Tweet
Hello all,
Someone recently pointed out I've been complaining for 5 years about "bright" speakers. He's right. I have. And I've looked everywhere for an answer, without finding anything.
At least I'm sort of stubborn. After years and years of measuring, designing, complaining, experimenting, etc., I think I've found a couple of guidelines regarding all those terms. I hope this is useful to someone. Suggestions & corrections are welcome - I'm still learning, and most of these concepts are assumptions on my part.
1) What is "harshness", "brightness", etc.?
I've been looking for an explanation myself. After looking back at the designs which I've built and thought of as "bright", I think that brightness has a lot to do with the sound power radiated on frequencies around 1 - 7 KHz and their relationship to the sound power above and below them.. It manifests itself first as extra detail, but it tires you very fast. In my case, it can make my ears ring.
2) What do I mean?
Whenever I build speakers, I usually get the following:
Two way speaker -> usually good results, no brightness.
3 - 3.5 way speaker -> brightness.
So, after a while, I decided I needed to start following some sort of method, to try and find out what was wrong. First, I did lots of experiments.
a. Lower the mids and highs(from 1 KHz up) -> no luck.
b. Raise the bass (from 100 Hz down) -> no luck.
c. Create a "house curve" -> no luck.
d. "Tilt down" the speaker's response -> no luck.
I hadn't noticed I never did anything to, say, 250 Hz to 1 KHz.
3) How I discovered my issue.
After redesigning dozens of times my crossovers, and blaming everything from the source material to the drivers' material, this is what I did.
a. I took a 2-way speaker I built that I love and took MLS measurements and RTAs at my seating position.
b. I took a commercial 2-way speaker that I really like (the Energy C-3) and which measures pretty good, and took the same measurements.
c. I overlaid both measurements on the measurements of a big 3 way I built, which has the "piercing midrange" voicing.
At first, I didn't notice anything out of the ordinary.... except for the fact that the lower midrange / upper bass region wasn't summing up as it should.
Upon closer inspection, I noticed a 2 ~ 3 dB difference from 300 Hz to 1.5 KHz between my 3 way and any of the 2-ways. This difference made the 3 way a bit depressed in that area. It was so smooth, I didn't really notice it while doing the sims. And besides, since the midrange / woofer curves followed a 2nd order curve, I thought I was doing things right.
Looking back on the 3 three ways I've so far "fixed", when I originally took their MLS measurements, I did so by raising them around 1.5 M above floor level, so I didn't even consider floor bounce. I re-took every measurement at floor level, and I got slightly different FR plots for both the woofers and the midranges. And yes, that was the reason I got my 300 Hz - 1.5 KHz "dip".
So I corrected the midrange / bass response in the crossover. Compared to the original measurements, they have a lot more power radiated on the midrange, lower midrange and upper bass. High midrange, treble and lower bass stayed the same. Once again, I compared the RTA measurements, and now the three ways are very similar to the 2 ways - except for a narrow dip at around 250 Hz (unavoidable, I guess, from what I've seen on my own designs and at Stereophile's JA measurements of floorstanders), and a smoother transition between the midrange and the tweeter (which was to be expected).
So far, so good. I've listened to my speakers for 4 days, and although I still sometimes think I hear a hint of glare, I'm pretty sure it's my mind playing tricks on me. The sonic balance seems to be restored.
In conclusion, it wasn't any extra power radiated in the upper midrange / low treble, IMHO - it was that the frequencies around the problematic area were too low! I suppose my ears "locked in" the 300 - 1.5 KHz frequencies, and then I thought the 1.5 ~ 6 KHz region was simply too high in power... therefore creating the impression of false detail and hardness.
I'll keep on listening and see if my theory is valid. So far, so good. If I end up fixing my 5 year issue, I'll have learned a huge lesson here - measure twice, build once. Better yet, measure the h*ll out of any design, and then start considering building anything.
Someone recently pointed out I've been complaining for 5 years about "bright" speakers. He's right. I have. And I've looked everywhere for an answer, without finding anything.
At least I'm sort of stubborn. After years and years of measuring, designing, complaining, experimenting, etc., I think I've found a couple of guidelines regarding all those terms. I hope this is useful to someone. Suggestions & corrections are welcome - I'm still learning, and most of these concepts are assumptions on my part.
1) What is "harshness", "brightness", etc.?
I've been looking for an explanation myself. After looking back at the designs which I've built and thought of as "bright", I think that brightness has a lot to do with the sound power radiated on frequencies around 1 - 7 KHz and their relationship to the sound power above and below them.. It manifests itself first as extra detail, but it tires you very fast. In my case, it can make my ears ring.
2) What do I mean?
Whenever I build speakers, I usually get the following:
Two way speaker -> usually good results, no brightness.
3 - 3.5 way speaker -> brightness.
So, after a while, I decided I needed to start following some sort of method, to try and find out what was wrong. First, I did lots of experiments.
a. Lower the mids and highs(from 1 KHz up) -> no luck.
b. Raise the bass (from 100 Hz down) -> no luck.
c. Create a "house curve" -> no luck.
d. "Tilt down" the speaker's response -> no luck.
I hadn't noticed I never did anything to, say, 250 Hz to 1 KHz.
3) How I discovered my issue.
After redesigning dozens of times my crossovers, and blaming everything from the source material to the drivers' material, this is what I did.
a. I took a 2-way speaker I built that I love and took MLS measurements and RTAs at my seating position.
b. I took a commercial 2-way speaker that I really like (the Energy C-3) and which measures pretty good, and took the same measurements.
c. I overlaid both measurements on the measurements of a big 3 way I built, which has the "piercing midrange" voicing.
At first, I didn't notice anything out of the ordinary.... except for the fact that the lower midrange / upper bass region wasn't summing up as it should.
Upon closer inspection, I noticed a 2 ~ 3 dB difference from 300 Hz to 1.5 KHz between my 3 way and any of the 2-ways. This difference made the 3 way a bit depressed in that area. It was so smooth, I didn't really notice it while doing the sims. And besides, since the midrange / woofer curves followed a 2nd order curve, I thought I was doing things right.
Looking back on the 3 three ways I've so far "fixed", when I originally took their MLS measurements, I did so by raising them around 1.5 M above floor level, so I didn't even consider floor bounce. I re-took every measurement at floor level, and I got slightly different FR plots for both the woofers and the midranges. And yes, that was the reason I got my 300 Hz - 1.5 KHz "dip".
So I corrected the midrange / bass response in the crossover. Compared to the original measurements, they have a lot more power radiated on the midrange, lower midrange and upper bass. High midrange, treble and lower bass stayed the same. Once again, I compared the RTA measurements, and now the three ways are very similar to the 2 ways - except for a narrow dip at around 250 Hz (unavoidable, I guess, from what I've seen on my own designs and at Stereophile's JA measurements of floorstanders), and a smoother transition between the midrange and the tweeter (which was to be expected).
So far, so good. I've listened to my speakers for 4 days, and although I still sometimes think I hear a hint of glare, I'm pretty sure it's my mind playing tricks on me. The sonic balance seems to be restored.
In conclusion, it wasn't any extra power radiated in the upper midrange / low treble, IMHO - it was that the frequencies around the problematic area were too low! I suppose my ears "locked in" the 300 - 1.5 KHz frequencies, and then I thought the 1.5 ~ 6 KHz region was simply too high in power... therefore creating the impression of false detail and hardness.
I'll keep on listening and see if my theory is valid. So far, so good. If I end up fixing my 5 year issue, I'll have learned a huge lesson here - measure twice, build once. Better yet, measure the h*ll out of any design, and then start considering building anything.
Comment