I would not expect miracles from one speaker, the increased level of harmonics in the low-frequency range is automatically multiplied throughout the frequency spectrum in the form of modulation, maybe someone likes it, but in general, this is a fundamental drawback of such solutions, also in such solutions it is quite possible to have an increased level of nonlinear distortion in the mid-high frequency range, and this cannot be dealt with from the amplifier side. If we want to hear a cleaner sound, then we need to separate the bands ...
Here it is, from the first, non-DSP, measurement:
There's also something that my ears pick up as a "hissing" sound in some mid to high friquencies that I can't put my finger on. I need to swap to a pair of decent headphones to see if I can hear it there as well.
It all makes sense. Judging by the level of nonlinear distortion, you should hear the problem in the midrange and high frequency range.
You have THD of 1% at 900Hz and 0.5% at 2kHz and 3.3kHz, and this cannot be fixed, so you will just have to put up with a certain level of lisping from the speaker.
Go to three bands, and then there will be no problems with sibilants.
The THD at 900 hz is driven almost exclusively by the 2:nd harminic. My intuition thinks that changing the dampening/stuffing may help, but mayby that's not how it works?
I'm feeling a bit daft, but could you explain your comment "go to three band"?
Most likely, this is a translation problem.
I wanted to say that in order to obtain a clean sound without sibilants and with a sufficient level of low frequencies that will not multiply harmonics across the frequency spectrum, it is best to use three different speakers, a low-frequency speaker, a mid-range speaker, and a high-frequency speaker. That is, go for three bands.
My intuition tells me that this is a consequence of the work of the auxiliary emitter in the middle of the speaker.
When you take measurements in REW, do you use acoustic synchronization?
And also, for harmonics it is important to see the noise level. This is activated by checking the box.
And also, for harmonics it is important to see the noise level. This is activated by checking the box.
I used acoustic synchronization "from start".
But I did not capture the noise floor. That's on the list for the next measurement then.
But I did not capture the noise floor. That's on the list for the next measurement then.
It's strange, usually the phase at a distance of 1m for a wide-range speaker looks more even than for you.
I admit to being somewhat disappointed: I had not thought to get significant distortion in with these drivers in a relatively large box.
Changing the stuffing config brought all harmonics down to <0.5%, however, the THD was still high and although DSP made it bearable, I don't want to spend this amount of cash on a pair of speaker elements only to have to hide the imperfections with DSP.
So, where to go from here?
* I could add a Tweeter able to be crossed down low to the existing box, such as the SB Acoustics TW29DN-B-8. Would a cross-over (if implemented correctly) at 1500 or 2000 hz actually help with the distortion?
* Try to redo the existing box, for example, since these drivers seems to like OB then maybe a more traditional Voigt Pipe slot is better than the current port for them? This is probably what I'll do today since I'm at the computer and bored :-D
* Bite the bullet and sell them. My design criteria for these speakers includes a slim width, so OB is not an option. Something like a 6'' coaxial is probably better for me. I just hoped I could get away without a cross-over for my first design.
In any case, huge thanks to @AllenB and @uriy-ch for all the imput so far, I really appreciate it!
Changing the stuffing config brought all harmonics down to <0.5%, however, the THD was still high and although DSP made it bearable, I don't want to spend this amount of cash on a pair of speaker elements only to have to hide the imperfections with DSP.
So, where to go from here?
* I could add a Tweeter able to be crossed down low to the existing box, such as the SB Acoustics TW29DN-B-8. Would a cross-over (if implemented correctly) at 1500 or 2000 hz actually help with the distortion?
* Try to redo the existing box, for example, since these drivers seems to like OB then maybe a more traditional Voigt Pipe slot is better than the current port for them? This is probably what I'll do today since I'm at the computer and bored :-D
* Bite the bullet and sell them. My design criteria for these speakers includes a slim width, so OB is not an option. Something like a 6'' coaxial is probably better for me. I just hoped I could get away without a cross-over for my first design.
In any case, huge thanks to @AllenB and @uriy-ch for all the imput so far, I really appreciate it!
Distortion shouldn't be significantly audible. Are you sure that's what you're hearing? What have you done with the DSP which reduces distortion? That's not normally how it works unless you've reduce excursion, for example.
I'm not sure at all since I'm a n00b 🙂 But it seemed to me like it made sense.
I changed the stuffing and found a config where the 2:nd order harmonic was tamed a bit, but the THD stayed at ~1% at 900 hz and with the same spikes at 2k and 3.5k. I perceived it as an improvement.
Then I used DSP to smooth the overall response curve from 100hz and upwards (like in this post: https://www.diyaudio.com/community/...quency-response-in-a-tqwp.417289/post-7783462). Again I perceived that as an improvement specifically in the "hissing" distortion-like sound I was hearing.
Could the status of the build - with the sidewall only held by clamps for example - impact the perception of distiortion? At this point I could seal the box properly, and also, if the "hiss" is still there, switch the elements around to see if there's a different between them in the pair.
I think you are connecting more importance to harmonic distortion produced by speakers, than you need to.
This move will help reduce distortion in the operating range of this speaker, if this speaker itself has less distortion. Usually it does. This is quite likely to improve the situation with sounds ssss, but the intelligibility in the midrange may remain the same, i.e. the sounds "hissing" will also be heard as unnatural. The reason is that you have an increased level of nonlinear distortion at 900 Hz.
Analyzing the distortion curve, this will be the most correct solution if you need to get low distortion in the midrange and high frequency range.
As a beginner, you have a very serious base, you very quickly applied DSP correction, and calculated the case for the speaker.
If you want to have a deliberately good result for low distortion, then you can't do without three way. Moreover, having the capabilities of DSP, you need to go in the direction of active amplification, i.e. put an amplifier for each speaker, and give the crossover function to DSP. Active speakers are very flexible in setup and do not require much time to reconfigure the crossover. This greatly facilitates the achievement of a positive result in a short time. Moreover, you can start small, buy any used music center with three-way speakers and assemble an active three-way speaker based on them. And then think about a larger speaker.
I disagree. 1% at 900Hz and 0.5% at 2kHz is sometimes quite audible on electronic music as weak intelligibility of the midrange, I myself suffered from this effect for a long time until I switched to active three way.
Last edited:
Not always, much depends on the properties of the selected speaker and the frequency range in which it operates. At the same time, if the speaker has increased distortion from the voltage source, and we do not want to change the speaker to another, then no one forbids us to use current sources as an amplifier, or to put additional reactance between the speaker and the voltage source.
The "hissing sound" is there. I'm using reference tracks I know well, and also have a pair of decent headphones at hand to double check myself. I'd love to find a way to reduce it within the boundaries of the existing design - whatever causes it 🙂