TFM Toolbox - About · TFM or AFM method that allows correction of the geometry irregularities...

OEM Functions for Advanced ImagingTFM Toolbox

[email protected]

Uses:

Toolbox for programming (C, C++, C#, Matlab, Python, etc.), in the shape of a DLL

Fast imaging using Nvidia GPU parallel computing (N.B. Nvidia GPU is mandatory)

Usable with real-time, open phased-array hardware (typically AOS OEM-PA)

Usable without hardware for post processing analysis in any

Windows environment

Compatible with various, constantly evolving acquisition schemes

(SAFT, FMC, etc.)

Perform your own FMC/TFM with the best featured program

using this DLL and the highest performance on the market

Features:

No image size limitation

2D/3D computations

Free selection of the resolution of the pixel/voxel

Linear/matrix probes

Pulse-echo/pitch-catch configurations

Flat/cylindrical pieces (any for Adaptive TFM)

Direct/Indirect/Corner modes (LL, L-LL, LL-LL, TT, T-TT, TT-TT)

Conversion modes (LTL, LLT, etc.)

About 100 configurations

Methods:

Delay and Sum (DAS) imaging methods (SAFT, TFM, AFM)

Migration imaging methods (SAFT, TFM, AFM)

Surface adaptive methods (in option)

High-level inverse methods: SAFTp, TFMp, AFMp (in option)

Envelope functions

Experts in Phased Array Ultrasound

TECHNICAL SIDE OF DLL

These code examples demonstrate how to use the TFM DLL in Matlab and C/C++. Our TFM DLL is designed to have the following proper-ties:

Extremely fast image reconstruction. The GPU-accelerated library is fully optimized to allow decent frame rate in real-time appli cations.

User-friendly APIs. The library brings easy- to-use APIs in Matlab and C/C++. Develop ers can employ our DLL in their existing ap plications with minimum modifications.

Complete configuration support. Our library provides full support of conventional scan and imaging configurations, as well as new migration-based imaging and adaptive imag ing for complicated/unknown surface geome try. Low level APIs are also provided for de velopers with knowledge in GPU computing.

...... % Inputs initialization % Load the dll % TFM('LoadLibrary','mxTFM.dll','TFM.dll') % Call the TFM function % pO = mxCall_TFM_PE_con_2D_flat(pData,Fs,... Nt,t0,Ne,pXe,pYe,pZe,Nz*Nx,pX,pY,pZ,c,H); % Envelope extraction % pOEnv = mxCall_envelope_columnMajor(pO,Nz,Nx); % Reshape the image % pOEnv = reshape(pOEnv,Nz,Nx); % Display the TFM image % imagesc(x,z,pOEnv); ...... % image display settings % Free the dll % TFM('FreeLibrary');

// Header and import library for the dll #include "TFM.h" #pragma comment(lib,"TFM.lib") ...... // Inputs initialization // Call the TFM and envelope functions if (useLowLevelAPI) {

DLL_CUDA_TFM_PE_con_2D_flat(dData,Fs, Nt,t0,Ne,dXe,dYe,dZe,Nz*Nx,dX,dY,dZ,c,H,dO);

DLL_CUDA_envelope_columnMajor(dO,Nx,Nz,FFT_Plan,dOEnv); } else {

DLL_TFM_PE_con_2D_flat(pData,Fs, Nt,t0,Ne,pXe,pYe,pZe,Nz*Nx,pX,pY,pZ,c,H,pO);

DLL_envelope_columnMajor(pO,Nx,Nz,pOEnv); } ...... // Further image display

Matlab Example:

C/C++ example:

You can use any TFM software you want. We have developed various and very effective, highly optimized TFM algorithms for you to implement.

Enjoy!


AFM

Shown in Image: Lack of Fusion Flaw in Weld

Steel Plate with Weld

AFMUses different waves from generic TFM, possibly different frequencies, this technique allows high speed, low data file size, excellent SNR, and in the range of 20 to 30 dB more sensitivity than Generic TFM. Typically more than 20 frame per sec.

AFM

AFM

Non Surface Breaking (1mm from the surface) Flaw

Shown in image:Inspection of crack in aluminum bar using 64 element 5 MHz probe

Surface Breaking Closed Crack Height < 3 mmAFM


ATFM

Curved Steel Pipe Section With Flaw shown with Flexible Wedge

Adaptive method applied on TFM or AFM

Adaptive methodTFM or AFM method that allows correction of the geometry irregularities without knowing the profile of the surface or the back wall. The FMC data provides all necessary information to correct the TFM or AFM processing in real time.

Real Time Detected Surface

Adaptive method on TFM and AFM

ATFM AAFM

Real Time Detected Surface

Steel Pipe With SDH In Immersion Tank

Adaptive method on TFMp

TFMp

Aluminum Block with Close SDH

TFMpTFM method based on the inverse problem that works with iterations, and gives improvement of the resolution from 1/10th to 1/20th of a wavelength. An image is processed in a few seconds.

Wavelength : λ = 2.2mm Depth = 40 mm

TFMpTFM

TFM and TFMp results obtained from the same FMC data file

Real Time Detected SurfaceATFMp



AFM

Generic TFM

Comparison of different TFM types

SAFT

FMC/TFMGeneric TFMThis recent, popular technique offers similar resolution to SAFT, but with a very good SNR. Generic TFM can be implemented in hardware or in software. The amount of data is very high and the speed is limited.Typically less than 10 frames per sec.

SAFTThis is an old technique from before phased array was technologically dupable, but can be applied to phased array. Fast, the resolution is excellent, but the SNR is very poor.Typically more than 20 frames per sec.

Below are a few different examples of types of TFM. There are many more types available. TPAC offers most TFM types available on the market.

SAFT

Migration TFM

AFMAFMAFM is a TFM process that works using different waves from cylindrical as well as multi-frame FMC-like process.The actual reduction of data and increased sensitivity by a factor of more than 10dB makes it very useful for most applications.

Test Block

Aluminum BlockWith SDH

Conventional Phased Array

Migration TFM MigrationMigration is a TFM process that can be applied to any FMC data where the probe is flat and parallel to the surface. Provides better contrast and resolution than generic TFM.Real time.

From File


SUMMARY OF BENEFITS AND CONSTRAINTS

Pulse Echo Wave Modes

Values are based on FMC/TFM acquisition using OEM-PA device from AOS

Pitch Catch Wave ModesAlso Available

TFM Type Speed * Data Size ** SNR Sensitivity *** Resolution ****SAFT High >20Hz Small 610 Poor 0dB Very Good ~0.5TFM Average ~5Hz Very Large 39,063 Good -‐6dB Very Good ~0.5AFM High ~20Hz Small 1,831 Good +20dB Very Good ~0.5ATFM Below Average 2Hz N/A N/A Good N/A Very Good naAAFM Above Average 10Hz N/A N/A Good N/A Very Good naTFMp Below Average 0.5Hz N/A N/A Excellent N/A Excellent <0.1

Table Legend* 128 ch, 512x512 grid** 128ch, 50µs, 5MHz, 2m with 2mm step*** Comparative sensitivity (+ is better than -‐)**** In wavelength

TFM Type Speed *SAFT High ~400HzTFM Average 12HzAFM High 400HzATFM Below Average 5HzAAFM Above Average 120HzTFMp Below Average 2Hz

TFM Software Services

What can TPAC do for you? • Customize TFM software to fit your application • Analyze FMC data • Help you design the best PA probe for TFM • Feasibility Study: Scan your parts so you can see the benefit of TFM

Linear Probe with TPAC wedge


About TPACCustom Software• Developing the exact application you want• Phased Array and Conventional UT• Solutions ranging from simple to complex

Custom Probe Design• Phased Array (Linear, Matrix, Annular, Pitch/

Catch)• Simulation, Manufacturing, Characterization• Selecting the right probe for your application

Training and Support• Teaching inspection techniques to technicians• Supporting all needs even if that means going

out into the field• Direct line to over 25 years of experience in

providing NDT services

Consulting• Deciding between phased array or

conventional ultrasonics • Determining the best equipment for the job• Utilizing our network of UT equipment

suppliers to get the best value

Equipment Customization• Integrating UT technology into a pre-existing

system• Building custom enclosures for any

environment

Matrix Probe with Flexible Wedge


9365 Allen Rd. West Chester, Ohio 45069, USAwww.ThePhasedArrayCompany.com

+1 (513) 785-0801

[email protected]

T PAC

TFM Toolbox - About · TFM or AFM method that allows correction of the geometry irregularities...

Documents

Transcript of TFM Toolbox - About · TFM or AFM method that allows correction of the geometry irregularities...