Which brings us back to this question and I quote> (Speaker A measures well but is not pleasing to listen to. Speaker B does not measure as well but is more pleasing to listen to. Which speaker is the more successful design?)
Anyone can have fun listening on a portable mac speakers, including me. or even on on a cell phone. If you measure the response it is as horrible a speaker can be, but our ears and brain has the ability to filter and hear what we want to ear.
Even with no bass or little, if the response around 100hz is acceptable or even 200 hz, our brain can cover up the loss of bass from the upper harmonics!! this is true.
So, anyone can build 'good sounding' speakers. But in the long term you need good sounding drivers, a good XO and a good box, and a good amp and music source if you want to have a better sound.
Every single piece of capacitor or inductor has a very great effect on sound, phase and impedance.
Today we have calibrated microphones, simulation softwares, and XO computer assisted simulator that will find the best XO options.
Is this replacing ears? no, some less perfect curves or xo points sound better, some curves sound better, unequal, non-mathematically perfect xo works better.
Are computer simulation accurate? No almost rubbish.
But our mind is limited, computers will help finding new types of circuits and open new ways to your design of the XO.
I am running computer simulation based on my drivers in their enclosures with measured Z curves and phase and response curves.
This is to me the only way, after you measure what the computer suggest at least 3 designs then you measure, then you ear, compare a to b to c, then make changes, measure again, put new stuff in the computer simulation etc until you know you did your job right.
I find especially useful tone generators and multi-tone generators, you can find unwanted XO resonances, and unwanted peaks and dips by ear, I like to go through the sweep curve manually one frequency after the other and note every dip and bump or resonances and give them a rating of annoyance level from 1-5, then I adjust the XO accordingly by entering modified target response curves in simulators.
To me, to be able to build a 3 order 3 way , with Z , notch and band pass filters, all from ear is nothing else than a miracle. And I don't believe in miracles for a good reason.
Even with no bass or little, if the response around 100hz is acceptable or even 200 hz, our brain can cover up the loss of bass from the upper harmonics!! this is true.
So, anyone can build 'good sounding' speakers. But in the long term you need good sounding drivers, a good XO and a good box, and a good amp and music source if you want to have a better sound.
Every single piece of capacitor or inductor has a very great effect on sound, phase and impedance.
Today we have calibrated microphones, simulation softwares, and XO computer assisted simulator that will find the best XO options.
Is this replacing ears? no, some less perfect curves or xo points sound better, some curves sound better, unequal, non-mathematically perfect xo works better.
Are computer simulation accurate? No almost rubbish.
But our mind is limited, computers will help finding new types of circuits and open new ways to your design of the XO.
I am running computer simulation based on my drivers in their enclosures with measured Z curves and phase and response curves.
This is to me the only way, after you measure what the computer suggest at least 3 designs then you measure, then you ear, compare a to b to c, then make changes, measure again, put new stuff in the computer simulation etc until you know you did your job right.
I find especially useful tone generators and multi-tone generators, you can find unwanted XO resonances, and unwanted peaks and dips by ear, I like to go through the sweep curve manually one frequency after the other and note every dip and bump or resonances and give them a rating of annoyance level from 1-5, then I adjust the XO accordingly by entering modified target response curves in simulators.
To me, to be able to build a 3 order 3 way , with Z , notch and band pass filters, all from ear is nothing else than a miracle. And I don't believe in miracles for a good reason.
Last edited:
The DBX Driverack PA and three digit (380, 480 etc.) use IIR filters, and are not capable of FIR implementation.
FIR filters do have an affect on phase, which can be used to compensate for the phase response deviations in loudspeakers. This process requires more signal delay the lower in frequency phase response is desired to be maintained "flat", not a problem in playback systems, but the latency to flatten phase below 500 Hz (or so) can be a problem in live use.
Thanks, good to know.....I know one of the features of the DriveRack is real time eq
would it also include phase correction in that mode?
If it did that would basically make it idiot proof.....that’d certainly appeal to me!
No, an automatic RTA EQ does not consider phase at all, if a phase variance causes an amplitude dip at the mic location, it will counteract with a boost of the same amplitude. This can be "very interesting", especially with the DBX Driverack PA+, which due to a design error, inverts the polarity of the mid output.
An idiot using the DBX auto EQ can make the system sound far worse than if the 1/3 octave filters were simply left "flat".
Absolutely not
But that experience is hard won (My first probably 25 years of speaker building was also by ear, and with shock horror , off the shelf crossovers!!!) I also got similar comments, and in fact everyone asked me to bring my speakers to parties because they were so much better than anything any of my friends had. In the end though they couldn't hold a candle to the speakers I have now built, which I designed using measurements and sims. I've no doubt there are people who can listen to a speaker and pick out the flaws, and know just what they need to do to fix it, experience is an amazing thing, and to someone without it, it can seem like magic!
Someone just starting out though doesn't have that experience, and measurements and sims will likely give them a massive head start
Your welcome! I really need to get the next installment done (it is 95%) but I've had other priorities lately.
Tony.
Seems as though that one might not be such a good choice after all.....any suggestions? I liked a lot of the features of the dbx.....is there anything comparable that does things correctly at or under $1k ?
The mini dsp looks like a good learning tool but I was hoping to avoid having to replace it later.
Thanks for the help,
Bob
Yes sir.
Although i would amend to say individual drivers are IIR filters, except when non pistonic behavior occurs.
And that loudspeakers, within the range(s) of crossover summation, are no longer IIR
Hi Bob, just depends on what you're trying to do....how far you're trying to go.
I think understanding IIR, in all traditional loudspeaker processors, is very helpful and necessary, before moving on to FIR.
imnsho, I think FIR is the name of the game today....that is, once we have a good grasp on what can and can't be fixed, and then how to use FIR appropriately.
But apart from miniDSP or something similar, or perhaps using a PC, ...i haven't seen a good FIR solution for under $1000.
Absolutely not
I've no doubt there are people who can listen to a speaker and pick out the flaws, and know just what they need to do to fix it, experience is an amazing thing, and to someone without it, it can seem like magic!
Someone just starting out though doesn't have that experience, and measurements and sims will likely give them a massive head start.
Tony.
Your one of the few I’ve come across that really gets where I’m coming from....
I am now to the point where I’m going to need to either take it to the next level and get setup with measurement tools and.... (cough,cough,choke) computer sim programming, or just be happy where I’m at and move on to active with RTA.
I do find my new Omni mic interesting, and most everything I’ve measured is right where I thought it was by ear.....so that’s kindly encouraging.
Another thing I’ve caught a lot of flack about is measuring at the listening position......I know for a commercial speaker or one not sure of placement designing anechoic flat is the only way, but for a diy that knows where the speaker is going to live I see a lot of benefit from measuring at LP.
Phase is a wall I keep having to hurdle.....I’m sure life would be easier to be able to sim that and I believe my current ‘audio tool’ app for the omnimic can measure it......just have to figure it out.
I do certainly understand the need for someone starting out to be led down the best path possible, but the flat out NO.....it’s impossible to hear any of these things, let alone make a intelligent call based on what your hearing is just plain wrong.
To me it relates to how they pass kids through drivers education class whilst never driving a car.....with simulators! Scary thought
Thanks again,
Bob
And that would be a solution to a problem that doesn't exist.
You can't hear phase and excepting possibly concentric drivers that are perfectly time-aligned, and once the signal gets to the drivers' terminals, all phase disparity is scrambled on its way to your ear and twice as scrambled on its way to both ears, and 16 times as scrambled comparing both stereo speakers.
B.
Last edited:
Thanks Mark,
What I’m wanting to do is three way two channel.....I’m setting up a new 2.1 (2.2) that will probably end up being my last total upgrade.
Speakers are 10” mtm (horn cd) with two lab 15’s , everything kindly leaning towards the use of pro techniques and equipment......a personal PA if you will.
Setting them up both passive and active for comparison with an idea of maybe ending up as a hybrid.
Bob
Last edited:
Might not be able to define it......but I can certainly tell when something is out of phase and manipulate it until it’s acceptable. Now whether or not it would measure the way it’s ‘supposed to’ ....idk.
Hi Bob, I tried measurements at my listening position when doing the active crossover from my MTM's to my subs. That didn't work out well. Probably largely due to the fact the crossover frequency was 270 Hz and I had to do massive smoothing to get something useful.
Outside measurements were much more successful, with respect to getting the integration working well. However the MTM's which have also been done using outside quasi anechoic measurements, do have a hump (I can't remember exactly where) somewhere between I think 500 and 800 Hz in room (they measure flat outside). So I understand the reasoning for measuring at the listening position. In my case I'm pretty sure it is a bit too much baffle step compensation, but something I've lived with so far. I'll get around to a revised crossover one of these days
Tony.
Outside measurements were much more successful, with respect to getting the integration working well. However the MTM's which have also been done using outside quasi anechoic measurements, do have a hump (I can't remember exactly where) somewhere between I think 500 and 800 Hz in room (they measure flat outside). So I understand the reasoning for measuring at the listening position. In my case I'm pretty sure it is a bit too much baffle step compensation, but something I've lived with so far. I'll get around to a revised crossover one of these days
Tony.
That's the route I've gone...personal PA...since setting aside an affinity for electrostats about 20 years ago.
Good point source PA has become more hi-fi to me, than any consumer gear I've been able to hear. This morning has me debating on whether to go to InfoComm next month to hear the latest prosound stuff.
The technical advances in PA gear are astounding... much of it is due to embedded active processing, often with FIR.
FIR has more benefits than many people realize because they often get hung up on arguments about phase, missing its other advantages.
What kind of I/O will you be using? If balanced, the tried and true DCX2496 is a decent low cost IIR processor.
I wrote myself crossover simulation software, and for the next 20 years (this began before I found the interwebs), I struggled to get anywhere near the sound I was aiming for (yes, it was accurate).
It seems easy enough to hear and describe a problem but not so easy to discover the reason. Then there are those that won't refrain from drawing conclusions and there is so much misinformation about.
Why simple crossovers, tuned by measurement, don't work either (without an open mind
).