Integrated Amplifiers

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


HK 990 Performance with a Subwoofer

Figure 4 follows the same approach as Figure 2, but this time for a subwoofer. Black is EQ off and red is EQ on. The room modes of the post EQ subwoofer are well suppressed but the low-frequency response (-6 dB) of the subwoofer has been shifted from 27 Hz to 21 Hz. The subwoofer has a 10 inch cone in a 12.6 inch cube. It produces high SPLs to 30Hz and then the show is over. The rolloff of the subwoofer exceeds sixth order. A 15dB boost was supplied by the HK 990 to move the low-frequency limit down to 22Hz. This is clearly seen in the electrical correction curve. The black curve is the EQ off, but 80Hz LPF bass management is in the signal path. 

This boost challenged the subwoofer. The frequency response of the subwoofer changed as the test level was increased. This indicates that a sliding EQ in the woofers overload prevention scheme was being activated.(I have not shown this measurement). In Figure 4, the test tone level is low enough to keep the subwoofer operation in its linear range.

The electrical correction curve is illustrated in Figure 5. The subwoofer is also attenuated by 15 db at 38 Hz, implying a 30 dB correction range from 23 Hz (the 15 dB boost) to 38 Hz!

As this review was going to press (as we would say in the old days), I finally gained access to the designer of the EQ system. Our dialogue occurred several months after the measurements had been completed. The cause of the issue described above is a control mode found in the HK7550 video GUI that was not transferred to the HK 990. This is the subwoofer cone dimension mode, which runs from 8 to 15 inches. Subwoofer cone size is a proxy for the 6dB cutoff of the subwoofer. Subwoofer specification for cutoff frequency can be very optimistic. Subwoofer cone size is not perfectly correlated to the 6dB cutoff but, in the absence of any other reliable data, it makes sense to use subwoofer cone size to determine the -6dB frequency point. With that frequency in the processor, the electrical correction is limited to the -6dB value. It is a really smart idea since most EQ systems do not ask the question and basically try to EQ the subwoofer down to 20Hz. Like full range speakers, this limit appears to be about 6dB.

I think many of you may have guessed what happened in the HK 990. With no way to enter the critical subwoofer cone dimension, the system defaults to 15inch. It then assumes the subwoofer is good to at least 20Hz and applies electrical correction unbounded. That is 15dB in the case of the 10 inch NHT B10-d subwoofer I used. NHT gives a -6dB down spec of -27Hz. It is hard to get an exact number from the uncorrected in room curve (Figure 4) because of the room modes but 27Hz is certainly plausible. Peak output (short term) is given as 109dB at 1 meter at 30Hz. If we accept the max SPL number (something I have no way to test without a visit from the homeowners association) this is clearly a subwoofer good for full-range music reproduction provided you are not enamored with 32ft organ pipes.

I bring this up because someone is going to question my use of this subwoofer with a state-of-the-art integrated amp. My intent was to use two, thereby allowing me to determine how well the BassQ system worked. Obviously, two subwoofers are doubly expensive as one and consume twice the space, so the NHT subwoofer appeared ideal. With something clearly amiss when I used one subwoofer, I never went on to two.

Figure 6 shows the acoustic response of the subwoofer with the level selected by the HK 990 and the level of the satellite speaker selected by the HK 990 (top curve). The subwoofer is 7 dB low. I made a correction (bottom curve) to archive a flat response across the frequency band. It is not clear how a consumer might implement this change without test equipment. I have not seen this issue with other room EQ systems when the subwoofer is properly set up, i.e., when the subwoofer level is not set too low (the HK 990 gives a warning message when this occurs). To be sure, I checked with an SPL meter to see if the level of the test tones to the main speaker matched the subwoofer. Owing to the brevity of the HK 990's tones, this test is hard to execute.

The setting of the internal crossover of the subwoofer can also create a level matching problem. Most systems like the internal crossover bypassed, but some prefer the filter to be set slightly higher than the desired crossover point. For these room EQs, the LPF of the subwoofer is cascaded with the LPF for the subwoofer in the DSP to save some DSP computations for the crossover which, in turn, can be used to correct other response errors in the subwoofer.

There is one other thing that can happen when the subwoofer crossover is bypassed: the SPL of the subwoofer is overestimated because the frequency range over which the uncorrected subwoofer is producing the test tone dwarfs the frequency range when the final LPF is incorporated. A good room correction system should compensate for this by filtering the subwoofers' spectra before calculating the SPL.

Needless to say, I tried to get the HK 990 to work properly anticipating all these potential problems, but could not make an improvement. The fact the subwoofer was set to the 15inch mode may have caused an internal SPL estimation error. The engineers at Harman could not confirm this idea, but given the 30dB correction range was introduced with the 15 inch subwoofer setting, such a problem would not be outside the realm of possibility. A revised HK 990 with a 10 inch setting should not show this problem. If it is a real problem, Harman is aware of it now and will put it on the punch list of corrections to be implemented if the software is updated.