• 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!

Suggestions for three-way crossover filters?

Looks good. But its just a 2 way crossover.
No, it is an eight-channel device. That's why they call it the Flex Eight.
I posted this earlier, please just scroll up or just go to the MiniDSP product page linked above and read it. It's an 8-channel device. It can be used (for instance) as an electronic crossover for a stereo 4-way active system.
 
Regarding SB15NBAC. What you are saying is exactly what I was thinking about. Is a steep filter enough? Maybe maybe not?

Revel seems to have done well in and of itself. Although it's Revel and not me. He he.
I think Tangband is talking about breakup modes of some drivers.
Here is the SB15NBAC, it has some stuff happening at high frequency, but not too strong.
sb15nbac-4-fr.jpg


Here is a classic example of a driver with a breakup mode, the Seas W18 EX-001, which has an extremely strong mode at 5kHz.
w18ex001-curve.jpg

This doesn't disqualify the driver, in fact Seas makes an entire series these woofers with similar breakup modes and are used in some really terrific speakers like:
But you will need to implement a notch filter and choose appropriate crossover point and slope, which is easy to do with DSP. As Tangband says, you will want to measure the actual peak in your drivers and box, which is easy with REW, etc.
As you suggest, the pros know because they know.;)
 
This suggestion is suboptimal for sound quality . I have tried it and you loose transparancy in the treble, no matter how good the passive crossover is, compared to a 3 way dsp crossover like dbx pa2.
You can add a resistor to the tweeter when you go active to get theoretical lower dynamic compression, but my listening test shows the opposite-you get a more dynamic sound with low impedance coupling in most cases.

Dont waste your time with passive crossovers, as Linkwitz used to say.:)



”Crossovers may be implemented either as passive RLC networks, as active filters with operational amplifier circuits or with DSP engines and software. The only excuse for passive crossovers is their low cost. Their behavior changes with the signal level dependent dynamics of the drivers. They block the power amplifier from taking maximum control over the voice coil motion. They are a waste of time, if accuracy of reproduction is the goal.”
Well I used active LR 4 filters for many years and do not agree that passive filters are worse for the treble.
 
Well I used active LR 4 filters for many years and do not agree that passive filters are worse for the treble.
Then you have to do more investigations, because your conclusion is wrong, or your gear is not good enough to hear the difference .
 
I think Tangband is talking about breakup modes of some drivers.
Here is the SB15NBAC, it has some stuff happening at high frequency, but not too strong.
sb15nbac-4-fr.jpg


Here is a classic example of a driver with a breakup mode, the Seas W18 EX-001, which has an extremely strong mode at 5kHz.
w18ex001-curve.jpg

This doesn't disqualify the driver, in fact Seas makes an entire series these woofers with similar breakup modes and are used in some really terrific speakers like:
But you will need to implement a notch filter and choose appropriate crossover point and slope, which is easy to do with DSP. As Tangband says, you will want to measure the actual peak in your drivers and box, which is easy with REW, etc.
As you suggest, the pros know because they know.;)
Correct - but my conclusion with dsp crossovers is that using a notch filter ( with dsp or analog ) is very often not 100% transparent .
In some cases , the sound will be better with a steeper crossover instead of a notchfilter.
Because music is ordered frequencys, with many tones playing at the same time causing IMD distortion , its very complicated to use notch filtering in a good way.

As an example , I used a notch filter with seas er18rnx for the hybrid loudspeaker . I had to use different notches for left and right speaker, because the breakup were at 4,3 KHz with the left speaker , and at 4,8 KHz for the right speaker . This is the difference of more than one musical tone - a very audible difference . The notch filters must be exactly applied to be useful.
I think its better to use drivers that dont need notch filtering .
 
Last edited:
Many people dont know how music works , theres tones in an ordered frequency way and has nothing to do with sine sweeps.

This is important to know when constructing loudspeakers .

 
Then you have to do more investigations, because your conclusion is wrong, or your gear is not good enough to hear the difference .

Tangband, you very well know that such claims need to be proven in DBTs, don't go there. It will not lead anywhere. As I said to Daniel, active whole the way is fine, but also making a good passive filter for the tweeter and mid using the active one for the bass/mid. There will be more work but the end results is less gear around. I have been using active tri-amp, parametric EQ for many years. The system I have now is no worse than that I had, and using less gear.
 
Thank you Tangband! Good tips. The fact that it can take time and that you need to learn is nothing that deters me. On the contrary, that's the fun. Then I have something to do in my spare time.

I have bought the UMIK-1 measuring microphone. Thought of doing some different measurements outdoors with the elements placed on a baffle.:)

Calculation programs for the construction of loudspeakers are, luckily, easy to come by these days.

General about used HiFi. Anyone who is interested can find out roughly what the prices of used hifi are. Approximately. You can buy and sell for roughly the same price. But DIY I know that you don't get that money back when you sell your stuff. But DIY for me because it's fun. It doesn't matter if I can't sell the stuff then. You should have fun for the money you earn.:D

Used active speakers a little more difficult to predict the price on the used market from what I think I see.

Measuring loudspeakers by John Atkinson ” and buy Dr Toole's book and read it. Sounds like a very sensible tip.


Regarding SB15NBAC. What you are saying is exactly what I was thinking about. Is a steep filter enough? Maybe maybe not?

Revel seems to have done well in and of itself. Although it's Revel and not me. He he.:D

With the sb adc26 and sb15NBAC - forget using a dsp-notch or analog notchfilter . I would try a crossover freq LP/HP of 2 KHz and linkwitz Riley 48 dB / oct . I would also try 36 dB / oct and 24 dB /oct - all in the acoustical domain ofcourse.

I would also play around with crossover freq. testing from 1,6 KHz to 2,5 kHz and then use the crossover frequency that sounds best with music . There are always tradeoffs and even if all crossovers looks good on measurements - there is only one freq that sounds the best .

You can try to eq individual drivers in the dsp so they are flat within +-0,5 dB one octave below and above the crossover frequency before applicating the crossover . This is best done measuring from 30 cm distance.

Regarding m105 - the breakup is clearly seen in the frequency response . This speaker should probably sound much better with a steeper, active dsp crossover.
B8CE41DE-4294-4557-AF66-391A4DD1AAFF.png
 
Last edited:
Hello again OP @DanieIT,

Please let me touch on one another general suggestion for you.

Prior to make your first step into multichannel multi-amplifier active system, I highly recommend you to establish your concrete and best "single amp passive reference audio system" to which you can roll back anytime during your coming DIY project.

Just for example, in my multichannel project the single amp setup of;
JRiver (all in DSDx2) --> ASIO USB driver --> OPPO SONICA DAC (or OKTO DAC8PRO as two channel DAC) --> ACCUPHASE E-460 --> LCR passive network in outer box --> all the SP drivers
has been my "reference sound system" all the way through my project thread.

I still keep, therefore, the above reference sound setup to which I can roll back my entire system very easily, within 10 minutes, by using the outer LCR box (#250) and the SP cabling/switching board ( #004, #137, #250).

You would please find the examples of such "roll-back and then go-advance" comparative listening sessions in my posts #253 #258 #265 (evaluating AHB2) , and #307 #308 #309 (evaluating A-S3000 and TA-A1ES).

You need to go forward step-by-step, I believe. You should not change multiple parameters at once; if you would change multiple parameters at once, then you would easily get into confusion since sometimes pros and cons would cancel-out with each other to give the pseud result of no improvement at all.

As Keith of Purité Audio kindly wrote here, "You must hear equipment in your own room in your own system, compare unsighted (close your eyes) if there isn’t an immediately apparent difference/improvement. To go further, if there isn’t a significant improvement then don’t change anything, the largest gains are speakers and room."

I summarized and shared my DIY multichannel project policy in my post here including the above points.

Edit: You would also please read these posts on "my policy and approach" on my project thread post #30, post #332 and post #341.

In any way, I wish much good luck and success for your DIY multichannel multi-amplifier active audio project!

Now, at least for the time being, I myself will go back to my enjoyment of "listening music" (not listening/analyzing sound) with my almost completed multichannel multi-driver multi-way multi-amplifier flexible-software-DSP fully active audio system giving wonderful and amazing total sound quality...

If needed, you would please awake me on this thread and/or please PM me anytime.
 
Last edited:
Agree its very important when developing DIY to have a very good reference speaker to compare with - it might be a passive or an active one .

Often I hear enthusiasts say they dont have a loudspeaker reference , and that real music is their reference . Thats a nobel approach, but in reality its very easy to get lost in the development getting worse sound than possibly.
 
With the sb adc26 and sb15NBAC - forget using a dsp-notch or analog notchfilter . I would try a crossover freq LP/HP of 2 KHz and linkwitz Riley 48 dB / oct . I would also try 36 dB / oct and 24 dB /oct - all in the acoustical domain ofcourse.

I would also play around with crossover freq. testing from 1,6 KHz to 2,5 kHz and then use the crossover frequency that sounds best with music . There are always tradeoffs and even if all crossovers looks good on measurements - there is only one freq that sounds the best .

You can try to eq individual drivers in the dsp so they are flat within +-0,5 dB one octave below and above the crossover frequency before applicating the crossover . This is best done measuring from 30 cm distance.

Regarding m105 - the breakup is clearly seen in the frequency response . This speaker should probably sound much better with a steeper, active dsp crossover.
Maybe or maybe not, is no exact match with the break-up so it could be diffraction effects or port resonance as well.
 
Last edited:
Maybe or maybe not, is no exact match with the break-up so it could be diffraction effects as well.
Can be. There are also some rumours that Revel dont use the sb 15nbac in the m105. I dont know.
 
There are some rumours that Revel dont use the sb 15nbac in the m105. I dont know.
From the M105 DIY thread there seem to be port noise causing some issues. Also diffraction. Not break-up.

h
 
From the M105 DIY thread there seem to be port noise causing some issues. Also diffraction. Not break-up.

h
From Amirs review of m105
CED867E8-8197-4869-8255-141C85545DCC.png


Amirms coment :

” Is our 5000 Hz hump created by the ringing/break up of the woofer as circled?”
 
From Amirs review of m105
View attachment 223776

Amirms coment :

” Is our 5000 Hz hump created by the ringing/break up of the woofer as circled?”
Please read

 
With the sb adc26 and sb15NBAC - forget using a dsp-notch or analog notchfilter . I would try a crossover freq LP/HP of 2 KHz and linkwitz Riley 48 dB / oct . I would also try 36 dB / oct and 24 dB /oct - all in the acoustical domain ofcourse.

I would also play around with crossover freq. testing from 1,6 KHz to 2,5 kHz and then use the crossover frequency that sounds best with music . There are always tradeoffs and even if all crossovers looks good on measurements - there is only one freq that sounds the best .

You can try to eq individual drivers in the dsp so they are flat within +-0,5 dB one octave below and above the crossover frequency before applicating the crossover . This is best done measuring from 30 cm distance.

Regarding m105 - the breakup is clearly seen in the frequency response . This speaker should probably sound much better with a steeper, active dsp crossover.
View attachment 223767

From the M105 DIY thread there seem to be port noise causing some issues. Also diffraction. Not break-up.

h
Tangband: It may be necessary, very steep filters that is. Plus to work with different crossover points. Resonances that Thomas is into can also cause problems.I will se henceforth how it evolves with my project and the problems/ challenges that lie ahead.:)

Then it remains to be seen if I'm a good enough DIY chef to whip up something tasty with the ingredients I have to play with. I want it to taste good, but at the same time I'm lucky, I'm not that picky about food. The fun is cooking the food yourself. Another analogy, home-cooked food tastes best. At least that's what the person who cooked it thinks.:D

The reason why I chose sb adc26 and sb15NBAC is partly because I read this thread and became curious:


McFly's well-built DIY with those elements. Lots of pictures and measurements in that thread. A lot about crossover. In addition, as McFly refers to in the thread, different types of instructions for 3 d printers so that you can fix your own waveguide. I also thought that seemed exciting. My friend has a 3 d printer so possibly some waveguide:


Plus I heard them (sb adc26 and sb15NBAC) together with Satori bass (Satori WO24P-4), in a three-way DIY speaker at a DIY Vintage fair, see attached picture. I thought they sounded good, really good.:)

Hello again OP @DanieIT,

Please let me touch on one another general suggestion for you.

Prior to make your first step into multichannel multi-amplifier active system, I highly recommend you to establish your concrete and best "single amp passive reference audio system" to which you can roll back anytime during your coming DIY project.

Just for example, in my multichannel project the single amp setup of;
JRiver (all in DSDx2) --> ASIO USB driver --> OPPO SONICA DAC (or OKTO DAC8PRO as two channel DAC) --> ACCUPHASE E-460 --> LCR passive network in outer box --> all the SP drivers
has been my "reference sound system" all the way through my project thread.

I still keep, therefore, the above reference sound setup to which I can roll back my entire system very easily, within 10 minutes, by using the outer LCR box (#250) and the SP cabling/switching board ( #004, #137, #250).

You would please find the examples of such "roll-back and then go-advance" comparative listening sessions in my posts #253 #258 #265 (evaluating AHB2) , and #307 #308 #309 (evaluating A-S3000 and TA-A1ES).

You need to go forward step-by-step, I believe. You should not change multiple parameters at once; if you would change multiple parameters at once, then you would easily get into confusion since sometimes pros and cons would cancel-out with each other to give the pseud result of no improvement at all.

As Keith of Purité Audio kindly wrote here, "You must hear equipment in your own room in your own system, compare unsighted (close your eyes) if there isn’t an immediately apparent difference/improvement. To go further, if there isn’t a significant improvement then don’t change anything, the largest gains are speakers and room."

I summarized and shared my DIY multichannel project policy in my post here including the above points.

Edit: You would also please read these posts on "my policy and approach" on my project thread post #30, post #332 and post #341.

In any way, I wish much good luck and success for your DIY multichannel multi-amplifier active audio project!

Now, at least for the time being, I myself will go back to my enjoyment of "listening music" (not listening/analyzing sound) with my almost completed multichannel multi-driver multi-way multi-amplifier flexible-software-DSP fully active audio system giving wonderful and amazing total sound quality...

If needed, you would please awake me on this thread and/or please PM me anytime.
You are quite right some sensible reference speakers to compare with would be good to have. I'm researching a few different options. Missed a couple fully functional, fairly new, used ones:


I thought I would buy them and have as a reference when I am doing my DIY. Maybe then keep them or sell them but the seller had set such a low price that they were sold in a few hours. :oops: But that does not matter. Sensible speakers always show up.:)

If needed, you would please awake me on this thread and/or please PM me anytime.
That was nice of you. Thank you for wanting to help. Of course, the same thanks to you others who write in this thread.:D
 

Attachments

  • IMG_4299 (2) (2).jpg
    IMG_4299 (2) (2).jpg
    243.1 KB · Views: 62
  • IMG_4302 (2).jpg
    IMG_4302 (2).jpg
    144.3 KB · Views: 82
Last edited:
Another advice to Daniel :
If you make a threeway, you can get inspiration from sammels building at faktiskt.se .

He uses the same mid and tweeter as in your project.
As subwoofer goes - A Dayton reference 10 subwoofer seems to be one of the best one can use for the money . Closed box , 30 liters each with linkwitz transformer.

Another wildcard is a new subwoofer from Reckhorn - d250. Much cheaper but maybe good ?


I suspect that about 33 cm wide baffle will be about sweet spot in such a speaker . Using the Troelsgravesen formula gives 11600/33 = 351 Hz crossover .
105-110 cm high cabinet makes this about optimal I think .
 
Last edited:
Yes, in;
https://www.minidsp.com/products/minidsp-in-a-box/minidsp-flex-eight
their description of "Flexible I/O versions for unbalanced/balanced/digital connectivity Array" at the top is really confusing and misleading, and it has inconsistency with their spec description of "Outputs: Unbalanced (RCA)"...
I got a response from MiniDSP regarding a possible balanced option on the Flex Eight (like the Flex has):
Thanks for your kind comments and interest in this platforms.
Unfortunately this Metadata is a typo that was carried over from the Flex series. The Flex Eight doesn't have a balanced option. I'm afraid that we can't share much of our roadmap as per our company policy but looking at the back panel, you can see that 8ch would be very tricky to fit.. :-(

Hoping this info helps and feel free to contact us if you have further questions.

Best Regards
miniDSP DevTeam
Seems a balanced option for Flex Eight is unlikely, or even physically impossible to fit... MiniDSP edited the text on the product page last night to limit the confusion.
Looking at the discussions online, we are not the only people wondering if there is going to be a balanced option. Lots of people saw the same erroneous text and are imagining and speculating!
 
I got a response from MiniDSP regarding a possible balanced option on the Flex Eight (like the Flex has):

Seems a balanced option for Flex Eight is unlikely, or even physically impossible to fit... MiniDSP edited the text on the product page last night to limit the confusion.
Looking at the discussions online, we are not the only people wondering if there is going to be a balanced option. Lots of people saw the same erroneous text and are imagining and speculating!
8 TRS would take the same space,as DM8 but don't know the depth needs,if it's packed then they are right.
There could be other solutions thought.
Too bad,it would be No1 candidate but my amps need close to 5V for full output (theoretically,but still).
 
8 TRS would take the same space,as DM8 but don't know the depth needs,if it's packed then they are right.
There could be other solutions thought.
Too bad,it would be No1 candidate but my amps need close to 5V for full output (theoretically,but still).
Sorry, let me ruin your day!:facepalm: I too was hopeful that balanced output was possible. It is not, it just won't fit. Here is the TRS version of the Flex with the Flex Eight below. As they alluded, it just isn't going to fit. If they do a balanced option in the future, it will have to be a Phoenix breakout...
1660322315618.png
 
Back
Top Bottom