After living with these for a while, several audiophile buddies independently have pointed out severe lacks in the midrange in both of my pairs (being more of a basshead i cant hear the same, but again i am not good at pointing out what exactly is the problem with my ears).
You posted several frequency response measurements.
Measured 10 cm from tweeter centre:
First of all, ignore the large peaks and dips below 300 Hz. It’s not possible to tell what is due to the speaker and what is due to reflections from the ceiling & floor, and opposite walls. Yes, that dip at ~140 Hz (in both of your graphs) is caused by floor bounce. Sound from the woofers, reflected off the floor, arrives at the microphone out-of-phase with sound directly from the woofers.
At 10 cm, your microphone is too close to the speaker. If you were measuring the tweeter alone, or the woofers alone, it would be alright, but with both working, 10 cm is too close.
I believe the very large dip centered at 2 kHz shows the crossover working. It appears as a large dip because within 1 octave of the crossover frequency (1 to 4 kHz), the tweeter and woofers are both producing roughly equal amounts of sound, but the microphone is closer to the tweeter and further away from the woofers. As a result, the tweeter’s and woofers’ sound arrive at the microphone out of phase with each other, and they cancel each other out.
Measured 1 meter from tweeter centre:
It’s better when you move the microphone to 1 meter. The 2 kHz dip disappears completely. What happens if you move the microphone further away, to 2 or 3 meters?
I do not understand why there is such a large difference in response from about 300 Hz to about 4 kHz. Something is clearly wrong. I can’t tell if it’s the speakers or your measurement methods. It rises centered at 500 Hz, falls from 600 Hz to 1 kHz, rises again centered at 2 kHz, only to fall again above 3 kHz. The loudness varies from a low of about 83 dB between 800-900 Hz, to as high as 89 dB at 500 Hz – an overall change of 15-16 dB. That’s very large!
The predicted responses when Dennis Murphy designed this crossover were flat (±3 dB) within two octaves of the crossover frequency. This flatness across the crossover range often causes people to think his speaker designs sound too bright because they are used to speakers which have mid range dips. But I never heard his designs lack in mid range.
So i found time... The difference is to my ears audible, more then i would have expected, but less then i hoped.
The measurements however are almost identical as previous, so the difference might be psychological...
Was this to test if rounded-over front baffles caused the large variation in mid range response? Your measurements (with and without rounded over baffles) were almost identical. I’m not surprised by that. Rounded over front baffles could not make such a large difference as your measurements indicate.
But i am still wondering what is causing this, i have a couple of theories:
Crossover: i may have done something wrong here, but cant really imagine what, the issue is the same on two different pairs built at different times.
Can more expensive caps be THAT audible? I have standard caps now.
I’m only guessing after seeing a few frequency response graphs, but the first thing I would look for is an error in the crossovers. It could be in part selection or wiring.
Your photos show you have the ribbon tweeters. There are different crossovers for the ribbon tweeter version and the dome tweeter version. Are you certain you used the correct crossover version? The reason I ask is because of the crossover frequency apparently centered at 2 kHz. It has been some time, but I remember the dome tweeter version has a lower crossover frequency (perhaps at 2 kHz?) than the ribbon tweeter version. The Fountek Neo CD3 tweeter cannot go down to 2 kHz, but the dome tweeter can.
When you purchased crossover parts, did they differ from the original design? By how much? Variations of ±10% are alright, but larger than that can cause problems.
More expensive caps do not make that audible a difference, if any at all. It is much more important to have cap values (as well as inductor and resistor values) as close as possible to those of the original design.
Stuffing: i have the exact amount polyfill specified, but have not tried to change the amount, can that have a big impact in the mids?
That would mainly effect lower frequencies, not the mid range. I don't think it's the polyfill.
