Room correction technology for audiophiles has existed for over a decade now, but this technology has not yet seen wide adoption. Many audiophiles just can’t stomach a processor messing with the signal. Never mind their speakers might have terrible frequency and phase response, at least when compared to their other electronics! On top of that, interaction with the room causes frequency response errors of well over 10 dB in all but the most acoustically perfect rooms. Still think your signal is “pure?”
The same DSP technology used in a room correction system can also be used to make a much more radical system, one that completely eliminates the passive loudspeaker crossover and corrects for individual driver response. DEQX is an Australian company that has pioneered this new type of technology. They have provided speaker correction systems for manufacturers including NHT. The well-regarded Xa speaker system uses a DEQX processor, but only for driver correction and crossover functions.
The new Tikandi speaker from Australian loudspeaker manufacturer Legend Acoustics is a complete system that uses the full power of the DEQX HDP-3 processor. In this review, I listen to the Tikandi speaker system, consisting of the Tikandi 3-way speaker system, powered by the 8-channel Marantz MM8003 amplifier and commanded by the DEQX HDP-3 processor. The HDP-3 can also be used with any loudspeaker, providing loudspeaker response correction, room correction and if desired, crossover functions in addition to preamp functionality. I also listened to the HDP-3 with my own Gallo Reference 3.1 loudspeakers.
- Tikandi Speakers
- Design: 3-way Sealed Enclosure
- Drivers: One 1″ Dome Tweeter, One 5.25″ Midrange, Two 10″ Woofers
- MFR: 20 Hz – 25 kHz ± 2 dB
- Sensitivity:88 dB/W @ 1m
- Nominal Impedance: 6 Ohms
- Recommended Power: 50 – 200 Watts
- Marantz MM8003 Power Amplifier Supplied as Part of Package (8 x 140 Watts RMS)
- Dimensions: 38″ H x 12″ W x 12″ D
- Weight: 100 Pounds/each
- DEQX HDP-3 Processor
- DEQX-HD™ Active Crossovers; 48dB/octave to 300dB/octave linear-phase
- Traditional Active 3-way Crossovers from 6dB/octave to 48dB/octave
- 32-bit, Floating-point, 240-MFLOPS Processing with Dual SHARC DSPs
- Analogue Volume Control Maintains Full DAC Resolution at Low Listening Levels
- Analog Inputs: Unbalanced (2 x RCA) and Balanced (2 x XLR)
- Digital Inputs: S/PDIF (1 x RCA) and AES/EBU (1 x XLR)
- Microphone Input has 48V phantom power for DEQX Calibrated™ Mics
- Outputs: 6 RCA Unbalanced
- Anodized Black brushed or Silver brushed Aluminum Front
- Dimensions: 3.5″ H x 17″ W x 11.5″ D
- MSRP for Tikandi Loudspeaker System as Tested, Including Amp, Processor and All Cables: $19,199
- MSRP for HDP3 processor with Standard Calibration Kit and Active Balanced Output Module $6,150
- Legend Speakers
The Legend Acoustics Tikandi loudspeaker is one of a new generation of loudspeakers that is designed to be used only with a digital processor. The Tikandi consists of two pieces: a small “pod” containing the tweeter and midrange, and a larger bass module containing a pair of 10″ woofers. Traditional binding posts are provided for the pod, while the bass module uses a Neutrik speakON connector. While the Legend looks normal from the outside, inside there’s something different. No passive crossovers. The pod has two pairs of binding posts, one directly wired to the tweeter and the other directly wired to the midrange. The speakON connector has 4 conductors, with each woofer wired independently. The system was provided with a complete set of Exakte loudspeaker and interconnect cable with the proper connectors. The loudspeaker requires at least 6 channels of amplification, but that would require the two woofers to be wired in parallel, presenting a minimum 1.7 ohm load to the amp. In this system, a Marantz MM8003 8-channel amp was used, with each of the four woofers powered by its own amp channel. This is one possible amplifer option. You can buy the system without amplification and provide your own, or soon you’ll be able to get the system with DEQXs new 4-channel digital amplifiers.
The 8-channel amp is driven from the 6 output channels of the HDP-3 processor. RCA T-connectors are used to split the low range output of the HDP-3 in two for driving the 4 bass amplifier channels. At that point, the magic starts. The HDP-3 processor has 6 channels of analog outputs (single ended, and two types of optional balanced outputs). One single ended and one balanced analog input are provided (digitized at 24 bits and 48 or 96 kHz), along with both coaxial and AES-EBU (aka AES3)digital inputs (also compatible up to 24 bits and 96 kHz). The HDP-3 processor provides 3 functions, apart from serving as a DAC, an ADC and Preamp…….. . The preamp volume control is fully analog and the channels are matched to 0.1 dB. First, the HDP-3 performs loudspeaker driver correction. Each loudspeaker driver is measured using a time-gated tone sweep. This tone sweep is generated by the processor and played through the system. A microphone connected to the processor then listens to this swept tone.
The HDP-3 then computes a correction to produce a flat driver response in both frequency and phase. This measurement is typically made with the microphone close to the loudspeaker to avoid room effects. The woofer offers two setup possibilities: The drivers can be measured in the near field, and then have room correction applied later, or the driver can be measured from the listening position, correcting the driver and the room simultaneously. Second, the user (or professional installer) sets up the crossovers. Several different crossover filter types can be used, including Linear Phase, with crossover slopes from 12 dB per octave up to 300 dB per octave. Once the drivers are measured, their response can be analyzed using the DEQX software, both to verify the correction, and tune the crossover points and slopes to best match the driver response and integration. One important tool is the ability to see the time domain response of the drivers, allowing the user to add time delays to get perfect driver time alignment.
Third, after the crossover and driver measurements, the microphone is moved to the listening position so room measurements can be made. The results of these room measurements can then be used to set the parametric equalizer function of the HDP-3 to provide room correction. Unfortunately, the parametric EQ cannot be set independently for the two channels; one PEQ curve is applied to both the right and left channels simultaneously. The parametric EQ can be set either automatically or by hand. The HDP-3 can store 4 separate configurations, with different crossover settings, driver correction settings and parametric EQs. One is typically used as a bypass with only the crossover function implemented, although it can be set to anything.
The description above is actually a simplified one. The HDP-3 software gives the user control over every last detail of the crossover, driver and room correction process. This is a double-edged sword, though. The learning curve is rather steep. But this is not a problem because all Tikandi systems come with professional installation, so the owner doesn’t have to concern themselves with all the details unless they want to. I personally love that kind of thing, but many owners will just prefer the professional install.
If you have lots of other analog sources (turntable, tuner, SACD player etc), or you want to integrate the system with a home theater, the small number of analog inputs could be limiting. This can be remedied with an analog switch like a Manley Skipjack, but that will run you an additional $700. There is no built in phono preamp. Also, for HT integration there’s another issue. The total time delay through the HDP-3 is a bit over 20ms, although in some configurations it can be as low as 8ms. You will want to be able to dial in that much delay into your surround sound processor’s center and rear channels to match the Tikandis. I could not do that with my old fashioned Rotel SSP. The delay mismatch wasn’t so much that it was clearly obvious, but the 20 ms difference between the center and the FR and FL was definitely reducing the precision of the sound when watching movies. With a newer SSP, with a wider range of delay options, this won’t be a problem. I also found the lack of a Toslink input rather annoying. I use a Mac laptop as my music server. It has a 24/96 capable optical digital output. For use with the HDP-3, I had to get a USB or firewire sound interface with a coaxial output, or use a toslink to coax converter. I did the latter, using a M-Audio CO2 bidirectional toslink to coax converter. This unit is completely asynchronous, so it should add little additional jitter. The HDP-3 completely regenerates the sampling clock driving its output DACs anyway, so the effect of clock jitter at the input of the HDP-3 should be nil.
The Tikandi system was delivered from DEQX’s US importer, KRC, by Larry Owens. Larry helped me set the system up, a process that took about 6 hours. After that, I was left to myself to play with the system, and also use the HDP-3 processor with my own speakers and amplifiers. The setup process I went through with Larry will help to explain just how the HDP-3 processor works. Several of the steps had already been done in advance. The basic configuration of the system was already in place.
The Tikandi is set up as a 2-way plus stereo subwoofer system, requiring a low pass filter for the woofers, low and high pass filters for the midrange and a high pass filter for the tweeter. The crossover points and slopes were already set to something known to work well, and the loudspeaker measurements of the tweeter and the midrange had already been made. We just used these settings. After positioning the loudspeakers and cabling everything up, we put the microphone at the listening position and hooked the laptop running the HDP-3 software up to the processor with a USB cable. We did a loudspeaker measurement of the woofers with the microphone at the listening position to provide both loudspeaker and room correction from 20-150 Hz. After making the pseudo-anechoic time-gated sweep measurements for both the left and right speakers, we windowed the measurement by looking at the time domain response to exclude as many reflections as possible while still giving us enough data to cover the bandwidth we needed.
We then chose the frequency range and allowed boost and cut amounts for the correction, and allowed the software to compute the correction filters. Making the choices for the time-window settings, frequency range and boost and cut limits is a bit of an art where Larry’s experience was essential. The resulting filters were trying for flat response from about 18 Hz to 150 Hz. Verification measurements showed that the result was within a few dB over the full range! We then made room measurements using both the woofers and the pods. These measurements are made separately and then plotted on top of each other. With the woofer room correction already done, we used these measurements to perform EQ on the midrange from 150 Hz to about 5 kHz. While the automated routine tries to correct every bump and wiggle, we applied the EQ manually, just correcting major deviations.
The key here is to have restraint, making subtle, broad-band corrections to get an overall flat response without trying to correct every little bump. Since the HDP-3 can’t measure the woofers and the pods simultaneously, we measured the speakers full range using another software tool and a professional Firewire soundcard to be sure the crossover region at 150 Hz was performing as expected. Everything was in order, with pancake flat frequency response from 18 Hz to just above 20 kHz. In addition, all three drivers were perfectly time aligned to 0.1 ms. With linear phase crossover filters used, the entire frequency range, including the crossover regions, is not only flat in frequency response, but flat in phase (i.e. time) response as well. This is quite literally impossible with passive crossovers.
I didn’t save screenshots of the setup, so Larry sent me some screenshots of the DEQX software when he was setting up another speaker system (Klipsh Heresys).
The system configuration window is where the basic speaker configuration is set, along with crossover settings, time delays and gain corrections for each channel.
After configuration the first step is loudspeaker measurement. The top figure shows the raw loudspeaker measurement with the chosen crossover settings.
The middle is the resulting correction filter calculated from the measurement. The bottom figure is the measured result after the correction filters are applied plotted over the original measurements (in the near field).
After speaker measurement, the room correction is taken care of. The above figure is the in-room response at the listening position with and without the correction applied.
Finally, while listening the control panel window allows real time control of the volume, saved configuration plus allows tweaking gain and time delay for each channel of output.
All this technology is cool, but how does it sound?
The Tikandi system is the best music reproduction system I have ever had in my listening room, regardless of price, and is one of the best sounding systems
I have ever heard in any room. Once the system was dialed in, Larry and I listened to some tunes. One of the things Larry played for me was an audiophile epiphany. He played a Japanese drumming track from the YG Acoustics demo CD. This track was recorded with gigantic dynamic range. In real life, drums are loud. Very loud. We turned up the Tikandis until Larry started to get uncomfortable about blowing something up. I just sat there like the Memorex guy with a stupid grin on my face. The power delivered by the drums was something I have NEVER heard from a stereo system. It wasn’t that it was just loud. Lots of speakers can play loud. It was the sharpness of the transients, the subterranean frequency extension, and the shocking dynamic contrasts that did the business.
As I listened to the Tikandis over the next few weeks, it became clear that these speakers were the audio equivalent of an electron microscope. You can hear, with startling precision, every last detail in every recording. This is a trick usually reserved for headphones. With the Tikandis, you can have that headphone-like super resolution and also get the realistic soundstage and physical impact loudspeakers deliver. This is another double-edged sword. Bad recordings are done no favors. You can hear every bit of compression, limiting, crappy artificial reverb and autotune in every one of your recordings. The super flat frequency response also is not kind to dynamically compressed recordings. Typical speakers and room interactions, with 10 dB or more of deviation from flat frequency response, create “fake” dynamic contrasts as they play sounds of different frequencies. The super-flat response of the Tikandis gives these dynamically compressed recordings no help.
Some of my favorite albums, like Bebel Gilberto’s Tanto Tempo Remixes, or Spoon’s Ga Ga Ga Ga Ga had their recording compromises laid bare. Not that they were unlistenable. It was just annoying that some of my favorite music was recorded badly, and in a way that was being covered up by my normal system. I always knew that the Spoon album was dynamically flat, and the remix album was, well, a remix album where all kinds of stuff had been done to the original source recordings. With the Tikandis, Spoon sounded even more flat than usual, and every sin visited on Bebel Gilberto’s voice by the remix artists was clear as day. On better-recorded albums, like Bebel Gilberto’s eponymous album, or any number of “audiophile” recordings like the 24/96 DVD release of Jon Faddis’ Remembrances, the resolution and accuracy of the Tikandis was intoxicating. The bass extension was ridiculous for such small speakers. They really were flat to 18 Hz, all from a pair of waist high floorstanding loudspeakers. The tonal smoothness, pitch accuracy and speed in the bass were truly spectacular, mainly because all the normally encountered room resonances and suckouts were non-existent. Imaging was as razor sharp as anything I have heard. The soundstage width and particularly depth, were also first rate, and in some cases unmatched. Naturally recorded tracks with lots of ambiance from the recording space were the most expansive. On Shake ’em on Down, by David ‘Honeyboy’ Edwards, the reverb and echos from the converted church used as the recording studio put its imprint on every part of the sound. It was possible to hear the back wall and sidewalls of the room in 3D in a way I’ve not heard before. The ambiance was always there and audible, but only through the Tikandis did it snap into a clear 3D sound model of the room.
Taken as a whole, I can’t imagine anyone who would not be thrilled with the Tikandis in their room, even at over $19,000. Actually, given this price includes the processor (which replaces the DAC and preamp), the amp and the speakers, its actually quite reasonable. All you need to add is a digital source.
The HDP-3 Processor Alone
If shopping for a new system in the $20k range, you’d be crazy to not consider the Tikandi system. Few audiophiles buy a system that way, though. Most, like me, upgrade component by component over many years. How would the HDP-3 processor fit into a more generic system with typical speakers? I tried the HDP-3 processor alone, replacing the Emotiva RSP-1 preamp and Bel Canto DAC-1.1 DAC in my normal system. The HDP-3 drove my Emotiva XPA-1 monoblocks, Gallo Reference SA subwoofer amp and Gallo Reference 3.1 loudspeakers. I tried this setup in three configurations: as a normal full range speaker using only 2 outputs from the HDP-3 driving the XPA-1s, as a 1-way plus stereo sub configuration using the XPA-1s to drive the main inputs to the Ref 3.1s from 125 Hz up, and the Ref SA to drive the woofer from 125 Hz down through the second voice coil inputs, and third as a hybrid using the XPA-1s full range and the Ref SA below about 40 Hz to beef up the bottom end. The final configuration is how you’d normally set up Gallos with the sub amp, even if it sounds a little weird.
I measured the Gallos in my room from a distance of about 0.5m, trying to move furniture out of the way to give the longest time before the first reflection hit. I was able to time gate out long enough to get good data down to about 80 Hz. Verification measurements made after the correction filters were created showed response flat to 1 or 2 dB! In the first configuration I had to do all room correction, including the bass, by hand. Only in the second configuration could I measure the entire frequency range of the woofers at the listening position as I did with the Tikandis and get the room correction built into the speaker correction filters. In the third hybrid configuration, I measured the bass at the listening position below 40 Hz, and had to apply manual correction above 40 Hz. Once completed, the measured frequency response for all three configurations was just about as good as I had gotten with the Tikandis. I saved each one of the three configurations to the three presets on the HDP-3, so I could hear the effects of all three setups.
I was quite impressed with the sound. I was actually expecting to lose more going back to my speakers. The sound was remarkably similar, with the main losses being subtle ones. The primary loss was the resolution and speed of the sound in resolving details. The sound through the Gallos, while vastly better than the Gallos alone, was a bit slower and less detailed than the Tikandis, with transients and fine detail smeared a just a bit. In addition, the bass was less smooth, less powerful and less deep. The Gallos began to roll of between 25 and 20 Hz in my room, a few Hz higher than the Tikandis. The best bass smoothness was achieved with the 1-way plus stereo sub configuration, but the best bass power and tone was achieved with the “hybrid” approach where the big Emotiva XPA-1s were able to exert their control over the Gallo’s woofers. The 1-way configuration was more difficult to deal with, since the parametric EQ correction does not work in stereo mode. The frequency response of the left and right channels in my room were very different, so it was impossible to apply a good correction for bass below 100 Hz for both channels simultaneously.
On The Bench
I measured the Tikandis and my own speakers using my Earthworks microphone and Spectra Plus FFT software. Frequency response measurements were made using a white noise signal played at 100 dB SPL at 1m distance. Measurements were made both at 1m distance and at the listening position.
Frequency response was measured at 1m on axis (top), and at the listening position (bottom). On axis, frequency response was within a few dB above 100 Hz, with bumps below 100 Hz. Room correction was calculated for the listening position, not on axis at 1m. The suckout just below 500 Hz is a room effect, and was not able to be corrected accurately due to the sharpness and the fact the frequency changes depending on the microphone position. At the listening position, the bass was much smoother, but the higher range was a bit more lumpy. Still, the frequency response was very flat compared to a typical loudspeaker.
1 kHz (top) and 10 kHz (bottom) sine waves showed 0.3 and 0.6% THD. This is a ridiculously low THD level for a loudspeaker. It would be reasonable for an amplifier much less a loudspeaker.
Frequency response of the Gallos was also measured at 1m on axis (top), and at the listening position (bottom). These measurements were actually flatter than the Tikandis. I believe this is because I set up, listened to and measured my speakers second, giving me more time and experience to set up, measure and adjust the speakers. The measurements shown here were for the “hybrid” mode I found to sound the best.
The 1 kHz sine wave showed 1.2% THD, 4 times higher than the Tikandis. I did not measure the 10 kHz sine wave with the Gallos.
The Legend Tikandis with the DEQX HDP-3 are a very special combination. I have never heard better sound in my listening room. The tonal accuracy, frequency extension, and most impressively the ability to pull detail out of recordings was nothing short of astonishing. If you have about $20,000 to spend on a complete stereo system, the Tikandis with the HDP-3 processor and Marantz 8 channel amplifier might be the best possible way to go. If you already have your own speakers, the HDP-3 processor can dramatically improve the performance of your loudspeakers. With the HDP-3 in the loop, my Gallo Reference 3.1s were able to deliver flat frequency performance from 20 Hz to 20 kHz +/- 3dB at the listening position. While they were not able to deliver the full resolution and detail of the Tikandis, the performance was quite close. My only complaints were convenience issues. I wish the processor offered more analog inputs, and a Toslink digital input. Other than that, the HDP-3 was virtually ideal, combining the roles of digital processor, DAC, ADC and preamp in one, for a very reasonable price.