Berning TF-12 Preamplifier
Stereophile, vol. 11, no. 7, July 1988, pp. 89
Hybrid stereo preamplifier with remote control. Frequency response: high level +-1dB from 4Hz to 50kHz; phono within +-0.25dB of RIAA standard from 20Hz to 20kHz. Gain: high level 32dB; phono 32dB. Sensitivity: 0.6mV phono or 26mV high level for 1V out. Noise: Phono 60dB below 1mV in; high level 84dB below 1V. Input impedance: 47k ohms phono, 60k to 80k ohms high level, depending on volume setting. Output impedance: 3k ohms, tape and main. Max. input: Phono 200mV at 1kHz, high level 9V. Max. output: 6V. Harmonic distortion: 0.03% 2nd & 3rd, 0.002% 4th and above at 1kHz. Remote control: Infra-red, controls volume, channel balance, and mute. Power consumption: 40W. Dimensions: 19"W by 3-1/2"H by 13"D. Weight: 11 lbs. Price $2950. Approximate number of dealers: 10. Manufacturer: The David Berning Company, 12330 McCrossin Lane, Potomac, Maryland 20850, 301-926-3371, 301-975-2069
Ever since it was recognized that the "tube sound" and the "solid-state sound" each had definite, characteristic strengths and weaknesses, electronics designers have sought ways of getting the best sound each had to offer while avoiding the weaknesses of both. Still, it took a surprisingly long time for anyone to realize that the best way of doing this might be hybridization-using tubes in combination with solid-state devices.
The first commercial attempt at this was, I believe David Berning's EA2-150 power amplifier, introduced during the early 1970s. This used solid-state devices in the voltage-amplifier and driver stages, and tubes for the output stages, the reasoning being that, while solid-state devices were "better: than tubed ones in general (a contention still argued), tubes were better suited for delivering the voltage signal required by electrostatic loudspeakers. But true hybridization, where tubes and transistors were, literally used in a complementary fashion, did not arrive on the audiophile scene until Berning's TF-10 preamp.
In the TF-10, every amplifying "stage" consisted of a tube and a FET arranged in a configuration so that their nonlinearities tended to cancel one another. (The transfer characteristic-output current vs. input voltage-of a tube and a FET is a gradual curve, rather than the ideal straight line. But as the curves are complementary to each other, when the two devices are used together correctly, the result is a desirable straight line.) The TF-10 was introduced in 1979, and became Stereophile's reference preamplifier for some years after we reviewed it. Subsequently, Audio Research "discovered" this hybridization trick1 and incorporated it into what has become their most successful preamplifier-the SP-11.2
(1) Berning's literature refers casually to their "patented tube-FET hybrid networks, which only now other manufacturers are copying." Some patent!
(2) The situation seems more complex than this. According to a letter published in June (p.25), whereas the Audio Research SP11 uses a tube and an n-channel FET in cascode, the original Berning patent shows the two devices used as each half of a differential amplifier. Actually, the circuit in the Berning patent shows a p-channel FET with its drain connected to V-, its source connected via a resistor to the cathode of a triode-tube, the grid of which is grounded. The input is to the FET gate and the output from the tube plate, this being connected by a load to V+. It is not, strictly, a cascode, neither is it a true differential amp, though feedback can be taken to the tube grid, which can be used as an extra inverting input.
JGH's "Some patent!" remark in footnote 1 again oversimplifies matters. As I understand it, the US patent system, unlike those of the UK or West Germany, where a thorough search for prior art is carried out by employees of the Patent Office, fundamentally allows anybody to obtain a patent for anything. If this new patent does infringe an old one, the US Patent Office assumes that the owners of the rights being infringed will take to the courts to contest it. This may be efficient, regarding the costs of granting the patent in the first place, but results in the litigants often expending large sums to protect their rights, as in the recent Magnepan/Apogee case (see Stereophile May 1988 p. 49). It is also worth reading the biography of Major Edwin Armstrong, Man of High Fidelity, which includes a full discussion of the decades-long case between Armstrong and RCA over the latter's cavalier infringement of the former's FM patents, to see how the US system works in practice. -JA
The new Berning TF-12 also uses tubes and solid-state devices, but not in a hybrid arrangement. Instead, all signal-amplifying stages are tubes, while the transistors are used only in "supporting" roles, as such things as voltage regulators and electronic switches. The preamp has a single phono input fixed at 32dB gain, and five 32dB-gain line-level inputs plus a tape loop.
In each channel, the phono stage consists of both halves of a 12AX7 tube in parallel, followed by half of a 12AT7. (Its other half is a cathode-follower isolation buffer for the Tape output.) RIAA equalization is all in a feedback loop between the first and second stage. Phone input termination is 47k ohms, but a pair of tiny plug-in receptacles, of the kind used in mockup breadboards, allows other resistors to be added to obtain lower termination values.
The line section is a pair of 6DJ8s in a differential configuration, to minimize power-supply modulation by the signal, and to cancel nonlinearities without having to resort to the use of negative feedback. There is no output capacitor; the output is DC-coupled, servo-circuits being used to minimize the DC offset, and the output impedance is an unusually high 3000 ohms. This limits the permissible length of output interconnects to about 30' of standard shielded cable or 50' of low-capacitance (15pF/foot) cable, which is hardly a severe limitation. (Either of these lengths will knock 20kHz down by 1dB.)
All circuit boards are made of DuPont Teflon, which is claimed to have dielectric properties second only to polypropylene for audio applications. (It is the poor dielectric properties of most circuit-board materials which accounts for the oft-told stories of circuit designs which sound great when connected with wire and lousy when is duplicated onto a pcb.)
The TF-12's power supply uses an improved version of Berning's unique "switcher" technology, here called a "Resonant Regulated Power Conversion System." Instead of using the 60Hz power source straight, it is rectified and chopped to produce a 70kHz pulse stream, which greatly simplifies several other design elements. The "power transformers" can be made much smaller because they are handling a relatively high frequency; the power-supply storage capacitors get recharged 140,000 times a second instead of 120 times, and there are no stray hum fields to foul the sound. All tubes heaters are DC-powered, and are run at a slightly reduced voltage to greatly extend tube life. (My TF-10 never needed tubes during the four years I used it.)
The TF-12 is claimed to be a dual mono design because there are separate storage capacitor banks, but this is not what is usually thought of as dual mono. Dual mono means total channel isolation, from the AC line on. A single power supply for both channels does not qualify, no matter how well it is regulated or isolated. Let's keep our terminology straight here, people!
This preamp is unique in that is does not contain a single potentiometer. Volume and channel balance are achieved by means of two digitally controlled matrices of electronically switchable resistors. (Each resistor can be connected to ground via discrete bipolar transistor.) One of these matrices, using four different-value shunt resistors which can each be connected to ground, is located at the input to the high-level section; the other, with seven different resistors, precedes the preamp's output stage. All high-level inputs pass through a voltage divider consisting of a 20k resistor and a 39k resistor to ground, with their junction feeding the first array of four switchable resistors. The second array of seven has a similar configuration. When a certain amount of volume change is required, different combinations of resistors are switched in or out by the transistor switches, which are controlled by external voltage sources. Although there are only 11 switchable resistors in each channel, by selecting different combinations of them it is possible to achieve a gain-control range of 80dB in 1dB increments, and to maintain channel balance within 1dB over that entire 80dB range. On the other hand, channel balance can be adjusted in the same fashion, and remain constant at the new setting through as much of that range as is still available at either end. (With one channel higher than the other, it will "bump" the wide-open point before the other. If the volume is increased further, the gain of the two channels will start to "slip," relative to one another, until at the full setting, both channels will have the same gain again.) Two front-panel alphanumeric LED arrays show decibels of attenuation in each channel, from 0 (full up) to -79 (fully down), making it a lead-pipe cinch to return later to any setting or combination of settings.
The front-panel volume control on the TF-12 is unique, too. It is rotatable through a full 360 degrees, with no physical stop points. One full revolution of the knob causes 16dB of level change, and it takes five revolutions in either direction to span the 80dB range of attenuation settings. The knob operates an optical "encoder," which tells the digital controller how many 1/16-revolution increments it has rotated through, and in which direction. Even more unusual, the same control serves for volume and channel balance-adjustment, the function selectable by a nearby toggle switch. Flipping this up causes the knob to adjust balance, in the ergonomically correct direction; that is, rotating the knob clockwise lowers the left channel while raising the right. Unlike the volume-control function, though, the channel balance function affects the channels alternately rather than simultaneously. As you rotate the knob slowly, one channel changes in level by 1dB while the other remains unchanged. Further rotation in the same direction changes the other channel by 1dB in the opposite direction. This allows for exquisitely fine balance adjustment, yet when the switch is returned to its Volume position, the same balance settings are once again maintained through the entire range of volume settings.
What makes this arrangement all the more attractive is that it is a natural for remote operation. The usual way of remotely controlling incremental parameters like volume and channel balance is by means of VCAs-voltage-controlled amplifiers, in which a bias voltage is varied to change the gain of the amplifying device. The problem with VCAs is that their changing parameters affect their distortion as well as their amplifications. The more you turn them down, the more their distortion tends to rise. To date, though, the only alternative to VCAs has been to use the remote control to operate a small bidirectional motor which rotates a conventional volume control. This has no more effect on the sound than the control would anyway, but it is hardly what one would consider an elegant solution. Berning's is. In fact, it is the first truly innovative development in the control of analog signal level since the less successful (for audio, anyway) VCA came along. My hat is off to David Berning-this is a brilliant design!3
The Linn LK1 preamp and the Mission PCM7000 CD player use switched-resistor remote volume controls similar to the Berning's, although both these use MOSFETs to accomplish the switching. And a DIY preamp design from my ex-colleague Ivor Humphreys, published in HFN/RR in 1980, used a 32-step volume control implemented mainly with switched series resistors (though this used op-amps as the gain devices rather than tubes and CMOS chips as the switches). What is ingenious about the Berning design both the way it uses just 11 shunt and 4 series resistors to obtain 80 identical steps and the fact that the audio signal does not travel through the shunt resistors and their associated transistor switches.-JA
The TF-12's remote unit has a diamond-shaped array of four buttons, for volume Up and Down and balance Left and Right, and a small red LED adjacent to them lights up when any button is depressed, to show that the unit is transmitting. The remote also has a feature the preamp itself lacks: a Mute On and Off. There's even a Mute status indication: A row of small red dots lights up under the preamp's LED array when Mute is on.
The remote behaves unusually, too. To begin with, it's the only one I've ever encountered that has a window of useful operating distances. With other infra-red remotes, maximum distance is the only limitation; you can get as close to the main unit as you wish. But the TF-12's controller won't work if you get it less than about 6 feet from the receiving sensor (located between the LED readouts). It seems the receiver goes into overload, and ignores the incoming signal. Berning's remote system also has a form of error-correction built into it (for reasons which escape me), and this requires that the sensor receive three bursts of the same command before it responds. You don't have to press the button three times; just holding it down causes the signal to keep repeating. But because of the required triple redundancy, it takes a moment for the preamp to respond to any remote command. You can't just tap a button and expect it to respond; it won't.
Equipment used for my listening tests included the Ortofon MC-3000 cartridge in the well-Tempered Arm, the SOTA Star Sapphire turntable, a Sony CDP-705 ESD CD player, a Revox A-77 15ips open-reel tape deck, the Threshold SA-1 power amplifiers, and Sound Lab A-3 full-range electrostatic speakers. Audio interconnects were Monster M1000s, speaker cables were MIT Music Hose, and the listening room is extensively treated with ASC Tube Traps. Program material was some of my own tapes, and CDs and analog discs from Sheffield, Opus 3, Telarc, and Reference Recordings.
My review sample TF-12 was allowed a 48-hour quiescent warmup before testing. The sound did not change during the ensuing 12 hours of listening, but it was evident from the start that something was amiss with the left channel. It had significantly more hiss than the right, and produced rather pronounced clicks each time its attenuation incremented through a 10dB point, as well as slight clicks between all increments above -10dB. At full gain, the muted rushing noise was formidably loud, but at normal listening levels with analog discs, the noise was well below the disc surface noise. (The other channel was dead quiet at full volume.) I notified Berning of the problems and he agreed to send me a second sample, which has still not arrived as of this writing, five days later. The noise was unobtrusive enough at normal listening levels, though, that I felt it would not compromise my listening tests, so I proceeded.4
The replacement TF-12 arrived the day after I finished this review. Both channels were as quiet as the right channel had been on the first sample, and the level-switching clicks were now negligible, but the sound of the second sample was no different from the that of the first.
The first thing I noticed about the TF-12's sound was its incredible liquidity. There is no spurious texturing whatsoever; the sound has all the limpid clarity of a dewdrop in the morning sun. It reproduces an impressively wide, spacious soundstage, while stage depth and front-to-back perspectives are rendered as well as from any preamp I've ever heard. But, unfortunately (sigh!), it isn't perfect.
Berning's literature supplied with the TF-12 makes a strong point of the preamplifier's neutrality. I beg to disagree. My first and subsequently enduring impression was that it sounded somewhat "tubey." Listening to the line section only, I found the bass to be slightly emphasized, the midrange slightly forward and very alive-sounding, and the treble definitely on the soft side-albeit beautifully sweet and musical. Lower-midrange instruments were reproduced with great power and naturalness, and massed violins were smooth and silky, even if deficient in guttiness. There was much in the way of delicacy at the top, but little airiness or openness. Cymbals were a shade deficient in sheen, triangle were shy of the usual razor-sharp attack, and that last bit of guttiness from bowed strings was absent.
With the phono preamp in-circuit, the bass and treble colorations were more pronounced. Through the high-level section, the LF heaviness and HF softness were relatively slight; when the bypass test included the phono preamplifier stages, the colorations were judged to be about twice as severe, which is to say quite noticeable. But the bass heaviness did not extend to the midbass, which is unusual for tubed components, and the lows has superb detail and heft-again, something rarely heard from tubes. In fact, heaviness or not, the TF-12 had by far the best low-bass performance of any tubed (or hybridized) preamp I have heard.
With the Well-Tempered Arm, the LF emphasis had one positive result: For the first time, the low end from analog discs was deeper and stronger than that from CDs. A quick look at the schematic showed why: there is apparently no low-end turnover at 50Hz where the RIAA characteristic calls for the bass boost to cease, I would hesitate, though, to recommend the TF-12 for use with the SME V arm, because that not need any LF augmentation. When I tried it, with the same cartridge I use now, bass from black discs had as much heft as CDs. With the TF-12, I doubt if I could live with the result.
I cannot explain these coloration. A soft high end usually indicates extremely low distortion along with a drooping HF response or a phase lag through the upper middle range, or both. With an upper-range spec of 1dB down at 50kHz, I fail to see how the TF-12 could have either, and I was only using 1 meter of cable on the outputs, so I couldn't attribute the softness to cable rolloff. (I rarely measure the frequency response of products any more; even five years ago, not one ever failed to meet that spec.) As for the fulsome low end, I can only speculate that it relates to somehow inadequate power supply regulation, which can be more of challenge with tubes than transistors because of the much higher voltages and current involved.
It must be emphasized that these deviations from absolute neutrality are very slight through the line inputs and only moderately noticeable from phono sources. This is, overall, a superb-sounding preamp; it's just not as neutral as some others, such as the Threshold FET-10, which is priced in the same ballpark as the Berning. Since the TF-12 is not entirely without coloration, it must join the ranks of the many which I can strongly recommend only on the basis of a personal audition.
If you're attuned to and admire the things that tubes still do best, the Berning TF-12 will probably be exactly what you're looking for. If, on the other hand, you seek the best of both the tube and solid-state worlds, you may find the tubey qualities of this preamp, slight as they are, to be less than satisfying.
The David Berning Company