Real Life Bi-Amplification: Some considerations

[TLS Guy]

on the speaker crossover design, i have seen a B&W 804 cut out and there are TWO set of crossover inside the cabnet. one at the bottom connected to woofer, and another one sit behind mid range for high/mid driver. in that case, even if you passively biamped the speakers, it should sounds better (in theory) than not biamping it.

another big problem of crossover is resistance swing. take B&W 804 as an example, from 20Hz to 20KHz, the resistance swing from lower 3 Ohm to as high as 30 Ohm. in that case, the ampifier not only have to ampify different frequency, but also need to work with very different indepence. of course, that result in distortion.

with that in mind, the perfect system would be active crossover, bi, or tri amp, and connect directly to the speakers. the cut off point on the crossover would be just before the driver's resistance starting to swing higher. in this case, the amp will remain constant in the resistance. actually, this approach has been using on commercial application for many years, i guess it's too complex to set up for average home user who want to "set it and forget it".

You have a lot of misconceptions. First the varying impedance does not cause distortion. There will be distortion if the amp can not provide the current at the low impedance point.

Now most of he variation impedance is due to the inductance of the driver voice coils. The crossover often reduces the impedance swing. So the amp does not see a resistive load even with biamping. Far from it.

 

[Antus]

You have a lot of misconceptions. First the varying impedance does not cause distortion. There will be distortion if the amp can not provide the current at the low impedance point.

Now most of he variation impedance is due to the inductance of the driver voice coils. The crossover often reduces the impedance swing. So the amp does not see a resistive load even with biamping. Far from it.

i don't see why vary in impedance won't add distortion. usually, an amp (for home use) are optimized at 8 ohm. they might have 20Hz to 20KHz at 8ohm at distortion at 0.05% or lower, some higher quality amp can get it even lower. However, the same amp certainly WON"T have the same super low distortion at 4 Ohm. (in fact, you will hardly see an amp munufacture quite their distortion at 4Ohm. i am sure it won't be 20-20K at 0.005% low) and if a speaker has a resistance swing from low 3ohm to high 30 Ohm (like B&W 804) it is nearly impossible (or very expensive) to design an amp that has 20 to 20K at distortion at 0.005% at any giving impedence from 3OHm to 30 Ohm.

yes, part of the variation is due to the driver cone. but how can a crossover reduce it? i can't think of an circuit that can reduce resistance. the best case is not adding any resistance, but in no ways reduce it.

 

[sparky77]

on the speaker crossover design, i have seen a B&W 804 cut out and there are TWO set of crossover inside the cabnet. one at the bottom connected to woofer, and another one sit behind mid range for high/mid driver. in that case, even if you passively biamped the speakers, it should sounds better (in theory) than not biamping it.

another big problem of crossover is resistance swing. take B&W 804 as an example, from 20Hz to 20KHz, the resistance swing from lower 3 Ohm to as high as 30 Ohm. in that case, the ampifier not only have to ampify different frequency, but also need to work with very different indepence. of course, that result in distortion.

with that in mind, the perfect system would be active crossover, bi, or tri amp, and connect directly to the speakers. the cut off point on the crossover would be just before the driver's resistance starting to swing higher. in this case, the amp will remain constant in the resistance. actually, this approach has been using on commercial application for many years, i guess it's too complex to set up for average home user who want to "set it and forget it".

That is pretty much what tlsguy has said in numerous posts before so I do agree with some of it. The fact is in any electrical circuit whether it's analog or digital is going to insert some kind of a delay or phase diffential, at lower frequencies it has more of an impact on what a human can hear, but at higher frequencies the human ear has a way of disregarding the differences, yet at the same time on the lower spectrum if the phases are close enough together such as with a high speed processing signal processing from digital crossovers it won't even be apparent to most people depending on the room that the audio system is set up in. There are an exhorbitant amount of factors at play in all of these situations, you can try to obtain perfection, but its not really needed, if it sounds good to you, that's what's important.

 

[mtrycrafts]

... With the strap in place, they are in parallel. Without it, they are completely separate networks that are totally isolated from each other.

Only if you are using another amp. But so what?

The "job" it would do would be to allow (in the denon's case) for 110W to be sent to the woofer and 110 watts to the midrange and tweeter, for a total of 220W watts of possible amplification.

Don't think so. since these are frequency dependent drivers, you are sending 110 watts to the low and 110 watts to the high, not additive. The same 110 watts would get there from 1 amp going to the two sets of terminals if that amp is rated for 110 watts from 20Hz-20Khz. That amp will deliver 110 watts to all the frequencies at the same time. that is what that spec is telling us. Just make sure you don't blow up the tweeter with that 110 watts as most tweeters are about 20 watts capable only, if you are lucky.

 

[sparky77]

Don't think so. since these are frequency dependent drivers, you are sending 110 watts to the low and 110 watts to the high, not additive. The same 110 watts would get there from 1 amp going to the two sets of terminals if that amp is rated for 110 watts from 20Hz-20Khz. That amp will deliver 110 watts to all the frequencies at the same time. that is what that spec is telling us. Just make sure you don't blow up the tweeter with that 110 watts as most tweeters are about 20 watts capable only, if you are lucky.

Don't think I could have said it much better myself, but you do have to consider that certain frequencies less power is required, don't remember all the numbers and ratios right off hand, but low frequencies in the 30hz range can require up to 10 times the power as frequencies in the 10khz range. The only problem is that if the power isn't consumed by the speakers it gets reflected back to the amplifier and causes feedback control circuit to go into parasitic oscilations which can sometimes cause accentuations at resonant frequencies, but is pretty much isolated in newer amplifier designs. I'm sure some one will correct me if I'm wrong on this.

The only way bi-amping is truly beneficial is with a pre-amp and seperate amplifies in conjuntion with digital crossovers, plain and simple. One amplifier per speaker element driven from a digital crossover, that will yield the closest thing to audio nirvana, as for the rest of us on a limited budget, feel free to enjoy your two or three way speakers with passive crossovers and remember that the majority of us will be totally ignorant of the phase response issues until it becomes noticeable with room modes that cant be corrected by acoustic treatments.

 

[TLS Guy]

Don't think I could have said it much better myself, but you do have to consider that certain frequencies less power is required, don't remember all the numbers and ratios right off hand, but low frequencies in the 30hz range can require up to 10 times the power as frequencies in the 10khz range. The only problem is that if the power isn't consumed by the speakers it gets reflected back to the amplifier and causes feedback control circuit to go into parasitic oscilations which can sometimes cause accentuations at resonant frequencies, but is pretty much isolated in newer amplifier designs. I'm sure some one will correct me if I'm wrong on this.

The only way bi-amping is truly beneficial is with a pre-amp and seperate amplifies in conjuntion with digital crossovers, plain and simple. One amplifier per speaker element driven from a digital crossover, that will yield the closest thing to audio nirvana, as for the rest of us on a limited budget, feel free to enjoy your two or three way speakers with passive crossovers and remember that the majority of us will be totally ignorant of the phase response issues until it becomes noticeable with room modes that cant be corrected by acoustic treatments.

Amp oscillations, are not due to power getting sent back to speakers. The speakers will take what they need up to the clip point. The oscillations are caused by the reactive element of the load. This is very rare with moving coil drivers and passive crossovers, but has been known. It is well known when driving electrostatics, and some amps will not drive them.

Now the phase and time delay problems that are serious occur in the crossover regions, whether analog and passive. Now there are time delays introduced by ports and the electromechanical characteristics of drivers, but I don't think these are of consequence. It is those introduced but analog crossovers that it would be nice to consign to history.
Only first order analog crossovers can be made phase and time coherent, and they have so much driver overlap, that only occasionally can first order crossovers be accomplished. Driver spacing issues also compound the problem. All of these issues have driven the full ranger devotees.

 

[Joe Schmoe]

Considering that a low-powered amp is more than enough for my speakers, I cannot imagine that the extra power from bi-amping would have any advantages whatsoever.

 

[TLS Guy]

i don't see why vary in impedance won't add distortion. usually, an amp (for home use) are optimized at 8 ohm. they might have 20Hz to 20KHz at 8ohm at distortion at 0.05% or lower, some higher quality amp can get it even lower. However, the same amp certainly WON"T have the same super low distortion at 4 Ohm. (in fact, you will hardly see an amp munufacture quite their distortion at 4Ohm. i am sure it won't be 20-20K at 0.005% low) and if a speaker has a resistance swing from low 3ohm to high 30 Ohm (like B&W 804) it is nearly impossible (or very expensive) to design an amp that has 20 to 20K at distortion at 0.005% at any giving impedence from 3OHm to 30 Ohm.

yes, part of the variation is due to the driver cone. but how can a crossover reduce it? i can't think of an circuit that can reduce resistance. the best case is not adding any resistance, but in no ways reduce it.

First of all the loads presented by speakers are not resistive. A moving coil driver presents a reactive load. Passive crossovers have both reactive and capacitive elements.

The next point is that the impedance curve of a driver, or complete speaker for that matter, has no correlation with its frequency response. At points of higher impedance the speaker is more sensitive at lower points less so. Now well designed amps will be able to maintain driving voltage and increase current in the face of falling impedance without adding excessive distortion.

Now pretty much every speaker has impedance humps in the bass region, unless compensated for in the crossover, which you usually you do not do. There is one impedance bass hump in a sealed box and two in ported and TL designs. In fact those humps are one of the checks of correct alignment.

Now every woofer has a rising impedance curve with frequency. This makes it almost impossible for a passive low pass crossover to even limit the HF cut off of the woofer. More often than not this has to be dealt with. Usually it is dealt with by shunting the woofer with a Zobel network. This is a resistor and capacitor in series that shunts the driver voice coil and prevents excessive impedance rise towards crossover. These can sometimes be fine tuned even further by placing tuned circuits of resistor, capacitor and inductor across the capacitor of the Zobel network. If you really work at it you can often make the impedance curve of a speaker very flat. However this adds greatly and usually unnecessarily to part count and cost. Also it may result in undesired consequences.

 

[3db]

These can sometimes be fine tuned even further by placing tuned circuits of resistor, capacitor and inductor across the capacitor of the Zobel network. If you really work at it you can often make the impedance curve of a speaker very flat. However this adds greatly and usually unnecessarily to part count and cost. Also it may result in undesired consequences.

PSB's new Synchron model 1 ( at least as advertised unless a misprint) and one other older model exhibited a very flat impedance curve across the entire audible spectrum.

 

[TLS Guy]

Considering that a low-powered amp is more than enough for my speakers, I cannot imagine that the extra power from bi-amping would have any advantages whatsoever.

Well it does. However the advantages of biamping a two way speaker with a crossover in the 2 to 4KHz range with an analog active crossover would be negligible. However doing it with digital crossovers with zero phase and time problems would be considerable. I'm sure this is the way things will go. The class D amps are beset with far greater speaker interface issues than the current class A/B amps. If the industry is correct in saying that moving to class D amps is the only way to maintain price points, then I think it will have to go this way otherwise we will have chaos. Really the speaker always has been the right place for the amps, this change in practice will likely make it mandatory.

In terms of analog biamping. These are the advantages. The only inductors in the amp circuit are the speaker voice coils. There are no caps on the amp speaker circuit. The level padding takes place without loss of efficiency. For low frequency low pass it eliminates the need for large costly inductors.

The biggest gains come in high power systems. Also in three way systems with crossovers below around 350 Hz passive crossovers are highly limiting and not a good solution.

 

[3db]

These can sometimes be fine tuned even further by placing tuned circuits of resistor, capacitor and inductor across the capacitor of the Zobel network. If you really work at it you can often make the impedance curve of a speaker very flat. However this adds greatly and usually unnecessarily to part count and cost. Also it may result in undesired consequences.

PSB's new Synchron model 1 ( at least as advertised unless a misprint) and one other older model exhibited a very flat impedance curve across the entire audible spectrum.

 

[TLS Guy]

PSB's new Synchron model 1 ( at least as advertised unless a misprint) and one other older model exhibited a very flat impedance curve across the entire audible spectrum.

Absolutely you can make the impedance curve of a speaker very flat. KEF have always made a big point of impedance correction in their crossovers.
I built these speakers with a very flat impedance curve.

http://mdcarter.smugmug.com/gallery/2424105#127080824

It is just one of many issues to consider and work through in the design of any speaker. The problem is it adds reactive elements to the crossover and you get into the realm of unintended consequences. The other issue, is that the impedance curve of the woofer can sometimes be made use of to tailor the driver responses and make the crossover simpler.

 

[trnqk7]

Only if you are using another amp. But so what?



Don't think so. since these are frequency dependent drivers, you are sending 110 watts to the low and 110 watts to the high, not additive. The same 110 watts would get there from 1 amp going to the two sets of terminals if that amp is rated for 110 watts from 20Hz-20Khz. That amp will deliver 110 watts to all the frequencies at the same time. that is what that spec is telling us. Just make sure you don't blow up the tweeter with that 110 watts as most tweeters are about 20 watts capable only, if you are lucky.

I know it isn't additive-I mentioned in another post that the lows could use a percentage of the 110W sent to them and that the mid/highs could use a percentage of the 110W sent to them...Clearly, the mid/highs probably will never use a full 110W (they just don't need enough current to reproduce sounds at the levels I listen at)...and in my system and my preferences, doubtful will the woofer either...Still, if it was the case, then there are 220W of total amplification available for the entire system (speaker), even if not for the individual portions of the system.

 

[mtrycrafts]

Amp oscillations, are not due to power getting sent back to speakers. .

There is one amp company that requires their special speaker cable, low capacitance, or the amp will go into oscillation, forgot the name right now.

 

[TLS Guy]

There is one amp company that requires their special speaker cable, low capacitance, or the amp will go into oscillation, forgot the name right now.

Well that would be an amp to avoid, but it is capacitance causing the problem as I stated.

 

[WmAx]

Digital active crossovers are another matter. They can be constructed with very high order roll off, and even brick wall, without phase or time delay problems. So that leaves the possibility of operating drivers well within their pass band and not have to worry about driver roll off and the gain of the two drivers at crossover.

I would not personally use a brick-wall filter at this point. Brick wall filters are still questionable in use of multi-way speaker systems. While on a narrow design axis, response can look superb, off axis, where the combining drivers alter in phase response, the time response can be something of a horrific nature(at least when you view the graphed impulse response) - resulting in severe pre-ringing artifacts. The audibility of this effect in conjunction with loudspeaker off axis response has not as of yet been studied in credible perceptual research to my knowledge. However, use of traditional analog filters(and nearly any digital crossover will emulate these) does not pose any substantial phase issues that cause any substantial difference in audibility, if executed properly. These have been subjected to credible perceptual research.

-Chris