Cary Audio Design CAD-805 monoblock power amplifier Page 2

Cary's 805 (at last)
With its ornate knobs and gold-on-black finish, the beautifully crafted Cary Audio Design 805 looks like a '30s period piece. A delightful feature is the "tuning eye" on the front panel. The eye, which monitors the AC output voltage, closes completely at the 50W output level. The open chassis is dominated by the large—almost 8" from head to toe—211 output tube. This true three-element tube has been heavily used over the years in RF amplifiers. The filament, a directly heated thoriated-tungsten type, requires a current of 3.25 amps at 10VAC. The 211's plate dissipation in the 805's circuit is 94W (100mA at 940VDC), which means that, when you add the filament dissipation to that of the plate, the tube heats up like a 120W light bulb. This is a power triode!

There was a time when General Electric built 211s, as they did just about everything; the Chinese are now the major supplier. (Although the 805's 211 is stamped "Philips," it is, in fact, Chinese.) I'm told that this is a tube they've been producing for over 20 years, so quality should be very good.

The first question anyone is likely to ask is: "Why is this thing so darn heavy?" The main reason is the size of the 805's output transformer. Because the bias current for the output tube flows through the primary winding of the transformer, core saturation becomes a serious problem. An air-gap transformer has to be used, with its core laminations typically spaced 2-16 mils apart, in a precise manner. The air-gapped design gives more inductance for the same number of turns (footnote 3), but its inductance becomes proportional to the core cross-section. Maintaining a large primary inductive reactance relative to the load at very low frequencies therefore mandates the use of a very large, heavy core and lots of wire. A primary inductance of over 60 Henrys may also be required.

Practically speaking, it's rare to see a single-ended transformer with a full-power bass response lower than 20Hz. Additionally, all that wire generates both ohmic losses and significant winding capacitance, which limits the HF bandwidth. Unless interleaved primary/secondary winding techniques are used to control winding capacitance, HF extension beyond 15kHz is not possible.

The 805's output transformer uses an E-I core with large air gaps between laminations. Output taps are provided for 4, 8, and 16 ohm loads. Note that these taps are designed for efficient power transfer to the load. The actual output impedance of a single-ended amplifier is typically around 2 ohms, which gives rise to a poor damping factor. Defined as the ratio of the loudspeaker impedance to that of the amplifier output impedance, the damping factor is intended as a figure-of-merit to describe the amp's control or braking action over a loudspeaker cone. A solid-state design with high negative feedback may have an output impedance below a tenth of an ohm, generating a damping factor of well over 100. In contrast, a single-ended design can only muster single-digit damping factors, even into a 16 ohm load. However, there's some controversy over the need for, or significance of, higher (ie, greater than 10) damping factors.

The following circuit description is based on information provided by Dennis Had. The input signal is DC-coupled to the grid of a 6SL7 dual-triode operated in parallel as a single voltage-gain amplifier. The 6SL7's plate is AC-coupled to the control grid of an EL34 pentode wired as a triode (footnote 4). The EL34 dissipates about 19W of class-A power and produces 4W of audio power to drive the grid of the 211 through an inter-stage transformer of air-gap design. The 211 is cathode-biased, and the output stage is operated in class-A1 (no grid current) up to 25W output. As the drive to the 211's grid is further increased, operation shifts to class-A2. The 211's grid goes positive (and starts to draw current) while the output power with increased efficiency reaches about 50W. The ability to drive a 211's grid positive is made possible by the use of the inter-stage transformer coupling between the EL34 and 211. Unfortunately, substantial class-A2 power entails significant harmonic distortion, as the tube is eventually pushed into non-linear operation.

The 805's power supply features a full-wave bridge rectifier using avalanche-protected diodes. The smoothing filter is a pi network using a series choke of air-gap design. Variable global feedback from 0-10dB (from the secondary of the output transformer back to the input stage) is offered via a pot and an impedance selector switch. The owner is thus given the chance to experiment with the amount of feedback—from none to moderate—within his or her own system.

The On/Off switch powers up everything but the 211 tube. The Standby/Operate switch controls the filament current to the 211. The proper turn-on sequence is to power up the amplifier in Standby. After waiting a few minutes to give the input and driver stages a chance to stabilize, the 211 may be powered up by switching from Standby to Operate. (I mention this because there's no formal manual for the 805.) The amp sounds good right out of the box, but reaches prime time after about 25 hours' break-in.

The chassis runs very hot to the touch because both the rectifier bridge and the 211's cathode resistors are heatsinked to the chassis.

Tube territory
With about 20 clean watts at my disposal, the choice of partnering loudspeaker became a crucial issue. The two factors paramount to the selection process are sensitivity and impedance magnitude. I say "sensitivity" rather than "efficiency" because the former properly describes a speaker's bottom line in terms of how loudly it will play for a nominal watt input. Efficiency has to do with conversion of electrical into acoustical energy: how much of an electrical watt at the input terminals is converted into acoustical power. But if, for example, I were to use ten inefficient drivers instead of a single efficient one, I might find that the sensitivity of the multiple-driver speaker was higher. In a typically sized listening room with only a moderate amount of damping, a sensitivity of 90dB/W/m would allow a stereo pair of loudspeakers to reach 100dB peak sound pressure level at the listening seat with about 10 electrical watts. This represents an acceptable dynamic range for many listeners.

Unfortunately, the great majority of audiophile loudspeakers are direct-radiator types with sensitivities of 88dB or lower, which makes them either marginal or simply unacceptable for use with single-ended amps. Ideally, the partnering loudspeaker should be blessed with a sensitivity of at least 95dB/W/m to make life easier for the amp...but that's horn-loaded speaker territory. Dennis Had reports good results with such average-sensitivity speakers as the Monitor Audio Studio Six and the ProAc Response One and Two. The key to using speakers with sensitivities of less than 90dB is a compatible (benign) impedance magnitude.

Footnote 3: An air gap actually reduces inductance, the converse of what is stated here, and the gap necessitates more primary turns to maintain an adequate low-frequency bandwidth. In electrical terms, this reflected inductance is the parallel source impedance driving the speaker load. It is the determining impedance for the low-frequency cutoff. More iron in the core helps reduce the peak flux density in the core, improving linearity and power handling.—Martin Colloms

Footnote 4: As this review went to press, we were informed that the EL34 driver tube was going to be replaced by a 300B triode. DO will comment on the change this makes to the CAD-805's sound in a future issue.—John Atkinson