RITEC RAM-5000

amplifier has been turned off. The ratio of thepulse amplitude to the CW leakage through theamplifier (On-Off ratio) is greater than 140 dB.This high On-Off ratio is important because avery small amount of leakage can overwhelmthe receiver input at high gains and causedistortions in the outputs of the quadraturephase sensitive detectors.This reduction in leakage of the CW signal is amajor advantage of using an amplifier that isgated as opposed to an amplifier that operatescontinuously amplifying externally produced lowlevel RF bursts. Another advantage is that theamplifier has been turned off and is no longeramplifying any noise that may present on thelow level CW input, thus improving the signal-tonoise. Other advantages are the economiesrealized in size, heat dissipation, and powersupply requirements.Because the amplifier is gated on and off, animportant factor to consider when determiningthe burst width and burst repetition rate ortrigger frequency is the duty cycle. This isdefined as the time the amplifier is turned on,specifically the burst width multiplied by therepetition rate, expressed as a percentage. Forexample, as shown above in Figure 3, at anoperating frequency of 350 kHz a burst width offive cycles results in a burst 14 microsecondslong. At a repetition rate of 100 Hz, the dutycycle would be 0.0014 or 0.14%. At a repetitionrate of 1kHz, the duty cycle would be 1.4%.difference between meaningful data and noobservable signals.Single Channel Frequency SynthesizerThe frequency source is provided through adirect digital synthesizer that providesinstantaneous changes in frequency as contrastto the settling times required in phase lockedlooped synthesizers. The synthesizer operatesup to 40 MHz in the broadband Mark III andMark IV configurations and up to 64 MHz in thesuperheterodyne Mark VI configurations. Thesynthesizer accepts an external clock signal atfrequencies up to 25 MHz and processesinternally resulting in a system clock as high as300 MHz.Frequency control is providedthrough 48-bit control, resulting in a typicalresolution of up to 0.5 x 10-6 hertz.In the broadband Mark III and Mark IVconfigurations, the synthesizer output is sent tothe timing circuitry and to the gated amplifier forgating and amplification. With the superheterodyne receiver in the Mark VI systems, theoutput of the frequency synthesizer is used as a"local oscillator" in the superheterodynereceiver. It also is mixed with the quartzintermediate frequency oscillator to produce theCW operating frequency that is sent to the gatedamplifier and the timing circuitry.Signal Channel TimingNo appreciable sag in output amplitude willoccur as the duty cycle limits are approached atfull power. If the limits are exceeded, however,an automatic shutdown circuit will be activatedand the power stage will be turned off withoutdamage. (Full power output can be restored byswitching the high voltage off, correcting theexcessive duty cycle, and turning the highvoltage on again.)All the digital timing functions of the RAM havebeen made coherent with respect to the CWoperating frequency. When a trigger is receivedfrom the computer, an external source, or theinternal rep-rate generator, the timing circuitrywaits for a positive zero crossing of the CWsignal before beginning the gating process. TheRF cycles are then counted to the desirednumber to produce a burst with the correctwidth.The high power available makes it possible todrive very low efficiency transducers such asEMATs and still have reasonable systemperformance. In addition, when materialsexhibiting high ultrasonic losses are to beexamined, the use of high power in conjunctionwith efficient transducers can mean theAt the same time the gating function begins, atrigger pulse is sent to the trigger outputconnector and a 10 MHz gated clock oscillator isstarted. This clock is thus made coherent withrespect to the CW and is used to generate thesignal processing gate delays and widths. Thecoherency feature avoids the gate position jitterRITEC, Inc. ) 60 Alhambra Rd., Suite 5 ) Warwick, RI 02886(401) 738-3660 ) FAX (401) 738-3661

that would be evident if a continuous clock wereused.Multiple Gated Amplifier SystemsIn some applications, the unique capabilities ofthe RAM-5000 system to accept multiple gatedamplifiers are particularly valuable. The abilityto widen the frequency range of a system withtwo gated amplifiers has already beenmentioned. This permits one RAM-5000 systemto cover the frequency range of 50 kHz to 7MHz, maintaining the 5 KW output, between twogated amplifiers. Another configuration, with upto four gated amplifiers, allows the system to fireseveral transducers simultaneously or in adesignated sequence.Multiple Channel SystemsWith the addition of a special triplesynthesizer/timing module, it is possible toconfigure a three channel Mark IV system, withthree gated amplifiers driving three transducersand three independently controlled receivers.Such a system has been particularly useful inonline industrial applications using electromagnetic acoustic transducers (EMATS). Thesetransducers are very versatile but have a lowefficiency in converting the RF electrical energyinto acoustic energy. With the high power, highcurrent output of the RAM gated amplifier andwith proper impedance matching in the transmitline, it is possible to produce tens of amperes ofcurrent in the EMAT.With the special triple synthesizer/timingmodule, it is possible to adjust the relative delaybetween the outputs of each gated amplifier in12.5 ns steps, with further control over therelative phases by adjusting the phase of thesynthesizer outputs. Unique timing circuitrymaintains coherency between each of thesynthesizer outputs and therefore between thehigh power RF bursts.Superheterodyne ReceiverSuperheterodyne circuitry has been used innearlyeveryradio-frequencyreceivermanufactured since the 1930s. The conceptallows the use of fixed tuning elements at theintermediate frequency (IF), a constantbandwidth independent of the operationfrequency, and rejection of out-of-band spurioussignals. In RITEC's implementation, theBroadband RF Receiver, Mixer and IF Amplifier,Direct Digital Synthesizer, and IF Oscillator andQuadrature Phase-Sensitive Detectors modulesare used in combination to complete the phasesensitive superheterodyne receiver. Selection ofthe transmitter frequency and receiver tuningare accomplished simultaneously by setting thesynthesizer. This feature greatly simplifies thecomputer control of the receiver, especially inapplications requiring the frequency to be swept.The adjustment of the phase sensitive detectioncircuitry is also simplified because themultiplication process occurs at the fixed IFfrequency. Specifically, the reference outputsfrom the IF oscillator are adjusted as closely aspossible to quadrature (90º) before beingapplied to the phase sensitive detectionmultiplier circuits.Broadband ReceiverThe digital control of gain in two dB steps makespossible accurate, repeatable measurements ofsignal strength, and low noise, high gainexternal pre-amplifiers are available for usewhen weak signals are encountered. Theaddition of adjustable IF band-pass, high-pass,and low-pass filters make this one of the mostversatile receiving systems available forultrasonic research.Quadrature Phase Sensitive Detection andAnalog IntegrationOne of the significant innovations of the RAM isa signal processing technique that involves theuse of both quadrature phase sensitivedetection and gated analog integrators.Quadrature detection allows signals to beprocessed in much the same way as is donewith a vector voltmeter. The two detector(multiplier) circuits produce two orthogonalvector components (real and imaginary) of theRITEC, Inc. ) 60 Alhambra Rd., Suite 5 ) Warwick, RI 02886(401) 738-3660 ) FAX (401) 738-3661

signal from which the amplitude and phaseangle can be calculated.To gain more insight into the operation of thephase-sensitive receiver circuitry, it is instructiveto consider the received ultrasonic informationas a monochromatic signal at frequency (fr)modulated by a term, which defines the width,amplitude, and shape of the received signal,Ar(t).Therefore, signal-averaging techniques may beemployed to recover signals from noise. InFigure 7, the signal to noise ratio of thebroadband receiver output was purposelyreduced to approximately one, but it is relativelyto detect the two echoes in the phase detectedsignals with a reasonable signal to noise ratio.Advantages of Quadrature Phase Detectionf(t) Ar (t) sin(2π f r t φ r )(1)Phase Detected Signal (50 kHz Low Pass Video Filter)D1 (t ) g 3 Ar (t) sin φ rvoltage (millivolts)400After conversion to the intermediate frequency,the received signal can be processed in thephase detectors. The output of the PhaseDetector No. 1 is given by:Phase Detected Signal (250 kHz Low Pass Video Filter)300Phase Detected Signal (2 MHz Low Pass Video Filter)200100RF Signal from Receiver RF Monitor(2)0where the total gain and conversion efficienciesare all included in the term g3. The output ofPhase Detector No. 2 is given by:D2 (t ) g 3 Ar (t) cos φ r(3)The two phase detector outputs are shownbelow in Figure 6, along with the broadbandreceiver echo.Phase Detector Outputs (mV)200100010601618In order to obtain accurate signal amplitude andphase information, the phase-detected signalsare processed with analog integration circuits.This method has the value of making the gatepositionnon-critical,removingtheRFcomponents, and improving the signal-to-noiseratio. The time limits of the integration arecontrolled by the integrator gate, and theintegrate rate (rI) is under computer control. Thegate is positioned so that it begins before andends after the signal. The integrator outputs aregiven by:t2t2t1t1I 1 r I D1 (t )dt A(t ) cos φ r dt657075Time (microseconds)Broadband receiver monitor and the two phasedetector outputs.This type of detection also has the advantage ofmaintaining excellent linearity even when thereceived signal is small; this type of detection isalso not affected by the presence of noise.(4)andt2t2t1t1I 2 r I D2 (t )dt A(t ) sin φ r dtFigure 620Figure 7Phase Detector 1Phase Detector 2Digitized RF Echo5514time (microseconds)-100-20012(5)where I1 and I2 are the outputs of integrators No.1 and No. 2, t1 and t2 are the start and stoptimes defined by the integrator gate, and D1(t)and D2(t) are the phase sensitive detectoroutputs. The integrator output voltages are readRITEC, Inc. ) 60 Alhambra Rd., Suite 5 ) Warwick, RI 02886(401) 738-3660 ) FAX (401) 738-3661

by internal 16-bit analog-to-digital convertersinternal to the RAM and then transmitted back tothe data acquisition computer. The phase angleof the received signal can be calculated from thetwo integrator outputs by: Iφ r tan -1 2 I1 (6)and the signal amplitude may be obtained from:A I 12 I 22 .been achieved. Unfortunately, steady stateconditions in the outputs of the phase detectorsare sometimes difficult to achieve as shown inthe typical situation shown in Figure 9.In this example, the RF signal (echo) wasproduced by transmitting a single cycle RFpulse to the transducer. It is clear that filteringout the RF components in the detected outputswill result in significant degradation of the riseand fall times.(7)Detection & IntegrationPhase Angle 0 degFurther improvement in the signal to noise canbe achieved using the gated integrators. Theintegrator outputs for the signals shown inFigure 6 are shown below in Figure 8.RF SignalVoltage0Unfiltered90 deg Phase Detector (M2)090 deg Integrator (I2)Unfiltered600Integrator Outputs (mV)0 deg Phase Detector (M1)0 deg Integrator (I1)400020002468101214161820Time0Figure 9Integrator Output No. 1Integrator Output No. 2-2005060708090100Time (microseconds)Figure 8Outputs of the Gated Integrators for the RFEcho and phase detected signals shown inFigure 6.However, if the signals are a very small numberof RF periods wide, measuring the magnitude ofthe vector components has some experimentaldifficulties. Typical signals are not flat toppedand the peak value may not be the best choicefor an accurate measurement of the echoamplitude. In an ideal situation, the RF termscan be completely filtered out, and theamplitude (A) and phase (φr) may be calculatedafter measuring the instantaneous value of thephase detected outputs at the center of thesignal or where steady state conditions haveOutputs of the RF Receiver, the QuadraturePhase Sensitive Detectors, and the gatedIntegrators for a Typical Signal with φr 0oThe outputs from the two integrators are alsoshown in Figure 9. Even though it is difficult todetermine the steady-state amplitude in thephase-detected signals, it is easy to determinethe steady-state amplitude for the outputs of thetwo phase detectors. In this example, theoutput of the 0 degree phase detector integratesto a maximum value and the output of the 90degree phase detector integrates to zero,indicating a phase angle, φr, of 0 degrees.In order to illustrate the effectiveness of thissignal-processing scheme and to determine ifattenuation measurements would be affected byburst width, a simple test was performed with aPlexiglas sample. The logarithmic ratio of theRITEC, Inc. ) 60 Alhambra Rd., Suite 5 ) Warwick, RI 02886(401) 738-3660 ) FAX (401) 738-3661

amplitude of two echoes, which is a measure ofthe attenuation, was measured as a function offrequency for a number of different transmitterburst widths. The integrator gate width andposition remained constant during theinvestigation. The results are shown in twodifferent formats in Figure 10.34Amplitude (dB)32301 cycle burst2 cycle burst3 cycle burst4 cycle burst5 cycle burst6 cycle burst7 cycle burst8 cycle burst9 cycle burst2826240.820.921.021.12Frequency (MHz)1.22burst width of a single cycle. However, the datataken with a single cycle burst is more sensitiveto noise and has a poorer signal-to-noise ratio(SNR) as expected.The case for phase sensitive detection andintegration is made even stronger in thisexample because the spectrum of the receivedsignal produced from a short burst is dominatedby low frequency components. This is truebecause there is always some low frequencycomponent in the driving pulse and theattenuation is relatively low at this end of thespectrum. The phase sensitive detection andintegration process is very effective in rejectingthese components. In fact, it can be shownmathematically that the quadrature phasesensitive detection and integration processproduces an amplitude equal to the value of theFourier Transform at the transmit frequency.The conclusion then is that this signalprocessing technique is the best method knownto us for obtaining either amplitude or phaseinformation from pulsed acoustic signals.Evaluation TestsIn order to illustrate the capabilities of theRITEC RAM system the results of a series ofsimple tests are presented.Measurements of the Absolute Transit TimeFigure 10Ratio of the Amplitude of Echo 1 to Echo 2 indecibels as a Function of Frequency and WidthThe figure shows that for each frequency aconstant value for the attenuation is obtainedwhich is independent of the pulse width. Theseresults could be repeated with a sample-andhold technique if the gate were placed within theflat-topped portion of a long echo and sufficientfiltering were used to remove the RFcomponents from the detected signal. Quitegood agreement can be found between the datataken with the maximum burst width of eightcycles and the data taken with the minimumAfter determining the slope of the phase versusfrequency curve for a signal, the total time ofarrival including delays through the electronicsand acoustic elements as well as the acoustictime of flight in the sample associated with thegroup velocity can be determined from therelation:T φ r2π F(5)where φr is the change in phase and F is thefrequency change.When two echoes are measured, an accuratemeasure of the acoustic transit time can beobtained by taking the difference between theresults and dividing by the appropriate numberRITEC, Inc. ) 60 Alhambra Rd., Suite 5 ) Warwick, RI 02886(401) 738-3660 ) FAX (401) 738-3661

of transits for the echoes chosen. In theexample shown in Figure 11, the first twoechoes were measured.may occur in the electronics or the acousticbond. However, the acoustic bond must remainstable over the course of the readings.In order to document the system's ability tomeasure small changes in signal arrival time, atest was devised using an adjustable length airdielectric coaxial line. Because the dielectric isair, the increase in time may be easilycalculated using the speed of light and thechange in line length. The block diagram of thetest is shown in Figure 12.RITECRAM-5000Figure 11Rec. InputRF Burst OutRF BurstAcoustic Phase Measurements in a Fused SilicaRod.A least squares data fit produced values for thephase versus frequency slopes of 61.533 and121.489 radians per MHz, and the acoustictransit time was than calculated as 4.771microseconds. It is not difficult to obtainbetter than four-place reproducibility in dataof this type. However, the researcher mustbe careful to include all relevant acousticeffects, such as diffraction and phase shiftsat the bond-transducer-sample interface, ifhe wishes to claim this level of absoluteaccuracy.Changes in Acoustic Transit TimeMany investigations are more concerned withchanges in acoustic velocity or time as afunction of some other parameter such astemperature or pressure than they are withabsolute times. These changes can also bedetermined with more accuracy and precisionthan absolute measurements. The calculationsare made from the relation:Change in Time φ r2π FHigh PowerDiplexer forPulse/echooperation6 dB 50 O

6000 7000 RMS Pulse Power (watts) into a 50 8000 Ω Load Typical RMS Pulse Power of High Power RAM-0.5-5 Specified RMS Pulse Power of High Power RAM-0.5-5 Typical RMS Pulse Power of Standard RAM-0.25-17.5 Specified RMS Pulse Power of Standard RAM-0.25-17.5 Figure 4 Available powers as a fun