Lately, I've been ruminating on the ideal qualities I would want in a tweeter amplifier for my 4-way 24db L-R quad-amplified system. (Thanks, Rod Elliot!) It had occurred to me that the tweeter amp will be spending an awful lot of time hanging around at low voltage, in the bias zone. They will be in class-A push-pull mode there, but frequently switching to either side. That sounds like a good thing, but isn't this also where class AB amps have the most distortion?
So, I thought perhaps I should go Class-A on the tweeters. I could build an Aleph-J! And I thought that I could just build a low-power version, to avoid a lot of the excess heat generation. 18V rails should be plenty, right?
So I wanted to find out. I connected my oscilloscope to the output of my tweeter amplifier (an Adcom GFA-535), and cranked the system up to VERY EFFING LOUD with some techno music. (Yes, this is a system performance requirement.)
I saw peaks on the scope of 24 volts or so! This is similar to the peaks I saw going to the mid-range driver. I was surprised.
So, it would seem that a Class-A amp with 25V rails is actually the minimum for this situation. But now, I'm starting to think that this Adcom is actually quite well suited. It has no trouble at all with the voltage swing, and I am surprised that it does actually get to stretch it's legs. However, at lower listening levels, the scope confirms that there is a lot of hanging around the push-pull zone. I know it's Class-A there, but I know there are also issues associated with that topology. Though I don't remember why, there are those who poo-pooh class-A push-pulls.
So, which is the greater sin? An occasional clipping at high volume, or lots of push-pull class-A?
So, I thought perhaps I should go Class-A on the tweeters. I could build an Aleph-J! And I thought that I could just build a low-power version, to avoid a lot of the excess heat generation. 18V rails should be plenty, right?
So I wanted to find out. I connected my oscilloscope to the output of my tweeter amplifier (an Adcom GFA-535), and cranked the system up to VERY EFFING LOUD with some techno music. (Yes, this is a system performance requirement.)
I saw peaks on the scope of 24 volts or so! This is similar to the peaks I saw going to the mid-range driver. I was surprised.
So, it would seem that a Class-A amp with 25V rails is actually the minimum for this situation. But now, I'm starting to think that this Adcom is actually quite well suited. It has no trouble at all with the voltage swing, and I am surprised that it does actually get to stretch it's legs. However, at lower listening levels, the scope confirms that there is a lot of hanging around the push-pull zone. I know it's Class-A there, but I know there are also issues associated with that topology. Though I don't remember why, there are those who poo-pooh class-A push-pulls.
So, which is the greater sin? An occasional clipping at high volume, or lots of push-pull class-A?
Last edited:
Hi,
FWIW a 4 way system with 4th order L/R crossovers is a PA
system at best and will be wrong for a proper hifi system.
Class A is best applied to the broad midrange, not the treble.
rgds, sreten.
FWIW a 4 way system with 4th order L/R crossovers is a PA
system at best and will be wrong for a proper hifi system.
Class A is best applied to the broad midrange, not the treble.
rgds, sreten.
I recall some discussion of power distribution in audio signals in threads here or on another DIY speaker forum. There have been studies (I can think of some data presented in Active Crossovers by Douglas Self) that show that the RMS power falls off above about 1kHz IIRC around 6dB/oct. But the peak power doesn't fall off. This means that your tweeter amplifier will on average be producing less heat as the RMS output will be lower, however, it still needs to have almost the same peak power (same rail voltage) as the amp for the midrange band. In this case, I think that your class-A idea is not a good approach. I hope that is more like the type of info you were interested in...
Thanks Charlie. That is interesting, and jibes with my casual experiment.
It really underscores the power advantage of electronic crossovers! That 24v peak would hit the rails if it were riding along with the bass.
It really underscores the power advantage of electronic crossovers! That 24v peak would hit the rails if it were riding along with the bass.
tweeter transient peak power.
I am not surprised.
I have promoted the design requirement that the peak transient SPL from the tweeter should be just as high as the mid range and bass range speakers.
Sreten, among many others, have rubbished this design philosophy repeatedly.
Average powers have little if anything to do with transient voltages and transient currents.
There is ample evidence that an optimally biased ClassAB amplifier is as clean as required at low levels, i.e. while still operating in the ClassA current region.
The push-pull transition into ClassAB is used when transients come along that require the higher current that ClassA (whether single ended or limited push-pull) cannot provide.
I built a tiny 5pair bc550/560 ClassAB headphone amplifier about 15 years ago. It sort of worked, but later I was to learn that compensation was where it fell down.
I am not surprised.
I have promoted the design requirement that the peak transient SPL from the tweeter should be just as high as the mid range and bass range speakers.
Sreten, among many others, have rubbished this design philosophy repeatedly.
Average powers have little if anything to do with transient voltages and transient currents.
There is ample evidence that an optimally biased ClassAB amplifier is as clean as required at low levels, i.e. while still operating in the ClassA current region.
The push-pull transition into ClassAB is used when transients come along that require the higher current that ClassA (whether single ended or limited push-pull) cannot provide.
I built a tiny 5pair bc550/560 ClassAB headphone amplifier about 15 years ago. It sort of worked, but later I was to learn that compensation was where it fell down.
Last edited:
Mmmhhh, I see from your site that those are 2nd order filters exept from the tweeter which is 3rd order.
Where the 4th order comes from ?
Where the 4th order comes from ?
I'd like to revive this thread with a related observation... I'm not exactly sure what this means yet, but it's making me re-think the requirements of a subwoofer amplifier.
So, it's clear that the tweeter amp needs to be capable of some serious voltage swing in order to not clip transients, even though RMS power is low. So I was wondering what the voltage swing required by my subwoofer amplifier might be.
My main system is a 4-way LR6 with crossover points at 70, 300 and 3K. I wanted to simulate what the signals to each of my amps like looks like.
For this informal experiment I chose a Steely Dan track called 'Lunch with Gina' in 24/192. It has a pretty typical tonal balance and good dynamic range. I realize other musics will have a different spectrum, but I thought this was a good use-case. In Audacity, I split it into four frequency bands at with 24db LR6 crossover points at 70, 300, 3K, much as they are in my system using active analog filters. (The actual system has some modest EQ going on.)
What surprised me is the low peak levels experienced by the subwoofer! Check it out... Remember these are voltage levels, not power.
Sub
Bass <70Hz
Midrange >300Hz & <3KHz
Highs >3K
The sub is actually being asked to deliver lower peak voltage levels than the other three bands, even as it's being asked to deliver more power. The midrange amp is the one that clips first!
Here are the waveforms in the same order...
Sub
Bass
Mid
Tweet
On bass-heavy music, obviously this gets flipped around. I'm using a Crown K2 on my subwoofer, and bassy music does get it swinging and I can clip it.
But it seems, for normal music, that even though the power drawn by the sub is high, the voltage swing is generally low.
I know my sample size is lousy—only one—but I feel the example shows a surprisingly clear contrast. Maybe I can take a larger sample and do some averaging...
So, what does this say about amplifier choice? I always think about high-wattage amps for subwoofer applications, but it seems that effortless current delivery is more important than maximum wattage, as it's actually spending most of it's time not swinging much voltage.
The Crown K2 I'm using certainly provides both, but I was surprised the other day by how well this little Adcom GFA-535 was able to drive my sub! (Actually, it's one I modified into a mono-bloc with a double power supply that delivers about 100W instead of 80W, but still not a huge amp.) I was expecting it to clip easier, but the midrange amp, also a GFA-535 was actually the first to clip!
Tentative conclusion: For a subwoofer amp, a high-current design is more important than high-voltage.
So, it's clear that the tweeter amp needs to be capable of some serious voltage swing in order to not clip transients, even though RMS power is low. So I was wondering what the voltage swing required by my subwoofer amplifier might be.
My main system is a 4-way LR6 with crossover points at 70, 300 and 3K. I wanted to simulate what the signals to each of my amps like looks like.
For this informal experiment I chose a Steely Dan track called 'Lunch with Gina' in 24/192. It has a pretty typical tonal balance and good dynamic range. I realize other musics will have a different spectrum, but I thought this was a good use-case. In Audacity, I split it into four frequency bands at with 24db LR6 crossover points at 70, 300, 3K, much as they are in my system using active analog filters. (The actual system has some modest EQ going on.)
What surprised me is the low peak levels experienced by the subwoofer! Check it out... Remember these are voltage levels, not power.
Sub
Bass <70Hz
Midrange >300Hz & <3KHz
Highs >3K
The sub is actually being asked to deliver lower peak voltage levels than the other three bands, even as it's being asked to deliver more power. The midrange amp is the one that clips first!
Here are the waveforms in the same order...
Sub
Bass
Mid
Tweet
On bass-heavy music, obviously this gets flipped around. I'm using a Crown K2 on my subwoofer, and bassy music does get it swinging and I can clip it.
But it seems, for normal music, that even though the power drawn by the sub is high, the voltage swing is generally low.
I know my sample size is lousy—only one—but I feel the example shows a surprisingly clear contrast. Maybe I can take a larger sample and do some averaging...
So, what does this say about amplifier choice? I always think about high-wattage amps for subwoofer applications, but it seems that effortless current delivery is more important than maximum wattage, as it's actually spending most of it's time not swinging much voltage.
The Crown K2 I'm using certainly provides both, but I was surprised the other day by how well this little Adcom GFA-535 was able to drive my sub! (Actually, it's one I modified into a mono-bloc with a double power supply that delivers about 100W instead of 80W, but still not a huge amp.) I was expecting it to clip easier, but the midrange amp, also a GFA-535 was actually the first to clip!
Tentative conclusion: For a subwoofer amp, a high-current design is more important than high-voltage.
Last edited:
Yes, ack... words not working...
I'm really thinking if one has to choose between two amps.
Assume you have an 8 ohm sub to power.
Amp A is a typical 1000W into 8 ohms amplifier that also happens to deliver 1500W into 4 ohms. It doesn't have enough juice to perfectly double power into half impedance. Most amps are like this.
Amp B is 500W into 8 ohms, but is a high-current design that unflinchingly delivers 1000W into 4 Ohms.
My guess is that Amp B is the one to get. Perhaps not delivering as much total SPL, but more effortless sounding 99% of the time.
I'm really thinking if one has to choose between two amps.
Assume you have an 8 ohm sub to power.
Amp A is a typical 1000W into 8 ohms amplifier that also happens to deliver 1500W into 4 ohms. It doesn't have enough juice to perfectly double power into half impedance. Most amps are like this.
Amp B is 500W into 8 ohms, but is a high-current design that unflinchingly delivers 1000W into 4 Ohms.
My guess is that Amp B is the one to get. Perhaps not delivering as much total SPL, but more effortless sounding 99% of the time.
Last edited:
And doubling wattage only gets you 3db more output from the speakers. Less with physical power compression losses in the drivers. In order to be perceived as twice as loud requires 6-10db.
Hi,
it´s also V²/R .. so what´s that gonna say???
The reason for high current demand of the sub and woofer is imho rather due to the high and widely varying phase shifts that occur especially with BassReflex systems.
Keep in mind,that at +-45° there´s already one Ampere of blind current for one Ampere of effective current.
The blind current ´wobbles´ between power supply and power transistors just stressing current paths and generating heat.
jauu
Calvin
it´s also V²/R .. so what´s that gonna say???
The reason for high current demand of the sub and woofer is imho rather due to the high and widely varying phase shifts that occur especially with BassReflex systems.
Keep in mind,that at +-45° there´s already one Ampere of blind current for one Ampere of effective current.
The blind current ´wobbles´ between power supply and power transistors just stressing current paths and generating heat.
jauu
Calvin
Although the spectrum analysis at source level shows low bass peak levels, when it comes to the sub amp actually reproducing those levels you will find that the voltage applied to the sub is high in relation to the other drivers simply because the sub is the least efficient driver.
You need to scope actual driver voltage levels for any meaningful analysis here.
You need to scope actual driver voltage levels for any meaningful analysis here.
I have said consistently throughout my years here that for speaker drivers of the same impedance and the same sensitivity, each should be driven by an amplifier of equal maximum voltage.
I have given some examples of this and the consequences in posts years ago.
Yet there are many, the vast majority of this Membership, that refuse to believe this.
They want to believe that bass needs extra power and treble needs less power.
I have given some examples of this and the consequences in posts years ago.
Yet there are many, the vast majority of this Membership, that refuse to believe this.
They want to believe that bass needs extra power and treble needs less power.
I think people do believe you Andrew.
The key part is the impedance and sensitivity as you have mentioned. If a tweeter is 3dB more sensitive then it only needs half the power. If a sub is 3dB less sensitive then it needs double the power.
This is still inline with your recommendations, as max SPL is the same across the board. It is just rare that drivers all have the same sensitivity.
The "vast majority" are only following articles, texts or forum posts here and elsewhere, that may well be in error or easily misinterpreted. You won't make headway here by shooting the readers and students of these docs who then get it wrong, but you could try publishing a positive and encouraging explanation of your own, illustrating the more correct way to go about it.
Perhaps a blog thread or even the DiyAudio management might care to put up an article at the appropriate pages that keep such worthwhile info. accessible.
Hi,
The problem is your argument has a serious consistency flaw you never address.
Your argument is independent of bandwidth and makes no sense.
By your argument all drivers in a active system need the same maximum
single frequency SPL. Which implies the x/o points don't matter, but they
do matter seriously in a cost effective active system.
The OP should try monitoring peak voltage across a tweeter x/o at
say 3KHz, 6KHz and 12KHz. By your argument the peaks voltages
won't change. By my argument they will, 0dB peaks are wideband.
rgds, sreten.
Fact is, its just a belligerent claim that doesn't stand up to the facts.
You say an active 2-way needs the same peak SPL per driver
independent of the x/o frequency, it is simply just wrong.
A 2-way at 300Hz and 3KHz are very different.
Whatever, with equal power amplifiers, for a 3 way active bass
to mid x/o and a passive mid to treble x/o is a good option.
Last edited:
- Status
- Not open for further replies.