Harman Kardon HK 990 Stereo Integrated Amplifier with Digital Room Correction and Dual Subwoofer Bass Management – Part I
- Written by Dr. David A. Rich
- Published on 20 October 2011
- Harman Kardon HK 990 Stereo Integrated Amplifier with Digital Room Correction and Dual Subwoofer Bass Management – Part I
- Page 2: What Makes the HK 990 Unique?
- Page 3: The Analog Electronics
- Page 4: HK 990 EzSET II Room Correction Introduction
- Page 5: HK 990 - Problems Identified in the Measurement of EzSet/EQ II
- Page 6: Best-Case Measurements for the HK 990
- Page 7: HK 990 Performance with a Subwoofer
- Page 8: Deep Dive to Examine the EQ 2 Setting
- Page 9: Understanding the Problem with EQ2: An Alternate Perspective
- Page 10: Conclusions About HK 990 Room EQ
- All Pages
Understanding the Problem with EQ2: An Alternate Perspective
The Anthem ARC room EQ (Figure 16; a full set of measurements will be presented for this system soon) has a variable max EQ frequency. I set it to 700 Hz, which is consistent with the peak of the uncorrected curve. The acoustic frequency-response plot from the Anthem ARC display is different than the image Acoustisoft RPlusD produced in prior Figures. The red curve is the uncorrected response (9pt average at one-sixth octave smoothing). The green curve is the corrected response. Note how the curve moves to match the uncorrected response curve (red) around 700 Hz. Above 900 Hz, the corrected curve tracks the uncorrected curve (red on top of green). For this to occur, the electrical correction curve must be at unity gain above 900 Hz (not shown).
Figure 17 illustrates the response of the HK 990 to this condition if its maximum correction frequency were 700 Hz. The HK 990 would brighten the speaker's sound by setting the level at the peak, which is about 2.5 dB.
With the same speaker and the maximum frequency of the Anthem ARC at 1.5 kHz, there is a dip (Figure 18). The Anthem ARC gravitates to the dip (red on top of green), signaling a shift of electrical correction curve to unity gain above 1.5 kHz.
Figure 19 shows what the HK 990 would do. The curve above 1.5 kHz matches the level of the dip and the electrical correction curve is 1.5dB down above 1.5 kHz.
The EQ 3 Setting: A Detailed Interpretation
Looking at the curves in Figure 10 - 15, it can be seen EQ3 (blue curves in the measurements) is reducing the top-end of the speaker, above 1 - 3 kHz depending on the speaker and room placement. The setup for figure 12 is the exception. EQ3 made the speaker sound different than its intrinsic voicing It is a matter of preference whether EQ3 is an improvement over no correction.
Since EQ3 appeared to be working properly for full range speakers from the data above I committed time to a nine-point average for the Infinity C336 in two new positions. The results pre- and post-EQ3 are shown below in Figures 20 for the speaker in position 4.
Figure 21 below is the pre- and post-equalized response for the speaker moved to position 5.
The figurers 20 and 21 illustrate the flat response of the Infinity C336 above 400Hz before EQ is confined to +/- 2 dB. Below 400Hz, my room dominates the response; however, once EQ3 is activated the response below 400Hz improves significantly. This is not the best I have seen at this price point, but it is close. Also note the push of the low end as was observed in the earlier Figures. This time the response is pushed down by about 15Hz as a result of the 6dB maximum lift in the electrical correction curve.
Harman engineers were not able to address whether EQ3 was performing properly. The step starting at 500Hz suggests a problem matching the near-field correction curve to correct variations in the speaker's response to the part of the curve corrected for errors due to the room. They also could not confirm if the downward slope above 1kHz is the desired result.
There is a lift above 8kHz at the post correction response. I expect, but cannot confirm, if this is the result of the microphone's top end response dropping. The system cannot tell a drop in level at the microphone from one occurring from the speaker. Accordingly, the software infers the speaker needs some lift in the high end. I have seen this problem in other room EQs with small un-calibrated microphones.
I am not overly concerned about the EQ3 issues. EQ3 is designed to correct a poor response curve (rough or significant octave-to-octave variation) of the speaker and it is unlikely someone would spend $2600 on the HK 990 and match it to bad speakers. Not all speakers that win the "best of the year" awards are flat on-axis. This is a result of the designers voicing of the speaker and how the designer handles the difference between the on-axis and off-axis responses. EQ3 removes the designer's voicing. I advise limiting the correction to a 500 -1kHz limit and leaving the voicing intact in stereo. Multichannel has a different set of issues and may benefit from full range correction. When working correctly, the EQ2 mode is a significant value-add for the HK 990 over a system that just does full correction.
Subjectively (and this is my sole subjective observation), the correction below 400Hz before the bass push is similar to my preferred EQ systems. The push in the bass was not as artificial sounding as I have heard with AVRs whose EQ systems also attempt to flatten the frequency response below the port frequency. The AVRs' amplifiers did not have the current sourcing ability of the HK 990. The parallel configuration of the Infinity C336's three woofers makes it difficult to drive when the gain is up 6dB. To confirm this observation, the distortion of the speaker in the area that the room EQ has pushed the bass must be measured. In my lab, these measurements are not feasible owing to my combined problem of high ambient room noise at low frequencies and close-in neighbors with little tolerance for 35Hz test tones. In any case, the C336 speaker sounded better with amplification when used with the Anthem ARC room EQ that prevents bass-push. Since the Anthem ARC is PC-based, post-corrected bass response can be panel controlled.
The EQ off is not normalized to EQ3 as can be seen be in figures 20 and 21. The off mode is about 2dB higher. You thus need to adjust the volume control to normalize the levels. This increases the time it takes hear the sonic change when EQ3 is activated.