Manufacturer's Specifications:
Usable Sensitivity: Mono, 11.3 dBf; stereo (with/without special circuitry), 16.3 dBf/34 dBf.
Fifty-dB Quieting Sensitivity: Mono, 16.1 dBf; stereo (with/without special circuitry), 21 dBf/37 dBf.
S/N: Mono, 82 dB (@ 85 dBf); stereo (with/without special circuitry), 74 dB/74 dB (@ 85 dBf).
Stereo Separation (Wideband I.f.): (With/without special circuitry) 45 dB/45 dB @ 1 kHz, 30 dB/36 dB @ 100 Hz, and 15 dB/36 dB @ 10 kHz.
Alternate Channel Selectivity: Narrow, 90 dB; wide, 35 dB.
Capture Ratio: 1.0 dB. AM Suppression: 50 dB, 65 dB with special circuitry activated.
Image Rejection: 110 dB. I.f. Rejection: 110 dB.
Spurious Rejection: 110 dB.
Power Requirements: 120 V a.c., 60 Hz, 15 watts.
Dimensions: 17 1/2 in. (44.5 cm) W x 3 1/2 in. (8.9 cm) H x 12 1/2 in. (31.8 cm) D, less rack-mount hardware.
Weight: 11 1/4 lbs. (5.1 kg).
Price: $549.00.
Company Address: P.O. Box 664, Woodinville, Wash. 98072, USA.
Fig. 1--Block diagram of Carver TX-11 tuner's special noise-reduction and
multipath-reduction circuitry.
Bob Carver seems to have done it again! This time he has turned his inventive genius to FM stereo and come up with a tuner which long-suffering fringe-area residents and those plagued by multipath distortion and interference have probably been praying for since the FCC approved an FM stereo transmission system that degrades stereo signal-to-noise by up to 23 dB! As is often the case with inventor types, Carver's TX-11 tuner was a bit late in reaching dealer's shelves, and even later than promised in reaching my lab. What Carver finally delivered to me was well worth waiting for. Furthermore, as is also fairly typical of inventor types, the version he brought me is really generation two, in that it incorporates modifications which are direct results of conversations that Carver and I had last June, at the Consumer Electronics Show, when Carver first described and demonstrated his tuner to me.
As originally designed, the TX-11 tuner offered a very significant reduction in background noise and multipath interference when tuning in to weak-signal FM stereo stations. Through a combination of four different circuit innovations, Bob Carver had managed to clear up these reception problems without losing audible or apparent stereo separation. He had, in fact, provided stereo reception with the noise and distortion levels of mono, no small feat. Upon further questioning, however, I learned that he had sacrificed steady-state, tone-measured separation significantly.
Product reviewers (such as yours truly), when subjecting the tuner to simple lab-bench measurement, would be shocked to find separation figures dropping as low as 10 or 12 dB at some frequencies. As Carver pointed out (and as many psycho-acousticians have confirmed over and over again), 10 to 12 dB of separation is all that is really needed to maintain a good stereo effect. But I countered that less knowledgeable audio enthusiasts would never accept a tuner with such low measured separation, no matter how good the stereo effect seemed to be. Apparently, Carver agreed, for he went back to the proverbial drawing board and has now come up with a version of the TX-11 that not only sounds great and improves weak-signal stereo reception by several orders of magnitude, but even offers respectable single-tone separation measurements on the lab bench. I'll get to the circuitry presently, but first let's take a look at the tuner's physical layout.
Designed to accommodate rack-mount adaptors, the Carver TX-11 gunmetal-finished front panel has a power on/ off switch and power indicator light at its left end. The center section of the panel is equipped with a smoked-gray transparent plastic "dial glass," behind which are a digital frequency display area, a six-LED signal-strength indicator, a stereo indicator light, and another indicator which illuminates when the frequency-synthesized tuner correctly locks in an incoming signal. Below the dial area are 16 FM station preset buttons, a "Memory" button for assigning favorite stations to specific presets, a manual/automatic scan selector, and a narrow/wide i.f. bandwidth selector button. The rightmost section of the panel contains "Up" and "Down" tuning buttons plus two buttons which activate the circuits that make this tuner so outstanding. Both are identified as "Asymmetrical Detector" buttons, with one of these further designated "Noise Reduction" and the other as "Multipath Reduction." In his owner's manual, Bob Carver calls his circuit an "asymmetrical charge-coupled FM detector." I haven't got the slightest idea what that name has to do with the way the circuits work, now that they've been thoroughly explained to me. But if it makes Carver happy to devise esoteric sounding names for his circuits, that's okay with me-so long as the circuits work as well as these do.
The rear panel of the TX-11 is equipped with 75and 300 ohm antenna terminals and a pair of output jacks. There are no external fuses, and output level is fixed.
Fig. 2--Mono and stereo FM quieting and distortion, wide-band mode, without
use of special Carver circuits.
Fig. 3--Stereo quieting and distortion, wide-band mode, with noise- and multipath-reduction
circuits activated.
Fig. 4--Distortion vs. frequency, mono and stereo, with and without special
circuits (wide band mode).
Noise- and Multipath-Reduction Circuitry
As Carver explains it, there are four separate circuit innovations which contribute to the noise reduction in weak signal stereo reception when the noise-reduction and multipath-reduction buttons are depressed. The block diagram of Fig. 1 shows all the circuit elements involved in the TX 11. With the exception of the block at the Lipper left of Fig. 1 (the Delta-Q Detector), all of the new circuitry occurs after detection of the composite audio signal via this detector and the following standard PLL multiplex detector. The Delta-Q is a variable-bandwidth FM detector. Simply stated, under weak-signal conditions, this detector's bandwidth is automatically narrowed to the minimum needed to pass FM sidebands for reasonable results. Under weak-signal conditions, an improvement of between 2.5 and 3.0 dB in signal-to-noise ratio takes place, at the expense of some increase in high-frequency harmonic distortion and some loss of high-frequency separation. This element of the system is perhaps the least important of the four and is also the simplest to understand: Reduce overall bandwidth and you reduce noise.
The remaining three elements of Carver's system are a result of his deep and penetrating investigations into both psychoacoustics and the standard sum-and-difference matrix techniques used in the transmission and recovery of the left and right stereo FM signals. Having recovered the L and R signals from a conventional PLL multiplex decoder, these signals are first recombined into L + R and L-R signals.
The L + R constitutes a mono signal, while the L-R signal conveys the "stereo" or "difference" information. In Carver's approach, the L-R signal can be thought of as containing two types of signals: L R information having random phase and L-R information having specific localizing phase information. The localizing information provides the left and right stereo image location in a sound field. The non-localizing information provides the stereo ambience contained in a stereo sound field. Carver maintains this non-localizing information is completely redundant with information that is already available in the L + R signal. But the L + R signal (mono) is, as we all know only too well, a lot quieter (about 23 dB worth) than the L-R information. It is also less vulnerable to multipath effects.
What Carver does, therefore, is to create a sort of synthesized L R signal, using the quieter L + R signal, to provide the randomized or ambience information that does not help in stereo localization. If you follow the line leading from the L + R output of the first matrix block, you will see that it leads to phase-randomizer and spectral-shaping blocks, the output of which has been designated as (L R)'. However, since stereo localization must also be derived from the real L-R signal, this signal is also directed to a series of four blocks (a pair of log amps, A and B, a differencing amp, and an anti-log amp) to establish the instantaneous ratio between true L + R and true L R. The true L R signal is also fed to a series of blocks called a leading-edge detector and a leading-edge amp. These blocks take advantage of a psychoacoustic phenomenon known as the precedence effect. When fast, short-term L-R information critical to the localization process occurs, these blocks allow that information to become part of the "mix" that occurs at the summing junction.
So, thus far we have the contributions of the artificially created (L R)' and the leading-edge detector at this summing junction, the output of which feeds the second matrix detector, along with the unmodified L + R signal. If Carver had stopped here (as he had, before our conversation last June in Chicago), the tuner would have sounded just fine.
However, critics and reviewers might have been disturbed when they fed in single-tone test signals to an FM generator and discovered that under those conditions, separation was reduced to 12 dB or even less at some frequencies.
To get around that, Carver added a few more blocks, shown at the lower right of the diagram, taking advantage of the fact that totally discrete L and R signals are available from the regular PLL multiplex decoder in the system. When L-only or R-only signals modulate the r.f. signal, that information is conveyed back to the summing junction via another pair of log amps, a differencing amp, an anti-log amp, and finally the a.g.c. amp. In this way, high orders of separation are maintained during single-tone tests.
I realize that I have skimmed over a complex combination of circuits, but space simply does not permit a more detailed explanation of the Carver TX-11 circuitry. What's important is that the system does what is claimed for it, both in terms of improved reception of weak signals and retention of good stereo imaging and separation. That it does both became apparent as I went through my usual sequence of tuner performance measurements and listening tests.
Measurements
I discovered very early on that the special circuits of the Carver TX-11 offer improvements primarily in the stereo mode. Improvement in mono is negligible (about 1 dB of signal-to-noise improvement), but of course, it's stereo that suffers most from noise and multipath. Accordingly, comparisons of performance with and without the special circuits are confined to stereo.
Figure 2 shows the usual plots of noise and distortion (at 1 kHz) versus input signal strength for mono and stereo operation, without the special circuits engaged, and using the wideband i.f. mode. Usable sensitivity in mono was 11.5 dBf, while in stereo it was 18 dBf. Fifty-dB quieting under these conditions was at 15 dBf in mono and 38 dBf in stereo. Ultimate signal-to-noise for mono was just short of 80 dB, while in stereo I measured 75 dB. Distortion at 1 kHz was an extremely low 0.032% in mono and an almost as low 0.06% in stereo.
Figure 3 is a plot of stereo noise and distortion versus signal input when the special circuit buttons are both depressed. I didn't bother to re-plot mono in this figure, since results are, as I said, essentially the same as they were for Fig. 2. But notice what happens at weak signal levels in stereo. For a 20-dBf stereo signal, for example (that's about 5.5 µV), the signal-to-noise reading obtained without the special circuits was 32 dB, making it a barely listenable signal. When the special circuit buttons were depressed, the same input signal strength resulted in a signal-to-noise reading of 48 dB-an improvement of 16 dB! And at 22 dBf, the 50-dB quieting point had been reached. Furthermore, as shown in Fig. 4, mid-frequency harmonic distortion remained virtually constant, while distortion at the frequency extremes actually improved slightly when the special circuits were switched in.
Fig. 5--Frequency response and separation, wide-band mode with and without
special circuits activated.
Fig. 6--Same as Fig. 5, but in narrow band mode.
Fig. 7--With special circuits off (A), crosstalk of 5-kH modulating signal
into unmodulated channel is typical of most tuners. With special circuits on
(B), separation diminishes, but crosstalk distortion components are no longer
visible
Fig. 8--Automatic variable bandwidth FM detector "opens wide" for
strong signals (A), but closes down to just over ±75 kHz linear range for weakest
signals (B).
Fig. 9--At weak signal levels, activating the noise- and multipath-reduction
buttons significantly reduces background noise, whether incoming signal is
modulated (A) or not (B).
That brings us to the question of stereo separation. In Fig. 5, I have plotted frequency response and separation versus frequency for both operating conditions (with and without the special circuits). The uppermost plot in the 'scope photo shows uniform response from 20 Hz to beyond 15 kHz; the lowest curve shows output in the unmodulated channel for that mode of operation (without the special circuits). The slightly displaced frequency response curve just below the flat one shows a minor dip at around 500 Hz caused by some phase-cancellation or "comb-filter" effects of the special circuitry, amounting to no more than 2 dB maximum. Of more importance is the fact that single-tone separation for a left-only or right-only signal remains more than adequate from one end of the audio spectrum to the other.
Specifically, at 1 kHz I measured 52 dB of separation without the special circuits and 41 dB with the circuits activated. At 100 Hz, separation went from 42 dB to 34 dB and at the most difficult 10-kHz test frequency, turning on the circuits reduced separation from 42 dB to a still satisfactory 26 dB.
All of these plots were repeated for the narrow-band i.f. operating mode and, surprisingly, mid-band separation actually seemed to be higher than before without the special circuits, but homed in at almost exactly the same separation values when the circuits were turned on (see Fig. 6). In conducting the usual 5-kHz crosstalk analysis measurements for this tuner (in which I modulate one channel with a 5-kHz signal and observe crosstalk and other distortion products in the unmodulated channel's output), I discovered another surprising benefit of the Carver circuits.
Figure 7A shows results obtained without benefit of the special circuits. The tall spike at left is the output (5 kHz) of the modulated channel. Sweep is linear this time, from 0 Hz to 50 kHz. The shorter spike contained within the taller one is the fundamental 5-kHz crosstalk output from the unmodulated channel output. Going further up in frequency, we see harmonic distortion components of 5 kHz as well as subcarrier product outputs at 19 and 38 kHz. In Fig. 7B the tests were repeated with the special Carver circuits turned on.
While 5-kHz separation is significantly less, notice that there are no longer any visible amounts of harmonic distortion or subcarrier products in the unmodulated channel output! Those measurements relating to the conventional portions of the FM tuner turned out to be pretty much as specified by Carver. Selectivity in the "narrow" bandwidth position measured around 88 dB, while in the "wide" mode it decreased to around 37 dB. Capture ratio was 1.0 dB, as claimed, while AM, i.f. and spurious rejection were all in excess of 100 dB (the limits of my test equipment). The oscilloscope photos of Figs. 8, 9 and 10 are all intended to show demonstrable operating characteristics of, and performance improvements provided by, the novel circuitry devised by Bob Carver for this tuner. Figure 8A shows the relative bandwidth of the FM detector when it is being fed a strong input signal, while in Fig. 8B signal strength has been reduced and the linear portion of the detector bandwidth has decreased accordingly.
Figure 9 tells the story of noise reduction about as well as it can be told in print. To believe it, you have to hear the difference for yourself when a weak signal is "cleaned up" by pressing those two front-panel buttons. The lower trace in Fig. 9A represents a noisily received test signal. The upper trace represents the same signal as it appears at the output of the tuner after the two special-circuit buttons are depressed. In Fig. 9B, the same sort of comparison is made, but this time there is no test tone modulating the r.f. carrier except the 19-kHz pilot signal needed to make it a stereo and therefore "noisy," weak signal.
The two 'scope photos of Fig. 10 are intended to show how the special Carver circuits help to reduce the effects of multipath reflections in received stereo signals. Notice the "pinched" ends of the stereo program display in Fig. 10A; they indicate the presence of rather severe multipath distortion. For the photo shown in Fig. 10B, the same received signal is displayed, but the Carver special circuits have once more been activated, effectively cleaning up the multi path distortion.
Fig. 10--Multipath distortion, indicated by compressed tips of this stereo
'scope display (A), is virtually eliminated (B) by TX-11's special circuitry.
Use and Listening Tests
As much as the 'scope photos, spectrum analysis charts, and graphs show about this novel Carver tuner, the significance of its design can only be fully appreciated by setting up the unit and tuning to the weakest, most unacceptable stereo signals you can find, then pushing in those two magic buttons. In my listening area, of the 50 or so stereo signals that I can pick up (using a good outdoor antenna, of course), at least 10 or 12 are so noisy in stereo that I normally have to switch to mono to merely be able to understand the content of the program. For every one of those 10 to 12 stations, depressing the noise-reduction and multipath-reduction buttons on the Carver TX-11 tuner made the reception quality acceptable and listenable! Separation was still there; only the background noise had been diminished, and with it, much of the sibilance and hissy edginess so characteristic of multipath interference.
To be sure, in my situation there are plenty of good, strong signals that really can't be significantly improved by the introduction of the special Carver circuits. Remember, it works best for weak signals but does little for strong signals that are noise-free to begin with. Still, I can easily see where many FM radio enthusiasts who live in less ideal areas will welcome this tuner as their first and only solution thus far to good, noise-free FM stereo.
Could I find anything to criticize about the Carver TX-11? Yes, one minor point. When you decide that an incoming signal is too noisy and want to press those buttons on the front panel, there's a time delay of about two seconds before the special circuits latch on and do their things.
Hearing memory being notoriously short, I am concerned not that users won't have the patience to wait for the improvement to be heard, but that with that interruption in sound, they may not fully appreciate the amazing degree of improvement that takes place when the sound finally comes back on. That would be a pity, since to my way of thinking, the Carver TX-11 is one of the few important circuit developments to come along in the FM radio field in several years.
And I would hate for anyone to be less than fully aware of its significance!
-Leonard Feldman
(Source: Audio magazine, Dec. 1982)
Also see:
Carver Compact Disc Player (Ad, March 1985)
Carver Model M-400 Magnetic Field Power Amplifier (Nov. 1980)
= = = =