They say that the main dis-advantage of an active crossover system, is that you need to buy more amplifiers. This is true, but also a lot of fun! We like amplifiers. Amplifiers are awesome, and more amplifiers are even awesome-er, right?

With an active system, you get to choose different amps for different jobs. One stereo amp each for the subwoofer, bass, mids, highs, etc. The common assumption is that since the midrange and tweeters don’t need as nearly much power as the subwoofer, then you can use a big 2000W amp on the subwoofer, and maybe just 50W on the mids and tweets, right?

Nope. Not exactly. You don’t need the power, but you still need the voltage! Transient spikes are the reason.

Here’s a diagram of my own active-crossover system. It’s a common approach for active systems. I’ve got a Crown K2 class-D switching amp pumping 1600W to the subwoofer, an Adcom GFA-555 delivers 200WPC to the woofers, a souped-up Adcom GFA-535 sends 90WPC to the midranges, and a normal GFA-535 sends 60WPC to the tweeters. It’s the typical arrangement used in active systems, whether for home systems or PA.

Huge amp, big amp, medium amp, little amp.

As it turns out, this is misguided. A little 100W amplifier showed me why.

And by the way, I am not the first person to have this revelation. Apparently Doug Self and others have written on the topic. Send me a link if you find it.

I happened to have a spare amplifier board for a GFA-535, and a chassis with two power transformers. So, I thought it might be fun to double-up those power supplies to create a little monster of a mono-bloc amplifier. I used one transformer for the positive rail, and the other for negative. It worked great! I got 104 watts into 8 ohms, and 207 watts into 2 ohms! (From what is originally considered a 60 watt amp.) What I had created, was an amplifier with a moderate voltage swing, but a very stiff, high-current power supply that does not flinch at high current demand, which is why the power holds up so nicely at 2 ohms. (A theoretically perfect amplifier with no voltage or current sag would quadruple it’s power to 400W into 2-ohms, given an 8-ohm rating of 100W.)

So naturally, I wanted to torture the little beast to see what it could do. I hooked it to my subwoofer in place of the mighty Crown K2 was and cranked some tunes… and… uh, wow… Holy crap, this thing pounds out the bass!

So what’s up? Why isn’t it clipping? I checked it on the oscilloscope, and yep, it’s not clipping. Peak transient voltage spikes are actually less than I am seeing on the bass, mid and tweet. Hmm… I put on some bass-heavy electronic music. Wubba-wubba-wubba, and oonce-oonce-oonce, and-boots-and-pants-and-boots-and-pants-and…  Wow, this little amp cranks the hell out of that big sub!

I had assumed that since bass uses the most power, that it would also have biggest transient voltage spikes. The clip-lights on the amps were telling me otherwise. The 90WPC midrange amp clips first!

I thought a good experiment would be to analyze some actual music in Audacity, dividing it into four frequency bands, to emulate the signals my amps actually receive. That way, I could see the size of the peaks, and how close they come to maximum voltage swing. Just as interesting, I could see the RMS power required of the amplifier.

For this informal experiment I chose a Steely Dan track called ‘Lunch with Gina’ in 24/192. I feel this track has a pretty typical tonal balance for pop music, and a good dynamic range. It’s an outstanding recording. I realize other musics will have a different spectrum, but I thought this was a good use-case. Plus, it’s funky as hell.

Here’s ‘Lunch With Gina’ chopped up into four frequency bands with crossover points at 70Hz, 300Hz, and 3KHz. This is basically the same signal that goes to my amplifiers…

Surprising, isn’t it!? The subwoofer has the LOWEST peak voltage level! The bass amp isn’t peaking much either.

What this says about amplifier choice:

The odd thing is, that even though the tweeter is using far less RMS power than the subwoofer, for most music, the tweeter amp actually needs to be capable of a higher voltage swing than the sub amp. This means you need a higher-powered amplifier, because that voltage swing only comes with high-powered amps. But you’re not even using all that power, you just need it for the voltage swing! The problem with my own setup, is that the voltage rails on the GFA-535 are only 45V, while the GFA-555 is 85V.

Amplifier manufacturers! I have an idea for you…. Sell an amplifier specifically made for tweeters and mids! It needs a high-voltage power supply so it can swing enough voltage to keep up with a typical 200 WPC amplifier—80v or so—but it only needs a small, low-current power-supply, with small and fast power supply filter capacitors. It only needs 10-20 watts RMS. Give it a high slew-rate and a frequency response that rolls off below 150Hz, but extends to at least 96KHz to cover the bandwidth of a 192KHz sampling rate.  (I’m not sure if that really matters, but if we’re going to have such high sampling rates, then why not have equipment that can actually reproduce it?)

These data also indicate that it is far more important that your subwoofer amplifier provide solid, un-flinching current delivery, rather than high RMS wattage. Super high voltage rails are unneeded. High current delivery and good power supply regulation are more important. Subwoofers consume a lot of current, but not much voltage swing.

So hypothetically, let’s pretend you have a pair of two-way speakers you wish to bi-amp with an active crossover.  I’m assuming the woofer and tweeter have the same efficiency, in-room and in its enclosure.

You only have two amps in your possession, and you need to choose which amp should drive the woofer, and which for the tweeter.

Amplifier A is a typical 100WPC into 8 ohms amplifier that also happens to deliver 150W into 4 ohms, and 170W into 2ohms. It has a moderately well regulated power supply. But it doesn’t have enough juice to perfectly double its power into halved impedance, and can barely muster any more power as impedance drops to 2 ohms. Most amps are like this.

Amplifier B is only 50WPC into 8 ohms, but is a high-current design that delivers 95WPC into 4 ohms, and 180W into 2 ohms. It nearly doubles its power each time the impedance of the load is halved.

Amplifier A has “more power” at 8 ohms, so you might think it should power the woofers but it’s actually better suited for the tweeter. It does not need to have high current to drive the tweeter. It just needs high voltage to handle transients.

Amplifier B belongs with the woofer, where it’s stiff power supply will be un-bothered by the large current demands of the woofer. Clipping will be rare, and hard to perceive anyway.

Also remember that the difference between 100W and 50W is only 3db SPL, and It takes 6-10db to for a human to perceive something as subjectively “twice as loud”. That 3db is almost nothing.

Choosing your amps:

Of course, everyone’s system is different, and far more complex than the hypothetical above. There are many factors to consider when sizing amplifiers for an active system.

• Efficiency of the drivers in their enclosures, and from actual listening position.
• Bandwidth of each crossover signal, and any equalization.
• Type of music
• Desired SPL

My system has a relatively efficient subwoofer, due to it’s large 14cuFt enclosure, and so a high-powered sub amp is apparently not required. But most modern subwoofers need more power than mine because they use small enclosures. The small enclosure causes the natural response to drop off in the bass, so this is counteracted by the use of a bass-boosting equalizer circuit to get flat response down in the 20Hz range. Many compact subs use 12db or even more boost. That’s 16 times the power requirement! You really do need a huge amp in that case. Perhaps this has contributed to the perception that bass amplifiers should always be larger. In many cases they should be.

Sizing the amps: Basically, we’re trying to prevent one band from clipping before the others, so we’re trying to get everything to clip at around the same point. You’ll need to know the relative peak signal levels of all the frequency bands of your active system. You can simulate the signals your amps will receive using a sound editor like Audacity. Open some music files that represent the kinds of recordings you’ll be playing over the system. Using the low-pass and high-pass filters, and equalizers, you can simulate your crossover’s behavior pretty easily. Observe the peaks, and the relative decibels between the different frequency bands, and do the math to determine what kind of voltage swings you’ll need for your amps. You’ll need to consider a number of factors including each driver’s sensitivity in their enclosures, and in the room. Perhaps even better, and easier, is just doing it experimentally. Just hook up your amps without speakers connected, play some music and watch their clipping behavior on the o’scope. If any one of your amps is clipping before the others, maybe it needs to be replaced with something higher voltage. The story may be a little different with speakers actually attached, as the amps will clip sooner with loads attached.

 

• Thanks to the folks at DIYAUDIO.COM, for helping me flesh out these ideas on this discussion thread.

UPDATE: 09-02-2015

I’ve tried this with a bunch of recordings now, and get pretty much the same result with most everything. Bass peaks are usually lower than mid and treble peaks with most recordings.

But here’s the most extreme example of bass voltage swing that I could find; Dieter Moebius – WOMIT.

As you can see, the bass peaks reach maximum, but so do the mids. So, even with bassy music, you’ll still need main amps that can swing the same voltage as the subwoofer amps.

Update: 09-03-2015

I am seriously considering changing my amp configuration now. Ideally, I’d have three GFA-555’s for bass, mid and treble, so they can all swing the same voltage. That would be expensive! And I don’t actually need 200W on my tweeters, I just need that 80V swing.

I’m thinking a better idea is to modify my GFA-535’s to run at 80V rails, but with a lower current power supply. There may be some slight circuit changes involved, and selection of higher-voltage parts… hmm…

Update: 08-25-2017:

Actually, I’m going to build myself a custom 4-channel amp based on the GFA-565.