This would be an incorrect assumption. I listen to music at very low volumes. I use a 20w Amp running at less than half volume with in efficient speakers.
I think I'm outside of the norm in that respect.
At the show most music was playing above levels that I would listen to but nobody was cranking it by any means. The listening rooms are all hotel rooms right next to each other. If you cranked it you'd seriously **** off the guys in the listening room next to you. I think it was a very good show of normal music listening levels. I very much appreciated that aspect as I thought it would be much too loud for my tastes when I walked in
Actually, my comment was for @perrymarshall, in his previous post.
Where he wrote "Especially in high dynamic range recordings where you can turn up the volume "
Where he wrote "Especially in high dynamic range recordings where you can turn up the volume "
I said "Especially in high dynamic range recordings where you can turn up the volume and it's not super loud but you're hearing nuances...."
I know... I just didn't feel like quoting the entire book...
My point is that when you lower the noise of the system AND the static coefficient of friction of the speaker drivers you are lowering the entire noise floor... and so you extend the absolute lower loudness level for the linear behavior of the speaker. Just make sure you got a First Watt type of amps..
So that if before you had linear behavior starting at 70 db and now that starts at 60... you've just gained 10 db... if you want 30 db of linear behavior... you can have 60 to 90 on one side... or 70 to 100 on the other. So the both have the same dynamic range but the former plays softer, no need to "turn up the volume" to hear those nuances.
I think it's a designer choice... do you want the speaker to have a louder or softer operating range...
My point is that when you lower the noise of the system AND the static coefficient of friction of the speaker drivers you are lowering the entire noise floor... and so you extend the absolute lower loudness level for the linear behavior of the speaker. Just make sure you got a First Watt type of amps..
So that if before you had linear behavior starting at 70 db and now that starts at 60... you've just gained 10 db... if you want 30 db of linear behavior... you can have 60 to 90 on one side... or 70 to 100 on the other. So the both have the same dynamic range but the former plays softer, no need to "turn up the volume" to hear those nuances.
I think it's a designer choice... do you want the speaker to have a louder or softer operating range...
Yes, C'mon indeed ....!!!
Way beyond any credibility, imso.
Maybe you're right, but I have my doubts that many Axpona speakers had phase linearization..
It would require both FIR hardware, and maybe more important, considerable experience at implementing linear-phase.
That's a great experiment that I've run several times on well-regarded passive and active analog commercial speakers, as well as on top my DIYs set up with traditional IIR DSP. I like to call the technique 'global' FIR phase correction. (meaning spanning an entire speaker already setup)
I've come to the conclusion it doesn't work, because it's so hit or miss, specific to a spot/area, and/or the luck of the measurement(s) used to base the corrections.
I think reports from others who have tried this, and how generally inconclusive they are, also supports the premise that global correction doesn't hold much promise. With opinions varying much like you are saying....
Last edited:
Agree. Over the last 3 years I have built an active 3-way and a passive 3-way, both with very similar drivers. Based on that experience, I agree that when the goal is very high performance, an active DSP architecture is so enabling and powerful that it should be viewed as a requirement.
Interesting observation. I did not realize that back in 2009 the trend was already evident. It certainly is today.
True. If you go to an optician not able to operate his diagnostic gear, then you'll hardly be happy with the derived pair of glasses either. Such is life.
In real-life and hopefully robustly knowing what you are doing then: Any measurement x/y/z point will yeald a very specific result for this x/y/z location inside the specific environment. And this result will no more be representative for the neighbour measuring point. That's the dilemma of any driver or loudspeaker measurement. And also of any design derived from a measurement. And consequently of any corrective measures derived meant to improve the response. We all know that, and this is the triumph of all non-measuring subjectivists.
Furthermore, this as you name it 'global' fir phase correction in a three way system means nothing but to try positively acting on a weird mishmash resulting from three passively filtered and individually potentially (and in most cases) misbehaving drivers. Seriously? So you happily will correct on this insane bouillabaisse? Better let's consult then your honorable expert about this one:
Nota bene that with the approach of DSP-izing passive multiways we talk about a delibarate and very rough use of DSP. If I would have to design a whole new system, I would hands down go DSP right from the beginning, starting by individually processing for every single driver. Actually even very fine multichannel DAC's got this ridiculously cheap, that there is no more excuse for not going DSP in search for a more advanced loudspeaker.
Last edited:
No. That's still subjective. Now, if you could get, say, 250 people to listen to various systems blindfolded and with the experimenter having no knowledge of what's being tested, then ran the appropriate statistical analysis then you'd possibly have objective data. But such an experiment is pretty difficult to conduct in a way that doesn't bias the results.
Three buddies going to a hifi show and agreeing on something is still personal opinion. And that's fine. But personal opinion is not the same as an objective measurement or a scientific fact.
I'm reminded of Solomon Asch's studies on conformity from the 1950s:
Tom
I have been demoing DSP crossovers for many years now. The perception is changing and clearly DSP is gaining momentum and acceptance.
This reminds me of other transitions in audio, vacuum tubes vs solid state, vinyl vs CD, etc. At some point, the newer tech becomes dominant. I think nostalgia plays a part in this. Our audio hobby is a big tent, you can like what you like.
I would claim that DSP can take a good speaker and make it better. A bad speaker is still bad. I will state that DSP is more than just math, implementation - hardware and software both matter, just like everything else.
Al Clark
Danville Signal
This reminds me of other transitions in audio, vacuum tubes vs solid state, vinyl vs CD, etc. At some point, the newer tech becomes dominant. I think nostalgia plays a part in this. Our audio hobby is a big tent, you can like what you like.
I would claim that DSP can take a good speaker and make it better. A bad speaker is still bad. I will state that DSP is more than just math, implementation - hardware and software both matter, just like everything else.
Al Clark
Danville Signal
a high quality passive is much more life like than dsp, more natural sounding, no question IMO, we have tested with dsp and tho you can shape things and it makes it easier but if done right a passive all the way
Hi Al,
Thanks for coming around here to comment. My personal opinion of your products is that they are of the highest quality, and my own conversations with you at various shows in the past makes it clear that you know what you are doing with hardware DSP!
My gripe is that your demo with the Magnepans is comparing the passive crossover, which you know (have told me personally) has some weaknesses, to an optimized DSP crossover applied to an active application of the same speaker. It's not surprising that the DSP crossover is the better one! But could we not also make a DSP crossover with much less expensive components but using the same crossover filters that you are using, and then would not that also sound better than the MFG passive crossover? My opinion is that I think it would. And this is where I think you are making a false comparison that promotes your DSP products as superior but which is simply an "apples to oranges" comparison that is designed to do so. Hey, that's marketing 101. I am not arguing the point that your product does not result in a better outcome, only that there are many other ways to achieve the same outcome that cost much less than 3.8 THOUSAND DOLLARS! Over here in Europe, the cost is more like Euro 4500! I would be willing to undergo some ABX testing to test my hypothesis... you can choose the dollar amount for the compoments that I will use to make up my own DSP system, as long as you share the DSP filters that you use with me in advance.
Certainly the DSP adds cost. The dspNexus 2/8 used in our demo is very powerful and comes with DSP Concepts' Audio Weaver that many diy'ers would like to have available. It certainly is not just a Maggie crossover box. The converters are based on AKM AK5578 & AK4499EX. We do not use the AKM reference design circuits. The DSP is a fifth generation SHARC running at 1GHz. It includes a FIR accelerator.
At Axpona, we have a bedroom sized room to work with so we use the 1.7i since they are reasonable for the room constraints.
The demo compared the passive implementation with identical power amplifiers to an active DSP system. Our interest is to sell the electronics.
At CAF, we EQ'd a passive Magnepan LRS. This will be supported in a soon to be released, less expensive product. We are not trying to compete with the raw parts cost of DIY. I like DIY too. You can see that I have been a member here since 2004. I don't generally share my i.p. either. I have payroll and a factory to pay for.
I think our product might be of interest to DIY speaker builders. I don't expect the electronics diy'ers to buy them. It would replace the fun of making their own.
As to the filters, there were a lot in play. I haven't counted them up by there are many parametric EQs and LR 8th order bandsplits. They are not implemented as direct form biquads which suffer from precision issues at low frequency.
IMO, virtually all passive speakers could benefit from better EQ and usually time alignment which is not particularly practical to implement.
Al Clark
Danville Signal
At Axpona, we have a bedroom sized room to work with so we use the 1.7i since they are reasonable for the room constraints.
The demo compared the passive implementation with identical power amplifiers to an active DSP system. Our interest is to sell the electronics.
At CAF, we EQ'd a passive Magnepan LRS. This will be supported in a soon to be released, less expensive product. We are not trying to compete with the raw parts cost of DIY. I like DIY too. You can see that I have been a member here since 2004. I don't generally share my i.p. either. I have payroll and a factory to pay for.
I think our product might be of interest to DIY speaker builders. I don't expect the electronics diy'ers to buy them. It would replace the fun of making their own.
As to the filters, there were a lot in play. I haven't counted them up by there are many parametric EQs and LR 8th order bandsplits. They are not implemented as direct form biquads which suffer from precision issues at low frequency.
IMO, virtually all passive speakers could benefit from better EQ and usually time alignment which is not particularly practical to implement.
Al Clark
Danville Signal
Hey.... I got a pair of 1.7 and 12.... and lots of amps laying around. What's this about an active crossover?
@AllenB, you wrote "Wouldn't static friction be harmonically related?"
Hmmm... at first blush, I don't see how. I think it's amplitude related more than frequency related. At low playback levels, static friction ( you know when the speaker slows down, stop and start again ) will affect a big chunk of the operating range - meaning the speaker will be affected for a large percentage of its usage. Whereas at loud levels, this will not be the case. And with higher voltages, the EMF in the coil will be much larger than the force generated by the static friction.
@AllenB, you wrote "Wouldn't static friction be harmonically related?"
Hmmm... at first blush, I don't see how. I think it's amplitude related more than frequency related. At low playback levels, static friction ( you know when the speaker slows down, stop and start again ) will affect a big chunk of the operating range - meaning the speaker will be affected for a large percentage of its usage. Whereas at loud levels, this will not be the case. And with higher voltages, the EMF in the coil will be much larger than the force generated by the static friction.
DSP is plastic surgery for speakers.
That's why I was really hoping you could share the measurements of your speakers so we could correlate them to what sounds better than 90% of stuff at the show. Apologies if I missed the links somewhere in your nearly 1000 posts
Maybe if a mini DSP type box was $150 or less, 6 channels or more, and near perfect, I might try some active speakers again. Not because I think it would be better, but because it might be fun, and possibly slightly better, and allow me to use some different driver combinations that aren't compatible due to sensitivities. The product would have to have a good return policy, because after a month, I'd know if I want to keep it. I might know the first couple days. I have amps. I have drivers. I also have 40 years of accumulated x-over parts. I'm pretty sure that my room, budget, and source equipment are the limiting factors on sound quality, not my passive x-overs. If I went active again, I would like it to work in my HT system. I would want to avoid analog to digital, and back conversions, so I'd like to start with digital from a CD player, not analog out from a budget AVR, and then convert to digital, and then back to analog, and then through some interconnects, and finally reaching the amps. People keep posting active is dirt cheap. Doesn't appear that way to me, but I don't shop for it, so maybe it's cheap. I recently looked into it. One product looked promising. Turned out that I would also need a new computer in order to make it work. A better unit costs more than the speakers I build. No thanks.
Last edited:
The "temporal equalization" possible with FIR DSP that Dave Gunness pioneered had already been incorporated by 2005 in the EAW (Eastern Acoustics Works) NT series, correcting linear, time-invariant problems including time smear from the compression driver/phase plug interface, horn and cone resonances, and crossover phase linearity between adjacent bandpasses.
Those are all "big deals" in sound reinforcement, as cleaning up those problems also tend to make loudspeakers sound less fatiguing or abrasive at the usual levels required from those systems.
The same speakers effectively have a wider usable dynamic range.
Speakers processed in this way can also be used together seamlessly, as each has the same flat upper phase response, while different speakers with standard crossovers can vary by hundreds (or thousands) of degrees depending on the number of frequency splits and slopes of each crossover region.
In 2006, EAW released the UX8800 DSP processor which allowed the "Gunness Focusing" to correct prior series of speaker systems.
Mackie Designs had planned to buy EAW around the turn of the century.
Around 2003 their name had been changed, EAW was now owned by Loud Technologies, differenciating the Mackie and EAW brand names.
In 2009 Mackie released the HD Series High-Definition Powered Loudspeakers incorporating Gunness Focusing.
https://d1aeri3ty3izns.cloudfront.net/media/2/27390/download_27390_1.pdf
That complete speaker with amps and DSP cost less than the 2006 UX8800 DSP processor.
Dave's research, patents and Moore's Law (exponential growth in computing power and efficiency) had "trickled down" to low cost, great sounding speakers available from many different manufacturers.
Meanwhile, people who had been paying attention to Dave's white papers, patents and the game-changing potential afforded by relatively inexpensive FIR DSP could create DIY solutions to problems impossible to correct with passive or IIR crossover filters.
As usual, the availability of a relatively new technology does not insure it's effective integration into speaker design, nor those familiar with it doing it for a living giving away all the "tricks of the trade".
Art
A Raspberry pi + 8 channel DAC is around £100. This gives you a computer as well as an 8 channel DSP active crossover. It's not only cheaper but more capable than a lower end proprietary mini DSP system (though not the very expensive systems). You will likely want to add other bits and bobs like screens, cables, boxes, power supplies depending on how you want to incorporate it into your speakers and audio system. All in around the price of a decent 3 way passive crossover and certainly less than an audiophile one. Noise and distortion doesn't match very expensive DACs but will be inaudible in use so long as interference is avoided. You can go cheaper but it will then almost certainly become a much larger more complicated and challenging DIY project (which some want) whereas there's loads of existing hardware, software and discussion for Raspberry pi-based DIY making it an appropriate place to start.
@temp25 Why the stipulation that a DSP has to be only $150?
Let's say you're designing a 2 way passive 2nd order crossover. If you're really good at VituixCad and measurements and simulation, MAYBE you'll get away with spending as little as $75 (total 4 inductors and at least 4 capacitors plus some resistors) - and that requires you to NAIL it the on the first take and we're assuming you don't have to to swap out any other differently valued parts to get it to sound right.
Odds are you will spend hours tinkering with the physical components, adding caps and resistors in parallel to get the values you're looking for. And fumble around figuring out you didn't wire it quite the way your schematic said.
I'm a pretty experienced speaker designer, and even with simulations I still have to tinker and substitute. It really requires a box of passive crossover parts of all different values. Which I have. But a box with a generous selection of passive parts costs many hundreds of dollars.
If you're building a 3 way with low crossover frequencies, the parts cost goes up 3-5X. The inductors in the Bitches Brews (which are mostly active DSP, partly passive) cost close to $100 each.
The DSP is infinitely adjustable, instantly adjustable, and fixed cost. I say a good user friendly DSP is easily worth the $250-500.
Let's say you're designing a 2 way passive 2nd order crossover. If you're really good at VituixCad and measurements and simulation, MAYBE you'll get away with spending as little as $75 (total 4 inductors and at least 4 capacitors plus some resistors) - and that requires you to NAIL it the on the first take and we're assuming you don't have to to swap out any other differently valued parts to get it to sound right.
Odds are you will spend hours tinkering with the physical components, adding caps and resistors in parallel to get the values you're looking for. And fumble around figuring out you didn't wire it quite the way your schematic said.
I'm a pretty experienced speaker designer, and even with simulations I still have to tinker and substitute. It really requires a box of passive crossover parts of all different values. Which I have. But a box with a generous selection of passive parts costs many hundreds of dollars.
If you're building a 3 way with low crossover frequencies, the parts cost goes up 3-5X. The inductors in the Bitches Brews (which are mostly active DSP, partly passive) cost close to $100 each.
The DSP is infinitely adjustable, instantly adjustable, and fixed cost. I say a good user friendly DSP is easily worth the $250-500.
Last edited: