Skip to content
No description, website, or topics provided.
Jupyter Notebook HTML Makefile Other
Branch: master
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Permalink
Type Name Latest commit message Commit time
Failed to load latest commit information.
doc
extract_temp_data/New_Data Submit For Contest May 17, 2018
src Add DBGMODE Jul 8, 2018
Data_Preprocessing_Degree.html Submit For Contest May 17, 2018
Readme.md Update Readme.md May 17, 2018
openocd.log Submit For Contest May 17, 2018

Readme.md

ARC Temperature Sensing Calibration Platform

In this contest, we aim to develop a machine learning method to calibrate the thermal sensor and to avoid the interference from the environment for higher accuracy level in temperature measurement.The computation complexity of our proposed method would be suitable for porting on to the ARC Embedded Starter Kit platform.

Introduction

In most acedemic research, most of those fabricated chips are verfied through testing machine to check the input and output pattern's functionality. However for sensor design, we need a platform to check the functionality.

Due to the process variation , sometimes our chip may not work as well as we do in our post-layout-simulation. So during system integration, we still need some software method (using some Machine Learning Algorithm) to calibrate the chip's functionality.

Therefore we adopt ARC Embedded Stater Kit as our platform to build up our system prototype.

  • Main Task Of Our Platform
    • Data Extraction
    • Chip Sensor Functionality Calibration

System-Architecture

Phase1: Data Extraction

Using ARC EMSK to build up the temperature monitor system to collect all the chip's response data in the testing chamber. All the data collected from EMSK would transmitted through UART inteface back to PC for analysis.

Phase2: Data Analysis & Algorithm Selection

After data preprocessing, we could divide those raw-data into training set and testing set, and start to build up our Machine Learning Model to get better temperature accuracy.

Phase3: Porting ML model to ARC EMSK Platform

Currently we only developed the Terminal Display Version Porting the Inference Part of our Machine Learning Model to the ARC EMSK platform and start to check our functionality of our ML model to boost up the accuracy.

Hardware-And-Software-Setup

Required-Hardware

ARC EMSK Platform with USB cable

Customized Temperature Sensor Chip

Picture
Technology:0.18u
Temperature Range: 0 ~ 100
Area 0.7 mm^2

Other Part for Sensor Read Out Circuit

Part Name Number Purpose
7404 Logic IC 6 Buffer from EMSK(3.3V) to CHIP IO(1.8V)
NTC Themistor $2 M\Omega$ 2 External Thermal Sensor
Capacitor 0.5uF 5 CHIP Voltage and IO PAD Voltage Stabilization
Dupont Line 16 ARC Connection
34Pin 2 Connector IDE Cable 2 For Long Connection to Breadboard in Chamber
PC platform
Intel Core i5-4570S CPU @ 2.90GHz (4Core)
RAM: 12GB
Windows 7 64-bit Operating System

Required-Software

Hardware-Connection

  • Chip Pin Definition
  • Thermal Sensor Chip on Breadboard
  • EMSK connection

User-Manual

before-running-this-application

  • Modify mux.c (/board/emsk/drivers/mux/mux.c)
line 201: change 
	set_pmod_mux(PM1_UR_UART_0 | PM1_LR_SPI_S	\
				| PM2_I2C_HRI		\
				| PM3_GPIO_AC		\
				| PM4_I2C_GPIO_D	\
				| PM5_UR_SPI_M1 | PM5_LR_GPIO_A	\
				| PM6_UR_SPI_M0 | PM6_LR_GPIO_A );
 to 
 	set_pmod_mux(PM1_UR_GPIO_C | PM1_LR_GPIO_A\
				| PM2_GPIO_AC \
				| PM3_GPIO_AC		\
				| PM4_GPIO_AC	\
				| PM5_UR_GPIO_C | PM5_LR_GPIO_A	\
				| PM6_UR_GPIO_C | PM6_LR_GPIO_A );

to change the default pmod setting from UART I2C SPI All to GPIO

  • makefile
BOARD ?= emsk
BD_VER ?= 22
CUR_CORE ?= ercem11d
JTAG ?= usb
EMBARC_ROOT = setting your root path of your embARC-osp without " "
Machine Learning Data Analysis Please Reference to the Technical Document with following links.

Link

run-this-application

run the "make run" command on cmd in directory where your makefile is

Result:

DemoVideo

Link

You can’t perform that action at this time.