Welcome, Guest
Username: Password: Remember me

TOPIC: WMMTMMW 3 way Active Speakers

WMMTMMW 3 way Active Speakers 4 years 6 months ago #24323

  • mr-marlen
  • mr-marlen's Avatar
  • Offline
  • Senior Boarder
  • Posts: 50
  • Thank you received: 7
  • Karma: 2
I present to your attention a fully three-way active speaker system "Sylvester" on the basis of the following speakers:
1) Mundorf AMT2510C x1
2) Fountek FW146 x4
3) Scan Speak 26W4867 x2
Immediately it should be noted that this project is laid out for experience exchange purposes, but not for a repeat. Too many nuances.
Acoustics was born slowly. First WTW concept was defined. In the course of experiments and models with shields were made very interesting conclusions. For example a set of AMT2510C and 26W4867x2 speaker is quite capable to live as decent quality two-way speakers, provided the use of FIR filters or IIR high cut slope (3-4y orders).
Unfortunately I was not happy about the use of FIR filters. I wanted to get to the possibilities of using acoustics for listening / viewing and for live playing / singing. Analysis of possible ways to get the result led me to the ideas of John Kreskovsky.
I recommend Googling the topic «A New High Slope, Linear Phase Crossover Using the Subtractive Delayed Approach» and read though, would be the first part.
In fact the principle of such crossover based on the selection of filters so that the overall impulse responses of the two drivers evolved mutually compensating for their tails. There are some conditions under which such sum will be able by using only IIR filters.
The only drawback of such filters - off-axis curves. In general, the article Kreskovsky offers several options to solve this drawback:
1) Using WMTMW configuration;
2) The use of filters of higher orders with maintenance of certain criteria (additional delays and certain types of slopes).
After simulations with actual measurements on someone else's system WMT I came to some conclusions:
- For WMT is necessary to use filters of higher orders with certain criteria. The fee for this is getting a little crooked impulse response because of incomplete compensation mutual-tails, but off-axis curves becomes like with a standard 3 way WMT configuration.
Here is a links to pictures of such systems:
cloud.mail.ru/public/4cZt/b8Dkhf7UB
cloud.mail.ru/public/Kbax/ArVBrxSWv
cloud.mail.ru/public/DhbG/KaBh4aW4b
- For WMTMW might well use a less steep slopes and get a nice impulse, but to sacrifice other things.
The picture cloud.mail.ru/public/Jvti/GRvjFCUrv shows the form of the pulse and frequency response of drivers with a conventional off-axis WMT. It is evident that in the off-axis, we have the terrible ups because drivers work in phases desynchronization.
But as soon as we make a configuration WMTMW we are geting the picture:
cloud.mail.ru/public/LmaU/nfYnv2i7z
Practically, peaks of drivers in crossover region can be made with lower level and with a moving slopes get excellent off-axis crurves.
System with such type of filters can be attributed to intermediate between "coherent phase" and "Linear Phase". Because the compensation of group delay (tails on impulse) occurs only to lower the cutoff frequency of the speakers work together, so to get on the woofer below 200Hz ideal Step will not work because of its physical characteristics of the motor (GD growth with decreasing frequency). Below 200Hz we can use electro acoustic/mechanical feedback with additional subwoofer or FIR filters. But I think it is not so nessary because masterings of songs makes on acoustics which in many studios are not linearphase.
The only disadvantage of such a filter – use of the speakers that can cover a large amount of octaves with low distortion.
Simulation of two way version WTW revealed difficulties in the implementation of IIR linear phase filter - tweeter pulls poorly on low freq. Therefore, the question arose the need for the addition of mid-range drivers.
It was seen a bunch of materials, expensive 5 '' did not consider. and the choice fell on the trail. candidates:
1) TB 75-1558SE very good, but below 700 Hz has high distortion. Purchase in Russia is difficult, the price is high.
2) Eton Symphony good, but below 700Hz has high distortion, the average price.
3) TB W4-1320SIF good, but there is a problem near 1 kHz and 500 Hz.
4) TB W4-1337SDF not bad, but there are problems in the 500Hz + met. resonance.
5) Fountek FW146 good, but it met. Resonance is already climbing at 2kHz.
6) Zaph Audio ZA14W08 good, but it met. Resonance is already climbing at 2kHz.
As a result, I decided to donate the horizontal direction and stopped at version 4 midranges in one speaker. With 2 in the horizontal section M. The reason is simple - the need to work mid-level section below (feature applied filters IIR) frequency and the desire to get a good mid-scale comparable the distortion of two-26W4867. Candidate from 1 dropped out due to a sharp increase in distortion below 500Hz.
Paragraph 2 also dropped due to increased distortion below 500Hz.
From 3-6 points initially inclined to paragraph 3 or 4, but I did not like the distortion in the area of 500Hz. Somehow I came across a series of articles by Dmitry Gorshenin about the features of the use of filters, in particular with metal diffusers. The article described in detail mechanisms of distortion and some useful conclusions.
After reading the article, paragraphs 3, 4 have been removed from the list and chose the 5 or 6. 6 was harder to get, and ran into the forum topic techtalk.parts-express.com/forum/tech-ta...-FW146=&p=449378 - ordered Fountek FW146.
My experiments in the "increasing output impedance of crossover" fully demonstrated positive results (described in the article by Dmitry Gorshenin solutions). 1,5mH Coil was enough to bring in a minimum of distortion. 0,82mH is not enough, 2,5mH greatly reduces freq amplitude levels and provides little improvements comparing to 1,5mH.
This THD shows in next pictures:
cloud.mail.ru/public/4M7r/cPPTw59m9
cloud.mail.ru/public/D9by/5yxRrMmjB
cloud.mail.ru/public/HQcE/EJRiMjpzS
cloud.mail.ru/public/GD75/YpcuF9NsH
cloud.mail.ru/public/JoPH/gFKXdViPh
Because I am fully aware of the need to increase the voltage level of the amplifier to 10-15dB (due to coil), I provided for the midrange amplifier in the D-class, which has acceptable distortion and a high voltage level. Woofers compared with midrange working with the weakening about 8dB (amplifiers MID and LF are similar).
The following are photograph speaker, drawing (not standard), graphics, simulations with step for a distance of 1.8 m and a graphics test measurements with 1.1 m.
Overall, I am satisfied with the result. But in the future I will try to optimize crossover between tweeter and midranges.
As a DAC I am using DIY based on Amanero / miniSHARC / 4xPCM1794A / Antecom control for 8 channels / Prev on two OPA4132UA - www.minidsp.com/forum/diy-hifi-projects/...c-amanero-4xpcm1794a
As amplifiers LM3886 for Tweeters, Iraudamp7S amplifiers for MID and LF.

You can find the original thread on Russian forum here: forum.vegalab.ru/showthread.php?t=72521

Thanks for reading!
Attachments:
Last Edit: 4 years 6 months ago by mr-marlen.
The administrator has disabled public write access.
The following user(s) said Thank You: devteam

WMMTMMW 3 way Active Speakers 4 years 6 months ago #24324

  • mr-marlen
  • mr-marlen's Avatar
  • Offline
  • Senior Boarder
  • Posts: 50
  • Thank you received: 7
  • Karma: 2
The 2nd post is to compare phase coherence version using LR2 and LR4 crossovers without and with FIR inverse all pass filter vs non phase coherence of my version.

Also i am using Linkwitz Transform for woofers to make more perfect with group delay and freq response on lower end. Pictures in post 1 and 2 without LT.
Attachments:
Last Edit: 4 years 6 months ago by mr-marlen.
The administrator has disabled public write access.
The following user(s) said Thank You: devteam

WMMTMMW 3 way Active Speakers 4 years 5 months ago #24669

  • mikekuehn
  • mikekuehn's Avatar
  • Offline
  • Fresh Boarder
  • Posts: 4
  • Karma: 0
Thanks for sharing your experience marlen.. I have a couple of questions.

I don't totally understand why you have 2 midranges horizontally. Just for power handling / lower distortion? Does this not create issues in the horizontal off-axis?

Do you prefer (listening impression) the FIR corrected version? If I understand correctly, you have tried IIR only?
The administrator has disabled public write access.

WMMTMMW 3 way Active Speakers 4 years 5 months ago #24742

  • devteam
  • devteam's Avatar
  • Online
  • Administrator
  • Posts: 8757
  • Thank you received: 1140
  • Karma: 163
What a great review of your system mr-marlen. We're glad to see it together with all these measurements.
We don't speak Russian so can't really pop in to help on that forum but thanks for sharing! :-)

DevTeam
miniDSP, building a DSP community one board at a time.
The administrator has disabled public write access.
The following user(s) said Thank You: mr-marlen

WMMTMMW 3 way Active Speakers 4 years 5 months ago #24770

  • mr-marlen
  • mr-marlen's Avatar
  • Offline
  • Senior Boarder
  • Posts: 50
  • Thank you received: 7
  • Karma: 2
mikekuehn wrote:
Thanks for sharing your experience marlen.. I have a couple of questions.

I don't totally understand why you have 2 midranges horizontally. Just for power handling / lower distortion? Does this not create issues in the horizontal off-axis?

Do you prefer (listening impression) the FIR corrected version? If I understand correctly, you have tried IIR only?

1) Yes, I have 2 midranges horizontally for lower distortion.
2) Due to good tweeter THD perfomance i have done low crossover near 2 kHz. With such crossover we have good horizontal off-axis for about 45 degrees. I have put here some pictures.
3) There was good measuring and listening of IIR version. Also there was time to make good measurements of IIR version.
There was not enougth time to simulate and measure full FIR (FIR for each driver section).
I have listened full FIR version but there was some issues with simulations (based on IIR version) vs fast measurements of FIR version. I have no no time to check what was wrong with FIR.
If we compare version of my full FIR vs my IIR, IIR was better.

There was no comparsation between "IIR with FIR inverse all pass filter" vs non phase coherence of my version. Becouse i am also have no time for this. I think if we compare versions showed in post 2 - "non phase coh. IIR" will be better due to better off-axis response.
If we make version with higher order slopes and better off-axis maybe it will be better than "non phase coh. IIR" but there will be delay of FIR filter which i can not accept.

So in future there will be 2 version:
1) IIR without FIR;
2) FIR on each driver section.
Attachments:
The administrator has disabled public write access.
Moderators: devteam