Active is better sounding than passive

[dfuller]

My experience is different and this needs to be differentiated more, soft and highly damped membranes like the typical polypropylene have often earlier break up modes in their passband, they just smear them to a wide region due to high damping. That's why I tend to prefer rather the sound of stiff membranes where the breakup mode is outside their passband and also additionally compensated with electric (like in above Purify paper) or mechanical filters (like KEF does).

In my experience, it's really crucial to have a really stiff membrane when the driver is doing lots of excursion - but above ~400hz the tradeoffs start to change. Metal midranges (which, yeah, very stiff) often have wild breakup modes near enough their passband that you need to start paying attention to and mitigating them - like that Purifi, for example, that 5k resonance is close enough to typical midrange passband that you either need to notch them (which complicates the crossover in a passive speaker, and produces more insertion loss - the more complex a passive filter network is, the more insertion loss there tends to be) or cross much lower or much steeper. In that case it makes a lot more sense to have something with better internal damping.

As an aside - PP is just not a great driver material IMO, it has good internal damping but it's both not stiff enough to make for a good cone driver and not flexible enough to make a good soft dome, where high internal damping is important. Mineral impregnation helps, but it's still not great and all other things being equal there are better options in every use case.

I think this is the main key to reducing IMD, reducing the octave span each driver section covers.

Simultaneously yes and no. Reducing coverage will inherently reduce IMD, this is true - but where those crosses are matter too.

 

[Thomas_A]

Yes it is - but this improvement is not something that always apply to every driver . Lilltroll at faktiskt.io did show with measurements that in some cases it can work and be an advantage .

In my opinion - its much better to use well behaved drivers without any breakups - then theres no need for notch filters. The sound will be better in both passive and active speakers .

All drivers break up more or less, it is just different Q. The key issue here is that the high impedance created by a series notch damps the motor nonlinearities such that the distortion of lower frequencies gets attenuated. 10 dB decrease is not a small achievement. You cannot do that with a normal amp hooked up directly to the driver.

 

[MAB]

To be clear - this is something that can be done in every active loudspeaker and this is not a dissadvantage with actively driven loudspeakers . Driving the tweeter with a slightly higher impedance from the amplifier in an active setup can give you all the benefits from both passive and active .

Passive loudspeakers still have the great dissadvantage in the EMF area , and this is why the bass quality often are superior in active designs .

I think you are missing the point. If you read the PuriFi paper and understood the circuit, you would realize that it isn't about "driving the tweeter with slightly higher impedance", they are demonstrating the dramatic increase in impedance at the notch frequency due to the presence of the passive filter!!!

Actually, you could implement a passive series filter in an active design, then your active speaker would have a passive filter wired inside of it and in the case demonstrated by Purif will have lower distortion.

We seem to be on a crusade here to bash passive crossover filters. Individually, passive filters have identical performance to active in most applications, even better in some, and limitations in others. The limitations are not typically because they have crap performance, it's because you can just make so many filters with passive active; some not practical in passive, and some not feasible. And if we had great performing transducers that had flat frequency response with no breakup modes, etc. there would be much less benefit for active, especially DSP. But we don't, and loudspeaker design is all about compromises, which is where active and active/DSP actually really differentiates. And it turns out some of the greatest drivers have some really nasty stuff that needs to be tamed. Which brings me to another point you have tried to make several times that is just not right.

In my opinion - its much better to use well behaved drivers without any breakups - then theres no need for notch filters. The sound will be better in both passive and active speakers

Some great speakers have drivers that need notches.
Like the Linkwitz Orion:

ORION-3

Elephant in the room: Linkwitz made a living developing speakers out of some of the most challenging drivers in applications that are not intrinsically flat!
And the Grimm Audio LS1:

LS1v2 - Grimm Audio

A good monitor combines a high degree of accuracy with untrammeled musicality. There aren’t many of those around – common wisdom has it that you can have one or the other but never both. The LS1 is the first monitor to prove conventional wisdom wrong. By integrating top quality digital signal...

www.grimmaudio.com

Both use drivers from the Seas Excel line that have massive legendary breakup modes.
For instance, Seas E00018:

I think both the Orion and the LS1 sound great, even if I personally prefer the LS1. (note, I do not know the exact Seas woofer in either design, but if you review the entire Seas Excel line midbass/woofer line, you will see all of the drivers have similar breakup modes, the E00017 is even worse, literally off the charts!) This should sink in, many of Linkwitz' speakers use drivers with massive breakup modes and irregularities requiring notches, among other filters. Convenient to cite Linkwitz when his arguments suit, then ignore his actual work in another argument... Turns out, if you look at drivers from Scan-Speak, Purifi, Morel, Accuton, etc., they all have breakup modes to varying degrees.

So, those were two really good papers that @thewas posted. I respect Linkwitz (I own two of his early Orion boards, used as active crossovers for my own projects), and I also respect Elliot Sound (I built a preamp, active crossover, phono preamp, etc. with his kit). But the 'amplifier to driver control' argument is non-physical and wrong per Toole, and I am siding with Toole on that.

Gosh, there are just so many reasons to do active DSP crossovers, many not even mentioned in this thread (I'll probably never mess around with analog active again, but I am going to try a passive series notch on my Seas E00018 drivers!)). But it's not because of this mythical 'driver control' idea, It's not because an individual passive filter has different performance than the same filter implemented active (except select cases like the Purif example). It's because a designer can implement so many filters trivially, and has access to filters that are not practical or do not exist in the passive domain. And these can all be done in software in moments where a passive crossover or even analog-active in most cases takes orders of magnitude longer time to implement. I think the Active vs. Passive vs. DSP subject is not nearly as black and white as the poll that started this thread.

I edited passive to active in the second paragraph. My bad...

 

[thewas]

In my experience, it's really crucial to have a really stiff membrane when the driver is doing lots of excursion - but above ~400hz the tradeoffs start to change. Metal midranges (which, yeah, very stiff) often have wild breakup modes near enough their passband that you need to start paying attention to and mitigating them - like that Purifi, for example, that 5k resonance is close enough to typical midrange passband that you either need to notch them (which complicates the crossover in a passive speaker, and produces more insertion loss - the more complex a passive filter network is, the more insertion loss there tends to be) or cross much lower or much steeper. In that case it makes a lot more sense to have something with better internal damping.

I don't agree there as shown in the above linked Purify paper the notch filter can even reduce the total distortion, or you can even do a mechanical filter for that break up like KEF does in the recent years. This doesn't mean that I don't like damping, but damping unfortunately brings down the modal frequency and just smears it more into the passband.

 

[Tangband]

All drivers break up more or less, it is just different Q. The key issue here is that the high impedance created by a series notch damps the motor nonlinearities such that the distortion of lower frequencies gets attenuated. 10 dB decrease is not a small achievement. You cannot do that with a normal amp hooked up directly to the driver.

Yet - measurements Amirm has done on Neumann and Genelec shows outstanding low distortion and they are active speakers with their midbass units directcoupled to the amplifier .

All drivers are different - some drivers have good motors with low inductance and dont need those passive tricks.
The same is true with tweeters, some of them has less dynamic compression ( often metallic tweeters, but not always ) and dont need a higher impedance power amplifier.

My own experience of this higher impedance amplification shows no improvement in sound quality , and in one case worse sound. This is not a miracle cure.

 

[thewas]

So, those were two really good papers that @thewas posted. I respect Linkwitz (I own two of his early Orion boards, used as active crossovers for my own projects), and I also respect Elliot Sound (I built a preamp, active crossover, phono preamp, etc. with his kit). But the 'amplifier to driver control' argument is non-physical and wrong per Toole, and I am siding with Toole on that.

Gosh, there are just so many reasons to do active DSP crossovers, many not even mentioned in this thread (I'll probably never mess around with analog active again, but I am going to try a passive series notch on my Seas E00018 drivers!)). But it's not because of this mythical 'driver control' idea, It's not because an individual passive filter has different performance than the same filter implemented active (except select cases like the Purif example). It's because a designer can implement so many filters trivially, and has access to filters that are not practical or do not exist in the passive domain. And these can all be done in software in moments where a passive crossover or even analog-active in most cases takes orders of magnitude longer time to implement. I think the Active vs. Passive vs. DSP subject is not nearly as black and white as the poll that started this thread.

Great summary, I don't think anyone questions the advantages of a DSP crossover, only the oversimplifying to a "sounds better" which is usually based on wrong reasoning which sounds logical on a first impression (like "direct driving must be better") but isn't true.

 

[thewas]

Yet - measurements Amirm has done on Neumann and Genelec shows outstanding low distortion and they are active speakers with their midbass units directcoupled to the amplifier .

But the low distortions are mainly resulting from the excellent drivers they use not the active driving as it was initially claimed.

All drivers are different - some drivers have good motors with low inductance and dont need those passive tricks.

Do you want to say that Purify has not a good motor?

My own experience of this higher impedance amplification shows no improvement in sound quality , and in one case worse sound. This is not a miracle cure.

Sorry, but as shown above such amateur anecdotal experiences have similar objective value like non blinded listeining impressions of different amplifiers or cables.

 

[Thomas_A]

Yet - measurements Amirm has done on Neumann and Genelec shows outstanding low distortion and they are active speakers with their midbass units directcoupled to the amplifier .

All drivers are different - some drivers have good motors with low inductance and dont need those passive tricks.
The same is true with tweeters, some of them has less dynamic compression ( often metallic tweeters, but not always ) and dont need a higher impedance amplifier power.

My own experience of this higher impedance amplification shows no improvement in sound quality , and in one case worse sound. This is not a miracle cure.

Yes they have good numbers. But even given that, it may still be possible to reduce those odd order distortion even further with some clever tricks, which could be implemented in an active amplifier, but then the output impedance needs to be adjusted. You could even go current drive. Lower distortion is better, not worse.

 

[voodooless]

Why not go real active with motion feedback All else is scaring the barrel.

 

[Tangband]

I think you are missing the point. If you read the PuriFi paper and understood the circuit, you would realize that it isn't about "driving the tweeter with slightly higher impedance", they are demonstrating the dramatic increase in impedance at the notch frequency due to the presence of the passive filter!!!

View attachment 225323
Actually, you could implement a passive series filter in an active design, then your active speaker would have a passive filter wired inside of it and in the case demonstrated by Purif will have lower distortion.

We seem to be on a crusade here to bash passive crossover filters. Individually, passive filters have identical performance to active in most applications, even better in some, and limitations in others. The limitations are not typically because they have crap performance, it's because you can just make so many filters with passive; some not practical in passive, and some not feasible. And if we had great performing transducers that had flat frequency response with no breakup modes, etc. there would be much less benefit for active, especially DSP. But we don't, and loudspeaker design is all about compromises, which is where active and active/DSP actually really differentiates. And it turns out some of the greatest drivers have some really nasty stuff that needs to be tamed. Which brings me to another point you have tried to make several times that is just not right.

Some great speakers have drivers that need notches.
Like the Linkwitz Orion:

ORION-3

Elephant in the room: Linkwitz made a living developing speakers out of some of the most challenging drivers in applications that are not intrinsically flat!
And the Grimm Audio LS1:

LS1v2 - Grimm Audio

A good monitor combines a high degree of accuracy with untrammeled musicality. There aren’t many of those around – common wisdom has it that you can have one or the other but never both. The LS1 is the first monitor to prove conventional wisdom wrong. By integrating top quality digital signal...

www.grimmaudio.com

Both use drivers from the Seas Excel line that have massive legendary breakup modes.
For instance, Seas E00018:

I think both the Orion and the LS1 sound great, even if I personally prefer the LS1. (note, I do not know the exact Seas woofer in either design, but if you review the entire Seas Excel line midbass/woofer line, you will see all of the drivers have similar breakup modes, the E00017 is even worse, literally off the charts!) This should sink in, many of Linkwitz' speakers use drivers with massive breakup modes and irregularities requiring notches, among other filters. Convenient to cite Linkwitz when his arguments suit, then ignore his actual work in another argument... Turns out, if you look at drivers from Scan-Speak, Purifi, Morel, Accuton, etc., they all have breakup modes to varying degrees.

So, those were two really good papers that @thewas posted. I respect Linkwitz (I own two of his early Orion boards, used as active crossovers for my own projects), and I also respect Elliot Sound (I built a preamp, active crossover, phono preamp, etc. with his kit). But the 'amplifier to driver control' argument is non-physical and wrong per Toole, and I am siding with Toole on that.

Gosh, there are just so many reasons to do active DSP crossovers, many not even mentioned in this thread (I'll probably never mess around with analog active again, but I am going to try a passive series notch on my Seas E00018 drivers!)). But it's not because of this mythical 'driver control' idea, It's not because an individual passive filter has different performance than the same filter implemented active (except select cases like the Purif example). It's because a designer can implement so many filters trivially, and has access to filters that are not practical or do not exist in the passive domain. And these can all be done in software in moments where a passive crossover or even analog-active in most cases takes orders of magnitude longer time to implement. I think the Active vs. Passive vs. DSP subject is not nearly as black and white as the poll that started this thread.

Thanks for this answer. I think you are right about going dsp when using active crossovers. One example that couldnt be done otherwise than with dsp is the Kef ls60 - a passive speakers with those drivers in that particular cabinet would have sounded awful.

 

[thewas]

Why not go real active with motion feedback All else is scaring the barrel.

My 1979 Philips active 3-way loudspeakers even do that for the woofers which measure and sound really good, especially for that time. I remember reading though that for mid and tweeters the advantages are too low compared to the disadvantages of the increased complexity.

 

[Thomas_A]

There are also other possibilities with active:

ACE-bass amplifier design

I am very much interested in the ACE-bass principle, invented by K.E. Stahl back in the seventies. I’ve found the patent of the principle and did a lot of searching on the internet. In this tread on DIYAUDIO I found a circuit of a prototype of Bjorno’s ACE-bass amp design. I am planning to...

www.diyaudio.com

 

[dualazmak]

Hello OP @Tangband and friends,

Even though I assume my story in this post is completely independent and apart from your current discussions, I would like to share the following with you, and your frank comments and/or impressions would be highly appreciated.

In the system diagram of my fully active setup shared in my above post #293, some of you are already aware of the three "tuning" 22 ohm "parallel" (not series) resistors in SP high level lines for Be-midrange, Be-tweeter and super-tweeter;

Incorporation of these three 22 ohm resistors (Jantzen Audio's 22.00 Ohm 10W Audio-Grade Non-Inductive Wirewound SuperRes Resistor) is based on my fully empirical, very careful, subjective experiment and observation (listening test) during the process of complete elimination of attenuators in the original LCR passive network. You would please find the details in my post here and here on my project thread.

I myself and even my wife who is always a good naive accompany in our audio listening sessions were (are) subjectively well aware of the fully reproducible preferable contribution of these tuning parallel 22 ohm resistors in better total sound quality, slightly better sound cleanliness and so called "sonority".

I well understand that these parallel resistors just always give a little bit (22 ohm) of extra workload to the amplifiers to work in slightly higher energy consumption, i.e. slightly above zero workload. I may discuss the preferable effect, therefore, in terms of amplifier's "zero crossing distortion", but I well aware of my HiFi excellent integrated amplifiers (Accuphase E-460, Sony TA-A1ES, Yamaha A-S301 in this setting) have no or negligible "zero crossing distortions".

After the introduction of these "parallel" 22 ohm resistors in my setup, I discussed this topic several times, at several places, in ASR Forum, but almost no people gave me clear-cut insights or understandings on the theoretical/technical background for the preferable effect.

Only our nice discussion with @DualTriode in our three consecutive posts here, here and here gave me some "relief feeling" since he wrote there;
"What this does is operates the amplifier at a higher power output where the amplifier operates at a higher SNR. Also the amplifier operates into a flatter, more resistive load."

In his post here, he also suggested me to try even additional "series" 11 ohm resistor together with my 22 ohm "parallel" ones, but I really would like to eliminate any of the "series" resistors in my fully active setup, as you may well understand. I would like to "directly" drive SP drivers by amplifiers except for the protection capacitors.

In any way, these three "parallel" 22 ohm resistors always give a kind of "very slight but magical preferable contribution" to the total excellent sound quality, at least in my setup with really efficient and sensitive, still amazingly excellent, SP drivers.

Similar to what you are currently discussing, I assume this preferable effect of "parallel" resistors would not be the general issue in active setup, but would be quite unique to my specific SP drivers in the specific cabinet.

For your kind notice, I do not have any of the sophisticated high-end measurement gears for further "investigation" of this issue, and at least I myself have no further motivation to objectively prove the effect; I just would like to enjoy the established improved preferable total sound quality.

 

[gnarly]

Very nice to see your configuration diagram, it is beautifully designed and easy to be understood; as I said "one picture/diagram is worth a 1000 words"!

BTW, can you, and how do you, implement precision (0.1 msec precision) time alignments over all of your SP drivers? If you actually do so, how can you measure, validate prove such time alignment in room air sound at your listening position using measurement microphone?

And very nice to see your diagram as well ! Easy to follow with more detail about what the actual drivers and xovers are, nice.
It's interesting that we both have real time volume controls with level monitoring on individual drivers for tone control. I love the technique. Seldom seen....



Yes, 0.1ms delays are easily implemented, as delays are adjustable in 0.02ms steps. (1 sample @ 48kHz)
I think 0.1ms of precision is a very good achievement with IIR crossovers.
With linear phase FIR xovers it's pretty easy to tighten timing up to 0.02ms alignment (other than for sub to main).

Timing validation is straightforward with either Smaart, ARTA, or REW, using the tried and true method of achieving phase traces overlay between drivers.
Smaart is the easiest to use, because it has real time phase tracking.
I've also played with wavelets / tone bursts, but only for sub to main timing, where phase traces are tougher to capture.

If you ever try linear phase xovers, you're not gonna believe how much easier it is to time align two flat level lines together, than align two sloping curves !
(This is true for wavelets / tone busts too...they have less assymetric deformity around the peak with lin phase xovers)
I think this comparative ease of alignment vs IIR is a major reason why 0.02ms is achievable.

Another cool thing about linear phase xovers, is if the xovers are also steep they don't alter either the phase or the acoustic xover frequency, as relative volume levels are changed between drivers. So using individual volume controls like we do takes on a new dimension of cleanliness. (i use 96dB/oct LRs)

 

[MAB]

My own experience of this higher impedance amplification shows no improvement in sound quality , and in one case worse sound. This is not a miracle cure.

Problem is, the Purifi case is not a miracle, it's real with data that is hard to refute without better data. And it's not even remotely similar to "higher impedance amplification" (you're repeating this and it is super-wrong, I tried to point this out earlier...)
The miracle is the driver-control concept, among other things.

 

[dualazmak]

Hello @gnarly,

I really thank you very much for your further insights and suggestions which greatly supplement our nice discussion the other day around there in the thread of "Time aligned speakers - do they make sense?".

Hopefully, if my daily tight schedule would allow me, I would like to carefully and intensively learn and consider possible implementation of linear phase Xovers in my fully active setup.

 

[gnarly]

The question of this thread seems to me to be missing the option of hybrid active/passive which I believe offers the overall best option.

While active offers big advantages in timing alignment, eq, crossover order & other areas, I believe it’s been demonstrated that best IM distortion needs the use of a series notch that raises impedance at driver breakup…something a dsp cannot do.

A hybrid may be the best of all solutions.....i have no clue, as i know very little about passives...but i do know fixing a problem at its source is always better than trying to correct the effects of the problem.

Maybe that's what the passive notch filter does best...for passives or actives as well....
for any design where the xover is low order enough to let out-of-band peaks in.

My active solution for such out-of-band peaks, is to not let them in to begin with...don't excite them with any appreciable level of drive signal.
I use steep complementary linear phase xovers, and out-of-band peaks just disappear from measured response.

Personally, I totally shun steep active IIR xovers (or passive if such were pragmatically doable).
Linear phase xovers via FIR via DSP is the only way steep xovers work imo/ime.
I find them a true blessing in achieving exemplary acoustic measurements on-axis, while tightening up off-axis measurements' consistency (and making sound I really like )

 

[gnarly]

Hello @gnarly,

I really thank you very much for your further insights and suggestions which greatly supplement our nice discussion the other day around there in the thread of "Time aligned speakers - do they make sense?".

Hopefully, if my daily tight schedule would allow me, I would like to carefully and intensively learn and consider possible implementation of linear phase Xovers in my fully active setup.

Excellent, dualazmak,
Thank you for the kind words !

 

[gnarly]

I agree. I implement this exact same method with my active 4 way setup.

what’s interesting from my perspective is to have several presets: 4 way active with brick wall filters, 3 way in full range as my main speakers go very low, stereo subs or mono summed bass. It takes my Lake DLP about 5 seconds to change between each preset so I can hear the difference quite quickly

Very nice. I've always admired Lake's products and capabilities. Although reading their manuals drove me to tears a few times Lol.

I like to do the same thing, such as switch between xover topologies like shallow IIR vs steep linear phase FIR, mono/stereo the subs, etc.
I guess you saw the Q-Sys presents I have for trying various center matrix strategies, as well as single/dual/tri mono etc ...

A cool thing about Q-Sys is that switching between presets is both instant and silent...seamless so far for everything i've tried, even for swapping multiple 16k FIR files.
My girl friend can be on any PC in the house, switching presets for me to try blind testing when I wonder if I'm jerking myself over what i think i'm hearing...haha
She's gotten good at not switching anything intermittently, to try to screw me up, bless her devious heart !

 

[thorvat]

With linear phase FIR xovers it's pretty easy to tighten timing up to 0.02ms alignment (other than for sub to main).

If you ever try linear phase xovers, you're not gonna believe how much easier it is to time align two flat level lines together, than align two sloping curves !

OMG not again with linear phase XOs..