[DIY] Passive Crossover in the DSP Era

And here we are, dear reader, at the trickiest part of this series of articles: the fateful measurement! Go back to Part I, in case you missed it!

But first...even if you stumbled upon this page by chance, listen anyway to this interview by Darko Audio with the renowned designer Andrew Jones. Good old Andrew talks about the practical and conceptual issues related to measurement, how cabinet design is carried out in the audio industry, and the budget differences a designer has to manage between speakers of different price ranges. It's VERY interesting.

Even though this is a mini-guide on passive crossovers, the measurement practices described here also apply to those using an active crossover or a DSP. Anyone implementing active filters without considering the driver's impedance and frequency response will indeed get a similar - and equally unsatisfactory - result to that, which we will see in the next article, obtained with passive filters calculated using standard formulas. But now let's get straight to the point, because there's a lot to cover.

Basic Equipment

To record the response of our speakers and drivers, we will need an adequately accurate omnidirectional microphone (you won't need a studio microphone; various models like Omnitronic, Dayton Audio UMM-6, UMIK-1, etc.. will suffice), a tripod, an amplifier - even a cheap one, the Fosi V3 or similar will work fine - a PC, and preferably a DAC, even a very inexpensive one.

For those who want to go beyond measuring just the frequency response, a tester such as the Dayton Audio DATS V3 LINK will allow us to also measure the impedance of speakers in the enclosure and quickly obtain the Thiele and Small parameters of our drivers. If you can find the impedance files (ZMA) for your speakers and you're just starting out with DIY builds, you can postpone the purchase...but it's not uncommon for fellow DIYers to upgrade their tool collection, so if you find one used, I suggest helping our colleague free up some space in their workshop.

There are many measurement software options, but if you're a beginner, my advice is to start with REW. Free and versatile, it supports microphone calibration files, making it possible to use even as a sound level meter. It is the most intuitive and manageable measurement program, and some volunteers have provided clear tutorials translated into various languages. After getting familiar with REW, I suggest trying ARTA.

Methods for measuring a loudspeaker

Even without an anechoic chamber, it is possible to obtain reasonably reliable measurements. However, we need to pay attention to the positioning of the speaker and microphone, as well as to reflective surfaces around them. The measurements described here should be performed with the speakers already mounted in the enclosures you have carefully designed and built. There are several ways to measure our speakers; the simplest are:

Ground Plane Measurement: Place the speaker outdoors on a wide, rigid, and as flat as possible surface, free from other objects that could introduce reflections. The microphone should also be positioned at ground level, aligned with the speaker. This setup allows you to capture the direct sound, minimizing floor reflections. Ground plane measurements provide a realistic evaluation of the speaker's response but only in the low and mid-low frequencies (ideally around 400 Hz).

Elevated Outdoor Measurement: Place the speaker on a stable support several meters above the ground, with an equally tall tripod for the microphone. Weather permitting (no rain or strong wind), this is an effective and reliable method. For example, assuming a height of 5 meters from the ground, and equal distance from other walls or large objects, you can measure the frequency response with reasonable accuracy starting at around 70 Hz. Increasing the distances proportionally improves low-frequency precision.

Compromise Measurement: Life is full of compromises, and as DIYers we must accept some as well. Measuring the speaker indoors using modern technology is the simplest method, and the one illustrated here. Let's get started!

Gated in-room measurement

Place your speaker at the centre of a room, as far as possible from walls and large reflective objects, maintaining a minimum distance of 1 meter. Raise it off the floor at least 1 meter if it's a bookshelf speaker, or 50 cm if it's a tower, using a stand, a stack of books, or any other stable support.

Connect the first speaker to be measured to the output of an amplifier (turned off), using simple speaker cables with alligator clips or any other system that ensures a solid and stable connection between components. The PC's audio output should be connected to the amplifier input, preferably from a DAC connected to the computer. Finally, connect the measurement microphone to the PC.

The microphone must be supported by a proper tripod - makeshift arrangements won't work and will just make you waste a lot of time. A decent microphone stand costs around €20. To capture the direct sound of the driver while minimizing reflections, the microphone should also be placed as far as possible from walls (at least 1 meter), aligned with the tweeter, and 1 meter from the speaker. Considering minimum safe distances between speaker, microphone, and surrounding walls, the measurement requires at least 3 meters of unobstructed space to ensure a sufficiently clean time window.

Once you open REW, click on Preferences in the top-left corner. The window shown below will appear. Select the output device - in this example, I chose the PC output. It is not perfectly linear, but if you don't have a DAC, this will do. Finally, make sure the signal input is routed to the amplifier output (in this case, R) to which the speaker is actually connected. Close the menu.

Now click the Measure button. A window like the one below will appear - follow the instructions highlighted in the image. Set the measurement to SPL (Sound Pressure Level) and adjust the frequency range of the sweep - the signal sent to the speaker for the measurement. If you are measuring a tweeter or midrange, set a higher minimum frequency than the usual 20 Hz, to protect it from excessive excursion that could be useless or potentially damaging. To protect the component, the measurement must also be carried out at a moderate Sound Pressure Level (SPL) (see below). If a tweeter has an fs of 800 Hz, it is prudent to start the sweep from approximately 1000 Hz, ensuring that the test volume is not high. Those who are particularly concerned about burning out the tweeter should insert a non-polarized protection capacitor in series with the signal - between the positive (+) terminal of the amplifier and the positive (+) terminal of the tweeter. A 20 µF capacitor will create a high-pass filter at 1000 Hz for an 8-ohm tweeter and 2000 Hz for a 4-ohm tweeter. You can always leave the maximum frequency at 20 kHz.

Next, set the SPL limit, beyond which the measurement will be automatically stopped. You don't need a high volume to capture the frequency response: you can safely measure at 75 dB SPL, setting the safety limit at 82 dB. This way, if the signal exceeds that threshold, the software will immediately stop the sweep, protecting the drivers, the microphone, your hearing, and your neighbours. If you have a microphone calibration file (Dayton Audio UMM-6 or MiniDSP UMIK 1 & 2 provide them), load it via the Calibration Files option. If your microphone doesn't have a calibration file, remember that to read an actual SPL value (dB SPL) you will first need to calibrate the system using an external sound level meter. Before starting the measurement, use the Check Level button: the software will confirm whether the signal level is appropriate relative to the room's background noise. In any case, keep the amplifier volume low before performing this check. Finally, double-check all settings and click Start.

And now...total chaos! Why? Because we have the speaker measurement plus the room with all its reflections. Fortunately, we have a sophisticated tool to deal with this problem: gating! Gating allows us to isolate the direct response by excluding from the measurement any signals that reach the microphone after a certain interval of time - the reflections, expressed in milliseconds. Understanding this step is important because it will let you adjust the gating depending on the environment where you measure. Sound takes 2.92 ms to travel 1 meter. If the speaker is 1 meter from a wall, the first direct reflection will reach the speaker after roughly 5.8 ms: 1 meter to the wall (2.92 ms) and 1 meter back (another 2.92 ms). Now you see why I suggested keeping a distance of 1 meter from walls. By setting the Right Width window to 5 ms, reflections will be excluded from our measurement. Click the IR Windows button, and set the three parameters circled in red as shown in the image. Finally, click Apply Windows to All and...

By correctly setting Ref Time, Left Width, and especially Right Width, you obtain a pseudo-anechoic measurement of the speaker's frequency response in a domestic room, with the limitation that low frequencies are excluded from the measurement.

Considerations

Measuring low frequencies is the main limitation of gated in-room measurements. A critical aspect of this method is that we cannot detect the baffle step: a phenomenon where, as frequency decreases, the speaker radiates sound more omni-directionally rather than just forward, reducing the level measured on-axis.

However, a compensation network is not always necessary: bass reinforcement from the room - such as placing speakers near walls - often masks the effect. This is why most commercial speakers do not integrate a baffle step, also to reduce costs and avoid loss of sensitivity.

The method I've described is the starting point for our measurements. It is also the most convenient and always available, even for those who cannot measure outdoors. Once you are familiar with this method and want to explore low-frequency behaviour, you can perform a second measurement using a ground plane outdoors. The IR Windows should be set as before, but the Right Width must be much wider depending on the available space, following the reflection-time rule, allowing you to measure much lower frequencies.

What we will do next is merge the two measurements. We integrate the two curves in the range where they overlap, typically around 300–400 Hz, the transition area. The simplest and fastest method is: select the two measurements in the ALL SPL window, click Actions → Align SPL, and choose a frequency within the transition area as the alignment reference.

Once the levels of the two responses are aligned, you will select Trace Arithmetic → Merge B (the ground plane measurement) + A (the in-room gated measurement). At that point, you will choose the frequency where these two curves should cross, giving you an accurate frequency response.

Once you are comfortable with this method, you can add a measurement of the reflex port if you have decided to use that type of loading. Additionally, it would be an excellent idea to measure the speaker from different angles. But how can we, humble DIYers with limited budgets, do this? Well, for that, you will have to wait for the fourth article, which will be dedicated to construction tips.

My suggestion is to proceed step by step, so you can steadily build your skills...and don't worry, you'll still have plenty of fun building your first fully designed speaker, even without measuring what happens in the low frequencies. Listening in your own room, after careful placement and/or later measurements, will tell you if a slight adjustment to your crossover is necessary, which you can easily do afterwards.

But for now, we must stop here, dear reader. Don't turn off your computer, because we'll need it shortly to design the crossover!!!

DISCLAIMER. TNT-Audio is neither a shop, nor a HiFi company or a repair laboratory for HiFi components. We don't sell anything. It is a 100% independent magazine that neither accepts advertising from companies nor requires readers to register or pay for subscriptions. If you wish, you can support our independent reviews via a PayPal donation. After publication of reviews, the authors do not retain samples other than on long-term loan for further evaluation or comparison with later-received gear. Hence, all contents are written free of any “editorial” or “advertising” influence, and all reviews in this publication, positive or negative, reflect the independent opinions of their respective authors. TNT-Audio will publish all manufacturer responses, subject to the reviewer's right to reply in turn.

Passive Crossover in the DSP Era

A Mini-Guide for DIYers: the measurement - Part II

Basic Equipment

Methods for measuring a loudspeaker

Gated in-room measurement

Considerations