CNXSoft – Embedded Software Development

Toradex Oak product family is a range of USB sensors enabling measurement of humidity, temperature, motion, orientation… and USB expansion boards with relays, digital I/O and more. Following feedback from customers who needed to customize the designs, Toradex decided to fully open source those sensors and interfaces by releasing hardware files and the source code under Creative Commons License ‘Attribution CC BY’.

Toradex Oak P Pressure and Temperature Sensor

This release brings 13 USB sensors to the open source community:

• Oak USB Sensor Atmospheric Pressure
• Oak USB Sensor 3 Axes Acceleration
• Oak USB Sensor Angular Rotation
• Oak USB Sensor Humidity
• Oak USB Sensor IR Distance Triangulation (10-60cm)


Toradex Oak Relay – 4 Channel Relay Output Card

• Oak USB Sensor IR Motion Detection
• Oak USB Sensor Luminosity (LUX)
• Oak USB Sensor Orientation (3-Axes Accelerometer & Magnetometer)
• Oak USB Sensor 3 Axes Tilt/Inclination
• Oak USB Sensor IR Distance Triangulation (10-80cm)
• Oak USB Sensor 2 Channels Thermocouple
• Oak USB Sensor 4 Channels Capacitive Proximity Switch
• Oak USB Sensor RGB Color (Prestudy)

and 5 interface boards:

• Oak USB 24 Digital I/Os at 3.3V or 5V Logic Level
• Oak USB 4 Channel Relay Output
• Oak USB Isolated 8 Channel Digital Inputs, 36V Tolerant
• Oak USB ± 10V Isolated 8 Channel A/D Converter
• Oak USB Sensor Generic 4-20mA Current (Isolated)

For each board, Toradex publicly released the technical datasheet, design data including PCB assembly drawings, Bill of Materials, manufacturing data (gerber), factory test program and procedures, schematics (PDF only), and firmware in source code and binary format.

The company also released documentation such as firmware programming and modification instructions using Cypress CY3217-MiniProg1 programming tool,  PSoC Programmer & Designer software, as well as Toradex Oak Studio, a software that processes Oak Sensor data in graphical and tabular form, and runs in Windows XP or Windows 7.

Source code and instructions to use the USB modules in Linux are available in Oak on Linux page.

Toradex explains that the chosen Creative Common license allows anybody to modify it, and modifications can kept private if you chose to do so. Following the decision to fully open source their sensors, the company has chosen to discontinue to manufacture and supply products of the Oak family. Or maybe it’s the other way around.

You can find full details for the sensors and interface boards on Toradex USB Sensors and Peripherals Developer’s Page.

“Le Labo Citoyen” is a recently founded French non-profit organization aimed at “promoting and experimenting with innovating and free technologies for the citizens and the environment”.  Their first project is to gather pollution data (NO2, O3, and SO2 levels) in Paris using 2 (soon to be) open source components:

• Gasser – Self-contained mobile sensor currently powered by the Raspberry Pi
• ThingStream – Open source IoT datastore which should be similar to iDigi Cloud, except you can just store data in your own server or on “Le Labo Citoyen” servers.

Gasser has four main parts:

• Sensor(s) – Alphasense B4-series sensors (black and red component in the top left of the  main box) with accuracy of up to <10 ppb (parts-per-billion). Cost: ~110 Euros. They currently only use the NO2 (nitrogen dioxide) sensor.
• ADC & Computer – Raspberry Pi (Cost: ~30 Euros) & Delta-Sigma ADC (Cost: ~30 Euros).
• Communication Medium – Huawei E220 GRPS USB dongle (White dongle on the left side). Cost: ~35 Euros. This is optional as Ethernet, Wi-Fi or writing to an SD card could also be used to gather data.
• Power Supply – 8Ah USB Battery Pack (Black box on the left side). Cost: ~30 Euros.

The Raspberry Pi may be overkill (in terms of processing power and possibly cost) for this application as the board just takes in information from the open-source delta-sigma ADC and sends data to the ThingStream server via a 2 Euros/month GPRS connection or writes it to the SD card.  The battery can last 5 to 6 hours on this system which seems very low considering the size of the battery, and I assume some optimizations could eventually be performed on the system to allow several days or weeks of operations depending on the sampling rate and server communications frequency.

As the project is open source (hardware and software), the developers expect to eventually release a Raspbian image running the necessary scripts to gather and process the data. Thingstream source code is still very early beta, and not open to the public. But if you want to participate as a tester or developer, you can contact Le Labo Citoyen by email at contact <@t> labocitoyen.fr.

As the casing costs 10 Euros, so the total BoM cost for the hardware shown above is about 255 Euros (~$330 US). This is version 2 prototype and costs should go down eventually. If you live in country with no VAT and source the components from Chinese shops, it should be possible to get this hardware for less than $250 US, and even less than $200 is your scrap the USB battery pack and the GPRS modem.

You can find further information about the Gasser on Labo Citoyen Wiki and their IoT server on ThingStream.com.

I’ve recently received a complementary book entitled “Texas Instruments FRAM MCUs for DUMMIES” sent by Mouser, that provides technical and practical information about FRAM (Ferroelectric Random Access Memory) – pronounced F-RAM – and Texas Instruments MSP430FR57xx MCU series which makes use of this relatively new type of memory.

FRAM is a non-volatile memory with power and write speed & endurance characteristics that almost matches SRAM capabilities, and leave traditional Flash and EEPROM memory in the dust in terms of performance, as you can see from the table and diagram below.

FRAM Memory Comparison

Write Speed & Power Consumption: F-RAM vs Flash/EEPROM

At constant speed, FRAM consumes 250x less than Flash/EEPROM. Please note the FRAM write speed also depends on the MCU used, and a MCU @ 8MHz can write the FRAM @ 1400 kBps (Source: TI).

However, you won’t see this type of memory in devices like smartphones anytime soon because the maximum size currently manufactured is 1MB, density is higher (130 nm process manufacturing) and the cost is higher than traditional DRAM + Flash combination. This is why FRAM are currently used in applications that require ultra low power consumption and non-volatile storage write capabilities such as data logging, sensor networks, batteryless applications and wireless networks where OTA firmware upgrade may be required (This lowers the maintenance costs).

This is why FRAM is currently inside some micro-controllers as a replacement to built-in SRAM and Flash. In the diagram below, you can see three type of MCU with embedded flash and MCUs with embedded FRAM.

Flash/EEPROM (left) vs FRAM (right) Memory Map

One key advantage of MCU with FRAM is that the memory map can be configured as RAM (R/W) or to store code and constant (Read only). For example, if your program requires 3KB or RAM and 10 KB to storage code and data, a 16 KB FRAM MCU would be suitable, but you may have to select a “traditional” MCU with 32 KB Flash and 6KB SRAM to achieve the same functionalities. FRAM MCUs are more flexible, and it may cost less depending on your requirements.

Finally, FRAM has also much better resistance to radiation than Flash or EEPROM and is immune to magnetic fields, which may be an advantage for some medical applications, for example.

MSP-EXP430FR5739 Experimenter Board

Texas Instruments is one of the few companies that provide FRAM 16-bit MCUs with their MSP430FR57xx series and (upcoming) MSP430FR58xx/59xx “Wolverine” MCUs which feature 4 to 64 KB FRAM, 1 or 2 KB SRAM and with MCU frequencies of 8, 16 24 MHz.

If you just want to evaluate the technology, Texas Instruments provides the MSP-EXP430FR5739 Experimenter Board with the following specifications:

• MCU – MSP430FR5739 16-bit MCU @ 8 MHz with 16KB FRAM,  1KB SRAM, 2x Timer_A Blocks, 3x Timer_B Block, 1x USCI (UART/SPI/IrDA/I2C ) Blocks, 16Ch 10-Bit ADC12_B, 16Ch Comp_D, 32 I/Os
• 3 axis accelerometer
• NTC Thermister
• 8 Display LED’s
• Footprint for additional through-hole LDR sensor
• 2 User input Switches
• Connections
  • Connection to MSP-EXP430F5438
  • Connection to most Wireless Daughter Cards (CCxxxx RF)

The board comes preloaded with out-of-box demo code with allows 4 Modes to test FRAM features:

• Mode 1 – Max write speed
• Mode 2 – Flash write speed emulation
• Mode 3 – Fast sampling with writes using accelerometer
• Mode 4 – Fast sampling with writes using Thermistor

The experimenter board can be purchased on TI e-Store or distributors for $29. There’s also a training workshop that you can attend online, visit MSP430 FR57xx Training Workshop for details.

Freescale announced the Xtrinsic Sensor Platform, a 12-axis sensor development reference platform for Windows 8.

Xtrinsic Sensor Platform connected to a Windows 8 / RT Tablet

This 12-axis sensor platform includes several sensors by Freescale and other companies:

• Xtrinsic MMA8451Q 3-axis accelerometer
• Xtrinsic MAG3110 3-axis magnetometer
• Xtrinsic MPL3115A2 precision altimeter, pressure, and temperature sensor
• An analog ambient light sensor
• A selection of 3D gyroscopes are also supported

This platform is powered by Freescale ColdFire+ MCF51JU128VHS MCU which combines, configures and processes sensor data with Freescale sensor fusion software to match the requirements of Windows 8/RT. Xtrinic sensor platform communicates with the host device via USB. It does not requires extra drivers as it uses standard HID drivers.

You can watch the video below which is an introduction (including some technical details) and demo of the system.

Strangely, I could not find a similar Freescale 12-axis reference design for Android or Linux, and I’m not sure why they would limit this hardware to only support one OS, especially they already have Android and Linux driver for the components (accelerometer, magnetometer…) part of this development kit.

Xtrinsic Sensor platform will be available in Q3 2012. You may find further information on Freescale Xtrinsic sensor platform for Windows 8 page. The platform is currently demonstrated at Computex, in Taipei, Taiwan.

Jen Costillo of Lab 126 discusses the Android sensor subsystem at the Android Builder Summit in February 2012.

Abstract:

This lecture will arm Android device architects with the tactical knowledge they need to navigate the Android Sensor subsystem and make knowledgeable design choices to improve user experience and improve battery performance. The talk will address:

• Hardware architecture and trade-offs including latency, power, and software architecture implications:
• Wake up events and power considerations
• Gesture Detection Algorithm processing location and considerations
• Testing methodologies (Creating tools to aid develop and collect data.

This talk targets the kernel/firmware developer responsible for the sensor architecture. They should be familiar with kernel drivers, embedded systems, hardware bring up, Android services, and the C language.

You can also download the presentation slides on linuxfoundation.org website.