MARCH 1988 @ $2.25 StereoReview SPECIAL TAPE RECORDING ISSUE UNUM U2 emi Seal DECKS GO HEAD TO HEAD RECORDING SYSTEMS/ TAPE BUYING GUIDE at Sete eee eee ene aOR eesti ay a Ue aSHead-to-head laboratory and listening tests _ of five of the world’s finest cassette decks BY GRAIG STARK COUPLE of months ago, STER, £0 REVIEW sent me five of the very finest cassette decks in the world. Lest my joy be unbounded, however, they arrived ‘with this injunction: “Now report which is best, and within a fort- night.” What a task! Four of the decks have been reviewed in these pages—the Na- kamichi Dragon ($1,995) in 1983, the Tandberg 3014 'A ($1,995) in 1984, the Onkyo TA-2090 ($900) in 1985, and the Revox B215 ($1,880) in 1986. Except for minor updates, all are mature rather than brand: new designs—the Dragon, indeed, dates back to November 1982. The same also holds for our fifth deck, Harman Kardon’s top-of-the-line D491. ($950), which was intro- duced in June 1983, ‘The need to return to these classic machines in search of the best reflects the fact that world-class cas sette decks are no longer econom- ical to design. They're a kind of con- cession to the relatively small pro. portion of potential buyers who are really dedicated audiophiles and en gineers and who tend to hate the whole business of building to a price SvrRH REVIEW Manca 1988 point. Unhappily, if there are still major analog-cassette-deck secrets waiting to be found, they're likely to remain undiscovered: The scientists who might have found them have been moved on to newer projects such as DAT All five of the decks use separate record and playback head el which enables the user to. make immediate, critical A/B compari- sons between the source and the just-taped signals, More important from a design point of view, howev er, it permits the playback head to hhave the very narrow magnetic gap, less than one micrometer, needed to resolve very high frequencies (short wavelengths) while also allowing the record head to have the relative- ly wide gap, three micrometers or 80, needed to penetrate the tapes uencies (long wavelengths) The Nakamichi and the Tandberg decks use physically separate record and playback heads; the other three use “sandwich” heads in which the separate record and playback cl ments are housed in a common case with @ shielding barrier between.Totally separate heads require (or a purists, would insist, permit) method of adjusting the azimuth, or perpendicularity, of either the re Cord or the playback head so that their gaps remain perfectly parallel despite some small amount of tape skew inside the cassette shell itself Common-case heads eliminate the need for azimuth adjustment, but they require an act of faith that the gaps were manufactured, and will remain, exactly parallel magnetica ly and that the shielding is adequate to prevent leakage of the record-bias signal into the playback electronics. ‘All five decks also use closed- Joop, dual-capstan drive systems to minimize irregularities in tape mo- tion, (Slow, periodic tape-speed Variations are heard as wow; more Tapid speed variations, which create J) a*eritty” or “edgy” character in the ound, are called flutter.) In a ‘closed-loop configuration the criti- eal portion of the tape—where it Dpasses across the heads—is isolated by the two capstans and pinch-rol- ‘ers from the tape packs on the two ‘reel hubs. The trailing capstan tries 0 rotate at a very slightly slower ‘speed than the pulling capstan, ‘eating a controlled amount of ten- ‘sion that presses the tape against the heads. In an optimum design, the diameters of the capstan shafts and the masses of their associated fly- wheels differ so that they have no common resonant frequency All five decks supply both Dolby Band Dolby C noise reduction; the ‘Onkyo also provides the alternative dbx system. In ac ion, the Revox, Harman Kardon, and Onkyo m chines incorporate Dolby HX Pro, a system that increases tapes’ high- frequency capabilities at high signal levels by dynamically adjusting the ‘amount of record-bias current being used. Tandberg uses its own Dyneq ‘system for the same purpose. To optimize performance from different cassette brands, each of the five machines provides a procedure using calibrated, internal test tones: for making fine adjustments of re- cord bias and sensitivity. With the exception of the Dragon, all the machines also incorporate tape counters that register playing time ‘as well as hub revolutions. And all support some varicty of memory- rewind/play options, provide for ex- ternal timer activation, and have defeatable FM-multiplex filters. (Switching the filter off is necessary for achieving 20,000-Htz response.) NAKAMICHI DRAGON ($1,995) AKAMICHT is pre-eminent in the design of tape heads, transports, ‘and ultra-low-noise electronics, ‘and the Dragon is the com: pany’s crowning casserte-deck achieve. iment, An antoreverse deck in playback it records in only one direction, and i tues completely separate record and Playback heads, The resulting prob lems of azimuth alignment are solved inan absolutely uncompromising ‘A fourchannel playback head is sed, 30 no head shifting 1s required for ‘autoreverse. At the same time, the right channel head element in each di rection is actualy spit into two ultra thin tracks whose separate outputs can be compared. / the playback head ts ‘ven slighty tilted toward one side or the other, there wil bea slight phase diffrence berween the outputs from the tivo halves of the right channel. This dliference, detected and amplified. s tused to drive a servomotor mechanism that readjsts the azimuth of the head fom the fy until the diference is elim nated. Though it was invented to dea! with the problem of tape shew within a Single cassete shell while the tape is ‘winding and unwinding, the Nakam hi Auto Azimuth Correction (NAAC) technique also serves to restore the highrequency response of tapes made ‘on ess accurate machines ‘Unique alo to Nakamichi isthe use of an old professional oper: ree! trick The Dragon's heads are pre-sloted at the tape edges $0 that a wear groove Which can play havoc with lefechannel reponse can never devi as he hheads themiscives wear The playback head has a 06 micrometer head pap the record head a 3 S-micrometer gap. The Dragon's transport uses a total | offive motors. Two are quartz-crystal Controlled and direcly drive the cap- Sians ina closed-loop configuration. These motors have been specially de- Signed to avoid the “cogging” that fre ‘quently induces flutter in directive systems. The transports precision is such thatthe cassette’ pressure pad is ‘actually pushed back and away from the playback head, thus eliminating it as a source of modulation noise. The Dragon's third motor powers the reel Pree and the fourth is used in place of see ads to control the head-gat func Rear imoochiy. The fifths used forthe wermatic azimuth-adjustment system Fymheel masses and capstan diame- ay have been selected to prevent com- mon resonance. tne Dragon incorporates a twent)- segmentiperchannel fluorescent peak: Seng evel display thats also used ‘optimizing bias and adjusting for vanations in tape sensitivity. Built-in Teo ded 1s 000s tone generators are Tra tthe tones are aatlable at the Out- ue eks for eaibrating other pars of xc fstern). and the adjustment proce ‘hare ts easy, though manual. Bias/EQ Saitching for different tape types is also ‘manual Other features include dud rie cine, duatepeed auto fading, a oulpulevelcOnroL Onkyo TA-2090 ($900) there's a feature missing from the Onkyo 74-2090, irs hard to imag: ine what it might be. In addition to its threehead, three motor, dial. capstan, closed loop transport desig, it ‘edas the option of dbx tothe usual oy B and Dolby C noise-reduction systems, and the Dolby HX Pro head roomvextension circuit is selectable rather than automatic The record and playback heads of he TA-2080, made of long-wearing Sendusi, are housed (as m most three- head decks) in a common case. The transport uses a direc-drive motor on the pulling capstan (the rear capstan ts bettcoupled) a second motor to rotate the reels, anda third to provide Smooth quiet operation ofthe head fate. pinch-rollers, and brakes. The ‘wow-andfutter on the unit we tested was actually the Towest we have yet en ‘countered Bias and equalization selection for the various tape types ts automatic, bt differences between brands of the same ‘ype can be eliminated by selecting the cbs fiom which wis intra tone generators to determine the proper ‘yas settings, then rewinds the tape 10 the beginning of ts adjustments. (A ‘manual adjustment to compensate for differen tape sensitivities ts also pro- vided.) The tape counter direct regs- {ers the remaining time per side (vouohare 10991 for the tape lege you're using), arid the sixteen-segment- per-channel level display provides clear Indication of signal peaks. Bidirection. al program search and a playback out- put-level control are provided, and the ‘Onkyo TA-2090 is one of the few decks that still provide microphone inputs. TANDBERG 3014 A ($1,995) 1 Tandberg 3014 A has recently been upgraded (hence the A desig- nation) in its electronics, with polypropylene coupling capacitors land a new variable-output stage, and las dual capstans have been roughened to,educe slippage, which can create modulation noise. Basically, how. ‘er 11 remains the same high-end, uris-oriented deck I reviewed here in 1984, Built like the proverbial battleship, with massive castings, a $-mimthick fase plate, and internal cross braces, the 3014 A uses a pair of servo-coupled direct drive motors to turn the reel /nubs: This arrangement allows con- stant-speed fast winding at sixty times playing speed (the speed can be re- ‘duced for cueing by holding the wind ‘button down). The closed-loop, dual- ‘capstan system uses large, diferent sized flywheels to prevent system reso- ‘nances, and itis Belcdriven by a third ‘motor. A fourth motor handles the hhead gate and associated functions. The record and playback heads of the 3014 A are totally separate unis, $04 record-head azimuth adjustment Is provided to insure absolute matching ‘of record and playback azimuth. The ‘adjustment makes use of a builtin 18,000-Mz testtone generator: The same generator. along with @ second generator that operates at 318 1 8 also used in the manual tape-optimiza- tion procedure. While dyed in-the-wool ‘audiophiles will not object the process ‘of optimizing bias and adjusting for tape sensitivity is somewhat cumber: some, requiring insertion ofa special Screwdriver into a sertes of four small holes on the face of the deck to engage the controls. Fortunately, optimization need be done only when changing brands within the same type. In nor ‘mal operation the 3014 4, like most decks, automatically senses which {ape type Is in se and switches its pre set bias and equalization accordingly. Tandberg is probably the last compa nny to use analog peak-reading meters instead of some form of LED level dis play, but meters have the unguestiona- ‘le advantage of providing more pre- ise readings. Moreover, in Tandbere’s case the meters are “equalized,” that is, they reflect the degree of record tre- ble boost being fed to the head and so rovide an accurate warning against ‘accidental overload. Tandberg pays articular attention to the record sig: nal, which is mixed with the record bias in a special amplifier stage (the Patented Actilinear I! circuit). A second nique recording stage dynamically varies the record treble boost t0 maxi ‘mize the signal-handling capability of the tape at high frequencies. Tandberg elaims that this Dyneg circuitry intro- duces less intermodulation distortion than the Dolby HX Pro system. The 3014 A provides memory set ‘ings or upto nine programed selec tions, and its rear panel contains con trols for manual rather than automatic tapetype selection and for record sensi tivity. A front-panel ouiput-evel control is also provided. REVOX B215 ($1,880) He solidity of design that marks Revox's professional open-ee! “decks 1s evident as well nthe Re ox B215. ts duat-capstan, closed loop transport uses a pair of quarts erystatcontolled Halkeer direct sine motor forthe can aecond ir of microprocessorscontolied di fective motors ts used to turn the reel hubs, which are electrically rather than mechanically braked. The record and playback heads in the B21S are separate elements in a common case. andthe record and playback head gaps are optimally sized for their respective function. Selection of bias and equalization for the diferent tape types ts automatic, but @ manual override is provided so that prerecorded chrome tapes designed to use 120-microsecond (nominally fer. ric) playback equalization can be ac commodated. Optimizing performance between brands within the same tape ‘ype ts fully automated: Press one but ton and the deck does the res, includ. ing rewinding to the starting point of its adjustments. Memory storage of op- timal settings is provided for two brands of feric, three brands of chrome-equivalent, and one brand of ‘metal tape. Record levels can be set ‘manually or with a single-touch level. | sensing button, ‘Signal levels are shown on a twenty four-segment-per-channel liquid-crystal display, and the electronic tape counter indicates elapsed time directly. A but ton is provided to create smooth fades | in or out. While there are rear-panel connections on the B21S for an acces sory remote-control device and even for 4 serial computer port, no playback utput-level control is provided. HARMAN KARDON CD491 ($950) te Harman Kardon CD#91 is an atiratively finshed deck that Seems designed to appeal oa highend audiophile who wants a ful range of oarares without a it of fuss te transports @ simpler afte. ‘than most of those reviewed here I tse only two motors one of which di rectly drives the pulling capwan inthe akcapstan closed: logp configuration lis separate record and playback heads are made ofsendust and ferrite, respec tine, housed ina common casing het needs no adustmente Signal feels are displayed om si: teensegmenr-perchannel peak indica torn, which can be switched toa weighted postion thar refets the re. ord equalization being fod 10 the heads am entirely sound approach While a complete biassenstvity ope limation procedure 1 incorporated In the C491. i involves insertng a spe, Cial screwdriver through a hole tn the front panel. Further, the fet tht ony one of adiiment proved ‘means tha if you ever ise the optim tion procedure, you can never he sure of getting back 10 the factory et. lings: My guess i that fo’ uses will take advantage ofthe feature The tape counter also diplays the playing time, which isa define conve. nience o the home recordist. Anather onvenlence 1s the inclusion (a rar these days) of microphone inputs Bid rectional program search anda plays back level controlar also proto STEREO REVIEWMARCH 1988 55,LABORATORY MEASUREMENTS c ; oe ae a. wa , : * ae 76 B52 88S measured using our reference cen- ter-line-sample TDK cassettes: AD (ferric), SA (CrO;-equivalent), and MA (metal). Since these are not the same tapes generally used by the five manufacturers for factory set- up, we checked the curves both at the IEC-standard O-dp level (250 nanowebers per meter) and at — 20 db, the customary level for frequen- ey-response measurements, both before and after following each manufacturer's optimizing proce- dure, (The curves shown in the accompanying graphs are all after optimization.) The difference be- ‘tween the pre- and post-optimiza- tion response was generally limited 10 a couple of decibels, but in one ‘extremes (+4 ds at with ferric and —3 ay Bion case it ran as high as 7 dp at 20,000 Hz. Certainly, in all cases the results Proved the importance of including Such calibrated fine-tuning proce- dures in a top-of-the-line deck. The effectiveness of the Dolby HX Pro and the Tandberg Dyneq | circuits in extending high-level high-frequency response can. be judged by comparing the O-ds curves. Only the Onkyo deck per- mitted the HX Pro system to be switched out so that the amount of high-end extension could be meas- ured, but_a comparison with the curves for the Nakamichi deck, which lacks HX Pro, is suggestive. In fairness, it should be noted that, the O-dB curves for all the decksA 7 3250 ee = eee, REVOX, "TANDBERG, 23% oni + + unwed Aswteh CIR ‘unwed, owed) CCIR Atthe conventional — 20-8 <—— iii etd 20, Cte Boe Wavand feller low on the Tandberg and Harman Kardon decksand extraordinarily low on the other three decks. mle —T STERFO REVIEW MARCH 1988| improved significantly in the 10,000- to 20,000-Hz range when— as would be normal practice—the Dolby C noise-reduction system ‘was switched in, ‘At the conventional —20-dB measuring level, all five decks had no difficulty reaching 20,000-42 sponse. At both levels, the bass response of the Harman Kardon and the Nakamichi units was su- pperb; the other three decks dropped Off by 3 10 5 dp in the lowest octave (20 to 40 Hz) and showed some Tesponse undulations, though they ‘were not audibly serious, Wow-and-flutter, measured both on a weighted-rms and on a weighted-peak basis, was extremely Jow on the Tandberg and Harman Kardon decks and extraordinarily Jow—the best I've ever measured— ‘on the Onkyo, Nakamichi, and Re- Vox. Pitch accuracy was exception ally good for all but the H/K deck, ‘whose 0.5-percent speed error, ak though certainly acceptable | for most purposes, was run of the mill. Listening Tests All the decks in this survey were included because they were already known to be among the best in the world. Tt should be no. surprise, then, that all of them could play back prerecorded material superbly and produce virtually flawless re- Cordings from all of the cb's and 1's we tried. The problem was to distinguish among winners, which required us to match signal levels from each unit 10 within 0.1 dv and to use an ABX comparator double blind switching system. The ABX comparator has a. microprocessor that randomly assigns the X'push- button to one of two audio sources, A and B, and records the assign: | ment. The listener then flips. be tween A, B, and X and decides whether X is really A or B. The computer records the answers over a series of trials, If a listener can Consistently distinguish which sis- nal is X, there is an audible differ ence between A and B—though which one is better is not automat cally determined. Since Dolby B noise reduction is ‘obsolete for anything but playback ‘of existing tapes, we concentrated ‘our listening tests on tapes recorded using Dolby C or, inthe case of the ‘Onkyo deck, dbx. And we used metal tape as well (Maxell MX, ‘optimized for cach unit). Starting | 38 STeERRO Review MaRcH 1988 with manual source-vs-tape switch ing and using the ABX comparator tox as necessary, we listened for the amount and character ofthe resid. al background noise in music witha wide dynamic range, the ‘overall Smoothness of response. the cay Of piano tones, the spatial imagin the presence or absence of gitmness (edginess) on. violin attack tran sients, the depth of bass response (organ pedals’ and the cannon ia Tehaikovsky's 1812 Ovenure) and the feeling of depth conveyed by a sense of air oF space around a instrument, The envelope, please. There were two fairly clear winners: the Nakamichi Dragon and the Revox B215. The Dragon had the lowest noise level, especial: ly high-frequency noise, by a clear margin, and in source-vs.-tape com- risons I judged it marginally more accuraie and realistic sound- ing even than the Revox. The Re- Vox, however, seemed to have a lit- | Te inore “presence” than the Na Kamicht without the tendency 10- ward. stndency (overly brassy eases, steey-sounding viol grows tapes made onthe oer three fnachines Both decks were virtually files and if deem the Dragon may deem it slightly dead jing The Tandberg deck was not faulty any respect: Its imaging was just te as sharp, the “air” around ruments not quite as clear as Nakamichi Dragon and the Revox B215. With Dolby C, the Onkyo seemed to have a slightly strained quality— a little too sharp and tending toward (but not reaching) tinniness. With dbx, on the other hand, the Onkyo had’ the quietest, clearest back: ground of any of the five units, although a little “pumping” was ap- parent low-level percussive sounds. (You'd be unlikely to notice it in any but the most stringent head-to-head comparisons, howev- er.) Also, on our sample at least, very slight buzzing was audible within its residual noise level with both Dolby and dbx. The character of the Harman Kar- don's noise low-frequency motor rumble and transistor shot noise as well as hiss—militated against it too. Its low-frequency bass repro- duction was superb, but it seemed overly bright and spatially less well defined compared with the Revok and the Nakamichi. Nonetheless the H/K is a fine tape deck—better than all but a few that I get in for testing. Your own ears may or may not agree with mine, of course: What is strident and unmusical to one may be “live” to another. Between the Nakamichi and Revox cassette decks you will certainly not go wrong, and among all five ous audiophile should fe Purist status if he chooses on the no loss of basis of diffe | price, nnces in features or