Over the years I've built a few speakers, ever nicer, always learning. Now, I want to build a very high-quality 3-way tower speaker, with a matching center channel (using the mid/high section of the tower). Madisound no longer designs 3-way crossovers, and my first and only attempt at designing a passive crossover modeled horribly... Electronic design is not my specialty, though I'm good at assembling existing designs on existing boards. I've since decided that, if I'm going to make this speaker really epic, I should use an active crossover and tri-amp it. And why not build those amplifiers!
This would be my first foray into high-power amplification (I've assembled many Seventh Circle Audio microphone preamp kits), so I want to run my ideas past some experienced DIYers.
I figured on using the ESP "P3A" amplifiers. There will be one mid, one tweeter, and two 8" woofers. All loads are 4 ohms. The woofers could be run in parallel, or I could drive them separately (the amp seems inexpensive enough, adding an extra pair of channels doesn't seem like a big deal) -- the woofer is available as an 8 ohm or 4 ohm version. So, in one chassis I would have 4x P3A stereo boards. I figured on the Dissipante 4U chassis that's 400mm deep; each of the four heatsinks should be good for a stereo pair of those amps.
I'd pondered using the P101 MOSFET amp for the woofers as it's more powerful, but I don't want to try cramming two separate PSUs into one chassis (unless that's easier than it seems), and I feel like 100 watts per woofer should be plenty in a tri-amped 3-way.
I'd have Rod's 15v power supply and P9 (?) Linkwitz crossovers in the chassis as well. I would use a soft start, Rod's or DIYAudioStore's. The P5 PSU would have its own, small, always-on transformer. The amp would power on with a 12v trigger from my Emotiva XMC-1 pre/pro (I understand the PCB for Rod's soft start circuit has provisions for this on-board that aren't shown in the schematic).
I figure the stereo towers would have one chassis, and the center channel would have its own, smaller chassis, unless there is an easy way to fit all three channels into one chassis. The amps would connect to the speakers via Neutrik Speakon NL8 connectors.
Am I on the right track, here? Doable?
This would be my first foray into high-power amplification (I've assembled many Seventh Circle Audio microphone preamp kits), so I want to run my ideas past some experienced DIYers.
I figured on using the ESP "P3A" amplifiers. There will be one mid, one tweeter, and two 8" woofers. All loads are 4 ohms. The woofers could be run in parallel, or I could drive them separately (the amp seems inexpensive enough, adding an extra pair of channels doesn't seem like a big deal) -- the woofer is available as an 8 ohm or 4 ohm version. So, in one chassis I would have 4x P3A stereo boards. I figured on the Dissipante 4U chassis that's 400mm deep; each of the four heatsinks should be good for a stereo pair of those amps.
I'd pondered using the P101 MOSFET amp for the woofers as it's more powerful, but I don't want to try cramming two separate PSUs into one chassis (unless that's easier than it seems), and I feel like 100 watts per woofer should be plenty in a tri-amped 3-way.
I'd have Rod's 15v power supply and P9 (?) Linkwitz crossovers in the chassis as well. I would use a soft start, Rod's or DIYAudioStore's. The P5 PSU would have its own, small, always-on transformer. The amp would power on with a 12v trigger from my Emotiva XMC-1 pre/pro (I understand the PCB for Rod's soft start circuit has provisions for this on-board that aren't shown in the schematic).
I figure the stereo towers would have one chassis, and the center channel would have its own, smaller chassis, unless there is an easy way to fit all three channels into one chassis. The amps would connect to the speakers via Neutrik Speakon NL8 connectors.
Am I on the right track, here? Doable?
I recommend one four channel amplifier and dedicate that to one 3 way speaker. A common PSU for all four amplifiers is a worthwhile simplification.
Use very short speaker leads and a long interconnect from the pre-amp/source.
4off 100W into 4ohms amplifiers driving a 3way can be very loud.
If all the drivers have very similar sensitivities, then you won't need much adjustment to balance the sound from the speaker.
With a 3 way you only need each driver to provide a pass-band of three to four octaves.
If you select drivers that have a smooth delivery at least one octave beyond your intended crossover frequencies, then that will greatly simplify the electronic (active) crossover filters.
Use very short speaker leads and a long interconnect from the pre-amp/source.
4off 100W into 4ohms amplifiers driving a 3way can be very loud.
If all the drivers have very similar sensitivities, then you won't need much adjustment to balance the sound from the speaker.
With a 3 way you only need each driver to provide a pass-band of three to four octaves.
If you select drivers that have a smooth delivery at least one octave beyond your intended crossover frequencies, then that will greatly simplify the electronic (active) crossover filters.
Last edited:
I have a very similar system.
I have a diy 3 way and of each way one diy amp.
P3A for high and mid and a P101 for low.
I'd go with one stereo amp per chassis. it might be more expensive but is easier to debug and upgrade. also go with a dsp active crossover. and get an
usb microphone to use with REW. room acoustics are very important.
Thanks
I have a diy 3 way and of each way one diy amp.
P3A for high and mid and a P101 for low.
I'd go with one stereo amp per chassis. it might be more expensive but is easier to debug and upgrade. also go with a dsp active crossover. and get an
usb microphone to use with REW. room acoustics are very important.
Thanks
Thanks for the responses! I don't feel like I need a DSP crossover, as my XMC-1 preamp/processor has built-in Dirac room correction.
The drivers I'm planning to use are all flat well within their intended passbands. I'm using Scan-Speak Revelator 8" woofers (flat to 600 Hz with a small bump up in response at that point), crossed at 2-300 Hz, a Morel Titanium Supreme midrange unit (flat from 73 Hz to 8 kHz!), and a Morel Titanium Supreme tweeter crossed at about 3kHz. The driver sensitivities are all around 90 dB.
brunogiovs, did you put the P101s and P3As in separate chassis? One chassis for stereo P101s, and one chassis for (2x?) stereo P3As?
The drivers I'm planning to use are all flat well within their intended passbands. I'm using Scan-Speak Revelator 8" woofers (flat to 600 Hz with a small bump up in response at that point), crossed at 2-300 Hz, a Morel Titanium Supreme midrange unit (flat from 73 Hz to 8 kHz!), and a Morel Titanium Supreme tweeter crossed at about 3kHz. The driver sensitivities are all around 90 dB.
brunogiovs, did you put the P101s and P3As in separate chassis? One chassis for stereo P101s, and one chassis for (2x?) stereo P3As?
you have 4 drivers (2 bass, 1 mid, 1 treble) in one speaker box.
You need 3, or 4 channels of amplification to drive that speaker box.
Put all three/four channels in one chassis.
You can locate the electronic crossover inside that chassis, or in a separate chassis.
These amplifiers and crossover would best be located at the one place right behind, or even attached to the speaker box.
Repeat that for the other stereo channel.
You need 3, or 4 channels of amplification to drive that speaker box.
Put all three/four channels in one chassis.
You can locate the electronic crossover inside that chassis, or in a separate chassis.
These amplifiers and crossover would best be located at the one place right behind, or even attached to the speaker box.
Repeat that for the other stereo channel.
I think it's worthwhile doing very different amps for mid & treble to what you need for bass.
If you look at a spectrum of most any music, almost all the power is concentrated in the first few hundred Hz.
For example:
https://www.researchgate.net/profil...spectral-density-of-a-typical-'Pop'-music.png
So if you need (say)100W to cover the woofer, you're only going to need (say) 10W for the mid, and you'll be fine with 3W for the tweeter, depending of course on where you cross from one to the next.
So use amps that maximise the performance where you're using them - little fast transistors for the tweeter, big slow transistors for the woofer.
An added bonus is that you can use the same power supply that supplies the mid and tweeter to run the signal processing circuitry that makes it all work.
One of the reasons I've been mad keen on designing poxy (but fast) 5W amps of late is that using 50W for a tweeter is slightly nuts, as is regulating the +/-40V supply that I run the amps on down to +/-15V for the opamps.
If you look at a spectrum of most any music, almost all the power is concentrated in the first few hundred Hz.
For example:
https://www.researchgate.net/profil...spectral-density-of-a-typical-'Pop'-music.png
So if you need (say)100W to cover the woofer, you're only going to need (say) 10W for the mid, and you'll be fine with 3W for the tweeter, depending of course on where you cross from one to the next.
So use amps that maximise the performance where you're using them - little fast transistors for the tweeter, big slow transistors for the woofer.
An added bonus is that you can use the same power supply that supplies the mid and tweeter to run the signal processing circuitry that makes it all work.
One of the reasons I've been mad keen on designing poxy (but fast) 5W amps of late is that using 50W for a tweeter is slightly nuts, as is regulating the +/-40V supply that I run the amps on down to +/-15V for the opamps.
Last edited:
only when you consider average energy.
This applies when using drivers of different sensitivity.
If you can find data showing the peaks of transients in the range of octaves covering the audio band, then I think you will also find that the peak levels across all the octaves is very similar. Except the top octave may be slightly lower in peak level.
When using drivers of the same sensitivity you are advised to use amplifiers in each of the Bass/Mid/Treble bands that have similar maximum output voltage peaks.
If you choose drivers with significantly different sensitivities, then scale the voltage/power levels to suit those driver sensitivities.
Last edited:
Not in my experience. I've listened to many days of music with my spec-an plugged in, on peak detect. I get exactly the same spectrogram.
Look at it from a different perspective. A typical tweeter will burn out if subjected to more than (say) a watt for a very short period. If all the amps in your active system are capable of driving the power levels needed for your woofer, then there's nothing to stop a malicious song from encoding some full-volume 22KHz tones, which you won't hear, to destroy your tweeters.
Or even just poorly written code on your player.
Wouldn't you prefer to protect your speakers against damage through malicious software or software that's lost the plot? Using amps that are appropriate for the drivers they're driving does this.
A 3-way crossover is almost always two 2-way crossovers.
Split the bass from the mids, say 500Hz. The 500-up side then splits mids/highs, say 3KHz.
While this sounds good in theory, in practice I don't want to deal with all those different amp types and power supplies. Keeping it all the same seems like it'd be easier -- construction, testing, etc. Also cheaper, for fewer enclosures.
Malicious full-volume 22 kHz tones could also wreck a speaker with a passive crossover. I suppose a small, slow-blow fuse could help protect the tweeters from damage; or a tungsten lightbulb. Some of JBL's commercial speakers had a light bulb in-line with the HF driver; basically causing power compression when the power levels got too high. Not sure if they still do that.
I agree with Andrew here, and I do have the data. I did an investigation and wrote an article on the topic...
Which amplifiers for an active-crossover speaker system? Some surprising observations… | Hoppe's Brain
I found that transient spikes in music are actually largest in the mid and high-frequency regions, despite the very low average power.
So your tweeter or midrange amp actually needs to swing at least as much voltage as your bass amp. I'm assuming equal driver sensitivities, so adjust accordingly for your system.
Amps with high voltage rails typically also happen to be high-powered. The power is wasted on a tweeter or mid. Instead, one could also use an amp with high voltage rails, but low average current. (Tubes?)
Which amplifiers for an active-crossover speaker system? Some surprising observations… | Hoppe's Brain
I found that transient spikes in music are actually largest in the mid and high-frequency regions, despite the very low average power.
So your tweeter or midrange amp actually needs to swing at least as much voltage as your bass amp. I'm assuming equal driver sensitivities, so adjust accordingly for your system.
Amps with high voltage rails typically also happen to be high-powered. The power is wasted on a tweeter or mid. Instead, one could also use an amp with high voltage rails, but low average current. (Tubes?)
Last edited:
I've just gone through a bunch of my favourite tunes (FLAC copies from CD source) with Audacity, and didn't find a single example where the peaks above 500Hz were above the peaks below 500Hz (24dB/octave filters).
I was trying to be representative of the different styles of music I listen to, so looked at various very different tracks:
Elton John: Benny and the Jets.
Aussie Crawl: Reckless (don't be so).
Alanis Morissette: Hand in my Pocket.
Simon and Garfunkel: American Tune (live).
Mark Knopfler: Going Home.
That said, I have overstated the amount of energy in the bass section - with the exception of live music (nothing at all after 1KHz - no wonder it's good to listen to in the car), the difference between woofer and mid is only about 6dB, not the 10dB I was assuming.
Plots attached.
I was trying to be representative of the different styles of music I listen to, so looked at various very different tracks:
Elton John: Benny and the Jets.
Aussie Crawl: Reckless (don't be so).
Alanis Morissette: Hand in my Pocket.
Simon and Garfunkel: American Tune (live).
Mark Knopfler: Going Home.
That said, I have overstated the amount of energy in the bass section - with the exception of live music (nothing at all after 1KHz - no wonder it's good to listen to in the car), the difference between woofer and mid is only about 6dB, not the 10dB I was assuming.
Plots attached.
Attachments
Interestingly, the tracks that I've found where the 500Hz high pass has absolute peaks close to or equal to the low pass (a good example being my recording of Reckless) are badly mastered to the point of clipping at the source.
My Knopfler stuff appears to be some of the best mastered source material I have.
My Knopfler stuff appears to be some of the best mastered source material I have.
Attachments
Ah, here's my domain! :-D I'm a recording engineer by hobby. It may look clipped in your screenshot, but it's probably not (unless you can actually hear the clipping distortion). It's hit hard with a limiter; all the peaks are basically the same volume, which gives it the appearance of being boosted past clipping.
Something interesting with regard to high power on HF... A commercial audio speaker (like you might find at a concert) has a horn-loaded HF driver that has a power rating of 60-100 watts (lower-end versions of these may only be about 15 watts). Seems like not a lot of power, when the woofers can take 500 watts or more... BUT, the horns are extremely efficient, producing often 110 dB or more 1w/1m. So, what amp do these sound contractors run on the horns? Something small and low-powered? Sometimes. But, they often find better value in just using the same amp for everything... Horns, mids, subs. If one fails, just swap in another. They DO usually turn the gain knobs down on the HF amp, and they use sophisticated DSP crossovers with limiters to prevent damage to the HF drivers.
Something interesting with regard to high power on HF... A commercial audio speaker (like you might find at a concert) has a horn-loaded HF driver that has a power rating of 60-100 watts (lower-end versions of these may only be about 15 watts). Seems like not a lot of power, when the woofers can take 500 watts or more... BUT, the horns are extremely efficient, producing often 110 dB or more 1w/1m. So, what amp do these sound contractors run on the horns? Something small and low-powered? Sometimes. But, they often find better value in just using the same amp for everything... Horns, mids, subs. If one fails, just swap in another. They DO usually turn the gain knobs down on the HF amp, and they use sophisticated DSP crossovers with limiters to prevent damage to the HF drivers.
If you have an amp that can crank high voltage, that's great... But it also needs to be able to deliver the current required to drive that voltage into the driver's impedance at that frequency. If it doesn't, your voltage will sag or, worse, it will clip.
Conversely, driving high current into the woofer or subwoofer requires a certain voltage driven into that driver's impedance.
What you're describing is peak power capacity, versus RMS power capacity. The peaks in HF occur faster (shorter wavelength), and so the instantaneous power demands are higher. So, yes, your ideal HF amp would have lots of small caps for quick, instantaneous-power delivery, and would have a high voltage slew rate.
You also have to consider the material you're analyzing. The Tron: Legacy movie, for example, has some very huge LFE transients. The scene where Sam Flynn first finds himself on the grid bottoms out my sub with some extreme subsonic thuds. Steely Dan, while being good music, isn't bass music. Giving that track a listen, I'm not surprised by the high peaks in the mids-highs. The snare drum is very prominent, and the kick drum has a pretty strong 3-4k boost. There's hardly any bass in this track at all, which explains the weak signals below 70 Hz. The snare also doesn't sound very compressed, so those peaks are intact.
The Morel tweeter I'm planning to use has a 10ms peak power rating of 1,000 (!) watts, 150 watts RMS, at 4 ohms. The midrange has a similar power rating; woofers are 170w RMS, no peak rating given.
Let me know if I've misunderstood any of this.
10ms for the transient peak (instantaneous) rating is quite a long duration.
A 10kHz wave takes 0.1ms for the full wave, or 50us for a half wave.
The near peak voltage/current takes very approximately 25us of each half wave transient.
A string of 25us peaks would need ~ 400 of them to reach the 10ms rating. ( the whole string @ 50% duty cycle would take 20ms to pass through the voice coil).
I wonder what the 1ms rating might be? Or even a single shot 0.1ms transient rating?
Well... When you consider the midrange unit has the same 10ms power rating, but a 3" voice coil (compared with the 1" tweeter voice coil), it makes me wonder if that's a real power rating, or something they made up.
A "100 watt RMS" rating on a hi-if tweeter is often "100 watts noise power", filtered at some frequency/slope. The average power reaching the tweeter is low. A 1 inch VC can handle about 6 watts of average power. One with a sophisticated motor (i.e. Scan Speak, Morel, etc) can take about 10.
And that difference between 6 and 10 watts is huge when it comes to how loud it will play before the smoke comes out.
And that difference between 6 and 10 watts is huge when it comes to how loud it will play before the smoke comes out.
OK, so to get back to the original topic... Would an acceptable amp arrangement for these speakers be all P3As? From all the other posts in this forum about the amp, consensus seems to be that it's very good, but maybe slightly harsh on top (with others saying it's not harsh on top, the listener was just used to a duller-sounding top end).
100w to tweeter, 100w to midrange, and... Either 100w to both woofers (2x 8 ohm version in parallel), or 100w each woofer (2x 4 ohm version driven independently).
Andrew T suggested building a separate amplifier enclosure for each speaker so the speaker wire runs are as short as possible. I may or may not do that, as it'd be handier to have one central enclosure to run power and trigger control to, despite the advantages of short speaker leads. The center channel speaker would have its own enclosure, regardless (probably).
On trigger control, has anyone used the ESP soft start? Rod mentioned in passing in the write-up on it that it has provisions for a 12v trigger input. Does it? Is it buffered, or is the relay driven directly off the trigger?
100w to tweeter, 100w to midrange, and... Either 100w to both woofers (2x 8 ohm version in parallel), or 100w each woofer (2x 4 ohm version driven independently).
Andrew T suggested building a separate amplifier enclosure for each speaker so the speaker wire runs are as short as possible. I may or may not do that, as it'd be handier to have one central enclosure to run power and trigger control to, despite the advantages of short speaker leads. The center channel speaker would have its own enclosure, regardless (probably).
On trigger control, has anyone used the ESP soft start? Rod mentioned in passing in the write-up on it that it has provisions for a 12v trigger input. Does it? Is it buffered, or is the relay driven directly off the trigger?
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.