Magazine Archive

Home -> Magazines -> Issues -> Articles in this issue -> View

Hi-Fi



One subject in Hi-Fi that has been shrouded — I believe deliberately — in mystery is crossover design. Every speaker construction project that I have read glosses over many of the finer design points. In an attempt to show that crossovers are well within the design capability of most enthusiasts I will explain how it's done.

The first thing needed when designing a multiway speaker system is the crossover frequency. Most woofers are not renowned for their treble response and tend to roll off at a few kHz. Tweeters, on the other hand, have a 'bass' resonance, usually around 1kHz. Naturally they cannot be used below this frequency. When choosing units the frequency ranges handled must overlap preferably by at least an octave. Having established this crossover frequency, it should be placed in the centre of the overlap. For example let's assume we have a woofer that will respond to 4kHz and a tweeter that goes down to 2.5kHz.

Subtracting 2.5kHz from 4kHz gives us 1.5kHz. Half of this is 750Hz so our crossover will be at 2.5kHz + 750Hz = 3.25kHz.

It is now necessary to measure the impedances of the drive units at 3.25kHz. To do this you will need a signal generator, a 100R resistor and an oscilloscope. The resistor is wired in series with the speaker and a sine wave at the crossover frequency, 3.25kHz, fed to it. The impedance of the speaker is then found from the relationship, Z=100 Vd / Vo-Vd Where Z is the impedance in Ohms, Vd is the RMS voltage across the speaker and Vo the RMS output voltage fed into the network.

When the impedances of the speakers have been determined, they should be mounted in the cabinet and the relative efficiencies measured. This is achieved by measuring the output of each speaker in isolation, when fed from the signal generator. The procedure is as follows:

Take the speaker cabinet outside and lay it on the ground with the drive units pointing upwards. Position a microphone about a foot away and between the two drive units. Feed a sine wave at the crossover frequency into the woofer and adjust the output to a convenient level observing the mic output on an oscilloscope. Feed the sine wave into the tweeter and adjust the output until the 'scope shows the same level as the woofer. By noting the inputs required for each unit the relative efficiencies can be found. Usually the tweeter will be much more efficient than the woofer. The simplest way to get the efficiencies equal is to put a resistor in series with the tweeter. It's best to use a 22R pot in series and adjust it, monitoring the tweeter's output with the aid of the mic and 'scope. Measure the resistance setting and use the nearest value component in series with the tweeter in the crossover network.

Remember though that this will now modify the impedance of the tweeter and this must now be remeasured as before with the resistance in series. Crossovers have various electrical slopes dependent on their complexity. Generally, as one might expect, the higher the slope the more difficult the design equations become. The nearest thing to a universal crossover is the second order type and I will concentrate on this type. Figure 1 details the basic circuit and the equations for determining the component values.

Figure 1. Basic crossover circuit and equations for determining component values.


That's all there is to it really, although you have to be careful when selecting components. Electrolytics are out. Apart from their wide tolerance they tend to explode when subjected to AC! Paper types are ideal although polyester types are probably easier to obtain. Awkward values can be formed by wiring standard values in series/parallel. Chokes can either be air cored, or wound on low hysteresis iron cores. Again, non-standard values can be produced by combining standard values.

Hopefully I will be able to come back to this subject at greater length at some future date.

I read with some interest Ben Duncan's comments on amplifier sound in the last issue of E&MM. And while I would agree with most of what he said, I would take issue with the impression left in the reader's mind that all transistor amplifiers sound the same. Transistor amplifiers do tend to sound ropey when overloaded but there is a solution. NAD in their excellent 3020 power amplifier have incorporated soft limiting. This effectively reduces the gain of the circuit as the overload point is approached. The result is that instead of overloading and producing large amounts of odd harmonic distortion, the circuit sounds like a valve amplifier in overload.

The other major difference between valve and transistor amplifiers is their modus operandi. Valve amplifiers almost invariably work in class A whilst transistor types tend to work in AB. If you listen to a Class A transistor amplifier it is very difficult to detect the difference between it and a valve type. Unbelievers should go to their nearest stockist and listen to the Electrocompanient amplifier. Although this only produces about 20W RMS its sound is very valve like. It operates entirely in Class A. The reason for the different sound is not hard to find. Crossover distortion is the major drawback of Class AB amplifiers. Overall feedback will reduce the level to theoretically imperceptable levels. Trouble is that crossover is a spiky waveform and its peaks are of much higher amplitude than the RMS level would indicate.

Probably, the advent of VFET output stages will render the above problem of academic interest only. I must admit though that the sliding bias amplifiers which the Japanese have foisted upon us this Hi-Fi season don't sound at all like Class A to me.

Next month I hope to discuss the ins and outs of active speaker systems.

To whet your appetites a little though, it may be as well to note briefly the advantages of the active approach over the passive one. As you will have realised having read this far, the design of a conventional crossover network is critically dependent on the impedances and relative efficiencies of the drive units used.

Further, these impedances are not simple pure resistances so there is always the risk of unwanted interaction between the various complex (reactive) parts of the networks. They absorb amplifier power, an insertion loss of 6db or more being commonplace. Last but by no means least, they prevent full advantage being taken of the damping factor available at the amplifier output.

Active crossovers suffer from none of these problems. The slope and crossover frequency are independent of the speaker impedance. The relative efficiency of the drive units can be equalised simply by means of a balance pot between the power amplifiers used. They are cheap, no costly inductors are used. A textbook response is automatically obtained assuming standard equations are used. In short they offer the ultimate performance possible at the present state of the art.



Previous Article in this issue

Organ Talk

Next article in this issue

Working with Video


Electronics & Music Maker - Copyright: Music Maker Publications (UK), Future Publishing.

 

Electronics & Music Maker - Sep 1981

Feature by Jeff Macaulay

Previous article in this issue:

> Organ Talk

Next article in this issue:

> Working with Video


Help Support The Things You Love

mu:zines is the result of thousands of hours of effort, and will require many thousands more going forward to reach our goals of getting all this content online.

If you value this resource, you can support this project - it really helps!

Donations for May 2021
Issues donated this month: 0

New issues that have been donated or scanned for us this month.

Funds donated this month: £21.00

All donations and support are gratefully appreciated - thank you.

Please Contribute to mu:zines by supplying magazines, scanning or donating funds. Thanks!

Monetary donations go towards site running costs, and the occasional coffee for me if there's anything left over!
muzines_logo_02

Small Print

Terms of usePrivacy