• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

Bi-amping 101

DonH56

Master Contributor
Technical Expert
Forum Donor
Joined
Mar 15, 2016
Messages
8,454
Likes
18,247
Location
Monument, CO
This article provides a quick look at how bi-amplification works. The conventional scheme is to split the signal into two frequency bands before the power amplifiers that drive the speakers. The bass (LF) amp sees only LF signals and drives only the LF driver (woofer). The treble (HF) amp sees only HF signals and drives only the HF driver. This reduces the signal for each amplifier, improving headroom and potentially allowing amplifiers to be chosen for their target frequency range without the burden of supplying the full-range audio signal.

Here is a regular single-amplifier system and passive speaker crossover. The crossover comprises a low-pass filter (LPF) for the woofer and high-pass filter (HPF) for the tweeter. The crossover is inside the speaker; the dashed outline shows there can be a single or split (high/low) set of speaker input terminals, but the connection is the same as far as the amplifier is concerned.

1619374086540.png


For this experiment both crossovers are second-order (12 dB/octave) Linkwitz-Riley designs and the same whether passive or active. The crossover frequency is 1 kHz. The speakers are represented by 8-ohm resistors in this simplified analysis. This is the transfer function (frequency response) of the crossover:

1619374096764.png


For analysis I am using two signals at 300 Hz (LF) and 3 kHz (HF) which on a log scale sit on either side of the crossover:

1619374137760.png


The output voltage from the amplifier includes both signals, of course:

1619374160294.png


You can see how the low-frequency signal and high-frequency combine to create the modulated output signal. Since both signals have the same amplitude, the combined signal is twice the level of the individual signals. While this makes it easy to see, note in practice HF signals are typically much lower in amplitude than the LF signals (see Equal Loudness Curves in Wikipedia or wherever).

The relative voltage, current, and power is in the table below. The numbers are low since I am using a small signal; everything is relative. The combined RMS voltage for this case is simply sqrt(LF^2 + HF^2). The power is the average power, the product of RMS voltage and current (RMS power is a meaningless term though often used mistakenly).

1619374751937.png


Now repeat the experiment, but place a line-level crossover before the power amps. The crossover is typically active but could be passive; the usurpation of the term “passive bi-amplification” the way AVR marketing defines it is an on-going source of confusion.

1619374798270.png


There are now two amplifiers but each only handles a portion of the signal divided into two (low and high) frequency bands. Since the crossover is not ideal, that is does not drop instantly to zero on either side of the crossover frequency, a little bit of the HF signal is still seen in the LF output and vice-versa. It is a little hard to see in the time-domain plot, but the frequency-domain plots clearly show this.

1619374807637.png

1619374835019.png

1619374845227.png


Now we can create a new table showing the parameters of each amplifier in the bi-amplified case:

1619374864296.png


The voltages are similar to the previous table except now they are at the output of the amplifiers and, with the filters, the RMS value is a little lower. The RMS current is reduced more significantly since the load is split between the two amplifiers, and power is also significantly reduced (by about 60%). Note power is not a linear function, but rather the product of voltage and current (a nonlinear function), so splitting the signal this way does not reduce the power by exactly one-half (it is better than that). You could use two amplifiers each one-half the rated power if the single amplifier and deliver the same power to the speaker. The amplifiers also directly drive the speakers so there is no crossover loss, and no chance of the signal from one speaker modulating the other through the crossover circuit, potentially reducing distortion.

Now consider “passive” bi-amping as implanted by most AVRs and shown below. In this scenario both amplifiers are driven by the same signal, so their outputs have exactly the same voltage. There is no benefit in voltage and, since most amplifiers are essentially voltage-mode designs, no increase in system headroom. Current is reduced from each amplifier, since the passive crossover in the speaker “blocks” current that is out of band, and there is a corresponding reduction in power (not as much as when an active crossover is used, however).

1619374882213.png


The voltage output of each amplifier is the same:

1619374898392.png


The current is lower since the load seen by each amplifier is only part of the speaker. The LF amp has almost no HF current in its output, and similarly the HF amp has little LF current, even though the voltage is the same.

1619374912785.png


Here is the parameter table for the “passive” bi-amping case:

1619374938888.png


And a comparison of all three examples:

1619374978179.png


A ratio less than one represents an advantage (decrease) in the parameter the amplifier is delivering to the speaker. In a conventional active bi-amp system, each amplifier delivers only 65% of the voltage and current and only 42% the power of a single amplifier providing the same signal to the speakers. A “passive” bi-amp system the way an AVR does it has no voltage benefit but does require less current from each amplifier, resulting in the same 65% reduction in current, but due to the higher voltage must deliver 65% of the power. A reduction, true, but less dramatic than for an active bi-amp system, and since there is no reduction in the voltage that must be delivered there is no (voltage) headroom advantage. Where an active system would allow you to replace a 200 W amplifier with two 100 W amplifiers, the passive scheme requires two 200 W amplifiers to maintain the voltage headroom required.

One other significant consideration is how we hear as reflected in the equal-loudness curves (see e.g. https://en.wikipedia.org/wiki/Equal-loudness_contour). It take much higher levels of low-frequency signals to sound as loud as midrange signals. At 80 dB SPL, a 100 Hz signal must be about 10 dB louder (10x the power) to sound as loud as a 1 kHz signal, and must be about 20 dB louder (100x the power) by around 60 Hz. That is one reason bi-amplified systems (of the non-passive kind) often use a much larger bass amplifier than for the tweeter.

A few key points about “passive” bi-amping, from the original version of this roughly ten years ago:
  1. Each amplifier requires the same voltage output as a single amplifier since they each have the same signal. There is no voltage headroom benefit.
  2. Because the speaker load is essentially an “open” in the unused frequency band, less current output is required from each amp.
  3. There is no net system power increase at the speakers assuming the amps have the same voltage rails (e.g. inside an AVR or multichannel amplifier with the same power voltage rails to all amps). If you had a 100 W amp before, and bi-amp with two 100 W amplifiers, passive bi-amping does not give you 200 W to the speaker. You have split the load into two frequency bands, but the maximum power is the same to the speaker. That is, 100 W to the lows and 100 W to the highs is the same as having a 100 W amp that covers the entire frequency range. It is not the same as driving the speaker with a 200 W amplifier; to increase the power, you need to increase the voltage rails. There is not an effective increase in power headroom as there is for an active approach.
  4. In fact, there is more power lost, since the amps are not 100% efficient. That is, it actually take more energy from the power supply to passively bi-amp than if you used a single amp. This is also true for active bi-amping, but in that case we can choose lower-power amps for the highs (which rarely need the same power as the lows) and realize net power savings. That does not happen with (typical) passive bi-amping.
  5. There is no damping factor improvement over a single amp since the speaker crossovers are still in-circuit. One of the benefits of active bi-amping is direct connection from amp to driver, providing better driver control; this is not true in passive bi-amping.
  6. There is no longer electrical interaction among drivers with passive (or active) bi-amping. (There may still be mechanical coupling if the drivers are not isolated from each other.) That is, if the woofer starts to distort the input signal through electromechanical forces, it no longer modulates the HF amp’s output. One plus for bi-amping, active or passive.
  7. If the amps share a power supply, as do most AVRs and many (most?) multichannel amps, then modulation between high and low amps can still occur through the power supply. This can also happen with active bi-amping, although separate amps are the norm in the pro world. At least when I have done it…
  8. There may be some distortion reduction since power output is lessened in the amps. I suspect this is not significant, but it should happen due to the lower current draw. The catch is that the voltage swing of each amp is unchanged, so any distortion related to voltage swing is not changed. Only distortion components depending on output current may be reduced. That is design-dependent, but since most amps are primarily voltage-mode amps, I suspect any distortion reduction is small.
  9. You have two amps now so presumably noise is a little higher since you have two uncorrelated noise sources. At the speaker outputs I suspect it’s a wash since only a reduced frequency band gets through the drivers to hear.
  10. Thermally it is a loss since no amp is 100% efficient. There is always a little “waste” power that gets turned into heat, both standing bias current (especially if not class D amps) and losses through the components in the amp. Thus passive bi-amping will cause your AVR/amp to run hotter than if using a single amp (assuming unused channels). It is worth noting that amplifiers are typically most efficient at maximum output; the HF amp is probably loafing most of the time and thus wasting power and generating heat.
So, there are some potential benefits using passive bi-amping, but I suspect they are inaudible (I have not tried passive bi-amping so cannot say). And a lot of drawbacks. The major benefit is mostly mental, IMO; users can now use their “extra” amp channels. Whether this benefits anyone other than the electric company I cannot say, but I suspect not… It does eliminate signal “bleed” through the crossover, so if the amplifier has high output impedance (like a tube amp) then passive bi-amping could benefit. If the amplifier is near clipping, or current-limited, then the reduction in current by a passive bi-amp scheme could reduce distortion. Again I suspect that is an insignificant improvement.

FWIWFM - Don
 
Last edited:
Thank you for this reasoned analysis. In item #8 above the reduction in distortion is mentioned. In the original article by Norman Crowhurst which caused me to have a "conversion experience" the distortion he mentioned being reduced is intermodulation distortion , since each amplifier now has a narrower band of frequencies to make powerful. I believe this is very audible.

Also quite audible is the improvement in damping factor which is caused by removing anything but sturdy cables between the amp output and the speaker drivers.

This would all be difficult to ABX test. I have never heard of anyone doing it.

Norman Crowhurst 's 1969 articles on the subject.
https://worldradiohistory.com/Archive-Radio-Electronics/60s/1969/Radio-Electronics-1969-03.pdf Page 32
https://worldradiohistory.com/Archive-Radio-Electronics/60s/1969/Radio-Electronics-1969-10.pdf page 42

For those of you who have never heard of him Norman Crowhurst was an audio hifi giant of the fifties and sixties and into the seventies. An early non snake oiler.
 
Norman Crowhurst was indeed an important figure in audio in the 1960s, along with Percy Wilson and others.
However, his references to improvements due to biamping need to be considered in the light of the amplifiers of the time, which would have been mostly valved, or early SS amps which had very limited current output.

Using modern SS amps, and by modern, I mean anything produced in the last 30 or more years, the will be no appreciable change in intermodulation distortion as such amplifiers have ample current capability, and remain linear regardless of the small changes in loading. That's why I've maintained that passive biamping is a con, designed to sell more amplifiers, and only active biamping effects an improvement, for the reasons Don mentions above.

S
 
@DonH56, your explanation is perfect and was identical to what I had thought earlier about the advantages of active biamping. However in a thread not too long ago (maybe 6 months?) I saw a proof that a high passed signal (noise?) actually had a higher voltage than the combined signal before the crossover (increased crest factor). This means that you still need power amps for the highs with similar voltage capacity, and if you look at Genelecs (older) 80x0 series (all 2-ways) you may notice that the amps for woofer and tweeter have the same power specification.
 
@DonH56, your explanation is perfect and was identical to what I had thought earlier about the advantages of active biamping. However in a thread not too long ago (maybe 6 months?) I saw a proof that a high passed signal (noise?) actually had a higher voltage than the combined signal before the crossover (increased crest factor). This means that you still need power amps for the highs with similar voltage capacity, and if you look at Genelecs (older) 80x0 series (all 2-ways) you may notice that the amps for woofer and tweeter have the same power specification.

Thanks.

I do not see how the signal after a passive filter is larger than the signal going into the filter. I can believe the crest factor may be larger for the HF portion, but that does not mean the maximum voltage is increased after the filter, just that the HF signal has greater peak-to-average ratio than the LF signal.

The actual power demand per amplifier is very dependent upon the crossover frequency, frequency spectrum of the source, and impedance of the speakers. Bass tends to be larger in level, but there is usually a lot more spectrum covered by the HF driver, so the actual power distribution does not always match the simple rule of thumb that bass demands the most. I had set up a 31-tone test and planned to weight it per the ISO 226:2003 curve but only so much time in the day... I have the analysis set up so may finish it at some point. It would make an interesting addition to this thread. But these posts take me quite a while to produce so not likely today. My original article, about ten years old, used SPICE simulations and I was never happy with the presentation (text or plots). Hopefully this one is easier to follow (the only part retained was the final list of pros/cons for "passive" bi-amping).
 
With passive bi-amping, any 'voicing' (frequency / phase manipulation in the passive crossover) which the manufacturer performs is maintained. If the passive crossovers are bypassed and the system converted to full active crossovers, this voicing is lost.

This is kind of a silly scenario and comes under the category of 'duh, why would I do that?', but I though I'd throw it out there anyway.
 
Very nicely done. Thanks, Don. An awesome explanation of the electrical characteristics of bi-amping.
 
Thanks.

I do not see how the signal after a passive filter is larger than the signal going into the filter. I can believe the crest factor may be larger for the HF portion, but that does not mean the maximum voltage is increased after the filter, just that the HF signal has greater peak-to-average ratio than the LF signal.
Think square waves and Gibbs.

[Edit] But admittedly, it is a corner case.
 
Norman Crowhurst was indeed an important figure in audio in the 1960s, along with Percy Wilson and others.
However, his references to improvements due to biamping need to be considered in the light of the amplifiers of the time, which would have been mostly valved, or early SS amps which had very limited current output.

Using modern SS amps, and by modern, I mean anything produced in the last 30 or more years, the will be no appreciable change in intermodulation distortion as such amplifiers have ample current capability, and remain linear regardless of the small changes in loading. That's why I've maintained that passive biamping is a con, designed to sell more amplifiers, and only active biamping effects an improvement, for the reasons Don mentions above.

S
I don't advocate "passive" biamping as the term is used now. Only active. The improvement is quite audible. I correctly identified a Bryston system as triamped instantly ( I was dozing off in their suite late in the day) when I heard an orchestral sforzando in a way you only hear it in triamped systems. ($45,000 system) But even "passive biamping" is likely a slight improvement even if not worth it.

Since a passive (ie RC or LC or RLC ) network between power amps and preamp is easy to do I don't see why anyone wouldn't do that.
I don't really believe in "voicing" I believe it is on the snake oil continuum or subjective terms BS continuum.

I started doing multiamping with tube amps and early solid state indeed benefitted by narrower spectra being fed to them.
But the last iteration which took place in 1977 had (has) amps that can compete with anything that is out today. I do have newer amps too.
 
Last edited:
Think square waves and Gibbs.

[Edit] But admittedly, it is a corner case.

Yah, I figured someone would bring that up, or resonance and ringing and such, but from the context I don't think that is the argument. Could be wrong, natch.
 
I don't really believe in "voicing" I believe it is on the snake oil continuum.

I 'voice' my actively bi-amped speakers using passive line-level filters between the output of the active crossover sections and the power amplifier inputs (a total of 3 filters, 2 in the HF chain and one in the LF chain), as the listening position frequency response would be unacceptable without this step.

Perhaps the term 'voicing' is being taken as a pejorative in the high-end-voodoo sense, but what I mean by it is correcting the driver responses/room interaction responses to yield the overall frequency response I desire at the listening position.
 
Hey.

I'm bi amped . I have a marantz 7005 and a bk205.5 amp. I run a asend acoustics center and vandy 3.5 sigs . I think when I went from non biamped to biamped I heard a big difference in the bass mostly but now it's been so long I think I would need to switch back to think if I really do hear a difference.

The more I read about audio and listen the more I think I'm confused about what I think I hear and what I'm hearing.

This God dam sport is a pain in the ass.

But I love it.
 
I 'voice' my actively bi-amped speakers using passive line-level filters between the output of the active crossover sections and the power amplifier inputs (a total of 3 filters, 2 in the HF chain and one in the LF chain), as the listening position frequency response would be unacceptable without this step.

Perhaps the term 'voicing' is being taken as a pejorative in the high-end-voodoo sense, but what I mean by it is correcting the driver responses/room interaction responses to yield the overall frequency response I desire at the listening position.
I have been spoiled because I have two Pioneer D-23 crossovers. They are audibly distortionless and flat. They can change slope, crossover frequency and shelving. I used to play with them alot. Now I just set them with an RTA and then fine adjust to suit and ocassionally to some program that I don't care for the balance of. If that is called voicing OK. I like drivers that measure pretty flat in the ranges I want to use them in. (From the get go. Drivers with tilted FR probably have distortion nonlinearities too.)

One of the things that I think is snake oil is when people tell me that the crossover in the monkey coffin box can do tricks that can't be done with the one that I am using, which is before the amps and can do tricks that no passive crossover in the speaker box can do. It may be that nowadays with better test gear the speaker manufacturers 100% test crossovers. But it wasn't the case in the past. (and I don't really believe it is the case now) Even back in the day bi and tri amping were the most economical way to improve the transparency and imaging of your stereo system.

Before the world went "high end insane" the best companies in audio , JBL, Bozak and others had bi and triamp systems. Marantz had tube electronic crossovers. Pioneer had multiple lines of active crossovers in the early seventies. (SF series and then the D23) Sansui and Nikko and many others had crossovers. Bozak had an inexpensive two way active unit to biamp their systems. I think I still have one of those although it may not have made the last move. Very similar design to the Crowhurst circuit in 1969. (but only 6db/oct rather than the 12 suggested by Crowhurst)

https://www.worthpoint.com/worthopedia/bozak-electronic-crossover-106b-107-26269752
 
Last edited:
This post continues our look into bi-amping using a more complex signal. Instead of two tones, the input is now 31 tones equaled spaced on 1/3-octave centers. This is often used to provide a more “musical” signal for testing. For testing it can show interdependencies leading to higher distortion; for this exercise, it provides a little more insight into the power distribution of the two frequency bands in a bi-amplified system.

Here is the input signal – the time-domain signal looks like noise; there are no distinguishable signal patterns at this level.

1619901149985.png

1619901160911.png


Applying the same second-order 1 kHz crossover, 31 tones clearly shows how the signals roll off above and below the crossover frequency. Again with so many tones the time-domain response is similar for the LPF and HPF output except for the time scale. Note the x-axis is 500 ms for LF and 5 ms for HF outputs.

1619901201630.png

1619901210426.png

1619901225159.png

1619901255080.png


The voltage is scaled so that the signal peaks at 1 V, leading to the individual tones being small. As before, what matters is the power ratio of the high and low frequencies, and the results are similar to the previous two-tone ratios. Active bi-amping reduces the voltage and current, and thus power, from the LF and HF amplifiers. The LF power is a little higher due to the choice of 1 kHz crossover; the ratio will change as you move the crossover frequency. Again there is no voltage benefit to AVR-type passive bi-amping, and the power benefit is much less than for the actively bi-amped design.

1619901273667.png
 

Attachments

  • 1619901127244.png
    1619901127244.png
    40.7 KB · Views: 172
To make things more interesting, and arguably more realistic, we can weigh the tones according to the ISO226 equal-loudness curves (see e.g. https://en.wikipedia.org/wiki/Equal-loudness_contour). With this weighting, each tone will “sound” as loud as every other. For this analysis I weighted the tones using the ISO226 coefficients for 80 dB SPL, a fairly loud level to me, but one used for mixing and playback when “realistic” levels are desired. The weighting is clear in the frequency response, and because now the LF signals are much larger than the midrange and upper frequencies, the time-domain response shows the LF tones with HF “fuzz” added.

1619901345756.png

1619901355312.png


Here are plots showing the signal with low- and high-pass filtering using our 1 kHz crossover. The input and LF output amplitude are essentially the same while the HF signal is much smaller. This yields a much greater “real-world” advantage when actively bi-amping the speakers. Again note the difference in time scale (x-axis) and now amplitude (y-axis) for the LF and HF time-domain plots.

1619901376326.png

1619901384758.png

1619901506891.png


1619901518168.png


Now the power ratios are significantly different for the LF and HF amps as we’d expect. Virtually all of the power is concentrated in the lower frequencies. As always, passive bi-amping does not improve voltage headroom, negating much of the benefit of active bi-amping.

1619901578908.png
 
Thanks.

I do not see how the signal after a passive filter is larger than the signal going into the filter. I can believe the crest factor may be larger for the HF portion, but that does not mean the maximum voltage is increased after the filter, just that the HF signal has greater peak-to-average ratio than the LF signal.
I think this issue was in a discussion about whether to run main speakers full range with a subwoofer, or to high pass the mains. So same idea but at different crossover frequency. High passing was shown to possibly increase the voltage in some cases. But at least it doesn't let you use a lower power amp for the mains.

Bi-amping a two way speaker would be a different situation since the crossover would be at a fairly high frequency where signal levels are somewhat lower compared to bass frequencies. Still it seems to me it wouldn't change requirements for the low frequency amplifier much. But the high frequency amplifier could then be lower power.
 
I think this issue was in a discussion about whether to run main speakers full range with a subwoofer, or to high pass the mains. So same idea but at different crossover frequency. High passing was shown to possibly increase the voltage in some cases. But at least it doesn't let you use a lower power amp for the mains.

Bi-amping a two way speaker would be a different situation since the crossover would be at a fairly high frequency where signal levels are somewhat lower compared to bass frequencies. Still it seems to me it wouldn't change requirements for the low frequency amplifier much. But the high frequency amplifier could then be lower power.
High passing the mains helps with reducing the LF excursion on the mains and therefore reduces distortion as the sub should be much more capable of the excursion at low distortion. Furthermore, removing the need for LF excursion allows that excursion to be used at higher frequencies, thus increasing the available 'undistorted' SPL. So far so good, as long as then the tweeter is capable of handling the higher SPL, but in general, rolling off the LF below what the mains can reproduce cleanly is a Good Thing.

S
 
High passing the mains helps with reducing the LF excursion on the mains and therefore reduces distortion as the sub should be much more capable of the excursion at low distortion. Furthermore, removing the need for LF excursion allows that excursion to be used at higher frequencies, thus increasing the available 'undistorted' SPL. So far so good, as long as then the tweeter is capable of handling the higher SPL, but in general, rolling off the LF below what the mains can reproduce cleanly is a Good Thing.

S
I didn't mean to say there is no benefit to high passing mains when using a subwoofer. Just that it doesn't let you use a lower power (voltage) amplifier. I think it was @RayDunzl who showed this? Sorry if I have the wrong person!
 
I didn't mean to say there is no benefit to high passing mains when using a subwoofer. Just that it doesn't let you use a lower power (voltage) amplifier. I think it was @RayDunzl who showed this? Sorry if I have the wrong person!
You remember correctly but I couldn't find his posting.
 
Back
Top Bottom