THE DYNAMIC RANGE POTENTIAL OF THE PHONOGRAPH By Ronald M. Bauman

his article describes a new transmission standards of even lower added to the quietest passages by the approach for analyzing the quality than our current CD standards. cartridge-preamplifier combination dynamic range of the phono- Unless these standards are dramatical- should be essentially inaudible. graphic playback system, in which the ly upgraded (in terms of information Similarly, the cartridge-preamp sys- cartridge and preamplifier are treated content), we may never have a source tem should be able to clearly repro- as an integrated system. I analyzed of music for our homes that sounds ducd the loudest sounds on record the dynamic range potential of several better than the phonograph. without distortion, compression, or combinations of phono cartridges and Are analog records inherently better clipping. preamplifier amplifying devices and in some sense? Your ears may already The same should be true of CD compared the results to CDs. be telling you that analog can sound playback. The quietest passages Additionally, I speculate about the better than today's digital. I will should be reproduced without added drawbacks of frequency domain char- provide quantitative reasons this may noise or distortion of the rnusic acterizations of musical audio compo- be so. caused by amplitude steps, or sam- nents and suggest that the time pling intervals that are too coarse, or domain may be a more natural frame Qualitative Requirements by filter phase shifts and ringing. The of reference for audio instrumentation The subtlety of detail in the grooves of loudest peaks encoded, as for analog development. an LP record is astounding. Quiet, yet records, must be reproduced without clearly audible, details (such as the distortions, compression, or clipping. lntroduction reverberation characteristic of Car- Why should anyone still be interested negie Hall) are represented by stylus Definitions in the analog phonograph? The reason motions of less than an ultraviolet For CD players, dynamic range is is simple: today's best record playback wavelength (1/100,000,000 of a essentially the ratio of the loudest pos- systems sound better than today's meter)-a dimension approaching the sible output signal to the quantization best CD players. (CD playback is size of a complex organic molecule. noise, i.e., the noise corresponding to improving, though maybe not faster This submicroscopic motion may be the round-off error of the least signifi- than turntable, arm, and phono car- superimposed on an orchestral cant bit. For the 16-bit standard CD tridge developments.) Moreover, crescendo involving stylus motions format, this is 96dB dynamic range. many analog record treasures are thousands of times larger. Over the To parallel the previous CD defini- unavailable on CDs or other available last ten years, the ability of modern tion, I define the dynamic range of a and anticipated digital music sources. cartridges, arms, and turntables to phono cartridge as the ratio of the After all, standards to transmit digital resolve this ultra-fine detail, particu- loudest sound to the background music (and video) directly to our larly during loud and complex music noise referenced to the preamp output homes are developing. This will open peaks, continues to improve beyond terminals. Crucial to this definition is a vast library of musical selections for all expectations. understanding that there is an inti- our potential enjoyment. Ideally, to reproduce the dynamic mate relationship between the car- I believe that commercial considera- range inherent in the groove modula- tridge and the first preamplifier stage. tions and the limitations of today's tion of a fine LP with the least possible Each interacts with the other to estab- digital technology will establish digital degradation, the background noise lish dynamic range; neither the car- tridge nor the preamplifier stand alone. Oddly enough, the preceding defin- ition is not the conventional one,

ABOUT THE AUTHOR

Hon Bauman received his BSEE from Lehigh University. He is founder and owner of inSound, which produces high- end audio amps, preamps, and, in partnership with Mapleshade Electronics, the Omega Mikro line of digital and analog interconnect, power, and speaker cables. Ron also specializes in personal and meteor burst communica- tion systems design. With his wile Marcia, he is raising his FIGURE 1: Simplilied noise model of phono cartridge and preamplifier. sons Marcus and Mitchell. His hobbies include music, beer-making, llying, and restoring old Volvos. l6 Audio Electronics 4/96 which is simply signal-to-noise ratio equivalent circuit parameters of the age of the cartridge resistance in volts; (S/N) at some arbitrary cartridge out- cartridge. For the above three car- e, is the equivalent input RMS put lcvcl wcll bclow thc louclcst. tridgc.s, thc significarrt cir.ctrit paralnc- voltagc noisc gcncrator of the pream- Often, preamplifier noise is character- ters are the series resistance and plifier in volts; ized with the input shorted, which, as point2 inductance. The values are: Blue i* is the equivalent input RMS cur- you will see, is an incomplete and 108C1, 190pH;MCZ 70Q, 9mH; and rent noise generator of the preamplifi- useless way to measure preamplifier MC 100 34, negligible inductance. er in amps. noise. This conventional definition led None of the manufacturers states a The output S/N ratio is (1) divided to the widespread notion that phono need for a loading capacitor. Small bv (2): dynamic range is only 60-70dB. If loading capacitors and cable capaci- true, this means LPs are about tance and inductance have a second- Vo2/No = ec2l[eR2 + N2er2 + [ir?Nltlz.l2]l 25-35d8 more compressed than CDs, order effect on the equivalent electri- (3) an amount of compression so severe cal noise anyway, so I neglected them that you would notice a real in the following analysis. Notice that the input impedance of "Muzak"-like loss of dynamic impact Figure 1 shows a simplified equiva- the preamplifier has dropped out, in LPs relative to CDs. lent circuit for analyzing the noise and, therefore, does not direitly affect In fact, serious listeners find the performance of a phono ca"rtridge and the S/N ratio. opposite is true. Using LP and CD preamplifier, or pre-preamplifier. The Shunt resistances placed across the issues of the same performance, they circuit model is based on the Rotl-re- preamplifier input increase the noise have carefully and empirically com- Dahlke3 theory of noisy fourpoles (in power of the voltage and/or current pared state-of-the-art (SOA) phono which the preamplifier has one input generators. Such resistors can only systems to SOA f;ljfback CD players. port and one output port). In this decrease dyrlamic range by increasing LP playback is generally considered model, amplifier noise originates, at noise. Motchenbacher and Fitchena either better or eclual to CDs in macro- audio frequencies, as two uncorrelat- show that the output S/N ratio is dynamic impact. LPs are usually supe- ed generators: an ideal noise voltage maximized if the rior following is true: in subtle microdynamic shadings. generator, e*, and an ideal noise cur- rent generator, i*. These generators N2[e*/ir]= lZ.l ohms (4) Examples interact with the cartridge impedance To help quantify dynamic range for to establish the basic noise level of the Substituting (4) real phono - into (3) and simpli- playback systems, I use phono playback system. fying yields: three modestly priced examples: a The figure includes a transformer to Surniko Blue Point, a high-output show the effects of coupling very low moving Vo2lNo = eczllen2 + [eri, lZ"l]l (5) coil; a Grado MCZ, a fixed coil output-moving coil cartridges to pre- with moving magnet; and an Ortofon amplifiers. You assume the trjns- Assuming optimum noise coupling, MC 1091 a very low output moving former is - ideal, but you can represent the output noise is dependent on the coil. Although better-sounding car-- its losses by increasing the power of product of the basic noise generators tridges are available, I chose theie for the preamplifier noise geneiators. To of the preamplifier and the magnitude modest price and convenience, not to apply the diagram to transformerless of the cartridge impedance. Equation mention that I own one of each. preamplifiers, simply set N, the turns (5) shows that the preamp,s e*in, To estimate the dynamic range ratio, to one. parameter is useful for choosing capability of the vinyl Lp, first i estab- To calculate the S/N ratio at the low-noise preamplifier; the lor.ier lish the cartridges' sensitivity and the preamplifier output, first calculate the e*i*, the greater the potential S/N. largest signal amplitude it must han- desired output signal as follows: Therefore, consider using e*i* as dle. The sensitivity figures in units of an appropriate figure for any ampUfy- mY /cm/s are: Blue Point, 0.85; MCZ, Vo2 --A,2 [e.2/N2] [tZ(Z+ ft)t2] votts2 (1) ing device in which low noise is 0.42; and MC 100, 0.018. The worst important. I find it more useful case (loudest) recorded than sound on a where: Vo is the desired output sig- noise figure because record in the the parameter literature is 105cm/s at nal in volts; contains information 7kHz.1 necessary to e-a is the open circuit output voltage _ noise-match to a source and compare Of these three cartridges, the largest of the cartridge in volts; two devices. signal available to the input of the N is the turns ratio of the coupling Of course, as with noise figure, preamp is 0.85mV/cmls x 105cm/s, eri* transformer; varies with frequency, bias cur.eiri, or 89mV. This establishes the upper Z is the input impedance of the temperature, and even signal level. limit of dynamic range requirea oiine amplifier in ohms; For my calculations, I account preamp for the Blue Point cartridge. for Za is the series impedance of the variations in frequency by breaking From a design point of view, the pie- cartridge, & + jXc, in ohms; the frequency band into small amp should cleanly handle an input segl _A, is the voltage gain of the pream- ments and assume typical bias signal of at least 89mV at ZkHz. cui- plifier. rents and temperatures. The output noise is: Noise Analysis Noise Matching Calculating the equivalent preamp No = [A12lN2] [er2 + N2er2l[tZ(z + z")12 For a low-output moving coil car- input noise is a bit more complicated. - +liNztN2lltzczl(z+ 26)t2ll votrsz (2) tridge,-the source impedance is very The input noise depends on the noise small, for example, 3C) resistive for th-e parameters of the preamplifier and the where: eo is the thermal noise volt- MC 100. For such resistances, it is Audio Electronics 4/96 l7 (The gains shown in Tnble 1 are more 120 than adequate to assure that second- stage noise is negligible.) Figures 3a t00 and 3b summarize the results of the g80a spreadsheet calculations. o c 3 oo LP vs CD Dynamic Range s E The vertical bars in Figs. 3a and 3b ioo represent the dynamic range obtain- o able with various combinations of car- tridges and preamplifier devices. The line marked with squares represents dad$s I U 3$*3358 5 6dItq 4i<^d the optimum dynamic range attain- 6ba t o R i oOq-a za? < B q oA - o r4 s E?$ 691 QZ T F's E f * * $ able with an ideal noise-matching F @ =v u F transformer. Ideal means it adds no 5 additional noise and varies its turns FIGURE 3a: Preamps arranged in descending order of er.i". ratio to match the ratio en,/iy as a function of frequency. Figure 3a arranges the preamps in descending example, for the Sumiko and Ortofon arranged in order of descending e*, order of their e*i* product, while Frg. cartridges, the parallel combination of the dynamic range of the relatively 3b shows the preamps in descending 2N4403s is best (highest dynamic low-impedance moving coil car- order of their noise voltage gener- range), while the C413N is best for the tridges, the Sumiko and Ortofon, fol- ators, e*. Grado. The larger impedance of the lows the trend in e*. The Grado, For the high-output cartridges, the Grado cartridge favors those preamps because of its significantly greater Sumiko Blue Point and the Grado with lower i*. impedance, does not always follow MCZ, the dynamic range possible In Fig. 3b, where the data is the e* trend. with good low-noise devices, such as the C413N or AD797, is almost opti- mum. For the low-output moving coil Table 1. Preamp Galn Requlrements for 3 Phono Cartrldqes cartridge, the MC 100, the dynamic range approaches within 3.5dB of opti- Frequencv Maximum Equivalent SUMIKO GRADO OFTOFON mum for a preamp with eight 2N4403 (Hz) lnput BLUE POIUT I\4CZ MC 100 transistors in parallel. Most important- Velocitv (cm/sec) lnput to 1st Staoe (Vrms) Iy, the dynamic range potential of any of the three cartridges uith its best preamp is 7 0.7 6 6.46E-04 3. 1 I E-04 I.37E-05 equal to or better than the theoreticalbest a 20 0.65 5.53E-04 2.73E-04 1.17E-05 CD can achieae! The best of these mod- 700 26.09 2.22E-02 1 .1 0E-02 4.70E-04 est cartridge-preamp combinations has 3 000 80.00 6.80E-02 3.36E-02 1.44E-03 16dB better dynamic range than the 70 00 I 05.00 8.93E-02 4.41E-02 1,89E.03 CD upper limit. 11000 40.00 3.40E-02 1,08E-02 7,20E-04 The best transformerless combina- tion, the Sumiko and eight parallel @ 112 __ 227 _ tr2sr- 2N4403s, has greater than 112d8 (10 Vrms Max Outpul) dynamic range, and is only 1.5d8 shy lnpu to 2nd St RIAA Weighting of the optimum combination of a 7 1.00 0,07 0,07 0.07 Sumiko with a transformer and a 2 1 .00 0.06 0.06 C413N FET. At 1l"1dB, the Sumiko/ 0.06 70 0.1 0 0.25 0.25 0.25 4D797 combination is only 1.7dB less 3 000 0.10 0.7 6 0.76 0.7 6 than optimum. 7000 0.10 1 .00 1.00 L00 In Fig. 3a the trend in optimum 11000 0.10 0.38 0.38 0.38 dynamic range performance for any of the cartridges follows the trend in the Maxlmum 2nd Stagg._9aln l_0 LO 10 e*i* product. Interestingly, the opti- 10 Vrms Max )utout) mum dynamic range is very nearly Prea rp Output (Vrms the same for all of the devices except RIAA Weiohtino for the HA5190. This means that car- 1 .00 0.72 0.72 0.72 tridge noise generators tend to domi- 20 1 .00 0.62 0.62 0.62 nate this set of optimally noise- 70c 1 .00 2.48 2.48 2.48 matched phono preamps. 3000 0.50 3.82 3.82 3.82 However, for nonoptimum cases/ 7 000 0.25 2.49 2.49 2.49 the trend is not so straightforward. 11000 0.16 0.62 0.62 0.62 The e./i* ratio interacting with the cartridge impedance determines Holman, Tomlinson equirements ol Ph onographic Preampl lers," which is the best combination. For ---AUDO. tr"|, 1rn r- l-- 20 Audio Electronics 4/96 tape hiss). You can decide if this factor I 20.00 adds an additional 10-30d8 to the

I 00.00 maximum dynamic range for LPs I previously calculated. @ o 80.00 No doubt, we can more readily hear o c music embedded in thermal noise G 60.00 s than in quantization noise. Why? E c 40.00 Digital noise destroys information, o whereas thermal noise adds innocu- 20.00 ous, unrelated information to music. Ironically, despite the low noise 0.00 83$ usually attributed to digital, quieter 9SO passages of records sound better than Ha*SE5BEEdEiSEpE$E*e=;3€5$F - , their digital counterparts. The greater Ee* 3* E dynamic range potential, and the abili- ty to hear music well below analog FIGURE 3b: Preamps arranged in descending order of e,u noise, and digital's quantization dis- tortion of quiet sound all partly The bottom line here is that if your Beyond the Numbers explain why I, and perhaps you, still goal is to achieve best dynamic range/ When comparing the sound of phono prefer the best phono playback sys- select the preamp and cartridge with playback with digital, note the differ- tems to their digital counterparts. knowledge of all the relevant parame- ence in the way noise combines with ters. Audio amateurs can do this dur- the signal and how humans react to Epilogue ing the design phase. Preamplifier, that noise. Phonograph thermal noise The intent of this article is to demon- low-output transformer, and cartridge is linearly added to the signal. Thus, strate that the LP is, contrary to popu- manufacturers should at least state the right kind of filter can retrieve sig- lar belief and the opinion of the audio their products' parameters so you can nals below the noise level. Our ears vanguard press, potentially superior to make intelligent choices. Of course, I are the right kind of filter. We can the CD in dynamic range. I used a rig- am not suggesting that we choose com- retrieve musical signals that are orous, ar-ralytic approach to do this. I ponents on these parameters alone. 10-30d8 below the thermal noise (or take little comfort in the results of the

lf you remember the worm sound of the originol Ho{ler MOS- Or, buy the whole foctory wired duol FET power omplifiers, you will be thrilled to know thoi ihey ore chonnel omplifier for $863 being monufoctured ogoin. SMART monu{octurers the world's lorgest line of cinemo sound products for professionol movie lheotres. We purchosed the inventory ond monu{octuring doc- umentotion from Rockford lnternotionol ond continue to pro- I duce these ompli{iers for cinemo use. For the post 3 yeors we hove been the only source for this product. Eoch module is o complete omplifier. All you need is your own power supply a. , 'l20 5945 Peoihtree iorners Eosl ond chossis. The omplifier is honestly roted ot Wotts into .l80 Norcross, GA 30071 8 ohms or Wotts @ 4 ohms. Eoch module is priced ot (800)4s-SlvtART o r (4041449 -6 698 only $230 ond includes o one yeor {octory worronty.

Audio Electronics 4/96 2l analysis because numbers rarely tell us ers, decomposing music into sinu- there is sornething more fundamental- much about how audio components soidal cornponents. At least for rnusic Iy wrong with digital. sound. In this case they seem to sup- our ears seem to be very sensitive, port what I already believe to be true. perhaps most sensitive, to distortions The Trouble with CD Sound The following colnrnents, althougl'r ir-r the time domain. We hear live To quote from an article by Fonte6: "lt speculative, support the notion that we music and listen to recorded music in is a mystery to me how basic concepts have much to learn about how humans the time domain; i.e., we process vari- can get so twisted around that they perceive music. An understanding of ations in air pressure as a function of become nearly meaningless. In this this process will help us to identify the time, and do not receive or perceive new world of digital sampling sys- significant parameters for audio repro- music in the frequency dornain. terns, you would think that tl-re duction systems and to develop tech- I believe fidelity in the time domain Nyquist theorem would be as familiar niques for measuring them. is paramount. Consider that great as Ohm's law Nevertheless, as time orchestra leaders breatl-re life into passes, I see more people and publica- Myth of the Frequency Domain music by controlling the flow of the tions misunderstand its implications. Literature suggests that perfect play- music in the time domain, i.e., by mak- The results of these misunderstand- back is possible if we could only ing adjustments in tempi and loud- ings are circuits that don't work well reduce measured distortion and noise ness. Not much training is required to and designs that arc inappro- to inaudible levels. Up to this point, I hear differences between first- and priate....Simply stated, tlrc Nyquist tlrc- focused on noise and dynamic range. third-rate orchestra conductors. orent tells yotr tlrc trrirrirrtutrt snntplirtg For the digital case, the term quantiza- This, of course, doesn't prove my rnte needed to recortstruct n sigrnl's fre- tion distortion is more appropriate time-domain hypothesis. It does quetrcy witlrout nl iasing." than noise, since quantization is underscore the idea that our hearing The theorem fails to directly men- essentially a nonlinear process. mechanisms are highly sensitive in tion recovering the phase or ampli- We generally consider nonlinear the time domairr. We can reasonably tude of the signal. If the sanrpled sig- distortion as harmonic and intennod- assume that similar perceptual mecha- nal has a sir-rusoidal sl-rape and also ulation distortion, which we measure nisms are at play when we perceive repeats for several cycles, phase and in the frequency domain, and use that something is not quite right with amplitude errors will decrease to sinusoids as the test signals. Almost today's digital music, and that digital small values. However, if the signal to all of our audio equipment is charac- is doing something wrong in the time be digitized is impulsive and does not terized in the frequency domain. Yet domain. We know that major im- contain rnany identical repeats, then modern sensory research suggests provements in digital happen when significant errors in phase and ampli- that our ears are not Fourier analyz- time-domain jitter is reduced, but tude can occur. Errors in the frequen-

A B c D E F G H I J 481 Table 2. Sample Calculations lror4he Spreadsheet 482 Nolse Analysls of a GRADO MCZ Cartrldge wlth 8 Parallel 2N4403 preamp 483 en tn Rc Lc t No Nt Sandwidth Vn Noise en'in 444 (volts/Hz) (ohms) (Henries) (Hz) ^ 1/2 (amps/Hz)^1/2 (volts^2/Hz) (volts^2) (Hz) (volts) (watts) 485 2.808-10 2.83E-11 70 9.00E-03 37.5 5.17E-18 1-29E-l6 25 1.14E-08 7.9E-21 486 2.80E-1 0 2.264E-11 70 9.00E-03 75 3.76E-18 1.88E-16 50 1.37E-08 6.3E-21 487 2.80E-10 1.981E-1 1 70 9.00E-03 150 3.1 I E-1 8 3.1 9E-1 6 100 1.79E-08 5.5E-21 488 2.80E-10 8.49E-12 70 9.00E-03 300 1.61 E-1 I 3.22E-16 200 1.80E-08 2.4E-21 48S 2.80E-1 0 8.49E-12 70 9.00E-03 600 1.67E-18 6.70E-16 400 2.59E-08 2.4E-21 49( 2.80E-10 8.49E-12 70 9.008-03 1200 1.928 B t.54E-15 800 3.92E-08 2.4E-21 491 2.80E-10 4.245E-12 70 9.00E-03 2400 1.66E B 2.65E-15 1 600 5.'t 5E-08 1.2E-21 492 2.80E-10 4.245E-12 70 9.00E-03 4800 2.65E 8 8.498-1 5 3200 9.21E-08 1.2E-21 49: 2.80E-l 0 4.245E-12 70 9.00E-03 960 0 6.64E- B 4.25E-14 6400 2.06E-07 1.2E-21 491 2.808-1 0 4.245E-12 70 9.00E-03 1 6400 1.68E 7 1.21 E-13 7200 3.48E-07 1.2E-21 495 fotal Nt volts^2 1.78E-13 1 9975 4.22E-07 496 497 rf a GRADO MCZ wlth a C4 3N JFE] 49t en Rc Lc Nt Bandwidth Vn Noise en'in 49! (Hz) volts^2/Hz) (voltsa2l tHz) (volts) watts) s0( 8.00E-10 3.1 E-l4 70 9-00E-03 37.5 1.80E-18 4.50E-17 25 6.71E-09 2.5E-2i 501 8.00E-10 3.1 E-14 70 9.00E-03 75 't.80E- B 9.00E-17 50 9.48E-09 2.5E-2a 502 8.00E-10 3.'t E-14 70 9.00E-03 150 1.80E- 8 1.80E-16 100 1.34E-08 2.5E-23 50i 8.00E-10 3.1 E-14 70 9.00E-03 300 1.80E- I 3.60E-16 200 1.90E-08 2.5E-2! 50( 8.00E-10 3.1 E-14 70 9.00E-03 600 1.80E- I 7.20E-16 400 2.68E-08 2.5E-2a 50t 8.008-10 3.1E-14 70 9.00E-03 1 200 1.80E- 8 1.44E-15 80c 3.79E-08 2.5E-2a 50t 8.00E-10 3.1E-14 70 9.00E-03 2400 1.80E- B 2.88E-15 1 60C 5.37E-0€ 2.5E-23 507 8.00E-10 3.1 E-14 70 9.00E-03 480 0 1.80E- B 5.76E-15 320C 7.59E-0t 2.5E-23 50t 8.00E-10 3.1E-14 70 9.00E-03 9600 1.80E- 8 1.1 5E-1 4 640C 1.07E-07 2.5E-23 50! 3.1 E-14 70 9.00E-03 1 6400 1.80E-18 1.308-14 720C 1 .1 4 E-07 2-5E-23 51 Total Nt volts^2 3.59E-l 4 1 S975 1.90E-07

22 Audio Electronics 4/96 cy domain can also occur for non- he ascribes to the CD sampling metrical and strikirrgly impulsive- repetitive sinusoids. process. On the contrary, I think Fonte looking shapes l-righly correlated in Fonte shows that time-domain has put his finger on one of the funda- time. I suspect that sinusoidally stim- errors are inversely proportional to the mental problems with CD sound: time ulated harmonic and intermodulation sampling rate and can be quite signifi- (and frequency) domain errors inher- characterizations tell us little about cant. He states that CDs can introduce ent in the sampling process. what we hear in the time domain. amplitude modulation errors to 307n, That notwithstanding, if the time In the absence of test equipment we amplitude errors of 0.436-1.84d8, and domain is important, shouldn't we be use our ears. Stories about the analog frequency errors up to 18% into single- able to characterize our audio compo- version of a performance sounding cycle events corresponding to the har- nents in the frequency domain and better than the digital are not unique, monics of instruments. He gives sever- then Fourier transform to predict their for they have been reported repeated- al arguments explaining why the ear behavior in the time domain? ly in audio publications emphasizing does not hear these aberrations in CDs: listening comparisons. Keep in mind Another Theorem these comparisons are for a digital r distorted harmonics are over- SuperpositionT is the basis of Fourier playback system, which should cer- whelmed by the strength of the fun- theory; i.e., the theory states that there tainly sound better because its fre- damental; is a weighted set (an infinite set in the quency domain distortion measures o the ear cannot hear an error of case of nonperiodic waveforms) of 90dB below the maximum signal. In 0.436dB; sinusoids, which, when linearly added contrast, the cartridge component of r the ear is not sensitive to phase together, can represent complex wave- the analog playback system measures errors in music signals; forms. But superposition only holds in somewhere between 30-60dB for o music signals, consisting of many the case of linear systems. Of course, sinusoidal references. Clearly, fre- repetitions of the same waveshape, we know our systems are aery nonlinenr. quency domain distortion measure- average out any of these aberrations As suclr, zoe cannot aalidly use Fourier ments are meaningless for predicting to zero. transforms to relate a chnracterization in which audio components will sound otrc donnin to the other. better than others. Wltile' I rcspcct Forrte's analysis, I When I scc music in thc tirnc Wc ncccl to rrrrclcrstarrcl irr rrrorc clisagree with his assurnption tl'rat domain (at tl-re output of an audio detail, and perhaps furrdamentally, CDs sound great, and that the ear component on an oscilloscope), I do how humans perceive and enjoy does not hear the rather large errors not see sinusoids, but rather nonsym- music. I predict that time-domain

Toroidal Line ru Transformers rupruruKu off-the-shelf ... Ask ru Typical Typical Secondary Secondary Dimensions Typical Secondary Secondary Dimensions Typical Load RMS Volts RMS Cunent lnches Weighl Load RMS Volts RMS Currenl lnches Weight Type VA A Dia. Ht. Pounds Price lype VA Dia. Ht. Pounds Price D4000 30 6+6 2.50 '160 D4030 18+18 4.44 $46 27 D4001 30 9+9 1.67 '160 $30.36 D403 1 22 +22 3.64 plus D4002 30 12+12 1.25 '160 plus D4032 25+25 3.20 45 1.8 4.19 50 D4003 '15 + $3 30 15 1.00 29 13 0.93 $3.00 D4033 160 30+30 lot D4004 30 18+18 0.83 for D4034 160 35+35 )ad shippjng & o4005 30 22+22 0.68 shipping & D403/A 160 12O + 12O 0.67 handling o4006 30 25+25 0.60 handling D4036 160 220 o73 D4007 30+30 0.50 o4040 230 25 +25 4.60 $54.97 D4010 60 9+9 3.33 o4041 230 30+30 3.83 $plus 0401 1 60 12+ 12 2.50 $35.s4 o4042 230 35+35 47 20 589 $4 50 o4012 60 15 +'15 2.OO plus D4043 230 40+40 2.88 for D4013 60 18+18 't.67 $3.00 o40464 230 12o + 12o 0.96 shipping & D40'14 60 22+22 1.36 3.5 1.4 147 lor D4045 230 220 1.05 han.llin. D4015 60 25+25 1.20 shipping & D4050 330 D4016 60 30+30 100 handling 25+25 6.60 $67 00 D4019A 60 l2O + 12O 025 D4051 330 30+30 5.50 plus D4018 60 220 o27 D4052 330 35+35 471 53 21 72A $5 00 D4053 330 40+40 413 for D4020 100 12+12 4.17 D4054 330 45+45 367 shipping o4021 100 15+15 & $38.68 D4057A 330 12O + 12o 1.38 handling o4022 100 18+18 2.78 plus o4056 330 220 150 D4023 100 22+22 $3.50 D4060 530 D4024 100 25+25 2.OO 3.7 l8 zb5 for 30+30 883 $84 02 D4061 530 35+35 7.57 D4025 100 30+30 1.67 shipping & plus D4062 530 40+40 D4028A 100 12O + 12O o.42 handling 663 $5 00 D4027 100 220 045 D4063 530 45+45 5.89 60 24 1103 for 04064 530 50+50 5.30 shrpping & Send orders with payment to: D4067A 530 'l2O + 12o 221 handlng D4066 530 220 2.41 47 South End Plaza AVEL iP! Primaries 120V + 120V 50/60H2 New Milford, fi 067/6 All rnounting hardware is included, UL recognized. Tel: 203 / 355-4711 TRANSFORMERS Fax: 203 / 354-8597 INC

Ads On Tapllffi Audio Electronics 4/96 23 measurements will be crucial and fre_ Of course, some analog D A 3 0 quency domain much noises are less so. Re_ not "thermal-like,, ancl- search that develops are quite a set of test tech_ important in audio. nlques more closely related to how the ..1. Record pops and clicks, cltre to ear perceives music could well lead to flsynchronous.digitol to dlrt or scratches. The effect of these on onolog rcvolutionary timc-dornairr instrtr_ convorsion uill tronsform every rrrusic is uot lincnr ancl acJclitivc. disk you otun. mentation breakthroughs. Instead, the effect appears to be a kind of burst All distortion, perhaps analogous Noise ls Not Created Equal to slew rate Il *y earlicr limiting because oithe quantification, I went steep.scratclr edge Heor digitol's trua pot@ntiol. along with thc thc stylus rncets. common notion that a Clcarly, irr gcncral, Heor - single-dimensional bettcr_sourrdins X Dlt 3.0. Haor tha music. parameter that is turntables audibly the sarne for botlr recording reduce the annovl tcclr_ rrrg. quality of pops and clicks niques represents Lp ancl boih Clinoise. wrth rcspect tcl thc car,s pcrception This is wrong for several reasons: of .Vour mong options noul includa the intensity of the initial diJtortion x - ltvr th@ doubla 1. Digital quantization filtered LCLC noise transi.ent and the ringing therca[ter. super qui@t pou.rer supp|,,r, sounds-and is-completely olso different 2. "Popcorn,, nois-e associated r@gutoror upgrodes for both DflC than analog thermal with ono or record surface a number of op amps. onolog sections. noise (or analog Unlike the hissv tape hiss). noise we find easy to filter 2. Within the cliss of analog ,,ther_ music through, "popcorn,, noise increases in mal-like" noises of equal powei maior power drfterences with lower frequencies and exist in how annoying occurs in fractional they sound. second bursts. This adds a gravelly, grungy 3. There is good reason sciund, . to believe particularly at Iow frequenciii. that not all quantization Audio noises, even 3. "Contact,, noise iypically Croflerr Gui ld at equal 16 (or n) bit resolution associ_ level, ated with dirty contacti and are or sound the same; poor con_ 5102 torl 38 plorc e.g., D/tr s6n_ nections, 918,627_50/gwht verters using or sometimes associated lulro different roundoff algo_ witl'r stranded Klltli ttil 9tA.6224J9Bfu rithms wire in the signal path. sound different. This hllp://**v. ltt. luho . nonlinear noise imparti nt nlruy@gohfu. cn on these very a digital . .\.expand fundamen_ c.haracter to analog sound tal differences, note that o. .r,Jk", Ads On Tap #105 iandom ana_ orgrtal sound log noise, uncorrelated with even drier and more the etched. Others hypothesize that ih; music, operates as another signal lin_ nonllnear elfect on music is the result SILVER SONIc, early_adding to the music ,ig.,ut. a of partially signal with amplitude shorted point_contact less tfiu" tn" drodes at the junction Audio Cables norse ts preserved between dissim_ and amplified just rlar material.r,.".q., as accurately copper and copper as a signal lirger tiran oxrde or nickel and Making the High End the noise. Thus, the nickel oxiie. ear can fllter out These diodes are highly Affordable! weakcr music from noise. components which severely"o"ti.,"u, The.effect of digital quantization distort "l low-level music. Distorted iow_level have found the Silver Sonic cables it entirely different and quite musical to l:tt: details submersed within be extremely musical over the long nonlrnear. As signals grow progres_ haul." - more faithful high-level music give Gary Galo, Audio Amateir' srvety smaller (relative to the "contact" riaxi_ noise a digital character." lssue #2/94 mum signal level the A/D converter can handle), fewer quantizing bits are Noisy avallable LPs to resolve them and smaller Scrakhed signals or dirty records are noisy. A thus become proqressivelv good record vacuuming system ereat_ Tof:.dit-!o'!gd by the A/o!ro."rr. i ly reduces limit is finally rhis noise. SOe .u.triJe;r, reached wh&e sisnals arms, and smaller turntables also reduce, b"v a than the smallest quuntirltion surprising step degree, the audibility'of are completely lost. dirt and The scratch effects. However, A/D p.o."s, sets all such sis_ even nals.to a pristine Lp can still emit hiss fixg{ amplitude. There is n"o 3 from the master tape and,/or record musrc srgnal for the ear to filter Silv_er Sonic T-14 out of surface noise. ^.. Speaker Cable the quantization noise. itrrver Sonic BL-1 Series ll I believe the Tape hiss lnterconnect facts that the . can be largely eliminated Silve.r Sonic D-1.t0 AES/EBU Digital-' ear filters weak music by noise from reduction prolessing (at the Silver Sonic D-75 Digitat Cabt6 noise and smaller amplituae sif_ Hoot( gxpense of a small loss in"overall up wire & Connectors just as u..,r.ut"ly u, nif" fidelity).or arel3lt,1*plily by_recording at higher lcv_ lmportant advantages of analJs els (at D.H. Labs over the risk of compressing Ioud p.O. digital. (Some of this advantaeE transients). Box 3i598 can At least 1OdE of hiss"reduc_ Palm Beach Gardens, be lost where Class B o, FL 33420 sturuJi tionis possible with either technique. (561 625-8998 (phone/fax) AB analog amplifiers, ) ,rr"h u, *uny When audible surface op amps/ are used noise occurs, because they dis_ it is due to indifference tort smaller signals.) on the pu.t oi Ads 0n Tap#1,30 the recording industry-poor_quality 24 Audio Electronics 4/96 vinyl or poor-quality stampers at the rc'kindled consumer ir-rterest in vinyl. I pressing plant. Lack of quality control hope that if this trend continues, and is abetted by the RIAA record indus- affordable analog records become try standard permitting surface noise available again, the recording indus- of 55dB below a lkHz sinusoid try will produce a quality record, free recorded at 7 cm / s (interpolating from of surface noise and warps, that RATION thc Maximunr Ilquivalcnt Input cxploits the fLrll dynarnic rangc potcn- Velocity column of Table 1, the equiva- tial of the phonograph. (We know SECUNffY ALENT lent RIAA dynamic range is roughly much bad vinyl was produced, espe- SIGNALLING SYSTEM 67dB at lkHz). cially near the end of the LP era, The above notwithstanding, clean, '80s, Designed lor the traveler to hang somewhere in the when it on the inside ol a hotel door, the unscratched, good quality records seemed that overnight all vinyl had Datawave KL-1 Security Alert played back through "first-rate" mod- been yanked out of record stores and emits loud beeps and llashing ern cartridges, arms, and turntables replaced by CDs and cassettes. During lights when someone knocks or il someone tampers with the sound as quiet as CDs to me. More this period, I remember trying four or door. ldeal tor hearing impaired, importantly, I find well-recorded LPs more warped records until I found the four high-brightness LEDs more dimensional, truer to the tonal one that played. Consumers with sim- are easy to see from far away. Can be switched lor audrble qualities and dynamics of live music, ilar experiences must have been alarm with lights or lights only. and very noticeably more relaxing huppy to see vinyl go away.) Requires 3 AA batteries than CDs, particularly during long lis- Having said few nice things about (not included).CAT # KL-1 tening sessions. Additionally, digital in this article, I must point out guests, most of whom are not audio-^y that I have great expectations. I prefer Alkaline AA batteries philes, listen to the same source mater- the convenience and ruggedness of CAT BAT-AA 3 for $2.25 ial played back from LP and CD, and CDs, and hope that over time their EDGEVIEW PANEL METER prefer the analog version without quality achieves or surpasses analog. I FS=l MICROAMP exception, much to their great sur- find it ironic that with all the technol- prise! ogy advances over the past 40 years, Mura analog tape recorders and records Beautif ully constructed of edgeview meter move- Observations the fifties can achieve a higher stan- ment. 2.75" X 0.8" face. I have shown that the phono playback dard of musical fidelity than today's Scale is labeled ( system is capable of greater dynamic technology. "BEATS/MINUTE" and is calibrated lrom 50-150. lt range than 16-bit digital. I am assum- But we understand a good deal can be easily removed or reversed to create ing that, with proper design and care, more about CDs' limitations, and a blank scale. The black bezel is 3.53" X the recording process (at least the technology has advanced since the 16- 1.33" and can be adjusted or removed. 0.1" thick clear plexiglass case. direct-to-disc method) should easily bit 4lkHz standard was Maximum case / established. size behind mounting plane: 2.75" X 1 .18" X achieve greater dynamic range than Affordable 20-bit A/D and D/A tech- 2.18". Solder lug terminals.""$5.*iln the playback system. Whether analog nology is here. Those few rvho have iliffiil; tape recording, which is the source of heard 20-bit digital attest to a sound most of our LPs and some of our best closer to analog masters (if also S"VffS.VIDEO COND CDs, can achieve greater dynamic played back at 20 bits). range than CDs or LPs needs to be To achieve analog quality, however, r/ tE evaluated using the methods of this digital standards need raising to per- \ article. haps 21 +'."ut"*im rlni"ru'r" 4 pin mini or.more bits. Also, sampling @ l Fortunately, there seems to be a rates must be significantly raised. DIN plugs on both ends. CAT# CB-369 (The CD standard sampling rate of 10 for $25.00 just over twice the highest frequency ACKNOWTEDGMENT 100 for $150.00 humans are reputed to be able to hear Thanks to Dr. Piene M. Sprey for many helpful suggestions (i.e., 20kHz) is suspect. A growing on organizing the manuscript. body of evidence suggests we can hear rates of change in time-domain REFERENCES signals that are greater than equiva- lent slopes for 20kHz sine waves.) 1. T. Holman, "Dynamic Bange Requirements of TERMST NO M/N/MUM ORDEF. Shipping and handling lor lhe Most importantly, the sampling rate 48 conbnental U.S.A. $5.00 per order. Ail others inctuding AK, Phonographic Preamplifiers,",4udb (July 1977). Hl, PR or Canada musl pay lull shipping. All o.dsrs delivered and in CALIFORNIA musl include local slate sales tax. Ouantities 2. Data provided quantization levels should match by Mr. Matthews at Sumiko. Limiled. NOCOD. Pricessubject our ears' demand for fidelity in the lo change wilhoul nolice. 3, H. Rothe and W. Dahlke, 'Theory of Noisy CALL, WRITE Fourpoles," time domain. Pru. I RE M, 8811-818 (June 1 956). FAX or E'MAIL Sooner or later the 16-bit/41kHz 4. C.D. Motchenbacher and F.C, Fitchen, Low Noise for our FREE standard be revisited, Electnnic Design (John Wiley, 1973). will if for no 64 eage other reason than to produce a small 5. "Audio Handbook,' National Semiconductor Corp. CATALOG (1976\:2-27. but truly state-of-the-art music Outslde the U'S'A' send $2.00 Postage' 6. G.C.A. Fonte, "Apply fundamentals to avoid surprises subindustry. I hope the next digital with umpled systems," EDN(June 24, 1993), standard is carefully and deliberately 7. S. Slein and J.J. Jones, Modern Conmunication set much higher than our de facto Pnnclples (McGraw-Hill, 1 967). analog standard of40 years ago. I Ads On Tapll0l Audio Electronics 4/96 25