Horn Response, Hiss and Equalization

Horn Response

When looking at a horn’s frequency response, a dip in high frequencies compared to mid frequencies might lead you to believe you’ve “lost” decibels (dB) and need to boost the highs with additional amplification.

However, this way of thinking is not accurate and we will see here why.

To gain a more comprehensive understanding of the concepts discussed here, particularly about horn loading (energy) and constant directivity, we recommend referring to our articles dedicated to these subjects.

Essentially, the bell response of a horn arises from the combined influence of its directivity pattern and acoustic loading effect. While constant directivity horns typically maintain a consistent directivity behavior up to around 7-8 kHz, the acoustic loading effect is inversely proportional to frequency. This means it has minimal impact on high frequencies.

Bell Response and Energy Balance

The interplay between driver characteristics, horn loading and horn directivity creates a phenomenon known as the “bell response”.

This describes a perceived roll-off in high frequencies. However, it’s important to understand that this doesn’t necessarily mean there’s less high-frequency energy overall.

The reasons of the bell response are:

Larger diaphragm compression drivers produce more low-midrange energy compared to smaller ones. So in comparaison the high frequency can look lower than a tinier diaphragm when in fact that is mainly the medium range that is upper.

The horn’s loading effect enhances this low-midrange presence, that is a very important aspect, as the loadding effect is strongly inversely proportional to the frequency, more we goes up in frequency less we have it. So for two horns that are constant at the same frequency, reducing coverage (see the point below) will principaly gives more SPL in low end.

Directivity distribution of the initial energy as we have seen upper, this initial energy will always remains the same, it’s the way to distribute it that change and affect the on-axis response.

This can make the high frequencies seem less prominent in comparison to medium, creating the bell response.

When we see these high frequency lower than medium in comparison we tend to thinks that we have “lost” dB in high frequency, we will see that in fact it doesn’t work like this in Horn Response and Hiss article,

In most cases we in fact have “gain” in the midrange area, it’s the high frequency level that doesn’t have moved.

Flattening Frequency Response with EQ

Typically, to achieve a flat on-axis frequency response, we choose a reference frequency above which we want a flat response. This is often set at 15 kHz.
Beyond this point, we allow the natural roll-off of the driver to occur, or use equalization (EQ) again if there’s a rise.

Negative or positive EQ can be used to achieve a flat response. The direction (positive or negative) isn’t crucial, the final result will be the same.
The primary goal is to avoid clipping (in the source, DSP, or amplifier) and maintain a proper gain structure to our desired maximum power output.

Minimum-phase EQ (IIR) is a valuable tool for correcting on-axis horn response because it addresses both the frequency response and the phase response simultaneously.

When a significant dip or peak occurs at a specific frequency, minimum-phase EQ (IIR) not only corrects the level at that frequency but also adjusts the phase response in a way that complements the level correction. This combined effect helps to achieve a more linear overall frequency and phase response.

Here are two EQ examples for an X-Shape X25 horn

BMS 5530 one:

EQ X25 BMS

18Sound 1095N one:

EQ X25 BMS

Our article about audibility can help to understand one of the reasons why we do nothing after 15 kHz in the 1095N case.

This region is also where the breakup occurs, that is the second main reason why we don’t try to increase volume here.

In each case 15 kHz is our anchor point, EQ are different but the main judge will be, as alway, the distortion after having EQ it flat below 15 kHz, as we have seen on our 1" compression driver test.

If we push constant directivity very high on the same horn, way after 7/8 kHz, the effective lose of energy in 12/17khz area will be in fact minimal, arround 1 or 2dB maximum.
Its due to the fact that only directivity will impact it and not the loading effect.

S/N ratio and Hiss

Hiss is a low-level noise often noticeable in high-sensitivity components in relation to the S/N (Signal-Noise) ratio of the system.

A common concern when comparing horns, especially regarding EQ usage, is increased hiss.

It might seem that extensive EQ required for some horns, compared to others, along with the gain adjustments needed to achieve a flat response, would exacerbate hiss, reduce the signal-to-noise ratio (S/N ratio), or even consume more power.

However, this vision is partial. Here’s why:

In conclusion, while EQ adjustments might influence how the DSP perceives the S/N ratio, the impact on hiss is still minimal.

L-PAD

To reduce hiss and protect the driver, an L-pad to reduce driver sensitivity is often recommended, it’s a simple resistor network placed just before the driver:

EQ X25 BMS

You can simulate the appropriate L-pad values using VituixCad software, considering the driver’s impedance and the horn on-axis response. For the X25 horn, the L-pad configuration might involve a combination of a 1.5-ohm resistor in parallel and a 12-ohm resistor in series, depending on the desired dB reduction to match the mid-woofer level.

It’s important to note that the L-pad resistor network doesn’t degrade audio quality, It simply absorbs some of the amplifier’s energy which will therefore not reach the compression driver or tweeter.

Conclusion

For a given driver, the amount of energy at 15 kHz is primarily determined by the driver itself, regardless of whether the horn has constant directivity or even a reduced coverage area.

In essence, differents constnat directivity horns have roundly the same hiss levels and S/N ratio when using the same driver.

The bell response is not a problem and is related to both directivity and loading effect:

copyright