m0tion said:
(JPK) Well if you work through the math you will find that there is a unique relationship between the transfer functions of an active filter and a passive filter that will yield exactly the same driver motion to within an arbitrary constant (insertion loss) . My analysis assumes that both the active and passive circuits are linear, i.e. no distortion. I don't consider driver nonlinearities because they are present in either case. So the bottom line is which introduces greater nonlinear distortion? That will depend on implimentation and in either case will likely be much lower than the inherent nonlinear distortion of the driver.
Of course, the fact that there is a unique relationship between the active and passive counter parts doesn't necessarily mean that it is possible to implement an active circuit that does exactly what a passive one does or vise versa. It just shows that there is no particular advantage. How the amplifier is connected to the driver is just one factor that must be considered in the analysis.
In a sense it is not unlike the discussion about constant current sources for speakers that was in a different thread. With a current source the amplifier has no control over the driver at all yet by shaping the drive current correctly the desired response is easliy obtained.
a different twist ...
I'd like to bring in a slightly different twist to the discussion ...
So far we have focused on active vs passive, but what about digital? I see this as a new frontier with many possibilities. Very quickly you can change crossover point / slope without soldering! You can also use very powerful contouring and assign digital delay, perform very accurate room correction. I have been enquiring about this with a mixed response.
Let's hear some views on digital vs active ...
I'd like to bring in a slightly different twist to the discussion ...
So far we have focused on active vs passive, but what about digital? I see this as a new frontier with many possibilities. Very quickly you can change crossover point / slope without soldering! You can also use very powerful contouring and assign digital delay, perform very accurate room correction. I have been enquiring about this with a mixed response.
Let's hear some views on digital vs active ...
It is so easy to compensate for the above-mentioned varying impedance without even using impedance-compensation circuits. No good DIYer should implement textbook active / passive filters -- we're not interested in the electrical transfer function -- good engineering calls for optimizing the acoustical transfer function.
Likewise I could say... Look at the basic equations for designing active filters and you will see that they do not take into account the acoustical response of driver units. While active filters let builders forget about impedance compensation, the problem of compensating for the acoustical response is still there.
Both types of filters (and digital filters) are easy to manually/automatically-optimize for transient perfectness, warmth, forwardness, zippiness... or whatever floats your boat.
Digital filtering should be all good! I think that people sometimes expect too much though, with empty promises of brick-wall filtering, room correction + speaker correction and perfect sound at the sweet spot. In a way, perfect sound from loudspeakers would be indistinguishable from headphones that use a good crossfeed system.
Digital filtering would mean that filter orders are not limited to some reasonable number like 4 or 6 as with analogue filters. IMO so-called compromise filters are the way to go. "Gaussian to 12dB" is one such filter. It's essentially a linear phase / Bessel filter in the cutoff region and has a sharp secondary cut-off in the stop-band at -12dB. Before the -12dB cutoff frequency, the filter has the best possible transient response with no ringing. The secondary cutoff at -12dB improves the frequency response, but produces ringing. Fortunately the ringing is referenced to a -12dB level, not -3dB, so it's much quieter than with classic filters such as Butterworth ones.
FIR filters with pre-ringing or IIR filters with sharp cutoffs should also be investigated. These could be most useful at low frequencies in cases where the relevant speakers are spaced closely enough to mutually cancel out all of the ringing. Designing such a system requires various skills that I doubt many designers of regular "hifi" equipment have learnt. Otherwise the market would be flooded with low-cost digital solutions.
IMO things like speaker correction and room correction are problematic. A bit of digital equalization is probably at the easier and most achievable end the scale, but apparently, people are disappointed that such systems often don't seem to improve the "quality" of the sound. Speaker correction can only make the most of the speakers that you already have, it can't stiffen speaker diaphragms or reduce panel resonances. If you correct the on-axis sound you make the off-axis sound worse, and vice versa. As for perfected "room-correction" with a resulting tiny sweet-spot, that's for headphone lovers in denial.
What I've got slowly simmering away on the back burner is a really big digitally controlled amplifier setup with the following attributes:
-Hybrid class A tweeter amps, with digitally controlled voltage rails for reduced power consumption. Possibly digital feedback and/or calibration to reduce distortion.
-multi-level switching amp for a subwoofer or 2 woofers.
-undecided on the midrange topology.
-microprocessor with multiple DACs to supply the filtered digital audio.
-a simple model of a speaker in a sealed box (with no passive crossover) still has a big LCR network with huge component values to represent the speaker's resonance and impedance-rise at high frequencies. I want to cancel out this LCR network so that the amplifier has maximum control over how the air in front of the speaker's diaphragm vibrates. This will require amplification with a negative output impedance and lots of clever tricks to make it stable.
-and it will have standard stuff like digital volume control - nothing less than relays of course, microphone feedback....
Digital filtering would mean that filter orders are not limited to some reasonable number like 4 or 6 as with analogue filters. IMO so-called compromise filters are the way to go. "Gaussian to 12dB" is one such filter. It's essentially a linear phase / Bessel filter in the cutoff region and has a sharp secondary cut-off in the stop-band at -12dB. Before the -12dB cutoff frequency, the filter has the best possible transient response with no ringing. The secondary cutoff at -12dB improves the frequency response, but produces ringing. Fortunately the ringing is referenced to a -12dB level, not -3dB, so it's much quieter than with classic filters such as Butterworth ones.
FIR filters with pre-ringing or IIR filters with sharp cutoffs should also be investigated. These could be most useful at low frequencies in cases where the relevant speakers are spaced closely enough to mutually cancel out all of the ringing. Designing such a system requires various skills that I doubt many designers of regular "hifi" equipment have learnt. Otherwise the market would be flooded with low-cost digital solutions.
IMO things like speaker correction and room correction are problematic. A bit of digital equalization is probably at the easier and most achievable end the scale, but apparently, people are disappointed that such systems often don't seem to improve the "quality" of the sound. Speaker correction can only make the most of the speakers that you already have, it can't stiffen speaker diaphragms or reduce panel resonances. If you correct the on-axis sound you make the off-axis sound worse, and vice versa. As for perfected "room-correction" with a resulting tiny sweet-spot, that's for headphone lovers in denial.
What I've got slowly simmering away on the back burner is a really big digitally controlled amplifier setup with the following attributes:
-Hybrid class A tweeter amps, with digitally controlled voltage rails for reduced power consumption. Possibly digital feedback and/or calibration to reduce distortion.
-multi-level switching amp for a subwoofer or 2 woofers.
-undecided on the midrange topology.
-microprocessor with multiple DACs to supply the filtered digital audio.
-a simple model of a speaker in a sealed box (with no passive crossover) still has a big LCR network with huge component values to represent the speaker's resonance and impedance-rise at high frequencies. I want to cancel out this LCR network so that the amplifier has maximum control over how the air in front of the speaker's diaphragm vibrates. This will require amplification with a negative output impedance and lots of clever tricks to make it stable.
-and it will have standard stuff like digital volume control - nothing less than relays of course, microphone feedback....
(JPK) Let's set this straight right at the start. I see this a lot. Preringing is not an artifact of FIR filters. It is an atrifact of linear phase filters of order greater than 2nd or 2nd order linear phase filters with Q greater than 0.5. FIR is just an algorithm which can be used to impliment linear phase, minimum phase or arbitrary phase filter responses. A minimum phase filter implimented using FIR has the same impulse response as a similarly implimented IIR filter.
FIR in its most basic form in noting more than convolution of the input signal with the filter's impulse response. Use any impulse response you like. If you don't want preringing, then don't use and impulse response that prerings.
paulspencer said:
My $0:02...
Aside from the usual problems associated with impedance not being flat for any real world speaker (which can generally be ameliorated by a parallel R+L+C), is what happens when the copper of the voice coil heats up. This can change the whole topology of the crossover.This, from "Design of Passive Crossovers" (http://sound.westhost.com/lr-passive.htm) says it pretty succinctly:
As an active crossover has split the signal prior to the amp, you won't get pronounced peaks near the crossover point when you've been driving your speakers at reasonable levels. On the other hand, if you use high efficiency drivers and don't drive them very hard, then this effect is not nearly so pronounced.
"I never bother with passive crossovers anymore; ever since I first built a bi-amped system some years ago and realized what a huge difference it made."
Amen, brother.
"As far as the future of sophisticated passive crossovers is concerned it's about as bright as that of the rotary dial telephone"
"Those who think that one would have to use six expensive monoblocks insted of one expensive stereo amp must not forget that multiple cheaper amps can beat more expensive single ones in active use.
Going fully active makes given drivers and amps play in a higher league. And the difference is NOT small !"
Cross fullrange drivers to stereo subs and all you need is two stereo amps...
"People talk about passive being transparant and less in the signal path, no noise, etc. but remain oblivious to the fact that the passive system is totally degrading the sound due to impedance mismatches, response irregularities and lack of damping."
"I could say so much, but suffice I went active in approx 1990 and have never looked back. however sometimes i found a combination of inductor on bass, cap resistor inductor on treble whose sound i still search for now. passive can yeild excellent results audibly but active is more precise and fullfilling I find."
Yup, yes, uh huh, yeah.
"I've been pricing here in Aus, and for $300 (+ about $100 for lead, wiring, boxes, etc) I can get a pretty good active x-over. Using Hovland caps and Hepta-litz inductors, I can make a 1st order with a couple of extras to tame inductance peaks for around $400, but this doesn't include any where to put them (except inside the cabinet)."
You can make em for way less than that - Marchand Electronics. Mine was about $100 in parts...
"I like active crossovers from a sub to satellite to take advantage of the amp's power distribution but above that passive to limit the amount of signal degradation due to active elements."
Well... maybe, but passive crossovers in the high mid sound like $*&^@#&% and do nothing more than degrade the signal.
Paulspencer, there is a lot of technical talk here but all you need to do is hear a good active system to know, and the talk becomes meaningless.
"Good speakers don't just sound good, they feel good."
NNK
Amen, brother.
"As far as the future of sophisticated passive crossovers is concerned it's about as bright as that of the rotary dial telephone"
"Those who think that one would have to use six expensive monoblocks insted of one expensive stereo amp must not forget that multiple cheaper amps can beat more expensive single ones in active use.
Going fully active makes given drivers and amps play in a higher league. And the difference is NOT small !"
Cross fullrange drivers to stereo subs and all you need is two stereo amps...
"People talk about passive being transparant and less in the signal path, no noise, etc. but remain oblivious to the fact that the passive system is totally degrading the sound due to impedance mismatches, response irregularities and lack of damping."
"I could say so much, but suffice I went active in approx 1990 and have never looked back. however sometimes i found a combination of inductor on bass, cap resistor inductor on treble whose sound i still search for now. passive can yeild excellent results audibly but active is more precise and fullfilling I find."
Yup, yes, uh huh, yeah.
"I've been pricing here in Aus, and for $300 (+ about $100 for lead, wiring, boxes, etc) I can get a pretty good active x-over. Using Hovland caps and Hepta-litz inductors, I can make a 1st order with a couple of extras to tame inductance peaks for around $400, but this doesn't include any where to put them (except inside the cabinet)."
You can make em for way less than that - Marchand Electronics. Mine was about $100 in parts...
"I like active crossovers from a sub to satellite to take advantage of the amp's power distribution but above that passive to limit the amount of signal degradation due to active elements."
Well... maybe, but passive crossovers in the high mid sound like $*&^@#&% and do nothing more than degrade the signal.
Paulspencer, there is a lot of technical talk here but all you need to do is hear a good active system to know, and the talk becomes meaningless.
"Good speakers don't just sound good, they feel good."
NNK
) with an existing speaker of yours. I bet this would already be convincing ........ And if you change to high-quality components later on ..........
I do have plans to go active, make no mistake about it! The cost isn't a huge issue, as I'm happy to use diy amps. First I need to get my first chip amp working well - hum issues to address. Even if active isn't better sounding, I think it's more realistic for me to achieve a good active crossover without spending years on it!
Now if anyone in Melbourne has an active system they would like to show me, please let me know! I'm yet to hear a good active diy system.
Further, active gives you flexibility that goes beyond passive. I have been thinking about a design for a surround speaker that would work best as active. It is essentially a hybrid monopole dipole speaker. It has a monopole TM facing the listening position and a dipole above 300 Hz. The idea is to be able to adjust the balance of the monopole and dipole parts so the sound is de-localised for the mid and above, but the bass is filled in by the monopole. It would also be able to play in monopole mode or an inbetween. I could experiment with filters to contour the response of each part. The aim would be to get a flat power response. It would be interesting to take this one step further to consider the "in-room power response" ie. the aim is to achieve a flat response at the listening position, and the room in fact could be used to eq the off axis response. I have a lot more power to play with these ideas with active.
Now if anyone in Melbourne has an active system they would like to show me, please let me know! I'm yet to hear a good active diy system.
Further, active gives you flexibility that goes beyond passive. I have been thinking about a design for a surround speaker that would work best as active. It is essentially a hybrid monopole dipole speaker. It has a monopole TM facing the listening position and a dipole above 300 Hz. The idea is to be able to adjust the balance of the monopole and dipole parts so the sound is de-localised for the mid and above, but the bass is filled in by the monopole. It would also be able to play in monopole mode or an inbetween. I could experiment with filters to contour the response of each part. The aim would be to get a flat power response. It would be interesting to take this one step further to consider the "in-room power response" ie. the aim is to achieve a flat response at the listening position, and the room in fact could be used to eq the off axis response. I have a lot more power to play with these ideas with active.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.