Audio Calibration

Anthem Room Correction (ARC) System - Part 1

ARTICLE INDEX

And Now . . . The ARC Review

I am not going to discuss the basic operation of the ARC panels and the basic measurement procedures at this point in the review. Some of this was covered in the other SECRETS reviews of Anthem products with the ARC system. Refer to the review of the Anthem MRX 700-7.1 AV Receiver, by Robert Kozel, Senior Editor, for example.  The value add for this review is to determine if ARC delivers on what the PC display predicts by performing independent electrical and acoustical measurements.

The verification divides into three sections:

  1. System Performance for Full-range Speakers
  2. Subwoofer Augmented Performance
  3. Use of the Advanced Options

I was given access to the Anthem D2 for this review. While the video cards have been changed for the current top of the line Anthem Pre/Pros, the analog sections are unchanged from what I tested. The Anthem Pro/Pros are designed and built in Canada. The look and feel of the D2 was more like expensive laboratory test equipment than consumer audio products. The internal construction was also at the standards of advanced test equipment.

The size of the filter bank is reduced in the Anthem MRX AVRs. The ARC PC software detects the maximum size of the filter it can construct and transfers the appropriate amount of coefficients to the MRX components. The Anthem ARC post EQ display, when the MRX is connected, is calculated using what will be the electrical room correction curve synthesized with the smaller filter bank.

Since the measurement system is identical for all Anthem models, Anthem engineers were able to convert measurement sets that I sent them so the corrected room frequency response plots represented the performance of the MRX device. I could then see how the performance was changed with a MRX AVR. The system still performs well in the MRX implementation.

The bulk of this review presents the actual performance of the ARC in one room, but I made measurements in three rooms. I had available main or satellite speakers from AR (a long discontinued Ken Kantor design), Infinity, KEF, Monitor Audio, NHT, Phase Technology, and Pioneer. Obviously, only a subset of these speakers was used in each room. The 3-way eight-order DSP based crossovers of the NHT speaker (model Xd) present a significant challenge since very high order filters are used, introducing significant group delay. Despite this hurdle, ARC preformed identically to the other tested speakers.

I also had subwoofers from two manufacturers (NHT and Sherwood). Note that I used no speakers from Paradigm. I have found some room EQs work best with speakers that come from the same company This is the reason I utilized numerous speakers and rooms to get a sense of the robustness of the EQ system, and it turned out to be very robust. Indeed I never saw any sign of the system failing to produce a good solution for correcting the room and speaker aberrations.

Obviously, you can create a condition that is impossible for any room correction system to correct. Under such conditions, the uncorrected system could hardly be characterized as a listenable audio system. ARC may issue a warning message during the microphone measurement process that the room response is not sufficiently accurate to correct. If the problem is speaker or seat placement, the ARC Quick Measurer tool can help you quickly find better placements.

The room's acoustic response was captured using my independent acoustical measurement system. This is the Acoustisoft RPlusD measurement system (www.acoustisoft.com). Having used Acoustisoft products for more than ten years, I have confidence in their accuracy, especially when averaging large data sets. My microphone is from IBF-Acoustic and it is recalibrated to a reference microphone every few years.

In using the ARC, I made nine measurements with the ARC microphone by placing the microphone around the perimeter of an 18-inch square to yield eight points. The ninth point was the center of the square. I then proceeded to make my measurements with the RplusD software. I placed my microphone in the same position as the ARC microphone and captured a response curve. I then moved my microphone to the next spot I had placed the ARC microphone and captured another response. I continued this process until I made nine coincident measurements, then averaged the results. Since the microphones were moved in space, this is referred to as a spatial average. ARC requires a minimum of 5 different placements of the microphone and allows up 10.

Some are going to criticize me for only using frequency domain analysis. The importance of time domain correction is controversial and the subject of current research. It is important to separate the time domain response of the speaker, which is not directly related to the speaker frequency response (technically this is called a non-minimum phase system), from the room response. The Anthem room correction system does not attempt to try to equalize the speaker to produce a linear phase post equalized response. Some electronic room equalizers will do this, although not at low frequencies. For example the Trinnov room equalizer as implemented in the Sherwood R-972 does time equalization.

Room response variations are minimum phase with a few exceptions, unlike a speaker, and the frequency and time domain response are directly related. Again significant research has been addressed to the validity of this statement and its implications for electrical room equalizers. A summary at the layman's level can be found in the text by Dr. Toole (Sound Reproduction, Loudspeakers and Room, Focal Press, 2008; chapters 12 and 13).

For those with the Toole text, it is best to carefully review the sections on waterfall and other time-domain graphics. The text clearly indicates that the parameters required to produce a plot with usable data are often not well understood (Section 13.5). Proper setup and interpretation of waterfall and similar plots are especially difficult to run and interpret. Some measurement tools designed for consumer use make very nice looking plots but do not provide the user with the ability to set up the measurement correctly. When using such tools, it is possible to convince oneself that large changes in the time domain can occur with no visible change in a properly set up frequency domain plot.

I spatially averaged over nine points to reduce noise in the curves and allow the smoothing of the curve to be reduced. I used 0.1 octave smoothing for these curves. 0.1 octave smoothing is more revealing in both the response variation of the speaker and the ability of the room EQ to squelch the variation than the more typical value is 0.33 octave smoothing. The ARC displays appear to be at a 1/6 (0.17) octave smoothing.

I also examined the accuracy of the frequency response display shown on the PC display of the ARC software. With an accurate display, one need not deploy an external room measurement system to observe the response before room correction. However, the post-correction plots will be optimistic since they are derived from the pre-correction measurement set as distinct from new acoustic measurements with the room equalizer in the circuit.

The post-correction display is generated by combining the pre-correction measured data and the electrical filter response curve that the Anthem hardware will produce as the signal travels into the device, through the DSP filter, and finally out the power amp (preamp out) terminals. Using the measured data, the parameters for the correction filter are generated by the proprietary ARC signal processing algorithms. Given the limitations on the ability of the DSP correction filter to mirror the uncorrected response, as a result of finite resources of the DSP in the AVR, the post-correction response curve is not going to be ruler flat in the speakers passband.

The ARC post correction curves do a good job of showing the accuracy of the algorithm which designs the DSP correction filter. In the formal verification section, I will also show measurements of the electrical correction that results when DSP coefficients are transferred from the PC to the Anthem D2 Pre/Pro. This reflects the DSP filters response with the ADC in front of it and the DAC at its output.

This electrical measurement is made by inserting a frequency sweep test tone into an analog input and measuring the change in response at the preamp output. I call these electrical inverse correction curves because they represent an approximate inverse of the room response above the speaker low frequency cutoff to the maximum frequency it is desired to correct.

Using the six-channel analog input, I could measure the electrical inverse correction curve of all channels except the rear channels since they have no inputs. All main channels showed the same performance.

I should note that not all AVR have ADCs on all the six channels of a multi-channel input. If the ADCs are absent, room correction cannot be applied to those analog channels. You must supply a multi-channel digital input, which for the consumer is via HDMI.

Below is the a screen shot of one frequency response curve shown on the PC with the ARC software running. The bottom curve shows that the scaling in the horizontal and vertical direction of the ARC PC frequency response curves can be changed.

One simply places the mouse at the starting point you want to window, and holds the right button down. The mouse then is moved over the section you want to expand. Finally releasing the right bottom yields the zoomed in graph. The Auto Scale button on the top menu restores the graphs to the original X and Y axis.

I used the zoom function for all screen shotss of the ARC display below. The ARC frequency response plots in the next section were windowed so as to match the independent acoustical measurements.

While I am showing the ARC display for one channel as a minimum, it will show two plots for a stereo setup, and it will show all eight plots for a 7.1 system.