Basic Information

  • Project Title / Domain: Supply Sensitivity Reduction of Ring Oscillator-based Temperature Sensor
  • Name: Anupam Sardar
  • Guide: Prof. Udayan Ganguly

Description of Abstract / Project

With higher density of transistors it is crucial to monitir the temperature of areas all over the chip. Normal temperature sensors output analog voltages to tell the temperature and that needs a huge ADC, wastes space. Instead, modern methods are to use ring oscillator-based sensors with some thermally sensitive element so that the frequency of oscillation changes with teperature, and read a frequency requires just a small frequency counter. But there is a little problem. The frequency is also affected by supply voltage fluctuations and it becomes hard to tell what the temperature is. My work was to figure out ways to desensitise the effect of supply voltage variations on the frequency output. Our sensor used a Band-to-band-tunnelling mode transistor that worked as the temperature sensing element. I trued current compensation to make up for the supply voltage variations, so that it appears to the ring oscillator that nothing has changed.

Whom did you work with?

PG students

Relevant Courses

EE618 CMOS Analog

How did you connect with the prof?

Via mail.

Expectations from Prof

I wanted a neuromorphic hardware project. But I settled with this cause that was not available.

Logistics of the Project

I used to meet the guide once every week.

Reading Material

Had to read and get ideas from a lot of paprrs. All ranged from circuit topologies on the top level to device physics in the deep level.

Tools / Simulations / Softwares / Hardwares Used

Cadence

Outcomes / Deliverables

It was not complete. But I could deliver the way of current compensation.

Extension to DDP

No. I did not want to work on this. I wanted to work on Neuromorphics under the prof. But all the hardware works had already been completed. Also there was little scope of mathematical and physical modelling. So the work had been more like hit and trial with ideas and techniques from various papers. I did not enjoy that process very much. So I switched my topic. My present DDP is more into analysis, modelling and more structured.

Type of Project

implementation-driven

Connection to DDP

nan

Advice

Professor Udayan Ganguly offers projects in devices and neuromorphics field. Most likely a semster long neuromorphic project is hard to get. There might be 1 year or 1.5 year long projects available. For that one should be prepared that they want to continue that into their DDP and should be fairly consitent and show significant effort. But I guess neitomorphic algorithms projects are readily available. My friends werein that field. All the best!