Simulate, build, listen then measure (except for impedance measurement, which must be done before initial XO design)- I use well-behaved filters and drivers, so high-Q resonances aren't really an issue, and I can identify most problems above a dB or two from listening. The measurement is then used as validation and to more precisely shift tonal balance, etc.
I have done both.
After doing a listening comparison between a dipole and an omni speaker I discarded omnis and have gone pure dipole.
Within the realm of dipoles I've been measuring a lot to try to get the perfect dipole pattern. I did try different tweeters. I had one setup with waveguided dome tweeters that measured extremely good but something just didn't sound right, they didn't sound as nice as my cone tweeters so I've continued down along that path.
In the end I've learnt what many others here already preach, to not overcomplicate or to fix too much. Simplicity is worth pursuing, especially when it is so easy to overdo it with for example DSP in an active setup.
After doing a listening comparison between a dipole and an omni speaker I discarded omnis and have gone pure dipole.
Within the realm of dipoles I've been measuring a lot to try to get the perfect dipole pattern. I did try different tweeters. I had one setup with waveguided dome tweeters that measured extremely good but something just didn't sound right, they didn't sound as nice as my cone tweeters so I've continued down along that path.
In the end I've learnt what many others here already preach, to not overcomplicate or to fix too much. Simplicity is worth pursuing, especially when it is so easy to overdo it with for example DSP in an active setup.
Does this thread belongs in The Lounge forum?
Now and then over the years at DIYaudio, we get the techies lambasting the golden-ear crowd. Maybe the "goal posts" for being golden-eared have moved.
But I would add a further complication. Measurements have their own peculiarities and easy to do it badly too. Even if done carefully, still rather cumbersome to define just how the measurements were done... smoothing, window, single speaker or dual mono, mic pick-up pattern.....
B.
Now and then over the years at DIYaudio, we get the techies lambasting the golden-ear crowd. Maybe the "goal posts" for being golden-eared have moved.
But I would add a further complication. Measurements have their own peculiarities and easy to do it badly too. Even if done carefully, still rather cumbersome to define just how the measurements were done... smoothing, window, single speaker or dual mono, mic pick-up pattern.....
B.
Last edited:
Such things can be learned, or at least that's the claim for this product. I haven't got it because I'm not really sure I want to hear 'this' well:
Moulton Laboratories :: Golden Ears
Not quite right to equate the unbelievable advertising puffery of an unknown company or the abilities some have posted anonymously above to the published research of Toole.
B.
Last edited:
At least a quick FR measurement is needed to make sure you have built the crossovers correctly and connected all the drivers in proper phase (and to the correct crossover section) and to make sure the left and right speakers match. It is surprising how a totally screwed up crossover can sound OK depending on the source material. I also check that the crossover frequencies and slopes match the simulation.
I'm into 4-way active designs where each section is first individually measured and adjusted for flat magnitude and phase over its intended passband.
Once that's done, linear phase crossovers are put in place.
Which leaves tying them all together, with proper polarity, timing, and levels.
Relative section levels can be done by ear.
But the only real clue that polarity or timing isn't correct, is a sound that says "something just ain't right".
IME, it's nearly always from mismatched phase alignments, which then induce magnitude variations.
So for me, it's kinda not even worth the trouble to listen, until I've achieved a smooth phase trace and know that timings and polarities are correct.
This also let's me know that magnitude is as smooth as it's gonna get without further work.
A kinda cool aside IMO (and only a recent discovery) is that when I have a smooth phase trace near zero degrees tying all the sections together,
then the relative levels between sections can be changed without changing the phase trace
Once that's done, linear phase crossovers are put in place.
Which leaves tying them all together, with proper polarity, timing, and levels.
Relative section levels can be done by ear.
But the only real clue that polarity or timing isn't correct, is a sound that says "something just ain't right".
IME, it's nearly always from mismatched phase alignments, which then induce magnitude variations.
So for me, it's kinda not even worth the trouble to listen, until I've achieved a smooth phase trace and know that timings and polarities are correct.
This also let's me know that magnitude is as smooth as it's gonna get without further work.
A kinda cool aside IMO (and only a recent discovery) is that when I have a smooth phase trace near zero degrees tying all the sections together,
then the relative levels between sections can be changed without changing the phase trace
Measure first. When you get it flat, listen. The problem with using your ears is oftentimes you will get something that appeals in the short term which just drives you away long term. Measurements invariably find the peaks etc. and crap that your ears adjusted to. Adjusting to constantly improve the measurements seems to provide the most long term pleasing speaker for my ears.
LOL. Thanks for the back-pedal, Ben - but I think your gears are stuck.
Or maybe you're just stuck in time.The pro world is far, far from your ideas of 5 band EQ and school gyms. We've used 31 band EQ for decades, and PEQ on the strips. Now that almost all consoles are digital, there is a mighty lot of DSP available right in the mixer. Graphic EQ, parametric EQ, notch filters, shelves, delays, spectrum analyzers, etc. And pro amplifiers have powerful built in DSP to tune the system that you can run from your phone, tablet or laptop, as it's all networked. Measurement is very important, and prediction software is used on all large systems. You might find DANTE interesting, as it has become the de-facto method of audio transport in the pro word.
Measurement and listening skills are used in conjunction with DSP every day in pro audio, in which order they are used isn't really important.
My MacBook mic is very close to my calibrated mic when then mic is in the same location.
Also, the kind of shortcomings the laptop mic has are likely to be unimportant except maybe the lowest freq band shortcomings. If you're getting polarity issues in the crossover region (and that isn't just a question of getting the red wire where it should be), the laptop mic will tell you.*
But if you understand research methods, it doesn't matter. By ear, you've settled on an OK "house curve". So hereafter, you are comparing the measurement of THAT curve, no matter what small error the laptop mic is making, to the original. Begging to differ with the golden-eared posters above, even a modest mic will reveal vastly more about tweeter behaviour in the 7-14 kHz octave than their ears**.
B.
*or not, since phase and polarity in your music room are nothing like the textbook or wannabes-engineer model.
**esp those PA guys who have been listening to hearing-destroying rock concerts much too long
Last edited:
I am only starting working with microcap 10 an when simulating you listen, then optimisze, listen, simulate and listen.
The file you listen to is a wave file of any complexity. What I now have available is a tool that can give me harmonic distribution which I can listen too, optimize and re listen and it is different than before.
Once satisfied that both the amplifier characteristics are both objectively as wel as subjectively on target I will commit it to a PCB, then one can measure to see if it corresponds to simulation and listen and compare recordings from bot simulated and actual hardware.
Quite fantastic, but I am still very much learning and it is slowly unfolding. Although you cannot create a high power sound card in your PC, you can creat a device under test that performs as it would in the real world.
The file you listen to is a wave file of any complexity. What I now have available is a tool that can give me harmonic distribution which I can listen too, optimize and re listen and it is different than before.
Once satisfied that both the amplifier characteristics are both objectively as wel as subjectively on target I will commit it to a PCB, then one can measure to see if it corresponds to simulation and listen and compare recordings from bot simulated and actual hardware.
Quite fantastic, but I am still very much learning and it is slowly unfolding. Although you cannot create a high power sound card in your PC, you can creat a device under test that performs as it would in the real world.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.