
emote: Total 268 (delta 0), reused 0 (delta 0), pack-reused 268
Receiving objects: 100% (268/268), 99.66 KiB | 0 bytes/s, done.
Resolving deltas: 100% (142/142), done.
Checking connectivity... done.
> cd esptool
> python setup.py install
running install
...
...
...
Finished processing dependencies for esptool==0.1.0
```
Download and unzip the latest esp-link release package, and start a commandline
in that directory. The command to run is pretty much the same as for linux.
Adjust the path to esptool and the COM port if you don't have the ESP on COM12. 460800
baud worked just fine for me, writing at ~260kbit/s instead of ~80kbit/s.
```
>python "../esptool/esptool.py" --port COM12 --baud 115200 write_flash \
--flash_freq 80m --flash_mode qio --flash_size 32m \
0x0000 boot_v1.6.bin 0x1000 user1.bin \
0x3FC000 esp_init_data_default.bin 0x3FE000 blank.bin
Connecting...
Erasing flash...
Wrote 3072 bytes at 0x00000000 in 0.3 seconds (79.8 kbit/s)...
Erasing flash...
Wrote 438272 bytes at 0x00001000 in 43.4 seconds (80.7 kbit/s)...
Erasing flash...
Wrote 1024 bytes at 0x003fc000 in 0.1 seconds (83.6 kbit/s)...
Erasing flash...
Wrote 4096 bytes at 0x003fe000 in 0.4 seconds (83.4 kbit/s)...
Leaving...
```

@ -61,6 +61,7 @@ SDK_VERS ?= esp_iot_sdk_v2.0.0.p1
# Try to find the firmware manually extracted, e.g. after downloading from Espressif's BBS,
# http://bbs.espressif.com/viewforum.php?f=46
# USING THE SDK BUNDLED WITH ESP-OPEN-SDK WILL NOT WORK!!!
SDK_BASE ?= $(wildcard ../$(SDK_VERS))
# If the firmware isn't there, see whether it got downloaded as part of esp-open-sdk

@ -1,545 +0,0 @@
ESP-LINK: Wifi-Serial Bridge w/REST&MQTT
========================================
Thorsten von Eicken
:toc:
:toc-title!:
:toc-placement!:
This firmware connects an attached micro-controller to the internet using a ESP8266 Wifi module.
It implements a number of features:
[options="compact"]
- transparent bridge between Wifi and serial, useful for debugging or inputting into a uC
- flash-programming attached Arduino/AVR microcontrollers, esp8266 modules, as well as
LPC800-series and other ARM microcontrollers via Wifi
- built-in stk500v1 programmer for AVR uC's: program using HTTP upload of hex file
- outbound REST HTTP requests from the attached micro-controller to the internet
- MQTT client pub/sub from the attached micro-controller to the internet
The firmware includes a tiny HTTP server based on
http://www.esp8266.com/viewforum.php?f=34[esphttpd]
with a simple web interface, many thanks to Jeroen Domburg for making it available!
The REST and MQTT functionality are loosely based on https://github.com/tuanpmt/espduino
but significantly rewritten and no longer protocol compatible, thanks to tuanpmt for the
inspiration!
Many thanks to https://github.com/brunnels for contributions in particular around the espduino
functionality. Thank you also to https://github.com/susisstrolch, https://github.com/bc547,
and https://github.com/katast for additional contributions!
[float]
Table of Contents
-----------------
toc::[]
Releases & Downloads
--------------------
- https://github.com/jeelabs/esp-link/releases/tag/v2.2.3[V2.2.3] is the most recent release.
It has a built-in stk500v1 programmer (for AVRs), work on all modules, and supports mDNS,
sNTP, and syslog. It is built using the Espressif SDK 1.5.4.
- https://github.com/jeelabs/esp-link/releases/tag/v2.1.7[V2.1.7] is the previous release.
- See https://github.com/jeelabs/esp-link/releases[all releases].
For quick support and questions chat at
image:https://badges.gitter.im/Join%20Chat.svg[link="https://gitter.im/jeelabs/esp-link"]
Intro
-----
### Esp-link goals
The goal of the esp-link project is to create an advanced Wifi co-processor. Esp-link assumes that
there is a "main processor" (also referred to as "attached uController") and that esp-link's role
is to facilitate communication over Wifi. Where esp-link is a bit unusual is that it's not really
just a Wifi interface or a slave co-processor. In some sense it's the master, because the main
processor can be reset, controlled and reprogrammed through esp-link. The three main areas of
functionality in esp-link are:
- reprogramming and debugging the attached uC
- letting the attached uC make outbound communication and offloading the protocol processing
- forwarding inbound communication and offloading the protocol processing (this part is the
least developed)
The goal of the project is also to remain focused on the above mission. In particular, esp-link
is not a platform for stand-alone applications and it does not support connecting sensors or
actuators directly to it. A few users have taken esp-link as a starting point for doing these
things and that's great, but there's also value in keeping the mainline esp-link project
focused on a clear mission.
### Esp-link uses
The simplest use of esp-link is as a transparent serial to wifi bridge. You can flash an attached
uC over wifi and you can watch the uC's serial debug output by connecting to port 23 or looking
at the uC Console web page.
The next level is to use the outbound connectivity of esp-link in the uC code. For example, the
uC can use REST requests to services like thingspeak.com to send sensor values that then get
stored and plotted by the external service.
The uC can also use REST requests to retrieve simple configuration
information or push other forms of notifications. (MQTT functionality is forthcoming.)
An additional option is to add code to esp-link to customize it and put all the communication
code into esp-link and only keep simple sensor/actuator control in the attached uC. In this
mode the attached uC sends custom commands to esp-link with sensor/acturator info and
registers a set of callbacks with esp-link that control sensors/actuators. This way, custom
commands in esp-link can receive MQTT messages, make simple callbacks into the uC to get sensor
values or change actuators, and then respond back with MQTT. The way this is architected is that
the attached uC registers callbacks at start-up such that the code in the esp doesn't need to
know which exact sensors/actuators the attached uC has, it learns that through the initial
callback registration.
### Eye Candy
These screen shots show the Home page, the Wifi configuration page, the console for the
attached microcontroller, and the pin assignments card:
image:https://cloud.githubusercontent.com/assets/39480/8261425/6ca395a6-167f-11e5-8e92-77150371135a.png[width="45%"]
image:https://cloud.githubusercontent.com/assets/39480/8261427/6caf7326-167f-11e5-8085-bc8b20159b2b.png[width="45%"]
image:https://cloud.githubusercontent.com/assets/39480/8261426/6ca7f75e-167f-11e5-827d-9a1c582ad05d.png[width="45%"]
image:https://cloud.githubusercontent.com/assets/39480/8261658/11e6c64a-1681-11e5-82d0-ea5ec90a6ddb.png[width="45%"]
Getting Started
---------------
### Hardware configuration
This firmware is designed for any esp8266 module.
The recommended connections for an esp-01 module are:
- URXD: connect to TX of microcontroller
- UTXD: connect to RX of microcontroller
- GPIO0: connect to RESET of microcontroller
- GPIO2: optionally connect green LED to 3.3V (indicates wifi status)
The recommended connections for an esp-12 module are:
- URXD: connect to TX of microcontroller
- UTXD: connect to RX of microcontroller
- GPIO12: connect to RESET of microcontroller
- GPIO13: connect to ISP of LPC/ARM microcontroller or to GPIO0 of esp8266 being programmed
(not used with Arduino/AVR)
- GPIO0: optionally connect green "conn" LED to 3.3V (indicates wifi status)
- GPIO2: optionally connect yellow "ser" LED to 3.3V (indicates serial activity)
If your application has problems with the boot message that is output at ~74600 baud by the ROM
at boot time you can connect an esp-12 module as follows and choose the "swap_uart" pin assignment
in the esp-link web interface:
- GPIO13: connect to TX of microcontroller
- GPIO15: connect to RX of microcontroller
- GPIO1/UTXD: connect to RESET of microcontroller
- GPIO3/URXD: connect to ISP of LPC/ARM microcontroller or to GPIO0 of esp8266 being programmed
(not used with Arduino/AVR)
- GPIO0: optionally connect green "conn" LED to 3.3V (indicates wifi status)
- GPIO2: optionally connect yellow "ser" LED to 3.3V (indicates serial activity)
If you are using an FTDI connector, GPIO12 goes to DTR and GPIO13 goes to CTS (or vice-versa, I've
seen both used, sigh).
The GPIO pin assignments can be changed dynamically in the web UI and are saved in flash.
### Initial flashing
If you want to simply flash a pre-built firmware binary, you can download the latest
https://github.com/jeelabs/esp-link/releases[release] and use your favorite
ESP8266 flashing tool to flash the bootloader, the firmware, and blank settings.
Detailed instructions are provided in the release notes.
_Important_: the firmware adapts automatically to the size of the flash chip using information
stored in the boot sector (address 0). This is the standard way that the esp8266 SDK detects
the flash size. What this means is that you need to set this properly when you flash the bootloader.
If you use esptool.py you can do it using the -ff and -fs options.
### Wifi configuration overview
For proper operation the end state that esp-link needs to arrive at is to have it
join your pre-existing wifi network as a pure station.
However, in order to get there esp-link will start out as an access point and you'll have
to join its network to configure it. The short version is:
1. esp-link creates a wifi access point with an SSID of the form `ESP_012ABC` (some modules
use a different SSID form, such as `ai-thinker-012ABC`)
2. you join your laptop or phone to esp-link's network as a station and you configure
esp-link wifi with your network info by pointing your browser at `http://192.168.4.1/`
3. you set a hostname for esp-link on the "home" page, or leave the default ("esp-link")
4. esp-link starts to connect to your network while continuing to also be an access point
("AP+STA"), the esp-link may show up with a `${hostname}.local` hostname
(depends on your DHCP/DNS config)
4. esp-link succeeds in connecting and shuts down its own access point after 15 seconds,
you reconnect your laptop/phone to your normal network and access esp-link via its hostname
or IP address
#### Notes on using AP (access point) mode
Esp-link does not support STA+AP mode, however it does support STA mode and AP mode. What happens
is that STA+AP mode is used at boot and when making STA changes to allow for recovery: the AP
mode stays on for a while so you can connect to it and fix the STA mode. Once STA has connected,
esp-link switches to STA-only mode. There is no setting to stay in STA+AP mode. So... if you want
to use AP ensure you set esp-link to AP-only mode. If you want STA+AP mode you're gonna have to
modify the source for yourself. (This stuff is painful to test and rather tricky, so don't expect
the way it works to change.)
### LED indicators
Assuming appropriate hardware attached to GPIO pins, the green "conn" LED will show the wifi
status as follows:
- Very short flash once a second: not connected to a network and running as AP+STA, i.e.
trying to connect to the configured network
- Very short flash once every two seconds: not connected to a network and running as AP-only
- Even on/off at 1HZ: connected to the configured network but no IP address (waiting on DHCP)
- Steady on with very short off every 3 seconds: connected to the configured network with an
IP address (esp-link shuts down its AP after 60 seconds)
The yellow "ser" LED will blink briefly every time serial data is sent or received by the esp-link.
### Troubleshooting
- verify that you have sufficient power, borderline power can cause the esp module to seemingly
function until it tries to transmit and the power rail collapses
- if you just cannot flash your esp8266 module (some people call it the zombie mode) make sure you
have gpio0 and gpio15 pulled to gnd with a 1K resistor, gpio2 tied to 3.3V with 1K resistor, and
RX/TX connected without anything in series. If you need to level shift the signal going into the
esp8266's RX use a 1K resistor. Use 115200 baud in the flasher.
(For a permanent set-up I would use higher resistor values but
when nothing seems to work these are the ones I try.)
- if the flashing succeeded, check the "conn" LED to see which mode esp-link is in (see LED info above)
- reset or power-cycle the esp-link to force it to become an access-point if it can't
connect to your network within 15-20 seconds
- if the LED says that esp-link is on your network but you can't get to it, make sure your
laptop is on the same network (and no longer on the esp's network)
- if you do not know the esp-link's IP address on your network, try `esp-link.local`, try to find
the lease in your DHCP server; if all fails, you may have to turn off your access point (or walk
far enough away) and reset/power-cycle esp-link, it will then fail to connect and start its
own AP after 15-20 seconds
Configuration details
---------------------
### Wifi
After you have serially flashed the module it will create a wifi access point (AP) with an
SSID of the form `ESP_012ABC` where 012ABC is a piece of the module's MAC address.
Using a laptop, phone, or tablet connect to this SSID and then open a browser pointed at
http://192.168.4.1/, you should then see the esp-link web site.
Now configure the wifi. The desired configuration is for the esp-link to be a
station on your local wifi network so you can communicate with it from all your computers.
To make this happen, navigate to the wifi page and you should see the esp-link scan
for available networks. You should then see a list of detected networks on the web page and you
can select yours.
Enter a password if your network is secure (highly recommended...) and hit the connect button.
You should now see that the esp-link has connected to your network and it should show you
its IP address. _Write it down_. You will then have to switch your laptop, phone, or tablet
back to your network and then you can connect to the esp-link's IP address or, depending on your
network's DHCP/DNS config you may be able to go to http://esp-link.local
At this point the esp-link will have switched to STA mode and be just a station on your
wifi network. These settings are stored in flash and thereby remembered through resets and
power cycles. They are also remembered when you flash new firmware. Only flashing `blank.bin`
via the serial port as indicated above will reset the wifi settings.
There is a fail-safe, which is that after a reset or a configuration change, if the esp-link
cannot connect to your network it will revert back to AP+STA mode after 15 seconds and thus
both present its `ESP_012ABC`-style network and continue trying to reconnect to the requested network.
You can then connect to the esp-link's AP and reconfigure the station part.
One open issue (#28) is that esp-link cannot always display the IP address it is getting to the browser
used to configure the ssid/password info. The problem is that the initial STA+AP mode may use
channel 1 and you configure it to connect to an AP on channel 6. This requires the ESP8266's AP
to also switch to channel 6 disconnecting you in the meantime.
### Hostname, description, DHCP, mDNS
You can set a hostname on the "home" page, this should be just the hostname and not a domain
name, i.e., something like "test-module-1" and not "test-module-1.mydomain.com".
This has a number of effects:
- you will see the first 12 chars of the hostname in the menu bar (top left of the page) so
if you have multiple modules you can distinguish them visually
- esp-link will use the hostname in its DHCP request, which allows you to identify the module's
MAC and IP addresses in your DHCP server (typ. your wifi router). In addition, some DHCP
servers will inject these names into the local DNS cache so you can use URLs like
`hostname.local`.
- someday, esp-link will inject the hostname into mDNS (multicast DNS, bonjour, etc...) so
URLs of the form `hostname.local` work for everyone (as of v2.1.beta5 mDNS is disabled due
to reliability issues with it)
You can also enter a description of up to 128 characters on the home page (bottom right). This
allows you to leave a memo for yourself, such as "installed in basement to control the heating
system". This descritpion is not used anywhere else.
Building the firmware
---------------------
### Linux
The firmware has been built using the https://github.com/pfalcon/esp-open-sdk[esp-open-sdk]
on a Linux system. Create an esp8266 directory, install the esp-open-sdk into a sub-directory
using the *non-standalone* install (i.e., there should not be an sdk directory in the esp-open-sdk
dir when done installing, if you use the standalone install you will get compilation errors
with std types, such as `uint32_t`).
Download the Espressif SDK (use the version mentioned in the release notes) from their
http://bbs.espressif.com/viewforum.php?f=5[download forum] and also expand it into a
sub-directory.
Clone the esp-link repository into a third sub-directory and check out the tag you would like,
such as `git checkout v2.1.7`.
This way the relative paths in the Makefile will work.
If you choose a different directory structure look at the Makefile for the appropriate environment
variables to define.
Do not use the source tarballs from the release page on github,
these will give you trouble compiling because the Makefile uses git to determine the esp-link
version being built.
In order to OTA-update the esp8266 you should `export ESP_HOSTNAME=...` with the hostname or
IP address of your module.
Now, build the code: `make` in the top-level of esp-link. If you want to se the commands being
issued, use `VERBOSE=1 make`.
A few notes from others (I can't fully verify these):
- You may need to install `zlib1g-dev` and `python-serial`
- Make sure you have the correct version of the esp_iot_sdk
- Make sure the paths at the beginning of the makefile are correct
- Make sure `esp-open-sdk/xtensa-lx106-elf/bin` is in the PATH set in the Makefile
### Windows
It is possible to build esp-link on Windows, but it requires a gaggle of software to be installed:
- Install the unofficial sdk, mingw, SourceTree (gui git client), python 2.7, git cli, Java
- Use SourceTree to checkout under C:\espressif or wherever you installed the unofficial sdk,
(see this thread for the unofficial sdk http://www.esp8266.com/viewtopic.php?t=820)
- Create a symbolic link under c:/espressif for the git bin directory under program files and
the java bin directory under program files.
- ...
### Updating the firmware over-the-air
This firmware supports over-the-air (OTA) flashing, so you do not have to deal with serial
flashing again after the initial one! The recommended way to flash is to use `make wiflash`
if you are also building the firmware.
If you are downloading firmware binaries use `./wiflash`.
`make wiflash` assumes that you set `ESP_HOSTNAME` to the hostname or IP address of your esp-link.
You can easily do that using something like `ESP_HOSTNAME=192.168.1.5 make wiflash`.
The flashing, restart, and re-associating with your wireless network takes about 15 seconds
and is fully automatic. The first 1MB of flash are divided into two 512KB partitions allowing for new
code to be uploaded into one partition while running from the other. This is the official
OTA upgrade method supported by the SDK, except that the firmware is POSTed to the module
using curl as opposed to having the module download it from a cloud server. On a module with
512KB flash there is only space for one partition and thus no way to do an OTA update.
If you are downloading the binary versions of the firmware (links forthcoming) you need to have
both `user1.bin` and `user2.bin` handy and run `wiflash.sh <esp-hostname> user1.bin user2.bin`.
This will query the esp-link for which file it needs, upload the file, and then reconnect to
ensure all is well.
Note that when you flash the firmware the wifi settings are all preserved so the esp-link should
reconnect to your network within a few seconds and the whole flashing process should take 15-30
from beginning to end. If you need to clear the wifi settings you need to reflash the `blank.bin`
using the serial port.
The flash configuration and the OTA upgrade process is described in more detail in [FLASH.md](FLASH.md)
Serial bridge and connections to Arduino, AVR, ARM, LPC microcontrollers
------------------------------------------------------------------------
In order to connect through the esp-link to a microcontroller use port 23. For example,
on linux you can use `nc esp-hostname 23` or `telnet esp-hostname 23`.
Note that multiple connections to port 23 and 2323 can be made simultaneously. Esp-link will
intermix characters received on all these connections onto the serial TX and it will
broadcast incoming characters from the serial RX to all connections. Use with caution!
### Flashing an attached AVR/Arduino
There are three options for reprogramming an attached AVR/Arduino microcontroller:
- Use avrdude and point it at port 23 of esp-link. Esp-link automatically detects the programming
sequence and issues a reset to the AVR.
- Use avrdude and point it at port 2323 of esp-link. This is the same as port 23 except that the
autodectection is not used and the reset happens because port 2323 is used
- Use curl or a similar tool to HTTP POST the firmware to esp-link. This uses the built-in
programmer, which only works for AVRs/Arduinos with the optiboot bootloader (which is std).
To reprogram an Arduino / AVR microcontroller by pointing avrdude at port 23 or 2323 you
specify a serial port of the form `net:esp-link:23` in avrdude's -P option, where
`esp-link` is either the hostname of your esp-link or its IP address).
This is instead of specifying a serial port of the form /dev/ttyUSB0.
Esp-link detects that avrdude starts its connection with a flash synchronization sequence
and sends a reset to the AVR microcontroller so it can switch into flash programming mode.
To reprogram using the HTTP POST method you need to first issue a POST to put optiboot into
programming mode: POST to `http://esp-link/pgm/sync`, this starts the process. Then check that
synchronization with optiboot has been achieved by issuing a GET to the same URL
(`http://esp-link/pgm/sync`). Repeat until you have sync (takes <500ms normally). Finally
issue a POST request to `http://esp-link/pgm/upload` with your hex file as POST data (raw,
not url-encoded or multipart-mime. Please look into the avrflash script for the curl command-line
details or use that script directly (`./avrflash esp-link.local my_sketch.hex`).
_Important_: after the initial sync request that resets the AVR you have 10 seconds to get to the
upload post or esp-link will time-out. So if you're manually entering curl commands have them
prepared so you can copy&paste!
Beware of the baud rate, which you can set on the uC Console page. Sometimes you may be using
115200 baud in sketches but the bootloader may use 57600 baud. When you use port 23 or 2323 you
need to set the baud rate correctly. If you use the built-in programmer (HTTP POST method) then
esp-link will try the configured baud rate and also 9600, 57600, and 115200 baud, so it should
work even if you have the wrong baud rate configured...
When to use which method? If port 23 works then go with that. If you have trouble getting sync
or it craps out in the middle too often then try the built-in programmer with the HTTP POST.
If your AVR doesn't use optiboot then use port 2323 since esp-link may not recognize the programming
sequence and not issue a reset if you use port 23.
If you are having trouble with the built-in programmer and see something like this:
--------------------
# ./avrflash 192.168.3.104 blink.hex
Error checking sync: FAILED to SYNC: abandoned after timeout, got:
:\xF/\x00\xCj\xCz\xCJ\xCZ\xC\xAÜ\xC\xAä\xC\xAÜ\xC\xAä\xC\xBì\xC\xBô\xC\xBì\xC\xBô\xC\xAÜ\xC\xAä\xC
--------------------
the most likely cause is a baud rate mismatch and/or a bad connection from the esp8266 to the
AVRs reset line.
The baud rate used by esp-link is set on the uC Console web page and, as mentioned above, it will
automatically try 9600, 57600, and 115200 as well.
The above garbage characters are most likely due to optiboot timing out and starting the sketch
and then the sketch sending data at a different baud rate than configured into esp-link.
Note that sketches don't necessarily use the same baud rate as optiboot, so you may have the
correct baud rate configured but reset isn't functioning, or reset may be functioning but the
baud rate may be incorrect.
The output of a successful flash using the built-in programmer looks like this:
--------------------
Success. 3098 bytes at 57600 baud in 0.8s, 3674B/s 63% efficient
--------------------
This says that the sketch comprises 3098 bytes of flash, was written in 0.8 seconds
(excludes the initial sync time) at 57600 baud,
and the 3098 bytes were flashed at a rate of 3674 bytes per second.
The efficiency measure is the ratio of the actual rate to the serial baud rate,
thus 3674/5760 = 0.63 (there are 10 baud per character).
The efficiency is not 100% because there is protocol overhead (such as sync, record type, and
length characters)
and there is dead time waiting for an ack or preparing the next record to be sent.
### Details of built-in AVR flash algorithm
The built-in flashing algorithm differs a bit from what avrdude does. The programming protocol
states that STK_GET_SYNC+CRC_EOP (0x30 0x20) should be sent to synchronize, but that works poorly
because the AVR's UART only buffers one character. This means that if STK_GET_SYNC+CRC_EOP is
sent twice there is a high chance that only the last character (CRC_EOP) is actually
received. If that is followed by another STK_GET_SYNC+CRC_EOP sequence then optiboot receives
CRC_EOP+STK_GET_SYNC+CRC_EOP which causes it to abort and run the old sketch. Ending up in that
situation is quite likely because optiboot initializes the UART as one of the first things, but
then goes off an flashes an LED for ~300ms during which it doesn't empty the UART.
Looking at the optiboot code, the good news is that CRC_EOP+CRC_EOP can be used to get an initial
response without the overrun danger of the normal sync sequence and this is what esp-link does.
The programming sequence runs as follows:
- esp-link sends a brief reset pulse (1ms)
- esp-link sends CRC_EOP+CRC_EOP ~50ms later
- esp-link sends CRC_EOP+CRC_EOP every ~70-80ms
- eventually optiboot responds with STK_INSYNC+STK_OK (0x14;0x10)
- esp-link sends one CRC_EOP to sort out the even/odd issue
- either optiboot responds with STK_INSYNC+STK_OK or nothing happens for 70-80ms, in which case
esp-link sends another CRC_EOP
- esp-link sends STK_GET_SYNC+CRC_EOP and optiboot responds with STK_INSYNC+STK_OK and we're in
sync now
- esp-link sends the next command (starts with 'u') and programming starts...
If no sync is achieved, esp-link changes baud rate and the whole thing starts over with a reset
pulse about 600ms, esp-link gives up after about 5 seconds and reports an error.
### Flashing an attached ARM processor
You can reprogram NXP's LPC800-series and many other ARM processors as well by pointing your
programmer similarly at the esp-link's port 23. For example, if you are using
https://github.com/jeelabs/embello/tree/master/tools/uploader a command line like
`uploader -t -s -w esp-link:23 build/firmware.bin` does the trick.
The way it works is that the uploader uses telnet protocol escape sequences in order to
make esp-link issue the appropriate "ISP" and reset sequence to the microcontroller to start the
flash programming. If you use a different ARM programming tool it will work as well as long as
it starts the connection with the `?\r\n` synchronization sequence.
### Flashing an attached esp8266
Yes, you can use esp-link running on one esp8266 module to flash another esp8266 module,
however it is rather tricky! The problem is not electric, it is wifi interference.
The basic idea is to use some method to direct the esp8266 flash program to port 2323 of
esp-link. Using port 2323 with the appropriate wiring will cause the esp8266's reset and
gpio0 pins to be toggled such that the chip enters the flash programming mode.
One option for connecting the programmer with esp-link is to use my version of esptool.py
at http://github.com/tve/esptool, which supports specifying a URL instead of a port. Thus
instead of specifying something like `--port /dev/ttyUSB0` or `--port COM1` you specify
`--port socket://esp-link.local:2323`. Important: the baud rate specified on the esptool.py
command-line is irrelevant as the baud rate used by esp-link will be the one set in the
uC console page. Fortunately the esp8266 bootloader does auto-baud detection. (Setting the
baud rate to 115200 is recommended.)
Another option is to use a serial-to-tcp port forwarding driver and point that to port 2323
of esp-link. On windows users have reported success with
http://www.hw-group.com/products/hw_vsp/hw_vsp2_en.html[HW Virtual Serial Port]
Now to the interference problem: once the attached esp8266 is reset it
starts outputting its 26Mhz clock on gpio0, which needs to be attached to
the esp8266 running esp-link (since it needs to drive gpio0 low during
the reset to enter flash mode). This 26Mhz signal on gpio0 causes a
significant amount of radio interference with the result that the esp8266
running esp-link has trouble receiving Wifi packets. You can observe this
by running a ping to esp-link in another window: as soon as the target
esp8266 is reset, the pings become very slow or stop altogetehr. As soon
as you remove power to the attached esp8266 the pings resume beautifully.
To try and get the interference under control, try some of the following:
add a series 100ohm resistor and 100pf capacitor to ground as close to
the gpio0 pin as possible (basically a low pass filter); and/or pass
the cable connecting the two esp8266's through a ferrite bead.
### Debug log
The esp-link web UI can display the esp-link debug log (os_printf statements in the code). This
is handy but sometimes not sufficient. Esp-link also prints the debug info to the UART where
it is sometimes more convenient and sometimes less... For this reason three UART debug log
modes are supported that can be set in the web UI (and the mode is saved in flash):
- auto: the UART log starts enabled at boot using uart0 and disables itself when esp-link
associates with an AP. It re-enables itself if the association is lost.
- off: the UART log is always off
- on0: the UART log is always on using uart0
- on1: the UART log is always on using uart1 (gpio2 pin)
Note that even if the UART log is always off the ROM prints to uart0 whenever the
esp8266 comes out of reset. This cannot be disabled.
Outbound HTTP REST requests and MQTT client
-------------------------------------------
The V2 versions of esp-link use the SLIP protocol over the serial link to support simple outbound
HTTP REST requests as well as an MQTT client. The SLIP protocol consists of commands with
binary arguments sent from the
attached microcontroller to the esp8266, which then performs the command and responds back.
The responses back use a callback address in the attached microcontroller code, i.e., the
command sent by the uC contains a callback address and the response from the esp8266 starts
with that callback address. This enables asynchronous communication where esp-link can notify the
uC when requests complete or when other actions happen, such as wifi connectivity status changes.
You can find REST and MQTT libraries as well as demo sketches in the
https://github.com/jeelabs/el-client[el-client] repository.
Contact
-------
If you find problems with esp-link, please create a github issue. If you have a question, please
use the gitter chat link at the top of this page.

@ -0,0 +1,163 @@
ESP-LINK: Wifi-Serial Bridge w/REST&MQTT
========================================
<img src="https://cloud.githubusercontent.com/assets/39480/19333951/73fcdcbe-90ad-11e6-8572-5e654377275a.png">
The esp-link firmware connects a micro-controller to the internet using an ESP8266 Wifi module.
It implements a number of features:
- transparent bridge between Wifi and serial, useful for debugging or inputting into a uC
- flash-programming attached Arduino/AVR microcontrollers and
LPC800-series and other ARM microcontrollers via Wifi
- built-in stk500v1 programmer for AVR uC's: program using HTTP upload of hex file
- outbound REST HTTP requests from the attached micro-controller to the internet
- MQTT client pub/sub from the attached micro-controller to the internet
- serve custom web pages containing data that is dynamically pulled from the attached uC and
that contain buttons and fields that are transmitted to the attached uC (feature not
fully ready yet)
The firmware includes a tiny HTTP server based on
[esphttpd](http://www.esp8266.com/viewforum.php?f=34)
with a simple web interface, many thanks to Jeroen Domburg for making it available!
The REST and MQTT functionality are loosely based on [espduino](https://github.com/tuanpmt/espduino)
but significantly rewritten and no longer protocol compatible, thanks to tuanpmt for the
inspiration!
The following people contributed significant functionality to esp-link:
[brunnels](https://github.com/brunnels) (espduino integration),
[cskarai](https://github.com/cskarai) (custom dynamic web pages),
[beegee-tokyo](https://github.com/beegee-tokyo) (lots of code documentation),
[susisstrolch](https://github.com/susisstrolch) (syslog feature),
[bc547](https://github.com/bc547) and [katast](https://github.com/katast) (misc contributions).
Esp-link is the work of many contributors!
Note that http://github.com/jeelabs/esp-link is the original esp-link software which has
notably been forked by arduino.org as [Esp-Link](https://github.com/arduino-org/Esp-Link) and shipped
with the initial Arduino Uno Wifi. The JeeLabs esp-link has evolved significantly since the
fork and added cool new features as well as bug fixes.
### Quick links
In this document: [goals](#esp-link-goals), [uses](#esp-link-uses), [eye candy](#eye-candy),
[getting-started](#getting-started), [serial-bridge](#serial-bridge), [contact](#contact).
Separate documents:
- [hardware configuration](FLASHING.md), [serial flashing](FLASHING.md#initial-serial-flashing)
- [wifi configuration](WIFI-CONFIG.md)
- [troubleshooting](TROUBLESHOOTING.md), [LED indicators](TROUBLESHOOTING.md#led-indicators)
- [flashing an attached uC](UC-FLASHING.md)
- [MQTT and outbound REST requests](RESTMQTT.md)
- [service web pages](WEB-SERVER.md)
- [building esp-link](BUILDING.md), [over-the-air flashing](BUILDING.md#updating-the-firmware-over-the-air)
- [flash layout](FLASH.md)
For quick support and questions chat at
[![Chat at https://gitter.im/jeelabs/esp-link](https://badges.gitter.im/esp-link.svg)](https://gitter.im/jeelabs/esp-link)
or (a little slower) open a github issue.
Releases & Downloads
--------------------
Esp-link uses semantic versioning. The main change between versions 1.x and 2.x was the
addition of MQTT and outbound REST requests from the attached uC. The main change between 2.x
and 3.x will be the addition of custom web pages (this is not ready yet).
- The master branch is currently unstable as we integrate a number of new features to get
to version 3.0. Please use v2.2.3 unless you want to hack up the latest code!
This being said, the older functionality seems to work fine on master, YMMV...
- [V2.2.3](https://github.com/jeelabs/esp-link/releases/tag/v2.2.3) is the most recent release.
It has a built-in stk500v1 programmer (for AVRs), work on all modules, and supports mDNS,
sNTP, and syslog. It is built using the Espressif SDK 1.5.4.
- [V2.1.7](https://github.com/jeelabs/esp-link/releases/tag/v2.1.7) is the previous release.
- See [all releases](https://github.com/jeelabs/esp-link/releases).
## Esp-link goals
The goal of the esp-link project is to create an advanced Wifi co-processor. Esp-link assumes that
there is a "main processor" (also referred to as "attached uController") and that esp-link's role
is to facilitate communication over Wifi. This means that esp-link does not just connect TCP/UDP
sockets through to the attached uC, rather it implements mostly higher-level functionality to
offload the attached uC, which often has much less flash and memory than esp-link.
Where esp-link is a bit unusual is that it's not really
just a Wifi interface or a slave co-processor. In some sense it's the master, because the main
processor can be reset, controlled and reprogrammed through esp-link. The three main areas of
functionality in esp-link are:
- reprogramming and debugging the attached uC
- letting the attached uC make outbound communication and offloading the protocol processing
- forwarding inbound communication and offloading the protocol processing
The goal of the project is also to remain focused on the above mission. In particular, esp-link
is not a platform for stand-alone applications and it does not support connecting sensors or
actuators directly to it. A few users have taken esp-link as a starting point for doing these
things and that's great, but there's also value in keeping the mainline esp-link project
focused on a clear mission.
## Esp-link uses
The simplest use of esp-link is as a transparent serial to wifi bridge. You can flash an attached
uC over wifi and you can watch the uC's serial debug output by connecting to port 23 or looking
at the uC Console web page.
The next level is to use the outbound connectivity of esp-link in the uC code. For example, the
uC can use REST requests to services like thingspeak.com to send sensor values that then get
stored and plotted by the external service.
The uC can also use REST requests to retrieve simple configuration
information or push other forms of notifications. (MQTT functionality is forthcoming.)
An additional option is to add code to esp-link to customize it and put all the communication
code into esp-link and only keep simple sensor/actuator control in the attached uC. In this
mode the attached uC sends custom commands to esp-link with sensor/acturator info and
registers a set of callbacks with esp-link that control sensors/actuators. This way, custom
commands in esp-link can receive MQTT messages, make simple callbacks into the uC to get sensor
values or change actuators, and then respond back with MQTT. The way this is architected is that
the attached uC registers callbacks at start-up such that the code in the esp doesn't need to
know which exact sensors/actuators the attached uC has, it learns that through the initial
callback registration.
## Eye Candy
These screen shots show the Home page, the Wifi configuration page, the console for the
attached microcontroller, and the pin assignments card:
<img width="45%" src="https://cloud.githubusercontent.com/assets/39480/19334029/f8128c92-90ad-11e6-804e-9a4796035e9a.png">
<img width="45%" src="https://cloud.githubusercontent.com/assets/39480/8261427/6caf7326-167f-11e5-8085-bc8b20159b2b.png">
<img width="45%" src="https://cloud.githubusercontent.com/assets/39480/8261426/6ca7f75e-167f-11e5-827d-9a1c582ad05d.png">
<img width="30%" src="https://cloud.githubusercontent.com/assets/39480/8261658/11e6c64a-1681-11e5-82d0-ea5ec90a6ddb.png">
<img width="45%" src="https://cloud.githubusercontent.com/assets/39480/19334011/e0c3fe40-90ad-11e6-9893-847e805e7b89.png">
<img width="45%" src="https://cloud.githubusercontent.com/assets/39480/19333988/c1858cec-90ad-11e6-8b1c-ffed516e1b7f.png">
Getting Started
---------------
To get started you need to:
1. prepare your esp8266 module for serial flashing
2. download the latest esp-link release image (you can build your own later)
3. flash the firmware
4. configure the Wifi in esp-link for your network
You can then attach a uC and upload a sketch:
1. attach a uC (e.g. arduino) to your esp8266 module
2. connect via the serial port to see a pre-loaded sketch running
3. upload a fresh version of the sketch
From there, more advanced steps are:
- write a sketch that uses MQTT to communicate, or that makes outbound REST requests
- create some web pages and write a sketch that populates data in them or reacts to buttons
and forms
- make changes or enhancements to esp-link and build your own firmware
### Serial bridge
In order to connect through the esp-link to a microcontroller use port 23. For example,
on linux you can use `nc esp-hostname 23` or `telnet esp-hostname 23`.
Note that multiple connections to port 23 and 2323 can be made simultaneously. Esp-link will
intermix characters received on all these connections onto the serial TX and it will
broadcast incoming characters from the serial RX to all connections. Use with caution!
Contact
-------
If you find problems with esp-link, please create a github issue. If you have a question, please
use the gitter chat link at the top of this page.

@ -0,0 +1,14 @@
Esp-link: Outbound HTTP REST requests and MQTT client
-------------------------------------------
The V2 versions of esp-link use the SLIP protocol over the serial link to support simple outbound
HTTP REST requests as well as an MQTT client. The SLIP protocol consists of commands with
binary arguments sent from the
attached microcontroller to the esp8266, which then performs the command and responds back.
The responses back use a callback address in the attached microcontroller code, i.e., the
command sent by the uC contains a callback address and the response from the esp8266 starts
with that callback address. This enables asynchronous communication where esp-link can notify the
uC when requests complete or when other actions happen, such as wifi connectivity status changes.
You can find REST and MQTT libraries as well as demo sketches in the
[el-client](https://github.com/jeelabs/el-client) repository.

@ -0,0 +1,36 @@
Esp-Link troubleshooting
========================
### Troubleshooting
- verify that you have sufficient power, borderline power can cause the esp module to seemingly
function until it tries to transmit and the power rail collapses
- if you just cannot flash your esp8266 module (some people call it the zombie mode) make sure you
have gpio0 and gpio15 pulled to gnd with a 1K resistor, gpio2 tied to 3.3V with 1K resistor, and
RX/TX connected without anything in series. If you need to level shift the signal going into the
esp8266's RX use a 1K resistor. Use 115200 baud in the flasher.
(For a permanent set-up I would use higher resistor values but
when nothing seems to work these are the ones I try.)
- if the flashing succeeded, check the "conn" LED to see which mode esp-link is in (see LED info above)
- reset or power-cycle the esp-link to force it to become an access-point if it can't
connect to your network within 15-20 seconds
- if the LED says that esp-link is on your network but you can't get to it, make sure your
laptop is on the same network (and no longer on the esp's network)
- if you do not know the esp-link's IP address on your network, try `esp-link.local`, try to find
the lease in your DHCP server; if all fails, you may have to turn off your access point (or walk
far enough away) and reset/power-cycle esp-link, it will then fail to connect and start its
own AP after 15-20 seconds
### LED indicators
Assuming appropriate hardware attached to GPIO pins, the green "conn" LED will show the wifi
status as follows:
- Very short flash once a second: not connected to a network and running as AP+STA, i.e.
trying to connect to the configured network
- Very short flash once every two seconds: not connected to a network and running as AP-only
- Even on/off at 1HZ: connected to the configured network but no IP address (waiting on DHCP)
- Steady on with very short off every 3 seconds: connected to the configured network with an
IP address (esp-link shuts down its AP after 60 seconds)
The yellow "ser" LED will blink briefly every time serial data is sent or received by the esp-link.

@ -0,0 +1,177 @@
Flashing an attached Microcontroller
====================================
In order to connect through the esp-link to a microcontroller use port 23. For example,
on linux you can use `nc esp-hostname 23` or `telnet esp-hostname 23`.
Note that multiple connections to port 23 and 2323 can be made simultaneously. Esp-link will
intermix characters received on all these connections onto the serial TX and it will
broadcast incoming characters from the serial RX to all connections. Use with caution!
### Flashing an attached AVR/Arduino
There are three options for reprogramming an attached AVR/Arduino microcontroller:
- Use avrdude and point it at port 23 of esp-link. Esp-link automatically detects the programming
sequence and issues a reset to the AVR.
- Use avrdude and point it at port 2323 of esp-link. This is the same as port 23 except that the
autodectection is not used and the reset happens because port 2323 is used
- Use curl or a similar tool to HTTP POST the firmware to esp-link. This uses the built-in
programmer, which only works for AVRs/Arduinos with the optiboot bootloader (which is std).
To reprogram an Arduino / AVR microcontroller by pointing avrdude at port 23 or 2323 you
specify a serial port of the form `net:esp-link:23` in avrdude's -P option, where
`esp-link` is either the hostname of your esp-link or its IP address).
This is instead of specifying a serial port of the form /dev/ttyUSB0.
Esp-link detects that avrdude starts its connection with a flash synchronization sequence
and sends a reset to the AVR microcontroller so it can switch into flash programming mode.
To reprogram using the HTTP POST method you need to first issue a POST to put optiboot into
programming mode: POST to `http://esp-link/pgm/sync`, this starts the process. Then check that
synchronization with optiboot has been achieved by issuing a GET to the same URL
(`http://esp-link/pgm/sync`). Repeat until you have sync (takes <500ms normally). Finally
issue a POST request to `http://esp-link/pgm/upload` with your hex file as POST data (raw,
not url-encoded or multipart-mime. Please look into the avrflash script for the curl command-line
details or use that script directly (`./avrflash esp-link.local my_sketch.hex`).
_Important_: after the initial sync request that resets the AVR you have 10 seconds to get to the
upload post or esp-link will time-out. So if you're manually entering curl commands have them
prepared so you can copy&paste!
Beware of the baud rate, which you can set on the uC Console page. Sometimes you may be using
115200 baud in sketches but the bootloader may use 57600 baud. When you use port 23 or 2323 you
need to set the baud rate correctly. If you use the built-in programmer (HTTP POST method) then
esp-link will try the configured baud rate and also 9600, 57600, and 115200 baud, so it should
work even if you have the wrong baud rate configured...
When to use which method? If port 23 works then go with that. If you have trouble getting sync
or it craps out in the middle too often then try the built-in programmer with the HTTP POST.
If your AVR doesn't use optiboot then use port 2323 since esp-link may not recognize the programming
sequence and not issue a reset if you use port 23.
If you are having trouble with the built-in programmer and see something like this:
```
# ./avrflash 192.168.3.104 blink.hex
Error checking sync: FAILED to SYNC: abandoned after timeout, got:
:\xF/\x00\xCj\xCz\xCJ\xCZ\xC\xAÜ\xC\xAä\xC\xAÜ\xC\xAä\xC\xBì\xC\xBô\xC\xBì\xC\xBô\xC\xAÜ\xC\xAä\xC
```
the most likely cause is a baud rate mismatch and/or a bad connection from the esp8266 to the
AVRs reset line.
The baud rate used by esp-link is set on the uC Console web page and, as mentioned above, it will
automatically try 9600, 57600, and 115200 as well.
The above garbage characters are most likely due to optiboot timing out and starting the sketch
and then the sketch sending data at a different baud rate than configured into esp-link.
Note that sketches don't necessarily use the same baud rate as optiboot, so you may have the
correct baud rate configured but reset isn't functioning, or reset may be functioning but the
baud rate may be incorrect.
The output of a successful flash using the built-in programmer looks like this:
```
Success. 3098 bytes at 57600 baud in 0.8s, 3674B/s 63% efficient
```
This says that the sketch comprises 3098 bytes of flash, was written in 0.8 seconds
(excludes the initial sync time) at 57600 baud,
and the 3098 bytes were flashed at a rate of 3674 bytes per second.
The efficiency measure is the ratio of the actual rate to the serial baud rate,
thus 3674/5760 = 0.63 (there are 10 baud per character).
The efficiency is not 100% because there is protocol overhead (such as sync, record type, and
length characters)
and there is dead time waiting for an ack or preparing the next record to be sent.
### Details of built-in AVR flash algorithm
The built-in flashing algorithm differs a bit from what avrdude does. The programming protocol
states that STK_GET_SYNC+CRC_EOP (0x30 0x20) should be sent to synchronize, but that works poorly
because the AVR's UART only buffers one character. This means that if STK_GET_SYNC+CRC_EOP is
sent twice there is a high chance that only the last character (CRC_EOP) is actually
received. If that is followed by another STK_GET_SYNC+CRC_EOP sequence then optiboot receives
CRC_EOP+STK_GET_SYNC+CRC_EOP which causes it to abort and run the old sketch. Ending up in that
situation is quite likely because optiboot initializes the UART as one of the first things, but
then goes off an flashes an LED for ~300ms during which it doesn't empty the UART.
Looking at the optiboot code, the good news is that CRC_EOP+CRC_EOP can be used to get an initial
response without the overrun danger of the normal sync sequence and this is what esp-link does.
The programming sequence runs as follows:
- esp-link sends a brief reset pulse (1ms)
- esp-link sends CRC_EOP+CRC_EOP ~50ms later
- esp-link sends CRC_EOP+CRC_EOP every ~70-80ms
- eventually optiboot responds with STK_INSYNC+STK_OK (0x14;0x10)
- esp-link sends one CRC_EOP to sort out the even/odd issue
- either optiboot responds with STK_INSYNC+STK_OK or nothing happens for 70-80ms, in which case
esp-link sends another CRC_EOP
- esp-link sends STK_GET_SYNC+CRC_EOP and optiboot responds with STK_INSYNC+STK_OK and we're in
sync now
- esp-link sends the next command (starts with 'u') and programming starts...
If no sync is achieved, esp-link changes baud rate and the whole thing starts over with a reset
pulse about 600ms, esp-link gives up after about 5 seconds and reports an error.
### Flashing an attached ARM processor
You can reprogram NXP's LPC800-series and many other ARM processors as well by pointing your
programmer similarly at the esp-link's port 23. For example, if you are using
https://github.com/jeelabs/embello/tree/master/tools/uploader a command line like
`uploader -t -s -w esp-link:23 build/firmware.bin` does the trick.
The way it works is that the uploader uses telnet protocol escape sequences in order to
make esp-link issue the appropriate "ISP" and reset sequence to the microcontroller to start the
flash programming. If you use a different ARM programming tool it will work as well as long as
it starts the connection with the `?\r\n` synchronization sequence.
### Flashing an attached esp8266
__Flashing another esp8266 module is possible in theory but real-world attempts haveso far been
rather unsuccessful due to Wifi interference. This section is left here in case someone else
wants to dig in and find a solution.__
You can use esp-link running on one esp8266 module to flash another esp8266 module,
however it is rather tricky! The problem is not electric, it is wifi interference.
The basic idea is to use some method to direct the esp8266 flash program to port 2323 of
esp-link. Using port 2323 with the appropriate wiring will cause the esp8266's reset and
gpio0 pins to be toggled such that the chip enters the flash programming mode.
One option for connecting the programmer with esp-link is to use my version of esptool.py
at http://github.com/tve/esptool, which supports specifying a URL instead of a port. Thus
instead of specifying something like `--port /dev/ttyUSB0` or `--port COM1` you specify
`--port socket://esp-link.local:2323`. Important: the baud rate specified on the esptool.py
command-line is irrelevant as the baud rate used by esp-link will be the one set in the
uC console page. Fortunately the esp8266 bootloader does auto-baud detection. (Setting the
baud rate to 115200 is recommended.)
Another option is to use a serial-to-tcp port forwarding driver and point that to port 2323
of esp-link. On windows users have reported success with
http://www.hw-group.com/products/hw_vsp/hw_vsp2_en.html[HW Virtual Serial Port]
Now to the interference problem: once the attached esp8266 is reset it
starts outputting its 26Mhz clock on gpio0, which needs to be attached to
the esp8266 running esp-link (since it needs to drive gpio0 low during
the reset to enter flash mode). This 26Mhz signal on gpio0 causes a
significant amount of radio interference with the result that the esp8266
running esp-link has trouble receiving Wifi packets. You can observe this
by running a ping to esp-link in another window: as soon as the target
esp8266 is reset, the pings become very slow or stop altogetehr. As soon
as you remove power to the attached esp8266 the pings resume beautifully.
To try and get the interference under control, try some of the following:
add a series 100ohm resistor and 100pf capacitor to ground as close to
the gpio0 pin as possible (basically a low pass filter); and/or pass
the cable connecting the two esp8266's through a ferrite bead.
### Debug log
The esp-link web UI can display the esp-link debug log (os_printf statements in the code). This
is handy but sometimes not sufficient. Esp-link also prints the debug info to the UART where
it is sometimes more convenient and sometimes less... For this reason three UART debug log
modes are supported that can be set in the web UI (and the mode is saved in flash):
- auto: the UART log starts enabled at boot using uart0 and disables itself when esp-link
associates with an AP. It re-enables itself if the association is lost.
- off: the UART log is always off
- on0: the UART log is always on using uart0
- on1: the UART log is always on using uart1 (gpio2 pin)
Note that even if the UART log is always off the ROM prints to uart0 whenever the
esp8266 comes out of reset. This cannot be disabled.

@ -0,0 +1,87 @@
Esp-link Wifi configuration
===========================
For proper operation the end state that esp-link needs to arrive at is to have it
join your pre-existing wifi network as a pure station.
However, in order to get there esp-link will start out as an access point and you'll have
to join its network to configure it. The short version is:
1. esp-link creates a wifi access point with an SSID of the form `ESP_012ABC` (some modules
use a different SSID form, such as `ai-thinker-012ABC`)
2. you join your laptop or phone to esp-link's network as a station and you configure
esp-link wifi with your network info by pointing your browser at `http://192.168.4.1/`
3. you set a hostname for esp-link on the "home" page, or leave the default ("esp-link")
4. esp-link starts to connect to your network while continuing to also be an access point
("AP+STA"), the esp-link may show up with a `${hostname}.local` hostname
(depends on your DHCP/DNS config)
4. esp-link succeeds in connecting and shuts down its own access point after 15 seconds,
you reconnect your laptop/phone to your normal network and access esp-link via its hostname
or IP address
### Notes on using AP (access point) mode
Esp-link does not support STA+AP mode, however it does support STA mode and AP mode. What happens
is that STA+AP mode is used at boot and when making STA changes to allow for recovery: the AP
mode stays on for a while so you can connect to it and fix the STA mode. Once STA has connected,
esp-link switches to STA-only mode. There is no setting to stay in STA+AP mode. So... if you want
to use AP ensure you set esp-link to AP-only mode. If you want STA+AP mode you're gonna have to
modify the source for yourself. (This stuff is painful to test and rather tricky, so don't expect
the way it works to change.)
Configuration details
---------------------
### Wifi
After you have serially flashed the module it will create a wifi access point (AP) with an
SSID of the form `ESP_012ABC` where 012ABC is a piece of the module's MAC address.
Using a laptop, phone, or tablet connect to this SSID and then open a browser pointed at
http://192.168.4.1/, you should then see the esp-link web site.
Now configure the wifi. The desired configuration is for the esp-link to be a
station on your local wifi network so you can communicate with it from all your computers.
To make this happen, navigate to the wifi page and you should see the esp-link scan
for available networks. You should then see a list of detected networks on the web page and you
can select yours.
Enter a password if your network is secure (highly recommended...) and hit the connect button.
You should now see that the esp-link has connected to your network and it should show you
its IP address. _Write it down_. You will then have to switch your laptop, phone, or tablet
back to your network and then you can connect to the esp-link's IP address or, depending on your
network's DHCP/DNS config you may be able to go to http://esp-link.local
At this point the esp-link will have switched to STA mode and be just a station on your
wifi network. These settings are stored in flash and thereby remembered through resets and
power cycles. They are also remembered when you flash new firmware. Only flashing `blank.bin`
via the serial port as indicated above will reset the wifi settings.
There is a fail-safe, which is that after a reset or a configuration change, if the esp-link
cannot connect to your network it will revert back to AP+STA mode after 15 seconds and thus
both present its `ESP_012ABC`-style network and continue trying to reconnect to the requested network.
You can then connect to the esp-link's AP and reconfigure the station part.
One open issue (#28) is that esp-link cannot always display the IP address it is getting to the browser
used to configure the ssid/password info. The problem is that the initial STA+AP mode may use
channel 1 and you configure it to connect to an AP on channel 6. This requires the ESP8266's AP
to also switch to channel 6 disconnecting you in the meantime.
### Hostname, description, DHCP, mDNS
You can set a hostname on the "home" page, this should be just the hostname and not a domain
name, i.e., something like "test-module-1" and not "test-module-1.mydomain.com".
This has a number of effects:
- you will see the first 12 chars of the hostname in the menu bar (top left of the page) so
if you have multiple modules you can distinguish them visually
- esp-link will use the hostname in its DHCP request, which allows you to identify the module's
MAC and IP addresses in your DHCP server (typ. your wifi router). In addition, some DHCP
servers will inject these names into the local DNS cache so you can use URLs like
`hostname.local`.
- someday, esp-link will inject the hostname into mDNS (multicast DNS, bonjour, etc...) so
URLs of the form `hostname.local` work for everyone (as of v2.1.beta5 mDNS is disabled due
to reliability issues with it)
You can also enter a description of up to 128 characters on the home page (bottom right). This
allows you to leave a memo for yourself, such as "installed in basement to control the heating
system". This descritpion is not used anywhere else.
Loading…
Cancel
Save