touchExtr

These slides present a fast capacitance extraction platform indented to obtain characteristics of mobile capacitive touch screens. The tool was built using COMSOL Multiphysics, showing good match to the experimental data and automated the process.

Transcript

1. Research experience Oleg Żero Investigation of edge-effects. Developing fast capacitance-extraction platform for touch-screen applications. CTO technology group FFFFF Corporation. ?
2. Research experience Oleg Żero challenge idea / strategy new skills
3. A part of FFFFF ’s portfolio was FFFFF – touch detection platform using a microcontroller unit. All hardware and software was in place, however, Understanding of underlying physics was insufficient! One year as application engineer Oleg Żero Before moving on to the project full-time. +5V local (PCB) ground C0 Human – Earth coupling Touch – intended coupling Circuit – Earth coupling (grounding) Human coupling to local ground (unintended) DC picture That was opportunity!
4. One year as application engineer Oleg Żero I was doing ”undercover” research. 0 10 20 30 40 50 60 70 80 90 100 0 10 20 30 40 50 60 70 80 90 100 sensitivity S[%] 23 24 25 26 27 28 29 30 31 32 33 34 0,0 0,1 0,2 0,3 0,4 0,5 0,6 0,7 0,8 0,9 1,0 sensitivity S[%] ’squareness’ parameter p [a.u.] Influence of the electrode’s shape on sensitivity Fill factor relative to 1 cm2 area footprint – FF [%]
5. Building FFFTouch Extractor Oleg Żero It turned out the whole corporation has a problem... Drawing each time a new pattern is drawn Solving time Drastic scaling with model size. From ~1.5 h to 1 week on large problems. Setting up required to update every boundary coondition, assign the parameters... Pre-processing Manual work Resources Using 48 CPUs on LSF not uncommon. Computation Data collection Large amount of data to handle. Formulating plots, graphs and reports takes time. Data analysis No inspection of physical qunatities such as electric fields or potential. Could they give us some hints? License cost Puts further limits on concurrent users. Post-processing Manual work 1-2 days 5-8 h ~1 day
6. Building FFFTouch Extractor Oleg Żero And I began to be the solution provider. I created a quick extraction platrom for anyone to use.
7. Building FFFTouch Extractor Oleg Żero It demonstrated to be useful. I created a quick extraction platrom for anyone to use. APP Server APP APP model LSF 1 license per seat up to 4 apps per seat 1 license – unlimited CPUs new app / update 1 full FNL Deployment scheme: APP Client Standard FNL NOK 169 900 GBP 13 761 AC/DC Module NOK 89 900 GBP 7 281 LiveLink to MATLAB NOK 35 900 GBP 2 908 Standard (FNL): NOK 259 980 GBP 21 059 Application: NOK 25 980 GBP 2 106 unit license list price per year:
8. Building FFFTouch Extractor Oleg Żero We started observing new patterns and behaviours. Stylus-X: A B Γ Δ A A B Γ Δ A Stylus-Y: A B Γ Δ A A B Γ Δ A A Γ Δ We can observe shift in sensitivity peak (edge-to-edge interaction). needle stylus needle stylus needle stylus Shaft Diam. 0.1 mm 3 mm Tip Diam. 0.05 mm 2.5 mm Tip Angle Z distance 45 deg 0.05 mm 45 deg 0.05 mm needle stylus X1 Y1 A B Γ Δ
9. Building FFFTouch Extractor Oleg Żero Got my colleagues to conduct experiments together. Sample (Stig Larssen) Measurements (Steinar Myren) Model 275 150 150 4455 4455 X-line Y-line Experiment
10. Building FFFTouch Extractor Oleg Żero My colleagues helped to conduct experiments. 4x4 – pattern using all approximations Practically no degradation to the results! All lines overlap. Self-Cap measurement and extraction comparison Self-Cap measurement and extraction comparison
11. Oleg Żero What have I leant? Impact on my philosophy. • We get paid by bringing value to the market place. • I definitely became a better planner. • Research seems only to make sense when you enjoy it.... the rule is confirmed. Impact on my skills. • Understanding the importance of customer focus. • I got even better at programming, especially MATLAB, but also java, python, C, ... • I got even better at COMSOL. • I fell in love with Linux. 2010 2013 2014 2016 Building FFFTouch Extractor