• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

Need help and advice: linear speakers sound very harsh

Maksims

New Member
Joined
May 19, 2023
Messages
2
Likes
0
Hello, I'm new at this forum, but I'm a proponent of the scientific approach in audio, and I hope some of you can help me with my issue.

I'm in the process of building my DIY loudspeakers. That is a 3-way (10-inch woofer, 5-inch midrange, and 1-inch dome tweeter) with cut-off frequencies around 400 and 2800 Hz made by 2nd order crossover. The on-axis frequency response I got is more or less linear, within +/- 3 dB, but the midrange and vocals sound very harsh. I would say the region around 2-4 kHz, where both midrange and tweeter are loud enough. I also feel a lack of high bass and low midrange. The responses were measured twice in different non-anechoic environments with CLIO and matched well.

I compared the speakers to the industrial models, for example, Dali Oberon 3. Despite both speakers being pretty linear on-axis, the sound difference is so dramatic, which, in my opinion, is hard to explain by just a couple dB deviations. I also tried to simulate this effect using a parametric equalizer. And it was hard to achieve similar results. The most similar sound I get is by reversing the polarity of one Dali speaker and putting one on another to simulate sound from a single loudspeaker. The midrange became highlighted, but it was also a bit noisy and distorted. Of course, the bass disappeared, which my speakers have. But I would describe my speakers as a 2.1 system with the reversed polarity of satellites and a low-setting subwoofer.

During listening tests, I also found another industrial loudspeaker with identical problems and sound character. It was a 3-way loudspeaker with two 6.5-inch woofers, a 5-inch midrange, and a 1-inch tweeter from Wavecor. Opposite to my wide classic speakers, these speakers are modern tall towers, but there is one common thing. They also require the same polarity for tweeter and midrange not to have null at the cut-off region, despite 2nd order. So woofers connected with +, but midrange and tweeter with -.

I tried to change the polarity of my tweeter, and it made the sound better, but probably because of the cancellation in this annoying region. Anyway, the phase characteristics matched well with the original polarity.

I thought about the midrange driver beaming, but it's just 5 inches, and the cut-off frequency around 2.8 kHz doesn't seem critical. I also made several off-axis measurements and didn't find significant deeps. Unfortunately, making anechoic polar response measurements is quite problematic for me. I thought about the geometry of the cabinet, but again, those towers are different speakers by shape. The only similar thing is the distance between the centers of the midrange and the tweeter, in my case - 135 mm.

Can anyone recognize the problem from my description? What are the most common and possible reasons, and what to start from? Thank you.
 

Attachments

  • on-axis-30degree.jpg
    on-axis-30degree.jpg
    42.2 KB · Views: 70
  • response.jpg
    response.jpg
    511.2 KB · Views: 75
  • cabinet.png
    cabinet.png
    51.5 KB · Views: 74

notsodeadlizard

Senior Member
Joined
Feb 18, 2023
Messages
403
Likes
362
Hello, I'm new at this forum, but I'm a proponent of the scientific approach in audio, and I hope some of you can help me with my issue.

I'm in the process of building my DIY loudspeakers. That is a 3-way (10-inch woofer, 5-inch midrange, and 1-inch dome tweeter) with cut-off frequencies around 400 and 2800 Hz made by 2nd order crossover. The on-axis frequency response I got is more or less linear, within +/- 3 dB, but the midrange and vocals sound very harsh. I would say the region around 2-4 kHz, where both midrange and tweeter are loud enough. I also feel a lack of high bass and low midrange. The responses were measured twice in different non-anechoic environments with CLIO and matched well.

I compared the speakers to the industrial models, for example, Dali Oberon 3. Despite both speakers being pretty linear on-axis, the sound difference is so dramatic, which, in my opinion, is hard to explain by just a couple dB deviations. I also tried to simulate this effect using a parametric equalizer. And it was hard to achieve similar results. The most similar sound I get is by reversing the polarity of one Dali speaker and putting one on another to simulate sound from a single loudspeaker. The midrange became highlighted, but it was also a bit noisy and distorted. Of course, the bass disappeared, which my speakers have. But I would describe my speakers as a 2.1 system with the reversed polarity of satellites and a low-setting subwoofer.

During listening tests, I also found another industrial loudspeaker with identical problems and sound character. It was a 3-way loudspeaker with two 6.5-inch woofers, a 5-inch midrange, and a 1-inch tweeter from Wavecor. Opposite to my wide classic speakers, these speakers are modern tall towers, but there is one common thing. They also require the same polarity for tweeter and midrange not to have null at the cut-off region, despite 2nd order. So woofers connected with +, but midrange and tweeter with -.

I tried to change the polarity of my tweeter, and it made the sound better, but probably because of the cancellation in this annoying region. Anyway, the phase characteristics matched well with the original polarity.

I thought about the midrange driver beaming, but it's just 5 inches, and the cut-off frequency around 2.8 kHz doesn't seem critical. I also made several off-axis measurements and didn't find significant deeps. Unfortunately, making anechoic polar response measurements is quite problematic for me. I thought about the geometry of the cabinet, but again, those towers are different speakers by shape. The only similar thing is the distance between the centers of the midrange and the tweeter, in my case - 135 mm.

Can anyone recognize the problem from my description? What are the most common and possible reasons, and what to start from? Thank you.

"What are the most common and possible reasons, and what to start from?"
Millions of reasons starting from the drivers selection and till infinity in DIY case.

If you started this project from scratch, you better stop now.

Take a good validated design and replicate it to begin with.
If this is your first speaker (I tend to think that way, otherwise you would not be applying to this forum), don't jump into multi-way designs right away, they are too complex for a first try. Of course, there is no problem making a primitive box and "sticking" drivers into it, but the result will be extremely far from even acceptable.

Try for example http://www.frugal-horn.com/fh3-plans.html
Here you won't feel lonely: https://www.diyaudio.com/community/threads/frugel-horn-mk3-picture-gallery.199849/

Go through all the stages, it will be fun, especially the damping :)
This experience will give you a lot.
(Including understanding why good speakers are so expensive)

If you don't have experience, don't start hard.
This is the path to disappointment.
 
OP
M

Maksims

New Member
Joined
May 19, 2023
Messages
2
Likes
0
"What are the most common and possible reasons, and what to start from?"
Millions of reasons starting from the drivers selection and till infinity in DIY case.

If you started this project from scratch, you better stop now.

Take a good validated design and replicate it to begin with.
If this is your first speaker (I tend to think that way, otherwise you would not be applying to this forum), don't jump into multi-way designs right away, they are too complex for a first try. Of course, there is no problem making a primitive box and "sticking" drivers into it, but the result will be extremely far from even acceptable.

Try for example http://www.frugal-horn.com/fh3-plans.html
Here you won't feel lonely: https://www.diyaudio.com/community/threads/frugel-horn-mk3-picture-gallery.199849/

Go through all the stages, it will be fun, especially the damping :)
This experience will give you a lot.
(Including understanding why good speakers are so expensive)

If you don't have experience, don't start hard.
This is the path to disappointment.
Thank you very much for your reply. I also thought about drivers until I heard the same sound character from other loudspeakers based on other drivers. So if it would be possible to redesign them based on these drivers, it would be great. If not, well, yeah, I think if worth trying something different :)
 

SIY

Grand Contributor
Technical Expert
Joined
Apr 6, 2018
Messages
10,383
Likes
24,749
Location
Alfred, NY
Hello, I'm new at this forum, but I'm a proponent of the scientific approach in audio, and I hope some of you can help me with my issue.

I'm in the process of building my DIY loudspeakers. That is a 3-way (10-inch woofer, 5-inch midrange, and 1-inch dome tweeter) with cut-off frequencies around 400 and 2800 Hz made by 2nd order crossover. The on-axis frequency response I got is more or less linear, within +/- 3 dB, but the midrange and vocals sound very harsh. I would say the region around 2-4 kHz, where both midrange and tweeter are loud enough. I also feel a lack of high bass and low midrange. The responses were measured twice in different non-anechoic environments with CLIO and matched well.

I compared the speakers to the industrial models, for example, Dali Oberon 3. Despite both speakers being pretty linear on-axis, the sound difference is so dramatic, which, in my opinion, is hard to explain by just a couple dB deviations. I also tried to simulate this effect using a parametric equalizer. And it was hard to achieve similar results. The most similar sound I get is by reversing the polarity of one Dali speaker and putting one on another to simulate sound from a single loudspeaker. The midrange became highlighted, but it was also a bit noisy and distorted. Of course, the bass disappeared, which my speakers have. But I would describe my speakers as a 2.1 system with the reversed polarity of satellites and a low-setting subwoofer.

During listening tests, I also found another industrial loudspeaker with identical problems and sound character. It was a 3-way loudspeaker with two 6.5-inch woofers, a 5-inch midrange, and a 1-inch tweeter from Wavecor. Opposite to my wide classic speakers, these speakers are modern tall towers, but there is one common thing. They also require the same polarity for tweeter and midrange not to have null at the cut-off region, despite 2nd order. So woofers connected with +, but midrange and tweeter with -.

I tried to change the polarity of my tweeter, and it made the sound better, but probably because of the cancellation in this annoying region. Anyway, the phase characteristics matched well with the original polarity.

I thought about the midrange driver beaming, but it's just 5 inches, and the cut-off frequency around 2.8 kHz doesn't seem critical. I also made several off-axis measurements and didn't find significant deeps. Unfortunately, making anechoic polar response measurements is quite problematic for me. I thought about the geometry of the cabinet, but again, those towers are different speakers by shape. The only similar thing is the distance between the centers of the midrange and the tweeter, in my case - 135 mm.

Can anyone recognize the problem from my description? What are the most common and possible reasons, and what to start from? Thank you.
A few general comments:

1. Putting in a nominally 2nd order filter does NOT mean that your response is actually 2nd order. In fact, it's almost certain that it isn't.

2. You need to examine off-axis response in steps to 90° both horizontally and vertically. Quasi-anechoic will help here. Lobing could be a significant issue.

3. It may be very worthwhile to also get in-room response using a good omni mike set up for diffuse field.
 

Slabshaft

New Member
Joined
Jul 15, 2023
Messages
1
Likes
0
I most certainly would not just give up. To a DIYer that’s insanity. Keep iterating and learning.

My inclination here, in looking at your sine sweep is that you really don’t want to listen to a speaker with a dead flat response. It will sound super bright and have no bass. Ideally it will have a response more like the Harmon curve with a downward slope as the frequency rises (similar to pink noise). Very dependent on your tastes, music style and hearing. So, in the interim, consider playing with Room EQ wizard and equalizer APO (if you’re using a PC to generate the sine sweeps) to equalize it until you like how it sounds. THEN figure out how you’ll tweak your design to work without the EQ. For me, I just EQ my problems away and move on the the next thing ;)
 

sigbergaudio

Major Contributor
Audio Company
Forum Donor
Joined
Aug 21, 2020
Messages
2,639
Likes
5,397
Location
Norway
how is the "response.jpg" measure done? In-room I assume? distance?

How do they measure at the listening position in your room?
 
Top Bottom