Akinori Maeda Verigy Japan akinori.maeda@verigy.com
September 2011
Abstract of Part 1 The Total Harmonic Distortion (THD) is one of the major frequency domain parameters for verifying dynamic linearity characteristics of devices like Analog to Digital Converter (ADC) Digital to Analog Converter (DAC) Amplifier and so on! Dynamic linearity characteristics of devices are varied "y output frequency and evaluated results at lo# frequency does not sho# the characteristics at high frequency! Therefore THD must "e measured at actual operating frequencies! $easurement instrument to evaluate harmonic distortion needs to have lo# harmonic distortion and #ide "and#idth characteristics! %n general it is difficult to reali&e "oth characteristics! The part 1 introduced the "asic measurement characteristics of Digiti&er and 'ampler! %f 'ampler has superior harmonic distortion performance it is the "est instrument to measure harmonic distortion! And then it provided the "est measurement condition of 'ampler to measure correct harmonic distortion #ith e(perimental results! )ith attenuated measurement method descri"ed in this *art + these measurement conditions ena"le 'ampler to measure correct harmonic distortion!
Attenuated Measured Results As descri"ed in previous section to increase harmonic distortion performance of 'ampler attenuated measurement method should "e used! %n this section ho# much of attenuation should "e applied to o"tain good measurement result is discussed!
,H- A). 'ampler /*- Attenuator
Figure 12: Configuration of the measurement
The -igure 1+ sho#s the measurement configuration of attenuated measurement method! %n this configuration you never use the attenuator in A).! This is "ecause distortion characteristics of A). may change #hen attenuator setting of A). is changed! This means that distortion may occur after attenuation "lock of A).! $easurements #ere performed #ith 10d1 12d1 and +0d1 attenuators! Averaging must "e implemented to remove noise! The purpose of this measurement is harmonic distortion measurement and averaging does not affect harmonic distortion result! %t used 100 times of averaging in these measurements!
+ 1d! Attenuator The -igure 12 sho#s data at 30$H& sampling frequency #ith 10d1 attenuator! The condition of this graph is same as the -igure 11!
-rom this result +nd harmonics distortion is a"out 4 to 10d1 improved and this means that attenuated measurement method is actually effective to improve harmonic distortion performance! The -igure 13 sho#s one of spectrums of a"ove data! This spectrum sho#s that noise level of this measurement is 5100d1, 65110d1,! The fundamental amplitude of this measurement is a"out 517d1, and measured harmonic distortions "elo# 83d1c is under noise level and they are not relia"le results! To lo#er noise level it must increase num"er of average!
'pectrum 9 10d1 attenuator *eriod 101 51+0 5100 580 570 530 5+0 0 0 100 +00 200 300 :00 700 400 800 ;00 1000 -requency ($H&) A m p l i t u d e
( d 1 , )
Figure 1$: Spectrum at 1d! attenuator
1"d! Attenuator The -igure 1: sho#s data at 30$H& sampling frequency #ith 12d1 attenuator! The +nd harmonic distortion of this measurement is improved a fe# d1 than the measured result #ith 10d1 attenuator! 1ut harmonic distortions "elo# 80d1c are not relia"le results! Therefore no relia"le data #ill "e taken if the attenuator values are increased more than 12d1!
&ncrease 'umber of A(erage <o# num"er of average is increased from 100 to 1000! The -igure 17 sho#s the spectrum at this num"er of average! This graph sho#s that noise level at 1000 times of average is less than 5110d1,!
'pectrum 9 10d1 attenuator *eriod 101 51+0 5100 580 570 530 5+0 0 0 100 +00 200 300 :00 700 400 800 ;00 1000 -requency ($H&) A m p l i t u d e
( d 1 , )
Figure 1): Spectrum at 1 times a(erage #ith 1d! attenuator
The Ta"le 2 and the -igure 14 sho# measurement results #hile changing attenuation value from 0d1 to +0d1! %n these measurements num"er of average is 1000 and num"er of periods in data is 101!
3 Harmonic Distortions and Attenuator 51+0!00 5110!00 5100!00 5;0!00 580!00 540!00 570!00 0 + 3 7 8 10 1+ 13 17 18 +0 Attenuator ,alue (d1) A m p l i t u d e
( d 1 c ) +nd Harmonics (d1c) 2rd harmonics (d1c)
Figure 1,: Harmonic +istortions at se(eral Attenuator (alues
As sho#n in the -igure 17 noise level is a"out 5110d1, and all of 2 rd harmonic distortion results and + nd harmonic distortion result at +0d1 attenuator are "elo# noise level! +nd harmonic distortion is decreased proportionally #ith attenuation values! After 12d1 attenuation value +nd harmonic distortion is not changed proportionally #ith attenuation values! This means that these +nd harmonic distortions are not caused "y non5linearity of 'ampler! The Ta"le 3 sho#s average and standard deviation of +0 times measurements #hen attenuation values are 12d1 17d1 and +0d1! 1ecause the standard deviation of + nd harmonic distortion result at +0d1 is larger than the standard deviation at 12d1 or 17d1 it is clear that + nd harmonic distortion at +0d1 is "elo# noise level!
*able $: A(erage and Standard +e(iation at 1"d!- 1)d! and 2d! Attenuator Average 'tamdard Devisation Average 'tamdard Devisation Average 'tamdard Devisation -undamental (d1,) 51;!07 8!8:=503 5++!1; 2!;2=503 5+7!14 1!12=502 +nd Harmonic Distortion (d1c) 58+!12 1!17 58+!73 1!10 5;+!83 7!+2 2rd Harmonic Distortion (d1c) 5;2!04 2!77 5;1!03 3!37 584!80 3!:8 17d1 +0d1 12d1
The + nd harmonic distortions at 12d1 and 17d1 attenuation values are almost same and the standard deviations of these conditions are small! This means that measurement results at these conditions are correct! The measured results 58+!1d1c and 58+!7d1c are very close to the measurement result "y 'pectrum Analy&er that is 58:!8d1c!
Summary %nput signal must "e attenuated do#n to a"out 5+0d1, for correct measurement of harmonic distortions "y the 'ampler at 1!+, range! 17d1 attenuator should "e used! This increases the relative noise level and average must "e performed to reduce this noise level!
.erification >sing o"tained measurement conditions several harmonic distortions of A). output #ere measured! %n this measurement output frequencies #ere varied from +0$H& to 40$H&? and no /*- #ere used! -irst A). output signals #ere measured "y 'pectrum Analy&er! Then same A). output signals #ere measured "y 'ampler! $easured results "y 'pectrum Analy&er and 'ampler #ere compared! %n these measurements 17d1 attenuator #as used and the num"er of average #as 100! The Ta"le : and the -igure 18 sho#s measurement and compared results!
: *able %: A/0 1utput Measure Result #ithout 2PF
Figure 13: A/0 1utput Measure #ithout 2PF
The ma(imum delta of measured data "y 'ampler and 'pectrum Analy&er is 1!:d1 of 2 rd
harmonic distortion 9+0$H&! @verall "oth of measured results "y 'ampler and 'pectrum Analy&er are #ell matched and it sho#s that o"tained measurement conditions are correct!
Conclusion The =;4+7A 7.H& 'ampler can "e used to measure harmonic distortions and measured harmonic distortion are #ell matched to the measurement result "y 'pectrum analy&er! The measurement conditions to measure correct harmonic distortion "y 'ampler areA
'ignal Amplitude is a"out 5+0d1, 9 1!+, input range (17d1 attenuation) 'ampling frequency is close to input signal frequency <um"er of periods in data is less than :B of num"er of data