Consumer products are usually packed with features but they often fall short when it comes to audio quality. These products deliver a level of performance that is acceptable to most consumers, and they do so at very affordable prices. Nevertheless there is often a large performance difference between consumer and professional audio products.
One of our customers, Jeff Switzer, owns a Marantz AV8801 pre-pro and he took a look inside to see how it was built. His detailed analysis shows how consumer product cost constraints limit audio performance. Please understand that we don't want to single out Marantz. The construction of the AV8801 is similar to most other consumer audio products, and it may even be better than most. These products are designed to deliver many features at a very low cost. Audio performance is not a primary goal of most consumer products, and this becomes clear as Jeff walks us through the signal path of the Marantz AV8801 and AV8802. Jeff opened the hood on his pre-pro, searched the internet for schematics photos, and data sheets and then sent us his analysis without our solicitation. His analysis was so good that I thought it deserved to be published in our application notes. Jeff graciously agreed to grant permission.
Jeff's teardown analysis is a bit technical, but I know that some of our readers will appreciate the detail. For the rest of our readers, let me summarize by saying that there are real differences between consumer products and high-end professional audio products.
John Siau, Benchmark Media Systems, Inc.
In investigating the new Marantz AV8802, I learned a bit about how typical consumer-grade AV products handle audio signals. The following text is based on my investigation of the AV8801 and AV8802. I don't mean to “pick” on these units, since it is likely that any preamplifier-processor (pre-pro) that sells for a similar price has to make similar design tradeoffs. My understanding of these units is based on my unit, partial schematics, pictures of circuit boards, and datasheets (which are publicly available online). The concepts that follow should be correct, although some of the details may miss the mark, given that I don't have access to the complete schematics and data sheets for the units.
It appears that internal audio signals are kept small, normally 1 V rms (1.4 V peak) or less. The low signal levels allow the use of low-cost low-voltage integrated circuits, but these low signal levels affects the signal to noise ratio (SNR) of these units. The SNR is also affected by many other design decisions, but low voltages always make it more difficult to achieve a high SNR. For example, if the level of noise in a circuit stays constant, but the signal level is reduced, then the SNR will be reduced. The 1 V rms internal signal level also makes attenuation necessary at the analog inputs.
In the AV8802 and the AV8801, the balanced inputs are attenuated by at least 6 dB (perhaps even 12 dB) as they enter the unit and are converted to single-ended signals. This means that balanced analog signals are reduced by a factor of 2 or by a factor of 4 before they reach the internal signal path. This attenuation reduces the SNR of the audio system, but it is necessary to prevent overloading of the volume-control IC.
The volume-control IC will handle 2 V rms signals, given the 5 V to 7 V supply rails used in these products. But, the volume control IC delivers its best distortion performance when signals are 1 V rms or less, with distortion rising rapidly over that level. I don't know if the single-ended inputs are reduced in level. There may be a 2:1 (6 dB) attenuation on the single-ended inputs, but it is hard to tell without a full schematic.
One qualification to all this; the volume control IC used in the AV880X units is from Renesas. It is a discontinued unit, Renesas no longer makes this volume control IC, and the datasheet is not available publically and perhaps never has been. I have based these comments on the published specifications for the NJW1299 volume-control IC from JRC which appears to be very similar to the Renesas chip used in the AV880X units.
After the volume control, the signal for each channel goes to an output buffer. The gain of the output buffer appears to be unity for single ended signals and 6dB for balanced signals. The output of the AV8802 is rated at 1.2 V rms single ended, and 2.4 V rms balanced. The higher output of the balanced circuit is achieved by an inversion of the single-ended signal, to provide a non-inverted and inverted signal to the balanced output connector. Curiously, the buffer has 12 V rails, but the active circuitry won’t see input signals close to those levels because of the limitations of the the volume-control IC. The output circuitry is composed of discrete transistors and associated components instead of the more-typical op-amp output stages found in many products. This may have been an attempt to introduce some higher audio quality, or it may have been a cost-saving measure. The product literature claims high slew rates for these output buffers and claims that they outperform the slew rates provided by op-amp buffers. This claim is somewhat dubious in that op-amps have 10 to 100 times the slew rates required for this application. These higher slew rates provide no technical benefit, but they may provide some marketing value.
The output buffers use 12 V power rails so that resistors can be used as cheap current sources. This cost-saving technique only works when the signal levels are small relative to the voltage rails. Even so, the lack of a proper current source will add distortion to the output. There are ten of these pseudo current sources in each buffer circuit. In the higher-end two-channel Marantz products these current sources are implemented with active transistor and diode circuits, to provide a higher quality current source. These more complicated circuits provide a current output that varies less with voltage. In the AV8802 there are 15 output buffers. Implementation of the higher quality current sources would have required a total of at least 150 added transistors and 150 added diodes, plus 3 to 6 square inches of added circuit board space for each buffer (channel). Circuitry that can be justified in a two channel unit, just adds too much cost to a reasonably priced surround-sound AV product.
Marantz rates the balanced outputs of the AV8802 at 2.4 V rms. This limit makes sense when examining the internal components, but it is a fairly low signal level for a balanced output. The single-ended outputs provide 1.2 V rms. 1.0 V rms will drive a typical power amplifier that has a gain of 29 dB to an output level of 100 watts into 8 ohms. If the balanced output of the AV8802 is used to drive the Benchmark AHB2 amplifier at the high-gain setting (23 dB), or say a Bryston amplifier at its lower gain setting (also 23 dB) the pre-pro will be very near its 2.4 V limit. At an amplifier gain of 23 dB, it takes 2 V rms at the amplifier input to deliver 100 watts into an 8 ohm load at the amplifier output. The noise levels of the amplifiers will be very good at this level, but the system noise performance will be limited by the noise performance of the Marantz pre-pro.
It takes approximately 3.5 V to drive a Bryston 4BSST2 amplifier to its output limit of 300 watts into 8 ohms at the 23 dB setting. With this amplifier, the Marantz can only drive the amplifier to full output when the amplifier gain is set to 29 dB. At this higher gain setting, full output can be reached at 1.73 V. The Bryston 6BSST2, 9BSST2 and the Benchmark AHB2 amplifiers all have 17 dB gain settings which cannot be used with the Marantz pre-pro. At a 17 dB gain setting, 4 V rms is required at the input to reach a 100-watt output, and 7 V rms is required to reach a 300 watt output from the 6BSST2. The Marantz, and most other consumer-grade products cannot deliver these high signal levels. The Benchmark AHB2 amplifier even has a very low-gain setting of 9.2 dB that provides the best signal to noise ratio. At that setting, an input of 9.8 V rms is required to reach full output power. Levels this high can only be delivered by a Benchmark DAC2, or much more expensive preamps from companies such as Mark Levinson, Simaudio or Boulder. Professional studio equipment also operates at these very high signal levels, and most professional devices will drive the AHB2 in the 9.2 dB gain setting. No medium priced pre-pro can do this. Using the lower gain settings can significantly improve the SNR of the overall playback system, but this improvement is only possible with signal sources that have high voltages available.
Lower amplifier gain settings allow higher voltages at the input of the amplifier. This allows the source component to operate at a higher voltage level and this produces a better signal to noise ratio at the output of the source component. Each of the amplifiers mentioned above provide excellent performance at the higher gain settings, but the performance of the amplifiers improve slightly at the lower gain settings. But, the lowest amplifier gain settings cannot be used when the amplifiers are driven by typical consumer-grade products such as the AV880X series pre-pros.
The AV8802 has an internal two-channel AKM AD4490 DAC. The AV8802 uses a single opamp at each output of the AD4490. This circuit appears to be a cost-reduced version of the triple op-amp circuit shown on the AD4490 data sheet. This simplification probably impacts the SNR and may add distortion. Nevertheless, it does lower the cost and reduce the power consumption. Given the headroom limitations in the volume control section, this simplification may have little impact on the overall performance.
Large-value resistors are used in the single op-amp DAC output stage, and these resistors may add some additional noise. The inputs to the single differential receiver appear to have different effective input impedances. The lower of the two impedances is driving the inverting input, and it appears to be close to the limit recommended for the AD4490 output pins. These compromises are some of the consequences of eliminating two or the three op-amps on each DAC output channel.
Many owners of the AV8801 like to drive them with the balanced outputs from an Oppo BDP-105, which can deliver over 4 V rms. In such a configuration, the Oppo may need to be turned down, and much of its performance will be lost, due to the added noise and distortion of the AV8801 signal path.
I don't own any Emotiva equipment, but I've looked at their pre-pro. The unit just passes video without processing and this avoids the huge investments required to process video. One interesting feature of the unit is that it has two Cirrus Logic CS3318 8-channel volume control chips that have 9V rails. The extra headroom provided by the higher voltage rails would certainly help in handling higher audio signal levels. Adding the 9V rails probably added some cost to the unit. Volume control ICs with 12 volt rails are also available, but this would have added even more cost to the Emotiva design. Cost and board space are driving factors.
A DAC like the ESS ES9018 provides an excellent internal 32-bit digital volume control, but an analog control would still be required for the analog inputs. Providing separate volume control systems for digital and analog inputs is a path to good audio quality. This technique is used in the Benchmark DAC2 HGC. Analog inputs are controlled by a variable resistor tied to the motorized volume control knob. Analog inputs do not pass through a performance-limiting volume control IC. Digital inputs only use the 32-bit digital control. The position of the volume control knob sets the gain of the 32-bit digital processor. This hybrid approach is too expensive a solution for mid-range consumer AV products such as the Marantz and Emotiva pre-pros.
Customers of Marantz, Denon, Yamaha, Onkyo, etc., likely aren't willing to pay the significantly higher prices for state-of-the-art sound quality in AV units. These customers seem to be more feature-driven. That doesn't mean the units don’t represent good value. They are likely designed with the goal of providing the best quality and features at a given price point. State-of-the-art audio performance is not the primary consideration. Furthermore, the AV units are typically used for both video and audio, where video plays the larger role.
Jeff Switzer, April, 2015
Copyright, Jeff Switzer, 2015
Reprinted by permission.
At Benchmark, listening is the final exam that determines if a design passes from engineering to production. When all of the measurements show that a product is working flawlessly, we spend time listening for issues that may not have shown up on the test station. If we hear something, we go back and figure out how to measure what we heard. We then add this test to our arsenal of measurements.
Benchmark's listening room is equipped with a variety of signal sources, amplifiers and loudspeakers, including the selection of nearfield monitors shown in the photo. It is also equipped with ABX switch boxes that can be used to switch sources while the music is playing.
Benchmark's lab is equipped with Audio Precision test stations that include the top-of-the-line APx555 and the older AP2722 and AP2522. We don't just use these test stations for R&D - every product must pass a full set of tests on one of our Audio Precision test stations before it ships from our factory in Syracuse, NY.
Paul Seydor of The Absolute Sound interviews John Siau, VP and chief designer at Benchmark Media Systems. The interview accompanies Paul's review of the LA4 in the December, 2020 issue of TAS.
"At Benchmark, listening is the final exam that determines if a design passes from engineering to production. But since listening tests are never perfect, it’s essential we develop measurements for each artifact we identify in a listening test. An APx555 test set has far more resolution than human hearing, but it has no intelligence. We have to tell it exactly what to measure and how to measure it. When we hear something we cannot measure, we are not doing the right measurements. If we just listen, redesign, then repeat, we may arrive at a solution that just masks the artifact with another less-objectionable artifact. But if we focus on eliminating every artifact that we can measure, we can quickly converge on a solution that approaches sonic transparency. If we can measure an artifact, we don't try to determine if it’s low enough to be inaudible, we simply try to eliminate it."
- John Siau