The 21 st Asian Test Symposium 2012 Toki Messe Niigata Convention Center, Niigata, Japan 21/Nov./2012 Two-Tone Signal Generation for Communication Application ADC Testing K. Kato, F. Abe, K. Wakabayashi, C. Gao, T. Yamada, H. Kobayashi, O. Kobayashi, K. Niitsu Gunma University Semiconductor Technology Academic Research Center
2/32 Outline Research Background Conventional Method Proposed Method Experimental Results Conclusion
3/32 Outline Research Background
Cost (Cents / Transistor) Research Background Mixed signal SoC, communication devices Analog digital converter (ADC) is key component Manufacturing cost Test cost Lower Higher Need for low cost test Year Cost of test trend (ITRS 2001) 4/32
AWG : Arbitrary Waveform Generator SFDR : Spurious-Free Dynamic Range 5/32 Research Goal Low distortion two-tone signal generation for communication application ADC testing with low-cost AWG Conventional method Proposed method Low-cost AWG IMD3 ADC DUT Low-cost AWG DSP program change ADC DUT Low SFDR High SFDR
Power Power Need for Two-tone Test Signal Communication application ADC Receive narrow band & high frequency signal Input Output ADC Single-tone test signal Analog filter DUT Two-tone test signal Analog filter Output spectrum Cannot test nonlinearity of ADC Can test nonlinearity of ADC 6/32
Two-Tone Signal Generation with AWG AWG DSP D in 01110 Nonlinearity DAC Y Nonlinearity ADC DUT High IMD Low quality test DAC : Digital Analog Converter IMD : Inter Modulation Distortion 7/32
Distortion by Nonlinearity DAC Ideal DAC Actual DAC Y = a 1 D in Y = a 0 + a 1 D in + a 2 D 2 in + a 3 D 3 in + Analog filter IMD3 D in DAC Y 2f 1 -f 2 f 1 f 2 2f 2 -f 1 2f 1 f 1 +f 2 2f 2 Cannot remove IMD3 with analog filter 3f 1 2f 1 +f 2 2f 2 +f 1 3f 2 8/32
9/32 Outline Conventional Method
Conventional Method AWG Y = a 1 D in + a 3 D in 3 DSP Input D in DAC Output Y CLK CLK D in X X X X X = A sin 2πf 1 nt s + B sin 2πf 2 nt s 10/32
11/32 Conventional Method Y = a 1 D in + a 3 D in 3 DSP Input D in DAC Output Y CLK IMD3 components appear IMD3 2f 1 -f 2 f 1 f 2 2f 2 -f 1 3f 1 2f 1 +f 2 2f 2 +f 1 3f 2 around fs/2
12/32 Outline Proposed Method
13/32 Proposed Method Phase Switching Frequency Switching Combination of Phase & Frequency Switching
14/32 Phase Switching Y = a 1 D in + a 3 D in 3 DSP Input D in DAC Output Y CLK CLK D in X 1 X 2 X 1 X 2 X 1 = A sin 2πf 1 nt s + π 6 + B sin 2πf 2nT s π 6 X 2 = A sin 2πf 1 nt s π 6 + B sin 2πf 2nT s + π 6
HD : Harmonic Distortion 15/32 Phase Switching Effects CLK D in X 1 X 2 X 1 X 2 Y = a 1 D in + a 3 D in 3 DAC Output Y Cancel IMD3 components Switching spurious appear IMD3 HD3 2f 1 -f 2 f 1 f 2 2f 2 -f 1 3f 1 2f 1 +f 2 2f 2 +f 1 3f 2 around fs/2
CLK D in Principle of Phase Switching X 1 X 2 X 1 X 2 X 1 = A sin 2πf 1 nt s + π 6 + B sin 2πf 2 nt s π 6 X 2 = A sin 2πf 1 nt s π 6 + B sin 2πf 2 nt s + π 6 Y = a 1 D in + a 3 D in 3 DAC Output Y Im +π/6 π/6 Re Phase difference π/3 Fundamental component DAC Y = a 1 D in + a 3 D in 3 Im 3 Re 3 Phase difference 3 π 3 = π Distortion component 16/32
17/32 Frequency Switching Y = a 1 D in + a 3 D in 3 DSP Input D in DAC Output Y CLK CLK D in X 1 X 2 X 1 X 2 X 1 = A sin 2πf 1 nt s X 2 = B sin 2πf 2 nt s
18/32 Frequency Switching Effects X 1 Y = a 1 D in + a 3 D in 3 f 1 DAC Output Y X 2 f 2 IMD3 components not appear IMD3 2f 1 -f 2 f 1 f 2 2f 2 -f 1 3f 1 2f 1 +f 2 2f 2 +f 1 3f 2 around fs/2
19/32 Combination of Phase & Frequency Switching Y = a 1 D in + a 3 D in 3 DSP Input D in DAC Output Y CLK CLK D in X 1 X 2 X 3 X 4 X 1 = A sin 2πf 1 nt s + π 6 X 2 = B sin 2πf 2 nt s π 6 X 3 = A sin 2πf 1 nt s π 6 X 4 = B sin 2πf 2 nt s + π 6
20/32 Phase & Frequency Switching Effects Canceled by frequency switching Y = a 1 D in + a 3 D in 3 D in Y DAC 2f 1 -f 2 2f 2 -f 1 2f 1 +f 2 2f 2 +f 1 IMD3 f 1 f 2 3f 1 3f 2 around fs/4 around fs/2 Canceled by phase switching
21/32 Merit of Proposed Methods Program change DSP CLK CLK D in Nonlinearity DAC D in No need for the DAC nonlinearity identification No need for calibration Only DSP program change No hardware change
Simulation Condition 3 Conventional Y = D in 0.005D in Proposed Input D in Output Y Phase switching DAC Frequency switching Phase & Frequency switching 12bit Input frequency f 1 99 Input frequency f 2 122 Amplitude 1 Sampling frequency 4000 Simulation by Matlab 22/32
Power [dbm] 2f1-f2 2f2-f1 3f1 3f2 2f1+f2 2f2+f1 Phase, Frequency, Phase Freq. Switching Conventional 0-60 -120 Phase Switching 0-60 -120 HD3 Around fs/2 Spurious Frequency Switching Phase & Frequency Switching 0-60 -120 0-60 -120 IMD3 Around fs/4 Spurious 0 0.1 0.2 0.3 0.4 0.5 Normalized Frequency f/fs 23/32
24/32 Outline Experimental Results
25/32 Experimental Condition AWG Signal Spectrum analyzer Agilent 33220A Frequency range 1uHz to 6MHz Amplitude resolution 14bit Max sampling rate 50MSa/s ADVANTEST R3267 Frequency range 100Hz to 8GHz RBW 10Hz to 30MHz RBW : Resolution Band Width Test Signal Two-tone signal Sampling rate Input voltage 200kHz, 220kHz 10MSa/s 0.8 to 2.0V pp (0.2V step)
Power [dbm] 6 3 Experimental Results Fundamental Conventional Frequency Switching 0 3 6 50 70 IMD3 Phase, Phase Freq. Switching Conventional Phase Switching Frequency Switching 90 Ave. Reduction : 12.3dB Phase & Frequency Switching 0.6 1.0 1.4 1.8 2.2 Input Voltage [V pp ] 26/32
SFDR [dbc] 27/32 SFDR Improvement 90 80 Conventional 70 60 50 0.6 1.0 1.4 1.8 2.2 Input Voltage [V pp ] Phase Switching Frequency Switching Phase & Frequency Switching +12.5 db +10.6 db +12.4 db Phase Switching Frequency Switching Phase & Frequency Switching
Phase Switching Power [dbm] Conventional Phase Switching 0 50 2f 1 -f 2 f 1 f 2 2f 2 -f 1 2f 1 +f 2 2f 2 +f 1 3f 1 3f 2 100 0 f 1 f 2 50 2f 1 -f 2 2f 2 -f 1 3f 1 3f 2 100 180 200 220 240 200 400 600 Frequency [khz] 28/32
Frequency Switching Power [dbm] Conventional Frequency Switching 0 50 2f 1 -f 2 f 1 f 2 2f 2 -f 1 2f 1 +f 2 2f 2 +f 1 3f 1 3f 2 100 0 50 2f 1 -f 2 f 1 f 2 2f 2 -f 1 2f 1 +f 2 2f 2 +f 1 100 180 200 220 240 200 400 600 Frequency [khz] 29/32
Phase & Frequency Switching Power [dbm] Conventional Phase & Frequency Switching 0 50 f 1 f 2 2f 1 -f 2 2f 2 -f 1 2f 1 +f 2 2f 2 +f 1 3f 1 3f 2 100 0 50 2f 1 -f 2 f 1 f 2 2f 2 -f 1 2f 1 +f 2 2f 2 +f 1 3f 1 3f 2 100 180 200 220 240 200 400 600 Frequency [khz] 30/32
31/32 Outline Conclusion
32/32 Conclusion For low distortion two-tone signal generation with low-cost AWG Proposed 3 methods No need for the DAC nonlinearity identification Only DSP program change No hardware change, no calibration Effectiveness of our proposed methods Verified by simulation, and experimental results IMD3 reduction and SFDR improvement
Question Q : 今後の課題 どのようなアプリケーション Q : 3 トーン 4 トーンに応用すると分解能はどうなるか C : IMD3 の取り方はいいが SFDR の取り方が良くない スカートを持つような図にするべき 33/32
Presented by Keisuke Kato 34/32