TECHNICAL NOTES
Crossovers, VUs and Kill Switches
Whilst answering some questions from valued customers, I ran some simulations to provide graphical information to help describe the different phenomena I was trying to describe. As is often the case, I went off on a tangent and ended up simulating a lot of things that really don’t get discussed that much and for which there is limited material available on line. So I thought I would share this material in the hope that it may help some people.
Crossovers and Kill Switches
A question that is asked occasionally is that some customers note that when running a multi-way set up, then it is possible to hear a small amount of output from one section on an adjacent section. Rest assured it’s not “leakage” or something going wrong. It is a perfectly normal attribute of crossovers, which split and feed the signals to the separate outputs. So firstly, you need to understand some crossover basics. Some people mistakenly think that the crossover point in a multi-way sound system is like a cliff edge; it is not. All crossovers have what is called a roll-off or slope characteristic. This is a db-frequency relationship that is more simply understood schematically in figure 1. The plots were produced in Spice and are validated simulations of actual crossover networks, so are a very accurate description of what is going on.
Figure 1. Sub, bass, mid and full range outputs in a 24dB/oct Linkwitz-Riley crossover network
Figure 1 shows an example of the sub, bass and mid outputs, with 85 and 466Hz crossover points. The frequency of the of the crossover points doesn’t really matter for what’s being discussed here. You can see that the sub frequency begins to roll off as it goes up and approaches the 85Hz crossover frequency. Similarly, you can see the kick bass frequency output increasing at the same time. At the crossover frequency, the output of both the sub and kick bass sections are equal. In this example, both sub and mid are at -6dB at the crossover frequency and both have a slope characteristic of 24dB/octave. This is a characteristic of a particular Linkwitz-Riley crossover filter design, but there are many more to choose from. You see exactly the same phenomena at the kick bass-mid crossover point (466Hz) and, if all these signals are then electronically added, very close to a flat response can be achieved across the full frequency range.
So now you have got your head around that, take note of the hatched area in Figure 2. This is the roll off from the sub signal and you can see it is extending into the kick bass frequency range. It is not leakage. It is a characteristic of a well designed crossover network. You can do a simple test to hear it on your preamp if you wish. Play some music and then kill everything except the sub using the kill switches. Then switch off the Sub, mid, horn and tweeter (or all tops anyway) power amps, just leaving the kick bass amp on. You should still be able to hear some music at a low level and skewed towards the lower frequencies. This is the hatched frequency content in Figure 2. It is the roll off from the sub section extending into the bass section, at a progressively reducing level and exactly as designed. You can observe exactly the same thing on the other sections by performing a similar test.
One might reasonably ask “but if I hit a kick bass kill for instance, why can’t that shut off the output from that section completely, sub roll off as well”. It is actually possible to do that if the kill switches are installed directly at the output of the preamp section. Dedicated crossovers with mute switches for instance do just that. But reggae preamps have different requirements and have additional functions such as mic and aux channels. We certainly do not want the kills to operate on these sections, just on the music, so we are forced to install the kill switches upstream in the preamp signal chain. Hence a “cliff edge” total kill is simply not possible with these additional functions.
A further note on crossover frequencies
A lot of people think that when you add all the sections together from a crossover (that’s using the same kind of filter throughout), you get a nice flat response. You don’t. The narrower the bandwidth, the lower the section output gets. As a rule of thumb, anything lower than an octave bandwidth is noticeable. So all you speaker builders who recommend running kick bass from something like 85-150Hz, take note. Don’t believe me? Then take a look at Figure 3 illustrating the progressive narrowing of a kick bass section bandwidth and the impact this has on section and full range output.
Figure 2. Extension of outputs into adjacent sections
Figure 3. Impact of narrow bandwidth on output
So why does this happen? Well, when you narrow the bandwidth of a section, the roll offs from the sections both above and below your section of interest begin overlapping at a level that has sufficient energy to have an impact. Combine that with the different phase characteristics of these different sections and you can see clearly what happens. If you have a variable crossover or DSP, try it for yourself.
VU meter operation
Whilst we are on the subject of crossovers and kill switches, it’s also useful to write something about VU meters because the subjects have some common ground. I’m aware that there are a lot of ways that people build these things into circuits and also a lot of ways that people use them. I think I have tried pretty much all options and, through many years of experience (as well as the destruction of a lot of speakers), I choose to set them up one particular way and there are some very good reasons for that, which I will try to illustrate. The first thing is to appreciate what these meters can show.
It is possible to build VU circuits for them to show a variety of things, including peak levels, average levels across frequencies, average levels on a rolling time basis and a whole bunch of other stuff. My preference is instantaneous peak level at any frequency at all times, and that measurement is made across the entire bandwidth of the crossover section. The reason I prefer this is because I run a reggae sound system and we are a particular breed who often push our sound systems to extremes. Therefore, for me at least, it is essential to know exactly how hard every part of the system is working instantaneously.
However, setting up this way does not suit everybody and some have other expectations for their VU meters. For instance (this is a common one), some people expect VU meters to drop to minus infinity (i.e. shut down) when the kill switch for that section is thrown. This is a risky type of set up in certain situations and I will explain why.
If you have read and understood the previous section on crossovers fully, you will understand the existence of the natural roll off that extends into adjacent sections. So, if I shut down a section with a kill switch, there is still some roll off signal from adjacent sections available. So, it is quite right to expect the VU to be measuring something, albeit at a lower level of course. Some of you may ask “well what is the point of that? You have shut the section down; it’s a running lower level; why do you need to monitor it?” Well here is why and believe me, I have seen plenty of sound systems destroy their kick sections because of this and because they had VU meters that stopped monitoring a section when it was shut down with a kill switch. Picture the following scenario……….
Figure 4. Transition from full range to sub only
The sound system is running. It doesn’t matter what the initial EQ is, but I’m going to illustrate with a flat EQ simply because it is easier. Then the operator kills everything except the sub on the music and everyone in the dancehall goes wild! So we go from scenario A to B in Figure 4. The preamp being used does not monitor the kick bass section in this instance, but you can see from the illustration, it’s 6dB down (the natural crossover roll off), so what’s the problem? Nothing will get damaged and the operator is watching the sub VU. It’s all about the sub and the crowd love it! So now the operator starts winding on a bit more EQ on the sub.
Figure 5. Increased signal levels on adjacent sections due to EQ increase
This is shown in Figure 5 A and this simulates a semi parametric EQ circuit being wound on towards the top end of the sub frequency. You can probably see where this is heading 😊. The sub is running big power hungry speaker drivers and the operator knows their limits, so they are comfortable with the hot signal. But what they don’t realise is that the kick bass amps are also being fed with an increased signal level too, even though the section is killed. A signal higher than it was prior to running sub-only! This is simply the natural crossover roll off from a heavily EQ’d peak which will clearly extend further into the adjacent section. But the operator doesn’t know any of this because they wanted their VU meters to shut off when the section was muted! The kick bass speakers are less resilient and were already close to their power limit. Unless there were limiters installed, the kick section risks being blown.
VU meters are there to monitor levels at all times. They are not “kill indicators”. This is why S&P VUs are always running and always monitoring peak signal levels. So even if you kill a section, expect to see some VU activity.
No doubt, someone will say this is why limiters, etc. are needed. Sure, but listening to compressed music all night is not an enjoyable experience (for me anyway) and there are an awful lot of people who: i) don’t use them because they prefer a realistic sound; and ii) experience difficulty setting them up such that they do not provide the necessary protection. You have been warned!