Factors Affecting Sonic Quality of Mid & HF Horns & Waveguides (Part #1 of 9)
This thread is an offshoot of the Handmade Ersatz M9500 thread, found at: http://www.audioheritage.org/vbulletin/showthread.php?t=12390&page=3
The thread evolved into a comparison of various horn contours, which included a pair of questions (from Rob H. and Ian Mackenzie) addressed to me. Rather than dilute the original thread with a long reply, I have opted to initiate a new thread dedicated to a discussion of the various factors which affect subjective sound quality of horns and waveguides intended for midrange and high frequency reproduction.
In this, and subsequent posts, I will show that the goal of high quality sound reproduction using compression drivers appears achievable if sufficient attention is given to;
- Holistic system design, including careful crossover parameters for optimum transducer integration,
- Applying proper horn & driver equalization and frequency response tailoring,
- Understanding the causes, and methods to attenuate, horn-honk
- Employing wideband CD waveguides, designed to be tonally neutral,
- And the value of using high quality transducers.
The original question on sound quality was posted by Ian:
Originally Posted by Ian Mackenzie
Ian’s question summarized my intent correctly; my posts do concentrate on improved sound quality of horns/waveguides as a central theme. Taking a step back, I would also add that I’m probably more concerned with overall sound quality of the full reproduction system, including: source, electronics, loudspeakers and room. I view the set of equipment and components in all four of these categories as a single unified system, and the optimization of the sound quality from the whole system should be the goal of any design exercise.
“Big Four Criteria”
From my experience, there are four criteria which impact a reproduction systems ability to render an impression of a live performance (in order of importance):
1) Flat frequency response, both on axis and total radiated power.
2) Wide frequency bandwidth. (20Hz -20kHz is sufficient)
3) Wide dynamic range, meaning realistic peak output SPL, with low electrical and mechanical noise floor.
4) Low distortion (all forms of non-linearity)
I’ll nickname this lofty set of goals as “the big four criteria”, for later reference. I won’t expand on any specific details of the big four criteria, because a thorough description of the quantitative measures, and measurement techniques to confirm performance would take a lot of time and space on this post. But I will mention that each time I have made an improvement in any of the big four criteria, I have subjectively noted a qualitative improvement in the sonic reproduction. For every measurable improvement I have managed on the one or more of these four criteria, I have always been rewarded with a system that sounds more realistic.
Clearly the big four performance criteria are acoustically, mechanically and electrically all highly inter-related, and this is the primarily motivation for a holistic rather than a disjointed approach to system design. In other words, a particular transducer sub-system, such as a compression driver and horn, should be integrated into the larger reproduction system in such a way that it does not draw attention to itself, in either a good or bad way.
I believe that the most important sonic aspect of any compression driver horn or waveguide combination is how well it can be integrated into the total system design, such that the outcome satisfies the big four performance criteria. For instance, the required performance characteristics of a horn aimed at use in a two way speaker deployed in a small well damped living room will be different from the requirement for a good mid or top unit in a fully horn loaded system of someone living in a house the size of a sports arena.
It’s not mandatory that each system must have its own individually designed horn, as it may be entirely possible to use the exact same horn/driver in both the previously mentioned situations. What would need to change are the specific details of crossover and equalization in the two cases, so that the device in question could be best integrated with the cabinet and other transducers to achieve specific sonic performance goals. This process requires access to specific engineering performance data for the transducers, along with a systematic approach to the design work:
A brief summary of the design steps for transducer integration:
1) Measure (or obtain) and review the power bandwidth and distortion across the power bandwidth of each prospective transducer in the trial system design. Does each transducer provide acceptable distortion and dynamic range limits within the intended bandwidth?
2) Will the chosen set of transducers (in the system configuration: 2, 3, or more way) provide sufficient frequency overlap between ranges above and below each preliminary crossover point? In the case of gaps, change transducers, or increase number of units and crossover bands. Can appropriate crossover frequencies and slopes be chosen to support adequate power handling, and restrict out of band distortion? In the case of a compression driver and waveguide combination, the crossover frequencies and slopes must restrict the bandwidth to operate above the cutoff frequency and below the band where diaphragm breakup causes excessive audible distortion.
3) Measure (or obtain) and review the (on axis and power) frequency response of each transducer. Can appropriate equalization be applied while maintaining acceptable distortion and headroom? Is the equalization practical with the chosen crossover topology?
4) Check the directivity (polar response) of each transducer at the preliminary crossover frequencies. Do the coverage patterns match correctly on either side of each crossover point? Do the crossover frequencies need to be adjusted to provide smoother directivity transition? Will the complete system provide good power response? Adjust crossover frequencies, slopes and/or change transducers to optimize.
I’m just scratching the surface of the tip of the iceberg in terms of system design steps, but by now you are getting the idea that a particular driver/horn/EQ combination which can excel at matching the big four criteria will be much easier to use compared to a narrow band device with rough out of band distortion and poor directivity. (Common sense really). I am emphasizing the importance of transducer integration into the total system design, because small changes in the crossover frequencies, slopes, and transducer equalization can have dramatic effects on frequency response, which will impact subjective opinion on the system, or individual transducer performance.
Compression driver attributes:
The preceding comments could be generically applied to almost any type of transducer; however horn/driver combinations have four acoustic aspects that definitely set them apart in terms of sonic characteristic (in order from best to worst):
a) The highest reference efficiency of any transducer type.
b) A wider range of designs for improved directivity control, compared to cone or planar transducers (arrays excepted).
c) Some requirement for equalization (either electrical tailoring, or by choice of horn contour).
d) The undesirable tendency to support longitudinal (and higher order mode) internal reflections within the horn/waveguide cavity.
High efficiency, along with high power handling can be used to build a system with prodigious peak acoustic output capability, and wide dynamic range. Careful use of constant directivity waveguides can also be used to configure a system that produces a flat frequency spectrum, and total power response in the final listening venue, however this depends on how well the compression driver & horn equalization is rendered.
In my experience, the equalization step is probably the most critical aspect of integrating a horn/driver combination into a loudspeaker design. Even though I use a full DSP crossover, with comprehensive EQ capability, calibrated by a laptop based acoustic test and measurement system using ETF software, (along with my own white/pink noise based spatially averaged spectral analysis technique), it still takes hours and sometimes days to use EQ for optimum response tailoring.
Errors in EQ and gain settings among and between transducers will negatively impact the sonic character of the system, moving further away from the most important goal of the big four, namely flat frequency response. Improper EQ does not necessarily mean that the horn/driver combination is at fault, it’s just that it takes time, and a certain amount of trial and error to sort out the interaction between equalization settings, crossover frequencies & slopes, and the effect on the overall frequency & phase due to the summation across the crossover overlap zone between various drivers in the system.
end of part 1 - Jack Bouska