Integrated Amplifiers

Harman Kardon HK 990 Stereo Integrated Amplifier with Digital Room Correction and Dual Subwoofer Bass Management – Part I


Best-Case Measurements for the HK 990

At the listening seat, I co-located my microphone with HK's. I could make measurements to 20 Hz by setting the gate time to 750 msec. With such a large gate time, the measurements are equivalent to an RTA, but with finer resolution than one-third octave. Figure 1 shows my acoustic measurements.

Normally, I would spatially average over nine points to reduce noise in the curves, but given the problems, I found myself making so many measurements with different speakers and rooms that there was no time to complete averages for each measurement. Without spatial averaging, the curves are too noisy to allow for 0.1 octave smoothing. As a workaround, I used a wide smoothing of one-third octave for the curves.

Figure 1 is a best-case result. The low-frequency room modes are well suppressed. Above 1.5 kHz the uncorrected curve and the corrected curve in the EQ2 mode match closely. The corrected curve was made with an Infinity Classia C336 speaker in the larger of the two rooms.

Figure 2 is the electrical correction of the HK 990 from the CD input (analog) to the preamp output. Figure 1 is repeated below the curve. At a given frequency, when the level of the preamp out increases relative to flat, the acoustic output of the loudspeaker increases by the same amount. For the HK 990, I show the corrected response for EQ2 and EQ3. EQ1 only involves the subwoofer not used here. Figure 2 identifies the 6 dB push in the electrical correction curve below 45Hz, well below the frequency that the port has unloaded for the C336. The net result is a cone flapping around when they try to reproduce a signal when the woofers are unloaded.

In Figure 2, I tried to normalize the scales of the acoustic measurements and the electrical correction curve measurements (Preamp out / CD in) to illustrate the one-to-one correspondence between the change in the acoustic response and the change in the electrical correction. Unfortunately, jamming two curves into one figure obscures the details. The electrical correction measurements alone are sized to take up the whole page in Figure 3. I show many electrical correction curves taken with different speakers in different positions. Flipping through the plots, please note the 6 -7 dB boost in most cases. Exceptions are the electrical correction curve on Figure 12, which shows a 1 dB attenuation, and Figure 15, where the boost is limited to about 1.5 dB. This boost is endemic in room correction systems unless they have special control settings (Anthem ARC for example). The room EQ has no way of establishing the frequency at which the speaker is no longer able to reproduce a clean signal at say 90dB SPL at the listening seat. The 6dB value is relatively constant across brands.