If you’ve been researching online about crossover design and tried inputting values into a crossover calculator, you might believe that you’re prepared to dive into the realm of professional audio. You might feel ready to embark on designing your own speakers and crossovers. However, once you start the actual process, you’ll quickly realize that theory and reality exist in separate realms, particularly true when it comes to passive crossovers and speakers.

So, what’s the challenge? The primary hurdle stems from the fact that nearly all online crossover calculators, textbooks, and the typical approach taken by beginners in crossover design make the assumption that both the Woofer and Tweeter possess an impedance of 8 ohms. Moreover, this assumed impedance remains constant across all frequencies, and their frequency responses are both flat and balanced. However, this is where reality intervenes. Once you begin examining the impedance measurements and frequency responses of the actual components you intend to use, the issues become readily apparent.

For our illustrative project, we’ve opted to work with a well-known 8″ PA Woofer: the B&C 8NDL51. Analyzing the graph below, we can observe that the impedance sits at 8 ohms around 800Hz. Consequently, at this frequency, our theoretical crossover derived from an online calculator should perform as expected in the real world.

Measured impedance response of B&C 8NDL51 8 ohm woofer

However, our goal is not to establish a crossover point at 800Hz. In order to align with a smaller tweeter, we aim for a crossover frequency around 2500Hz or slightly higher. At the 2500Hz mark, the woofer’s impedance measures around 11 ohms and continues to rise beyond this point. If we were to input 11 ohms instead of 8 ohms into a crossover calculator, we might achieve a more favorable outcome. Nevertheless, due to the fluctuating impedance, predicting the results becomes challenging, making it difficult to determine the optimal choice.

If the objective is to stabilize the impedance load of the woofer, there are potential solutions available. However, these solutions can become intricate and potentially costly. Delving into the intricate details of crafting the ideal crossover leads us to the realm of Zobel Networks. While we occasionally employ Zobel networks, our focus primarily centers on crafting uncomplicated crossover designs that remain cost-effective to produce.

When it comes to crossover design, we start by examining the impedance and frequency responses. This analysis aids us in determining the best approach. We have already measured impedance, our next task is to measure the frequency response. This step assists us in selecting a starting filter frequency. However, it’s common for us to experiment with various filter configurations before pinpointing the one that works the best.

The frequency response measurements have all been conducted utilizing a clio system paired with an appropriate microphone. As these measurements predominantly serve the purpose of illustrating comparative outcomes for educational intents, we haven’t taken the effort to meticulously calibrate the settings to precisely 1W/1M. The measurements were taken within a standard room, where minimal sound-absorbing materials have been strategically positioned on the walls to curtail reflections. It’s important to note that all the measurements presented below capture room reflections and anomalies induced by the environment. Consequently, they don’t offer a true representation of the exact response.

Displayed below is the on-axis measured frequency response of the 8NDL51 without any filters. The response exhibits a natural decline from approximately 4kHz; with a distinct peak at 3.5kHz. This peak aligns itself within the midrange vocal region, which coincides with the area of greatest sensitivity for the human ear. Naturally, we aim to mitigate the prominence of this peak in the final outcome. If the 8NDL51 and Tweeter were to sum constructively at this frequency, the overall sound could be a little harsh. Consequently, our strategy involves selecting a low-pass filter slope that effectively attenuates the 3.5kHz peak.

Measured on-axis frequency response of B&C 8NDL51 8 ohm woofer.

For the purpose of comparison, we started our experiment with a conventional Butterworth 12dB low pass filter at 2500Hz, calculated using default 8 ohm value for impedance. It’s worth noting that the bump down at 30hz was caused by background noise whilst measuring, and should be disregarded. The measured frequency response doesn’t look like the filter is actually working at 2500Hz -this is partly due to the peak at 3.5khz, but also due to the fact that the impedance of the woofer isn’t 8 ohms at the crossover frequency, so the calculated component values wont actually make a Butterworth filter, and will also shift the filter frequency a little higher. This was expected, but we took this step anyway as part of the process to show that default crossover values calculated for 8 ohms will often be incorrect.

B&C 8NDL51 with 2.5khz low pass filter

To progress further, we conducted another measurement using a standard Butterworth 12dB filter at 2500Hz. However, this time the calculations were based on an 11 ohm impedance, which is the actual impedance value we’re working with. The outcome of this measurement is illustrated in green in the graph. This adjustment has led to a subtle reduction in the range between 2000Hz and 3000Hz. However, the peak around 3500Hz persists, and the response doesn’t exhibit a substantial decline until after this peak. In order to achieve a satisfactory final outcome, we need to address this peak.

B&C 8NDL51 with 2.5khz low pass filter calculated for 11 ohm load impedance

From experience I’m aware that peaks within the range of 1500Hz to 4000Hz could result in a somewhat shrill sound. While it’s acceptable to rectify this issue using a graphic equalizer, I prefer, whenever possible, to devise crossover designs and opt for components that preemptively mitigate these unfavorable peaks.

At this point, it’s pretty common to start playing around with a bit of trial and error to find the right filter that works for getting the woofer’s sound just right. Normally, I kick things off by tinkering with various filter frequencies, gradually going lower until peaks are sufficiently suppressed, and also experimenting with either Butterworth or Linkwitz-Rileyalignemtn . This step-by-step process helps us dial in the woofer’s sound.

Multiple measurements of B&C 8NDL51 with different low pass filters

The orange plot is a Linkwitz-Riley 12dB filter at 2500Hz, calculated with 11 ohm impedance, and the blue plot is Linkwitz-Riley 12dB filter at 2000Hz, also calculated for 11 ohm impedance. The peak at 3.5kHz is being gradually suppressed, and getting closer to the frequency response I want from the 8″ woofer.

Multiple measurements of B&C 8NDL51 with different low pass filters

In the final test, I experimented with a 12dB Butterworth filter set at 1.65kHz, with calculations based on an 11 ohm impedance. The result is represented by the purple plot. It’s getting a bit busy, with multiple results, and I omitted a 4 or 5 other test results which would have made the graph undecipherable. Here’s what’s happening in the final test: this response is a tad louder around 2000Hz compared to the blue plot, but then it takes a steeper plunge after that 3.5kHz peak, which is about as good as I can hope for.

Instead of the typical 2.5kHz low pass filter calculated for 8 ohms, we opted for a 1.65kHz low pass filter calculated for 11 ohms. And already, you can see this is shaping up to be something more bespoke than your standard off-the-shelf crossover. A 2.5kHz low pass would utilise a 0.72mH inductor and 5.6uF Capacitor, we ended up with 1.5mH and 6.3uF – significantly different.

To make things a little easier to see, in the graph below the red plot is the original frequency response of the woofer, and the purple plot is the final low-pass filter we decided to use. We needed the woofer response around 2.5kHz to be lower compared to 1000Hz and under , as the woofer’s response will couple with the tweeters response around this frequency.

B&C 8NDL51 frequency response unfiltered (red) and with 1.65khz low pass butterworth (purple)

Using the lower filter frequency of 1.65kHz has had the added benefit of attenuating the response around 500-1000Hz and bringing it more level to the response between 50hz and 200Hz. This will benefit the final result by allowing for a more balanced mid-range response.

So reaching this decision on the low pass filter for the woofer has been influenced by the impedance measurements of the woofer, which resulted in a different impedance being used for the crossover calculations. In addition to this I have looked at the frequency response and aimed to smooth the mid response and suppress any large peaks on the response. I have used past experience to judge the result from the 8″ I am aiming for, and settled on a solution once the response is looking close to what I am aiming for.

Now we move on to the tweeter. For this project we chose the B&C DE12TC, and repeated the same process of measuring impedance and frequency response. You’ll notice the frequency response plots dont got down to 20Hz, we changed the cut-off frequency as we don’t want to put bass frequencies into a tweeter.

DE12TC Impedance measurement:

Measure impedance of B&C DE12TC tweeter

Again we have bumps in the impedance of the tweeter, however tweeters are more problematic, as the impedance bumps are often very close to the crossover frequency. The vary impedance cant interact badly with a passive crossover. These impedance bumps are caused by natural resonant properties of the tweeter, and personally I find if the crossover point is too close to one of these bumps the sound is not particularly nice. So I try to keep the crossover points away from the impedance bumps. Again we have the same problem of not being sure what the impedance is at the crossover point, we have a peak of around 10.5 ohms just below 3khz, dropping down to 7.5 ohms and then slowly rising again. This is why impedances are stated as ‘nominal’ – they are just a vague average over the operating frequency range.

Below is frequency response of the DE12TC.

Measured frequency response of B&C DE12TC

This might not be the smoothest response, but it’s quite par for the course with most compression drivers. The good news is that nothing here is beyond refinement with a touch of EQ. However, I usually try to work in attenuation and EQ directly into the crossover. This helps harmonize the woofer and tweeter in a way that minimizes the need for extra EQ down the line.

But let’s take a step back. Our first task is to examine the frequency response of both the woofer and the tweeter, and then compare the two. This comparison will give us a solid foundation to figure out the best approach for smoothing out the frequency response of the tweeter so that it blends seamlessly with the woofer.

Measured frequency responses of B&C DE12TC and B&C 8NDL51

The first thing that stands out is how the DE12TC is considerably louder than the 8NDL51 from 1kHz to 5kHz – up to 12dB louder (if dB isn’t your thing, just know that 3dB is like doubling the measured sound pressure, and around 10dB is what our ears sense as twice the loudness). If we leave this as is, the end result will likely be a pretty intensely harsh tone with an excessive amount of midrange. To tackle this, we’ll need to dial down the tweeter’s volume to get things balanced.

It’s common, in fact recommended, to utilize an L-Pad arrangement for attenuating high frequencies. This technique is theoretically sound as it facilitates both effective attenuation of high-frequency response and stabilization of the load impedance within the High Pass Filter circuitry. I rarely use L-Pads, and I’ll explain why.

When I build high power crossovers, for speakers of perhaps 600W-1000W, there can be 50-200W of power being directed to the compression driver, sometimes even more in extreme cases. If the compression driver is 9-12dB more sensitive than the woofer (not unusual) you may need to dissipate 100-150W of this power in the resistors. I have cooked enough aluminium clad resistors to have reached the conclusion this is just a bad idea. Dissipating lots of power in resistors invariably leads to reliability issues caused by overheating.

My solution? Use a series resistor. There will be heaps of people that will tell you its a bad idea, but its commonly used in many commercial applications, and power dissipation is the reason. If you need fairly heavy attenuation on an 8 ohm compression driver, you can put another 8 ohm resistor in series with it. The series resistor takes half the power, tweeter takes the other half – but then you have a 16 ohm load. If you know your power calculations, the maths is simple, double the impedance, halve the current. Power = Impedance x current squared. With half the current, you have a quarter of the power, which very effectively reduces total power without having to dissipate loads of heat in resistors.

For serious watts, you can use 2 resistors, spreading the power dissipation over both resistors, and managing the power tidily. There are a couple of drawbacks – first is the fact that the amount of attenuation is proportional to the impedance of the tweeter, so the attenuation will work nicely on this tweeter above 2.5kHz, but not so well below this, as the tweeter has variable impedance (not 8 ohms!) Additionally the combination of impedance bumps, and series resistors messes with the crossover frequencies, particularly the knee in the bend of the high pass filter, as you have variable impedance in the region the filter is transitioning to the pass band, so you end up having to redesign your high pass filter around the attenuation and the new load impedance

We did some trial and error tests, to work out roughly what attenuation would help balance the sound. The orange plot is with approximately 12 ohms series resistance, and the blue plot is with 24 ohms. There is a noticeable peak at 1.6khz, but we would hope to eliminate this once we have applied a suitable high pass filter. For comparison, the purple plot is the response of the 8″ woofer with the 1.65kHz Butterworth filter applied, and its this purple plot we need to get the tweeter to align to smoothly

Measured frequency responses of B&C DE12TC with varying attenuation resistors

Having worked with compression drivers like these before, I skipped a few steps here and just went straight for a filter alignment I expected to work which was a 3.3kHz Butterworth filter This should flatten the response between 1khz and 3khz. You’ll notice also the response is tailing off from around 4khz, due to the fact we are attenuating the high frequencies too much, so we added a 3.3uF bypass capacitor in parallel with the series resistors. This has the effect of adding a hi-shelf EQ to the response, the idea being that you get attenuation below 3-4khz, but at higher frequencies, the attenuation is bypassed by the capacitor, allowing the tweeter to operate at its natural level. If you check out the grey plot, you can see this has worked quite effectively. I could have bored you with multiple iterations of different filters here, so we basically skipped 5-6 tests by utilising past experience.

Measured frequency responses of B&C DE12TC with  attenuation resistors and hi-shelf EQ

At this point we have added a filter and some EQ, and converted the natural response of the tweeter (yellow-green) to a flatter response (grey). Things are looking quite promising here, but what we have are two independent measurements, purple for woofer, grey for tweeter, but this is not a system measurement of the two together. The next result shows the frequency responses with woofer and tweeter together as a system.

Measured frequency responses of B&C DE12TC and 8NDL51 as a complete system

Taking a closer look at the crossover region, there are 2 plots, one green and one blue. Green was with the tweeter in standard polarity, and blue with the tweeter with inverted polarity. Inverting polarity is usually necessary due to phase shift cause by a filter. If we had used a 2.5khz Butterworth high pass and low pass, we would normally run the tweeter in reverse polarity to counteract the phase shift in the crossover. The frequencies chosen: 1.65khz and 3.3khz, are an octave apart, and in the past we have found that having the filter frequencies exactly an octave apart means you don’t need to invert the polarity of the tweeter, and the phase response across the crossover region remains fairly linear. Sadly in this instance it didn’t work. In fact neither configuration worked particularly well. There are other factors, the physical position of the horn makes a difference -you want the voice coils of woofer and tweeter in the same vertical plane to avoid the need for time alignment. Despite trying to ensure the components were arranged correctly, and trying a few subtle variations of crossover components, we couldn’t quite get the flat response we were after. The end result was by no means awful, but there is a noticeable dip at 2.5khz which we just couldn’t rectify using this design approach. The best final frequency response we could achieve is below.

Measured frequency responses of B&C DE12TC and B&C 8NDL51 on a dedicated custom crossover

So the response has a few bumps, but we have a working design, which is potentially good enough for some applications. In fact I have had people request that we intentionally put in a slight dip in the frequency response between 1khz and 3khz to give a slight smiley face EQ effect. In this case, the dip is around 10dB, a little more than I would like. I have had very good results with this approach in the past, and got some very smooth sounding speakers, but in this instance it just wasn’t quite right so I decided to go back to the beginning and take a different approach. I went back to the frequency response of the tweeter, to take a closer look, and consider an alternative approach. At 5kHz and above, the tweeter is approximately equal in output to the median level of the woofer. Below 5khz, it gradually rises in output, until its around 12db higher at 2.5khz. This is a pattern I have seen before, and already know how to deal with.

Comparing frequency responses of B&C DE12TC(unfiltered) and B&C 8NDL51 on 1.65Hz Butterworth low pass filter

The downward slope of the tweeter’s response is approximately 12db over an octave, exactly the opposite of the slope you would get from a second order high pass filter. Applying a 12db/octave high pass filter at around 5khz, we can effectively rebalance the response of the tweeter, flattening it below 5Khz. Here we have the plots of the 1.65kHz low pass applied to the woofer (purple) and a 5kHz (approx) high pass applied to the tweeter (green). You can clearly see the high pass filter has done its job and flattened out the response from the tweeter.

Comparing frequency responses of B&C DE12TC with 5kHz high pass Butterworth filtr and B&C 8NDL51 on 1.65Hz Butterworth low pass filter

This is looking very promising, shifting the filter point up to around 5kHz has created a fairly flat response from 2.5kHz to 10kHz on the HF device. There is a small peak just above 10kHz, which isn’t a bad thing usually, and will give a little HF sparkle at the top when listening directly on-axis. I did do a few iterations of small filter tweaks and measurements after this, and ensuring the physical positioning of components was just right, but eventually got the result we wanted in the red plot below:

 B&C DE12TC with 5kHz high pass Butterworth filtr and B&C 8NDL51 on 1.65Hz Butterworth low pass filter - measured together as a speaker system

Yes we know its not completely flat. As mentioned at the outset, the environment was such that room anomalies would feature in the measurements. The aim here was to make a crossover that balanced woofer and tweeter, and had a smooth transition from woofer to tweeter without any massive peaks or troughs, which has been achieved.

Going back to the reason for writing this article basically to show how standard off the shelf crossovers don’t usually work, we did a measurement with a standard 2.5khz Butterworth crossover calculated for 8 ohm components, you can see the result in the orange plot:

 B&C DE12TC and 8NDL51 measured with 2 different crossovers

If you understand the graph, you’ll be getting the earplugs ready. It has the potential to sound pretty awful with a 12-15db hill across the midrange and lower hf which will require significant EQ adjustment to rebalance.

The yellow-green plot below is with attenuation resistors added to the tweeter on thedefault 2.5khz crossover, whichjust seems to be introducing more problems. The orange plot is slightly different here, in the previous graph I recalled a measurement with a different mic position. The fact that the frequency response has changed a fair bit from just a small change in mic position is an indicator of phase response problems in the crossover. Basically the woofer and tweeter are not aligned, and in some listening positions will couple constructively, and in other positions couple destructively, causing hot spots within the listening area. The key point here is that the route we are going down, with the off the shelf crossover is not shaping up to be a good solution, despite us trying to make adjustments to rebalance the sound

 B&C DE12TC and 8NDL51 measured with 2 different crossovers, and with added attenuation

Whilst I knew I was heading down a dead end, for purposes of completing the test, I added a 3.3uF bypass capacitor on the attenuation resistors, to bring back the upper HF, this is shown in the green plot. Its a little better, but there is a sharp dip at 1.8kHz and a lump at 2.5kHz which are both undesirable, and not as good as the red plot. A lower value capacitor, such as 2.2 uF or 1uF would tame the 2.5kHz bump, but I already knew there were phase response issues with this solution. A little more refinement, and potentially I could have got the 2.5kHz off the shelf default crossover sounding OK on-axis, but I know it would have had issues off-axis.

 B&C DE12TC and 8NDL51 measured with 2 different crossovers, added attenuation and hi-shelf EQ

So its clear that the best frequency response has been obtained using a customised crossover following non-standard practices. At the outset, we were considering a standard 2.5kHz crossover. We ended up with a 1.65kHz low pass and high pass around 5kHz, but with the intent of flattening the frequency responses of the devices. There will probably be critics of this approach, as it doesn’t follow the text book methods everyone has been preaching for years. However, its a simple solution, requiring fewer crossover components, keeping size and cost to a minimum. Also, it works, and works quite well – so why not use it? I have seen similar crossovers used in ‘audiophile’ speakers costing many thousands of pounds. It may be possible to make a better crossover incorporating a zobel network, a large oversized L-Pad attenuator, and possibility using third or fourth order filters -the resulting crossover board could easily end up being larger than the speaker cabinet. This design approach provides for a compact crossover, which can be easily fitted inside a typical 8″ + 1 ” speaker cabinet.

Using a circuit simulation tool I wrote, I put the final design components in and produced the graph below. The tool calculates the attenuation of the filter from the component values and impedance entered, allowing you to calculate the theoretical frequency responses of the filters, and the summed acoustic power response. The result shows the attenuation relative to 0dB of the high pass (green) and low pass (blue) sections. It also shows the summed response in red (ignoring any phase response issues). The ‘crossover’ point works out to be around 3kHz – this is where the high pass and low pass have equal attenuation. Usually the crossover point is defined as the -3dB point of the high pass and low pass filters in a standard crossover. The net result of this circuit design is to put a 7-8dB attenuation across the mid band centered at 3kHz, effectively counteracting the bump in frequency response from the woofer and tweeter when used with a standard off the shelf crossover.

Just to bring things to a conclusion, I thought I would make some extra measurements, to check the speaker performance off-axis. Usually – if there are issues in the phase response of the crossover, which prevent the woofer and tweeter working in harmony, when you move the measurement mic off axis and repeat measurements, you will encounter peaks and troughs at various frequencies. Below are the various responses measured: Red is on axis, Yellow is approx 10 degrees off axis, Green approx 20 degrees, Blue around 30 degrees and purple around 45 degrees. The plots follow the expected results, where the wave guide on the HF device is controlling the dispersion, and the higher frequencies gradually get quieter off axis. The slightly higher response of the HF above 10khz is tamed when listening off axis to help maintain a good sound from multiple angles. Overall I am pleased with the result. All that’s left is to build a prototype cabinet and get some listening tests done.

On-axis and off-axis measurements of B&C 8NDL51 and DE12TC as a speaker system

The process of measuring, trial and error testing, simulations, and repeated iterations can take anything from 15 mins to several hours for each crossover. It depends on how well the woofer and tweeter interact with each other, and whether different approaches offer suitable solutions. Sometimes things just don’t work, and wont work, and no amount of iterations gets a good result – this usually means the wrong drivers have been selected that just don’t want to work together. Initially I started doing crossover designs this for personal research only, but I do offer custom crossover design as a service, with limited availability. It can be a fairly involved process, so the time and effort required needs to be justified, and its simply not viable to do it for everyone who asks. Additionally, I can only do this for components I have access to. If I don’t have the components, I cant measure them, and I can’t test the possible solutions. For best results, I also need the finished cabinet; the position of the drivers with respect to each other, the front baffle design, and other factors all affect the frequency response. Even a grill attached to the front can have an noticeable effect on the HF response, so to get this ‘exactly’ right – its best to have everything exactly as it would be in a finished cabinet. Its is absolutely impossible to design a custom crossover theoretically, you have to be able to measure and test in order to refine the solution.

For anyone who fancies building themselves a super high quality speaker, the components for this project are:

B&C 8NDL51 (8 ohm)
B&C DE12TC (8 ohm)
B&C ME15 horn flare

and a custom crossover:

Circuit diagram of a simple 2-way 2nd order crossover

L1 = 1.5mH

C1 = 6.3uF (5.6uF + 0.68uF)

L2 = 0.51mH

C2 = 1.8uF

This design isn’t intended for massive bass response, but you can get sensible levels of bass down to around 50-60Hz at medium volume with the cabinet tuned right. For best results at higher volumes you would need a separate subwoofer.


The Human Ear

Posted By Andy Kos

Something often overlooked, but I believe to be an important part of designing, building and configuring loudspeakers systems is understanding some of the basics of the human ear, and the effects of sound on the human body. This article is intended as a brief introduction, and is by no means exhaustive.

The smiley face equaliser


I’m sure you’ve seen this used, and possibly even done it yourself at some point in time.

Some would argue this is wrong, others that it is right.

The ‘smiley’ face curve often seen on graphic equalisers is similar to the effect achieved by the ‘loudness’ button on many hi-fi systems. It boosts the bass and treble to make it sound ‘better’ – but why do we think it sounds better?

Is it the speakers arent working properly? Maybe… we’ll discuss this later

But is something else wrong?

You might assume your ear works like a high quality studio microphone, with a flat frequency response across the audio spectrum, research has shown this is not the case. The way the ear responds to different frequencies varies considerably.

human ear


The graph above shows lines of perceived equal volume. First thing you will notice is that the smiley face equaliser curve is remarkably similar to the frequency response of the ear, but offset a little with the centre point around 3kHz and more emphasis on low frequencies. To some extent the smiley face can be explained as just naturally compensating for the human ear, making lower volume program material sound like we would expect it to sound at high volume.

Key Points:

Essentially deaf to bass frequencies:  This goes some way to explaining the loudness functions on hi-fi systems, at low volume, we find bass very difficult to hear, and it needs boosting significantly. As the volume increases the curve flattens, requiring less bass boost. In effect the loudness function is giving our ears the same balance as ‘loud’ music, but at low volume. Many people are unable to hear detail in bass frequencies, and some actually prefer the sound of distortion in bass frequencies, as they feel the sound is ‘warmer’

Most sensitive to mid-range frequencies peaking at around 3-4 kHz: Approximately the same frequency as a human high pitched scream or yell, which is not dissimilar to a baby’s cry. This means our ears are most efficient at detecting important sounds, research suggests this is down to years of evolution. Many alarm designers utilise these frequencies to maximise effectiveness. With out ears being so sensitive in the mid frequencies, poor quality sound, particularly distortion will be extremely noticeable, perhaps this goes some way to explaining the smiley curve; a way of masking problems in the mid-band by overpowering with bass and treble? Many people find distortion in the upper-mid frequencies painful, and this is often linked with occurrences of tinnitus.

Response varies with volume: As the volume increases, our ears hear differently. This is one of the reason many high-end large scale PA Systems utilise Dynamic EQ, where the equalisers are programmed to change as the volume increases. If you do apply equalisation to your sound system, you may need to adjust it for low/high volume.

So is the smiley curve correct? In my opinion, most of the time it isnt, particularly if you are playing back pre-recorded music the original recording will have been tweaked by the engineer to sound ‘right’. What is definitely correct is to equalise your system to make it sound right at the volume it is being used, and the room it is being used in, and the type of program material being played through it. If this happens to be a smiley curve, so be it, but as a system operator you should resist the urge to just boost bass and treble in the hope it will sound better. If you find you are doing this a lot, you might want to consider upgrading your sound system.



Be the first to comment

Totally Addicted to Bass?!?

Posted By Andy Kos

Its well known that a heavy bass line is in dance music is often very popular, and many people believe it is absolutely essential in order to create the best atmosphere for certain styles of music, but could we actually be addicted? Some people think it may be possible, and there is some research to suggest they are right.

To understand how this may be possible, we need to understand how sound affects our bodies. In modern life, one of our primary needs to hear is to communicate, often at moderate volume, but our ears can be much more useful than this, allowing us to be aware of things further away than we can see, and some of these things may help explain how our bodies react to sound.

Thousands of years before we had amplified music, bass frequencies, and how we reacted them, could have been critical to our survival. In nature, loud sounds, with an emphasis to low frequencies are often connected to danger. Just think of the sounds created by a stampeding herd of animals, an earthquake or a volcano erupting. Research suggests that years of evolution have developed the ‘fight or flight’ response in humans when presented with danger, this stimulates the production of adrenaline, enhancing the bodies ability to react to the danger.

You’ve heard of adrenaline junkies right? Well, it is possible that the brain associates high levels of bass with pleasure due to the mild adrenaline rush that bass frequencies may cause, and over time, coupled with other stimulants, could contribute to an addiction.

Another field of research suggests exposure to very high sound pressure levels (commonly found in bass frequencies) damages our ears and causes ‘pain’, however our bodies naturally react to this pain by creating numerous chemicals within the body, including adrenaline, endorphines and encephalons, collectively acting to blunt pain, but at the same time causing a pleasure enhancing morphine-like effect. This has yet to be proven, but the theories seem to hold true, and could also contribute to this concept.

One researcher has even gone as far as to suggest that extreme bass frequencies that penetrate the human body, causing you to literally ‘feel the bass’ may cause temporary damage to cellular structures within your body, cause the same pain blocking chemicals to be produced. These chemicals make you ‘feel good’ and may go some way to explaining the positive feeling experienced by high intensity bass frequencies.

So, is it possible to be totally addicted to bass?




Be the first to comment

What’s up with the Watts?business path choice

Choosing the right driver for your speakers can be something of a minefield, particularly if its your first time. There are a number of things to be aware of, one of which is the power rating (specified in Watts)

Technological advances in materials are allowing Loudspeaker drivers to be developed with larger power handling, at the same time we have seen some loudspeakers drivers have their power ratings changed, with increases of 25% or more, but with apparently no change to the driver. Add to this the confusion of RMS Power, Continuous Power, Program Power, and Peak Power and it’s not suprising some of you are getting confused. This article aims to explain some of terms used, and dispel a few myths.

Q: My amp is rated at 400W per channel, will a 600W driver break my amp by drawing too much power?

A: NO. The power rating of your amplifier is a measure of how many watts it can deliver to the speakers before it reaches the limitations of its internal power supply and starts to clip/distort.  The power rating of your speakers is the maximum power they can accept from an amplifier before they are in danger of overheating and burning out. Providing you have the corrent impedance load, your speakers are unable to draw more power than the amplifier is willing to give. https://speakerwizard.co.uk/impedance-faqs/

Q: If I replace my 400W speakers with 450W speakers will they go louder?

A: Not necessarily, the limiting factor is generally your amplifier, if you amp is rated at 400 Watts, you wont get any more than 400W output without severely distorting the sound, and potentially damaging your amplifier. If your amp can output 450W, then an increase in power handling may make your speakers go a little louder, but its possible they may be no louder, or in some cases quieter. The key factor here is efficiency, some speakers are more efficient than others. If you have two speakers operating at the same power level, and one is more efficient at converting electricity into sound, it doesnt take a rocket scientist to figure out which of the two will be louder. Most manufacturers give an indication of efficiency using the sensitivity figure measured in db@1W/1M (link to sensitivity)

Q: Which power rating should I look at?

If you’re reading this, chances are you are a novice, so for simplicity we suggest you use the continuous RMS power rating. Luckily, this is the one most manufacturers specify. You may also see Music Program Power Ratings, typically these are double the  RMS power rating. If you see a peak power rating, its often meaningless, and serves little purpose. Peak power ratings are often four time the RMS rating, so if you see this anywhere, divide it by 4 to give you an idea of the real power rating.

Q: What do the power ratings mean?

RMS Power.

Sometimes referred to as Average Power, or Continuous Power. The term RMS here is incorrectly used, it is not RMS power in the true sense of the term, as Power does vary from positive to negative, it is the Voltage. The RMS Voltage is used in the power calculation, hence giving rise to the term ‘RMS Power’.

Over the years, various standards have been used in this ‘RMS Power’ calculation, including;

The IEC268-5 (1978) standard (IEC = International Electrotechnical Commission)

The EIA RS-426-A (1980) standard (EID = Electronic Industries Association)

The AES2-1984 standard (AES = Audio Engineering Society)

The AES2-2012 standard – which is becoming the most commonly adopted standard. Previously many manufacturers used the EIA standard, re-ratings some speakers using AES saw increases in power ratings of 25%, sometimes more. One example we are aware of was a high power 18″ driver, which was previously rated at 600W, and had the power rating changed to 800W overnight – with no changes whatsoever to the speaker design.

How is this possible? The AES power rating has a higher crest factor than the EIA rating, and is also for a shorter duration 2 hours for AES whereas the EIA rating was over an 8 hour test period. Some would argue this is not a totally realistic test. However for the layman, it is a useful benchmark for matching up driver power to speaker power. We’ll explain why:

Music Power:

Also know as Program Power, this is an indicator of the power rating of speaker use with ‘typical program materal’, which most of us call music. The RMS Power test is not done with music, it is generally done with bandwidth limited pink noise, which is a continuous signal. When was the last time you played music that sounded like static? Not often I bet. Music power takes into consideration that a bass beat is not continuous, it is a series of pulses. In between the pulses, there is no power being applied to the speaker, so when you average out the power over time, you can handle much more power.

What does this mean? Well you can exceed RMS Power for short periods of time, but not with a continuous signal. So if you play average music, you can usually run above RMS Power level quite happily in most instances, and with appropriate limiters in place, somewhere between RMS power and Music Power is generally a safe level. This is why I’m saying the Music Power is a useful benchmark – it helps determine amplifier power choice ‘roughly’ and is a good level to aim at to keep your drivers operating within safe parameters, if like most people, you try to push things just a little harder every now and then, you should still be fine, just as long as you don’t start treating music power as a long term continuous power rating, as you will almost certainly cause your speakers to fail if you do this.

Peak Power:

This is the maximum short term power that can be applied to the driver, and is typically calculated to be four time the RMS Power. I recommend you dont use Peak Power for anything except bragging to someone who doesnt know anything about sound.

Should I buy the most powerful speakers I can afford?

Probably not – It’s best to get speakers appropriate to your requirement. This is partly due to how speakers work. High power speakers are designed to have a high excursion (thats lots of cone movement) – in order to be able to handle the extra power, they typically have stiffer components, particularly with regard to the suspension. These high power speakers require a certain amount of power to overcome the mechanical resistances of the suspension. Comparing two extreme examples, of say a 100W speaker and a 1000W speaker running bass. The 100W speaker will have a low output for say, the first 5-10W put into it, once you are putting in 50W or so, the speaker will be at half its rated power, and will be (depending on sensitivity) fairly loud, ramp it up to 90-100W and the driver is giving all it can, potentially operating at its most efficient point. Lets suppose you only have a 100W amplifier, the 100W speaker will give you more output than the 1000W speaker. If you put your 1000W woofer on the end of your 100W amplifierm the first 40-50W of power will be used inefficiently, just to overcome the stiffness and resistance of the suspension, ramp it up to 100W, and you are still only tickling the 1000W woofer, you will find that you may need 200W running through the 1000W woofer to be as loud as the 100W woofer. As you keep ramping up the power, the 1000W woofer will ultimately create a lot more sound output than the 100W speaker ever will, but if you only have a small amplifier, you are better off with an appropriately matched driver.

So what if I exceed the power ratings?

You run the risk of overheating the voice coil of the speaker, and causing it to fail – but be careful – just keeping an driver within it’s recommended power rating is no guarantee of longevity. You also need to be away of excursion (often specified using Xmax and Xlim) as you can damage an driver through over-excursion without exceeding the power rating.

What about Power Compression?

Power compression is the little bugbear that can upset your best laid plans, and give you reason to throw the manufacturer’s specifications out the window. Many manufacturers choose to ignore power compression, some actively avoid specifying it or even mentioning it.

The sensitivity (efficiency) manufacturers specify is measured at a power level of 1W at a distance of 1m from the speaker. At 1W, very little heat is lost within the voice coil, so the driver is more effective at converting electricity into sound.

Its very common for loudspeaker voice coils to be wound from copper, which has a positive temperature co-efficient of +0.393% per degree C. It’s quite feasible for the voice coil of a high power speaker to reach 200 degrees Celsius, which could mean a change in the resistance of the wire to increase of 50% or more. Your 8 ohm driver will no longer be an 8 ohm driver, and the nominal impedance could rise to as high as 13 or 14 ohms. At full power, many drivers are no longer operating at their stated impedance, and they generating a lot of heat.

A good quality driver, with low power compression at its full rated power, could be 3-4 dB louder than a driver that suffers badly from power compression. Many modern designs are taking account of this, and making efforts to ensure driver cooling is maximised to counter the effects of power compression.










Be the first to comment

Power Compression

Posted By Andy Kos

When selecting speakers, it’s common for people to just look at maximum power handling, and many manufacturers make a point of specifying seemingly unbelievable power handling capacity of 1000W or more. Its quite rare for manufacturers to specify power compression though, and it seems to be often overlooked by system designers.

It seems that loudspeakers to handle what appear to be insanely high levels of power compared to 10 or 15 years ago. Has there been some amazing technological breakthrough? Do we need to re-write the physics text books? No, it’s still just basic physics – so what are the changes?

Firstly, modern materials used in the construction of voice coils are able to withstand significantly higher temperatures before failing.  Why is this important? Well Cone loudspeakers are in fact very inefficient, with even the best transducers only converting around 5% of the electrical energy supplied into sound, the majority of the remainder is converted into heat. So a 1000W bass speaker running at full power may well be converting only 50W into acoustic power, and 950W of heat. Thats like having a 1kw bar heater in your bassbin! That’s a lot of heat, which can cause big problems.

Aside from improving construction materials, manufacturers are also refining designs to maximise heat transfer away from the voice oil, this also contributes to the increases in power handling capacity we are experiencing.

What’s all this got to do with power compression?

Enabling speakers to handle much higher temperatures might seem a good thing, as it increases maximum power handling, but it also has a detrimental effect. Most voice coils are made from copper or aluminium wire, both of which have a positive temperature co-efficient of around 0.4% per °C. What’s the significance of that? You will have heard of superconductors, which operate at extremely low temperatures in order to try to reduce and minimise resistance.  Loudspeaker voice coils  unfortunately work in the opposite way: as the temperature increases, the resistance also increases.

A modern state of the art voice coil is designed to withstand extremely high temperatures, often operating at up to 3000C or more when driven at full power. 0.4% may sound insignificant, but remember this is per °C – at only 2300C the voice coil DC resistance has almost doubled which causes the voice coil impedance to increase accordingly. Some simple maths and you can quickly see that the increase in temperature  can make your 8 ohm speaker start behaving more like a 16 ohm speaker.

So after setting your sound system carefully at the start of the night, an hour in, and it doesn’t sound as loud – you might wonder whats going on. Two things: firstly, your ears have a self defence mechanism: there are 2 tiny muscles in the middle ear that will contract when the ear is exposed to loud sounds. This contraction will reduce the loudness of the sounds reaching the inner ear, thereby protecting the inner ear against exposure to loud noises. This isn’t power compression, but it’s something to be aware of, as you may well be tempted to turn up the volume, I know from experience that a typical DJ will certainly try this, and end up running his mixer into overdrive in the attempt to get more volume.

The second factor is power compression, a typical loudspeaker can lose 3-6 dB of volume once power compression kicks in.

You could think of power compression a bit like the aerodynamics of driving a car. When you start moving, a certain level of power from your engine sets you hurtling forwards at high speed, but as you go faster, wind resistance increases, so you stop accelerating. You need to apply more power to increase speed, but wind resistance keeps increasing too, so you have to apply even more power.

If your amplifiers have headroom, your instincts will make you want to turn them up, to restore the original volume level. To some extent this will work, if you’re familiar with the maths, you’ll see whats going on. Your 8 ohm speaker at room temperature happily accepts 1000W from your amplifier, and gradually reaches an operating temperature of say 250°C. Your resistance has doubled, and your ‘new’ 16 ohm speaker will probably only be receiving around 500W from your amplifier. In a way, as the speaker reaches temperature, it ‘protects itself’ by reducing the power it is operating at, stopping it getting any hotter. If it were to cool a little, the power would increase again, causing it to heat up.

Lets suppose you turn the gain up on your amplifiers, determined to try to push 1000W through your speakers. As you apply more power, you will generate more heat,  perhaps reaching 350°C or more, with your speakers resistance continuing to increase to perhaps 20 or more ohms. Essentially you are fighting a losing battle, as you turn the gain up, the speaker fights back with a higher resistance. You will eventually reach a limit, either your amp will run out of headroom and you cant turn it any louder, or the other possibility, which happens all too often, is your speaker will overheat, and burn out causing catastrophic failure.

Now you know about power compression and the fact that speaker resistance increases with heat, you’ll probably realise that you actually have to push a speaker very hard in order to cause it fail – so if your speaker suddenly fails and you smell burning, the only person to blame is YOU, as you now know better than to try to fight power compression by applying more power.

Now consider what effect power compression will have. 3-6dB loss at full operating power is almost like switching off half your PA system. To achieve the same consistent volume you will need twice as many speakers!

What’s the solution? Either buy speakers with headroom, e.g. if you want to operate at around 500-600W, you might want to consider purchasing speakers rated at 800W or more. At 75% of rated power, the effects of power compression should be much less significant. Also, try to select speakers with improved cooling technology, that suffer less from power compression. Avoiding power compression could make your speakers twice as loud, meaning you could take half as many to the gig!

There are other side effects from the increased levels of heat in a speaker, T/S parameters can vary, bass can sound boomy and mid frequencies can sound muffled. For the best sound quality, its best to try to  minimise power compression effects,




Impedance – FAQs

Posted By Andy Kos

How do I know what impedance load I have?

Most manufacturers will specify impedance, and will include it in the product specifications, often printing it on the speaker itself. If you don’t have this information, you can measure the DC resistance using a multi-meter (please note Resistance is NOT Impedance – find out why here: https://speakerwizard.co.uk/impedance-and-resistance-whats-the-difference/

You should only measure the resistance of speakers when they are not in use, and not connected to an amplifier. By putting your multi-meter probes on each terminal of the speaker you will get the DC resistance, which can be used as a guide to get the impedance. A DC resistance of 5-6 ohms is normal for a driver with 8 ohm impedance, around  12-13 ohms  is common for  a 16 ohm impedance driver, and  3 ohms DC resistance would be normal for a 4 ohm impedance. You may notice that moving the cone whilst checking the resistance will make the reading change, this is because the voice coil is moving in a magnetic field, which will induce a voltage in the  coil, which in turn will affect the multimeter’s measurement.

Many loudspeaker manufacturers will label the drivers to make identification easier, Eminence for example include a suffix on the drivers, for example the Beta12A is the standard model, and is 8 ohm impedance, the letter A designated 8 ohm impedance. The Beta 12B is 16 ohm impedance, and the Beta 12C is 4 ohm impedance. This same letter designation is used through the range of Eminence speakers.

I have more than one speaker in parallel – what’s the impedance?

First, let’s clarify what we mean by parallel, this is where the electrical paths through the drivers from + to – run in parallel to each other. If you trace a route from + to – you go through either one driver, or the other. The diagram below shows two speakers wired in parallel:


 To wire speakers in parallel, all you have to do is connect the + (positive or red terminal) on each speaker to the + (positive or red terminal) on your amplifier, and the corresponding – (minus or black terminal) on the speaker to the – (minus or black terminal) on your amplifier. If you plug several speakers into one amplifier, unless you have unusual cabling, this would be the standard way you would run several speakers off one amplifier.

Its normal to put speakers of the same impedance in parallel with each other, mismatching impedances isn’t a great idea unless you have a fairly advanced knowledge of speaker systems and are doing this for a specific purpose.

So what does this do to the impedance?

The impedance of each speaker stays the same, but the impedance load the amplifier sees will change. In the diagram above, if the two speakers were both 8 ohm impedance, the load the amplifier would see is 4 ohms. To think of this in simple terms, you could think of one loudspeaker as a busy road with a specific amount of traffic travelling along it, if you have two roads, the traffic can travel along either road, which presents less ‘resistance’ to the same amount of traffic. With a basic knowledge of maths, and using this analogy of two routes between start and finish, you can guess what the resistance of two parallel 8 ohm drivers would be, it’s half that of one 8 ohm driver, and is 4 ohms.

The formula for calculating parallel resistances is as follows:


R1, R2, R3, are the individual resistances, the formula works for as many, or as few resistances there are in parallel, for two drivers in parallel, you use R1 and R2 only, for three drivers you use R1, R2 and R3.

RT is the total parallel resistance. For equal parallel resistances, the formula becomes very simple, as the table of parallel 8 ohm impedances shows:

No drivers Parallel Impedance Fraction
1 8 ohms 1/1
2 4 ohms 1/2
3 2.6 ohms 1/3
4 2 ohms 1/4
5 1.6 ohms 1/5
6 1.3 ohms 1/6

As you can see, 3 drivers gives a combined parallel impedance of one third of the original impedance of 8 ohms, and 4 drivers gives a combined parallel impedance of one quarter of the original impedance.

Very few amplifiers will run happily into impedances below 2 ohms, and there is a strong possibility you can damage the amplifier by plugging too many speakers into it. Some amplifiers will not work safely below 4 ohms, so it’s quite important to ensure you have the correct load on your amplifier.

How do I wire speakers in series?

The term series where things are arranged in sequence implies how you would arrange speakers in series, as per the diagram below you can see that the positive (+) terminal of the first speaker is connected  to the positive (+) of the amplifier as normal, but the negative  (-) terminal goes the the positive terminal of the second speaker. The last speaker in the series has it’s negative (-) terminal connected to the negative (-) terminal of the amplifier.


Series impedances work opposite to parallel, going back to the comparison with traffic, if your busy road has traffic lights in it, every extra set of traffic lights adds more resistance to traffic flow. In the same way, each loudspeaker in series adds to the impedance. To calculate the total impedance, simply add together the individual impedances, as shown in the table below. In most instances, its rare to have more than 2 drivers wired in series, as the increase in impedance will mean most amplifiers are able to deliver very little power to the drivers.

No drivers Series Impedance
1 8 ohms
2 16 ohms
3 24 ohms
4 32 ohms

 If we get less power, what’s the point of connecting drivers in series?

If you just have one pair of speakers, there isn’t much point, but it gets interesting when you have multiples of speakers. If for example you have four speakers, that are 8 ohms, and you want to run all four speakers off one amplifier, you could wire all four in parallel, to give a 2 ohm load, or all 4 in series to get a 32 ohm load. But what if your amplifier wont work below 4 ohms?

The solution is simple, a series-parallel combination:



Assuming all drivers are 8 ohms, some simple maths and you can see that each of the two series combinations has an impedance of 16 ohms. Two 16 ohm impedances in parallel have an overall impedance of 8 ohms. What this allows you to do is use four speakers where you would previously have only used one, giving you a significant increase in power handling.

Variations of series-parallel configurations are common in guitar speakers,  4 x10″ and 4 x 12″ cabinets are common, with different wiring to suit specific applications and impedance requirement. Many guitar cabinets utilise 16 ohm drivers in order to achieve the desired results.

Its sometimes advised that its best to avoid using series configurations with speakers, due to the fact that that you have two coils or inductors which can induce unwanted voltage and cause distortion. Series configurations are rarely used in hifi or studio systems.








1 Comment. Join the Conversation

Why does my 8 ohm speaker read 6 ohms when I measure it on a multimeter? It must be faulty right?


I’ve heard this so many times I’ve lost count, but there is a difference between impedance and resistance. When you measure resistance with a multimeter you are measuring DC resistance. The DC resistance is determined by the copper (or sometimes aluminium) wire in the voice coil of the speaker, and is actually as the name suggests; resistance to the passage of electric current through the copper. The key point here is that the electrical current travels in one direction only, and is fixed and does not change.

Impedance is equivalent to resistance, but for circuits where the voltage and current change, such as in a loudspeaker. An extra factor comes into play, which is the fact the the loudspeaker is based on a coil of wire. This coil of wire acts as an inductor. Without getting too involved in the science part of this, its sufficent to know the inductor creates an additional ‘reactance’ to alternating signals, which when added to the DC resistance of the voice coil, gives the overall Impedance.

To complicate matters further, the Impedance varies with frequency, so the 8 ohms specified for loudspeakers is not totally accurate, it is referred to as ‘nominal impedance’ – a kind of ‘average’ impedance figure that can be used for typical calculations involving loudspeakers. The graph below show a typical 18″ subwoofer, the impedance is shown on the scale on the left hand side.


For purposes of being able to run your own sound system, or building your own speakers, it’s sufficient to accept the manufacturer’s quoted impedance as being correct for your application. You don’t need to be concerned with the finer points of impedance unless you get into more serious aspects of speaker design, and if you’re at that level, I highly doubt you will have bothered read this far, as you will know all of this already!


Be the first to comment

With current technology, it’s  impossible to have a single transducer that is able to reproduce the entire audio spectrum effectively, different types of loudspeaker driver are better suited to different speakers. Typically those handling lower frequencies are cone drivers, and commonly known as woofers. Drivers handling higher frequencies are usually much smaller, and are often known as tweeters. In many basic speakers its common to have a woofer and a tweeter in order to cover as wide a range of the audio spectrum as is possible.

A crossover is a device which splits sound/music into two or more frequency bands. In the case of a basic two-way system there would be two bands, one band covered by the woofer, and one by the tweeter.

Why can’t we just put a Woofer and Tweeter in parallel without a crossover?

Tweeters can not handle bass frequencies, lots of bass into a tweeter would destroy it. Tweeter must have high frequencies only, limited to the frequencies the tweeter is designed to handle.

In a very simple speaker, you could just use a high pass filter in series with a tweeter, in parallel with the woofer. The high pass filter would remove damaging bass frequencies and keep the tweeter operating safely.

For purposes of simplicity, all diagrams will be simplified, and a 1st order filter is assumed to be in use. A 1st order filter does not require a connection to negative (-) and can be simply put in series as per the diagram. Any filter 2nd order or higher will require a connection to negative from the filter, which will be explained in more detail in another article: https://speakerwizard.co.uk/passive-crossoversfilters-how-do-they-work/

Simple speaker with Passive High Pass Filter in series with tweeter:


HPF in the diagram is the High Pass Filter, which must be fitted in series with the Tweeter in order to protect it.

For basic speaker designs, this solution may sometimes be acceptable, but you need to be aware of the fact that below the filter frequency, the impedance of the circuit will be 8 ohms, as the amplifier will only see the woofer as the load, but at high frequencies, the amplifier will deliver power to both the woofer and the tweeter, this will present a 4 ohm load to the amplifier at higher frequencies. (8 ohm impedance of woofer and tweeter assumed in this example)

If you only intend to put one cabinet on the output of the amplifier, this wont present a problem, but if you use multiple cabinets you may find the overall impedance drops too low, which is undesirable.

Many woofers are also very inefficient at reproducing high frequencies, whilst they will readily allow power from the amplifier, they wont necessarily turn that power into anything useful, in effect wasting the power from the amplifier.

The final thing to consider is that some woofers really dont sound good outside their designed operating range, so whilst putting high frequencies signals through the woofer wont damage it, the woofer may just sound completely horrible when it tries to produce those frequencies.

The Solution? 

The solution is to put a Low Pass Filter (or LPF) in series with the Woofer, this filters out high frequencies, so that the woofer is only producing sounds that are in its operating range.

2-way crossover

The above diagram shows a passive LPF in series with the woofer, and a passive HPF in series with the tweeter.

A matched LPF and HPF that usually share the same cut-off frequency form a system known as a crossover. With a simple two-way system, crossover frequencies of between 1200 Hz and 3000 Hz are common, depending on the components used.

The cut off frequency is the point in the audio spectrum at which the filter begins to take significant effect, in the case of a Low Pass Filter, frequencies significantly below the cut-off frequency should be passing through unaffected. Just slightly below the cut-off frequency the filter begins to take effect, and starts blocking. The cut-off frequency is generally regarded as the point where the signal is at -3dB, and is in the middle of the ‘knee’ or bend in the response graph. Just above the cut-off frequency, the level begins to drop off rapidly, blocking higher frequency signals from passing. The HPF filter works the opposite way around.


By aligning the cut-off frequencies to be the same on the HPF and LPF circuits, the system impedance will stay more or less the same over the audio spectrum. Overlapping the cut-off frequencies of passive filters will cause the impedance to drop in the overlapped range. Leaving a gap between the cut-off frequencies will cause the impedance to rise in the gap.

It is possible to create more elaborate passive crossovers, such as three way crossovers that split the sound into bass, mid and treble. For smaller applications, such as hifi or studio speakers this is fairly common, but this becomes less common in high power PA speakers, as the component costs can increase significantly and in some instances it becomes difficult to source parts that can handle sufficient power

So far, we have only tackled passive crossovers..

So what is an active crossover? and whats’s the difference?

Passive crossovers do not have their own power source, all they can do is block the signal, they are regarded as passive as they can not increase it or amplify it. Passive crossovers/filters are placed between the amplifier and the speaker driver(s).

Active crossovers work quite differently. DO NOT ever fit an active crossover between the amplifier and driver, they are designed to go BEFORE the amplifier.

An active crossover will split the signal at line level, before it reaches the amplifier. The amplifier will then only amplifier the desired frequency band and deliver those frequencies to the speaker. This is a better solution, as it is more efficient – any passive crossover will have losses in it due to the components used to do the filtering. The losses amount to wasted power. Also, in cheaper crossovers, distortion can be introduced from cheaper components. Low losses and minimal distortion can be achieved with passive crossovers, but the cost of components can become astronomical, making active crossovers a better solution. There are also physical limitations with what can be achieved with passive crossovers, and as the overall system power is increased, passive crossovers become a less desirable solution.

There is a significant difference with using active crossovers, you need more amplifiers.

By splitting the signal BEFORE the amplifier, you then need a separate amplifier for each frequency band. In the case of a two-way system you will need two amplifiers, for a three-way system you will need three amplifiers.

Each amplifier will only be used to run speakers within a specific frequency band, as per the diagram below.


An active crossover also gives a much greater level of control over the system, with a typical crossover allowing boost or gain of each frequency band, and adjustable crossover frequencies. Some more advanced active crossover also allow variation of filter type (Butterworth, Linkwitz-Riley, Bessel, etc) to give precise control over the system configuration.

For large HIGH POWER systems, active crossovers are the preferred solution, with a separate amplifier for each band.

For small-medium size systems, a hybrid crossover solution is common. A 2-way active crossover is used to split bass from the mid and high frequencies, this requires one amp for Bass, and one for mid-high. The Mid-High cabinet then utilises an internal passive crossover to split between Mid and Treble. This solution is something of a compromise, it doesn’t quite give the total control of a fully active system, but it reduces the number of amplifiers needed, by not requiring a dedicated HF amplifier, and also simplifies cabling a little – eliminating the need for four core cable to run to the mid-high cabinet. This is a very common solution, as it provides a good balance of versatility and cost.



 Whats best active or passive?

Its generally regarded that active crossover solutions are best, for a number of reasons:

1. Passive crossovers are always lossy, even the best passive crossovers lose some power within the crossover, primarily due to the DC resistance of the inductors.

2. Sound quality. Passive crossovers using cheaper components can often suffer from sound quality issues, to achieve better sound quality costs often escalate with passive crossovers.  Generally, active crossovers offer better sound quality than passive crossovers.

3. Active crossovers allow for a more accurate predictable response, there is always some manufacturing tolerance with inductors and capacitors with variation of values of +/- 5% being common. This can often mean (more so with cheaper components) that no two passive crossovers will produce exactly the same response, so if your system comprises of numerous speakers, they could all be producing a different sound around the crossover frequency.

4. Better control. With active crossovers its much easier to balance different frequency bands. Its common with passive crossovers to require attenuation of high frequencies, through the use of attenuation resistors. With an active crossover you can just reduce the gain.

5. Easier scalability. Active crossovers make it easier to increase the size of your system, you can simply add more amplifiers and more speakers, and run them off the same signal from your crossover.



Be the first to comment

To save you time, we’ve put together a calculator for one of the most common passive crossovers, a 2nd order Butterworth. You can configure the Impedance of the woofer and tweeter, and the crossover frequency.

C1, C2, L1 and L2 correspond to the relavant capacitors and inductors in the diagram below:

2nd order Crossover medium

Where Vin is connected to the positive terminal of your amplifier, and 0V is the negative terminal. The dots on the circuit signify where components/wires are connected. L2 connects to the input, and NOT to 0V, hence the ‘loop’ in the wire in the diagram.





New improved version of the crossover calc this now includes a graphical plot of the frequency response. Due to the size of the graphics, the form below will submit to a full page version of the calculator. You can select 1st order or 2nd order slopes, with the option of Linkwitz-Riley on 2nd order. We will add 3rd order and 4th order in due course. This calculators works two ways, you can enter the frequencies and impedances and calculate the component values, or you can enter the component values to get the crossover frequencies and see the frequency response. This version also allows different impedance and frequency between Low Pass and High Pass, as well as different slopes. So you could for example have the Low Pass section with a 8 ohm woofer, crossing over at 1200 Hz, and the High Pass at 16 ohms crossing over at 1800 Hz. Combinations like this are becoming increasingly common, as using a 16 ohm HF driver often negates the need to put attenuation in the HF part of the circuit. Also, a typical 1600Hz Butterworth crossover can often have a peak in response around the crossover frequency, particularly if the HF driver is highly efficient – offsetting the crossover frequencies may seem counter-intuitive as it might appear you are leaving a hole in the response, but often the coupling between LF and HF counteracts this. If you already have a crossover, you can simulate the response using the lower part of the controls. Please check you have component values correct, Capacitors should be specified in microFarads (uF) and Inductors in milliHenries (mH). Most pre-built crossovers will have capacitor values printed on the components, unfortunately very few divulge the Inductor values, to get these you will need an appropriate measurement meter.

2nd order Crossover Calc
To get the component values for a crossover, enter the impedances and crossover frequencies for the high pass and low pass sections and then click ‘CALC’
Low Pass Fc:
Woofer Impedance: Ohms
High Pass Fc:
Tweeter Impedance: Ohms
To see the response and crossover frequencies for known component values, enter these in uF and mH in the boxes below and click ‘CALC’
C1: uF
L1: mH
C2: uF
L2: mH