• 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!

Shefffield

Member
Joined
Nov 30, 2021
Messages
44
Likes
42
Location
Munich, Germany
Hello, Amir! Hello, Forum!

I found my way to this website through Floyd Toole's book and Amir's excellent YouTube channel. I'm hoping to find a similarly grounded approach to our fascinating hobby here.

Who am I?

I guess my 'hifi career' has been fairly typical until roughly 12 years ago. The hifi bug got me in my teens, I couldn't afford the dream speakers, and Germany had a lifely DIY scene in the 90s. So I built several concepts designed by speaker brands and DIY magazines and spent a happy time listening without much demand for more. Once settled in a job, different city and own flat I began testing different concepts in earnest. At some point in my journey towards the 'best speaker' for myself I came to the conclusion that passive crossover design wouldn't cut it.

I pretty much skipped the passive XO design stage and tried to start directly with active crossovers. Before I could heat up my soldering iron, I stumbled upon Uli Brüggemann's phenomenal measuring and filtering tool Acourate. And saw a whole new world of potential.

What followed was a heated, but brief discussion with the (paid) hifi forum I was a member of. The forum hosts completely denied to even look at digital speaker development. But they hosted a yearly, open competition for hobbyist speaker developers. Nothing to win, beside praise or friendly critique and a good time among like-minded people. I took the bait and felt challenged to enter with a fully digital concept. With valuable help from a friend (wouldn't have been possible without him!) the speakers got ready at the last minute. And the massive computer, multi channel DAC and amping setup needed to run it. The rest is history (see Avatar pic).

For the contest I wrote a lengthy paper about the design concept, but that's in German. I can sum up the basics for you: One target for Zoé was a combination of drivers and cabinet that showcases the possibilities of super steep, linear phase, substractive digital crossovers. And also massive cone area with a small footprint.

Tweeter: Ciare 1.38 TW2, 38 mm, fs 1300 Hz - crossed over between 1600 and 1750 Hz (different setups)
Midwoofer: B&C 8NW51, 20 cm, fs 74 Hz, crossed between 100-120 Hz and 1600-1750 Hz
Subwoofer: Beyma 15G450/N, 38 cm, fs 44 Hz, crossed over at 100-120 Hz, enclosed in ca. 30 l (closed), EQ'd for a flat response down to 25 Hz

Since 2010 Zoé went through multiple software upgrades and received refined filters that also allow time alignment for all drivers.

Ah, and of course Acourate offers clever room correction.


What am I looking for?

Zoé still is a well-behaved speaker that is great fun with good recordings and brutally reveals flawed ones, but I am well aware of its limitations. With my next project I'm aiming to optimise the parameters that I can not (or only to a small degree) correct digitally: resonances and directivity.

I'm looking forward to sharing ideas and experiences with you!

Servus from snow-covered Bavaria,
Axel
 
OP
Shefffield

Shefffield

Member
Joined
Nov 30, 2021
Messages
44
Likes
42
Location
Munich, Germany
Welcome!
Do you mean dsp controlled by fully digital?

Yep.

The DSP in this case is completely done in software. The process is in general:

Step one: Define Crossovers. Initially I used Neville-Thiele 2nd order (200 dB/oct), but recently Uli has upgraded Acourate's capabilities with a new filter family by his own design. Slopes are comparatively steep ('brickwall filters').

Step two: Time alignment between drivers by near field measurements.

Step three: Define a target response curve at the listening position and apply appropriate ilters (handled by macros).

Step four: Correct for preringing and other artifacts, if necessary.

The filters are FIR and therefor subtractive and linear phase. Each driver is fed by an individually filtered input signal to apply XO and room correction together.

To make such a speaker run, you need measurement equipment, a computer to apply the filters on the fly, a multi channel DAC and enough amping channels. And you have to get a long with a typically ca. 700 ms delay between music signal input and speaker output.
 

ppataki

Major Contributor
Joined
Aug 7, 2019
Messages
1,216
Likes
1,358
Location
Budapest
Just a side-question if I may: 200dB/octave filters in linear phase mode - don't they introduce pre-ringing?
I have tried several linear phase VST filters and even a 24dB/octave linear phase HPF causes an audible (and measurable) pre-ringing artifact
 

fluid

Addicted to Fun and Learning
Joined
Apr 19, 2021
Messages
691
Likes
1,196
Just a side-question if I may: 200dB/octave filters in linear phase mode - don't they introduce pre-ringing?
I have tried several linear phase VST filters and even a 24dB/octave linear phase HPF causes an audible (and measurable) pre-ringing artifact
Complementary crossover filters cancel both pre and post ringing perfectly electrically, if the acoustic slopes match as well then it should work as well acoustically except for far off axis angles where the sum won't be quite right.

A single filter (or multiple) for EQ or high pass is not the same (non complementary) and all of the pre-ringing will be present and can become audible if it's high enough.
 

ppataki

Major Contributor
Joined
Aug 7, 2019
Messages
1,216
Likes
1,358
Location
Budapest
Thanks @fluid, are there any VST plugins that can create complementary crossover filters?
If I use one plugin on one driver (HPF) and the same plugin with exactly the same settings (but LPF) on the other driver then wouldn't this be complementary too?
 
Last edited:

fluid

Addicted to Fun and Learning
Joined
Apr 19, 2021
Messages
691
Likes
1,196
I would imagine all of the ones that can create filters that can be used as crossovers could do it. In this sense complementary just means the same HP and LP type, order and frequency are used. In reality matching the acoustic slopes to those of the desired electrical ones will need other equalization unless the drivers are extremely flat over the crossover region which is rarely the case. The steeper the slope of the crossover the smaller the flat bandwidth needs to be.

A good way to generate FIR filters for free if you don't want to pay for Acourate is rephase. Then you can use a convolver plugin or media player with built in convolver.

https://rephase.org/
 

ppataki

Major Contributor
Joined
Aug 7, 2019
Messages
1,216
Likes
1,358
Location
Budapest
I do use the same HP and LP type, order and frequency already so it shall be fine if I understand correctly then
 

fluid

Addicted to Fun and Learning
Joined
Apr 19, 2021
Messages
691
Likes
1,196
I do use the same HP and LP type, order and frequency already so it shall be fine if I understand correctly then
If the acoustic slopes of the crossover are equally complementary then yes. If they are not the same then the cancellation of ringing won't work as well.
 
OP
Shefffield

Shefffield

Member
Joined
Nov 30, 2021
Messages
44
Likes
42
Location
Munich, Germany
Thank you for your great input, fluid!

Same type/slope and frequency crossovers are the default in Acourate. If you bypass the macros you could however mix anything, the possibilities to put together "hand-crafted" filters are pretty much endless.

Acourate has a macro function to simulate the convolution and check for any pre-ringing. This can be addressed by modifying the filters, if necessary. I still have to play around with all the features though.

Greetings from South East Asia...
 

fluid

Addicted to Fun and Learning
Joined
Apr 19, 2021
Messages
691
Likes
1,196
Thank you for your great input, fluid!
You are welcome
Acourate has a macro function to simulate the convolution and check for any pre-ringing.
Acourate has really good pre-ringing compensation, pre-ringing of any kind can be completely eliminated.

If you haven't seen it Mitch has a really good video where he describes his use of Acourate
 

ernestcarl

Major Contributor
Joined
Sep 4, 2019
Messages
3,106
Likes
2,313
Location
Canada
And you have to get a long with a typically ca. 700 ms delay between music signal input and speaker output.

That is a lot of delay… have you tried to create a “minimal delay” working version of the same system? I mean, still using convolution along the way, sure, but way more economical yet still sounding good.
 
OP
Shefffield

Shefffield

Member
Joined
Nov 30, 2021
Messages
44
Likes
42
Location
Munich, Germany
That is a lot of delay… have you tried to create a “minimal delay” working version of the same system? I mean, still using convolution along the way, sure, but way more economical yet still sounding good.
Never felt the urge to do that. The delay doesn't bother me at all. It's just a small gap between pressing 'play' and hearing the music.
 

ernestcarl

Major Contributor
Joined
Sep 4, 2019
Messages
3,106
Likes
2,313
Location
Canada
Never felt the urge to do that. The delay doesn't bother me at all. It's just a small gap between pressing 'play' and hearing the music.

Ah. Okay... I was thinking more in the lines of multipurpose system where online streaming of both and audio and video could also be performed without a hitch whenever desired.
 

Zinda

Member
Joined
Apr 30, 2022
Messages
19
Likes
3
Hello, Amir! Hello, Forum!

I found my way to this website through Floyd Toole's book and Amir's excellent YouTube channel. I'm hoping to find a similarly grounded approach to our fascinating hobby here.

Who am I?

I guess my 'hifi career' has been fairly typical until roughly 12 years ago. The hifi bug got me in my teens, I couldn't afford the dream speakers, and Germany had a lifely DIY scene in the 90s. So I built several concepts designed by speaker brands and DIY magazines and spent a happy time listening without much demand for more. Once settled in a job, different city and own flat I began testing different concepts in earnest. At some point in my journey towards the 'best speaker' for myself I came to the conclusion that passive crossover design wouldn't cut it.

I pretty much skipped the passive XO design stage and tried to start directly with active crossovers. Before I could heat up my soldering iron, I stumbled upon Uli Brüggemann's phenomenal measuring and filtering tool Acourate. And saw a whole new world of potential.

What followed was a heated, but brief discussion with the (paid) hifi forum I was a member of. The forum hosts completely denied to even look at digital speaker development. But they hosted a yearly, open competition for hobbyist speaker developers. Nothing to win, beside praise or friendly critique and a good time among like-minded people. I took the bait and felt challenged to enter with a fully digital concept. With valuable help from a friend (wouldn't have been possible without him!) the speakers got ready at the last minute. And the massive computer, multi channel DAC and amping setup needed to run it. The rest is history (see Avatar pic).

For the contest I wrote a lengthy paper about the design concept, but that's in German. I can sum up the basics for you: One target for Zoé was a combination of drivers and cabinet that showcases the possibilities of super steep, linear phase, substractive digital crossovers. And also massive cone area with a small footprint.

Tweeter: Ciare 1.38 TW2, 38 mm, fs 1300 Hz - crossed over between 1600 and 1750 Hz (different setups)
Midwoofer: B&C 8NW51, 20 cm, fs 74 Hz, crossed between 100-120 Hz and 1600-1750 Hz
Subwoofer: Beyma 15G450/N, 38 cm, fs 44 Hz, crossed over at 100-120 Hz, enclosed in ca. 30 l (closed), EQ'd for a flat response down to 25 Hz

Since 2010 Zoé went through multiple software upgrades and received refined filters that also allow time alignment for all drivers.

Ah, and of course Acourate offers clever room correction.


What am I looking for?

Zoé still is a well-behaved speaker that is great fun with good recordings and brutally reveals flawed ones, but I am well aware of its limitations. With my next project I'm aiming to optimise the parameters that I can not (or only to a small degree) correct digitally: resonances and directivity.

I'm looking forward to sharing ideas and experiences with you!

Servus from snow-covered Bavaria,
Axel
I have gone through some similar exploration seeking to find the absolute best sound possible for under $2000. I too went with an active crossover simply because it's the easiest way to make changes and have immediate results, but there are other bonuses that come along with it but I found the DSP route (tried mini DSP, Beringer UltraGain and UltraCurve and a Brazilian DSP) to have flaws that diminished the sound dramatically.

I heard such a huge loss in all the parts of the music that made it sound real and the tiny trailing sounds like a whisper or the end of a lone string fading was being compromised far too much. All 3 were unexceptable and the next step up was far too expensive and on paper offered nothing more than a bit more processing on hand.

The change from analog to digital and back was the main problem. My conclusion is, if you're using a computer or any DSP and you can't hear the difference between all analog meaning analog EQ (if even needed, my 31 band deviates slightly in a few places) and an analog active xover.

I tried the 3 way as you have mentioned, I tried 2x 8" low mids (70Hz to 660Hz) a bunch of different low bass drivers before settling on a 10". Of course the only high frequency driver I'll ever use is an electrostatic driver, in my case 14 panels per side run from 600Hz to 24KHz. After countless hours of testing I found that the low mids were just causing problems and I removed them. I added a 12" infinity Kappa woofer and a 10" Monitor Audio sub per side, 2 huge old QSC amps power them.

The point I was making is if you can't hear the difference when using the DSP, (as most people can't) your system may not be able to produce those realistic sounds I'm speaking of. In my system the DSP is intolerable and I have sold off most of my digital sound stuff that I could, still have a biamp audia thats been posted for sale for 6 months with no interest! I also found noise introduced when my computer is running so I will never use my computer for anything audio.

My system is dead silent and as others have stated, I don't have to worry about blowing any high frequency drivers since electrostats can run till the roll off naturally (at least mine can). There is no comparing tweeters (especially 2 of them) to electrostat panels (especially 28 of them).

Still in the experimental enclosures for testing but ready to start making full cabs this week. I'm definitely building with removable motorboards per driver in sealed cabs, in case I change drivers, I can easily replace the entire piece and add sand bags or a new back piece inside to alter volume.

I will make sure that my motorboard and back are not parallel and 1 side will have an angled piece mounted on standoffs and braced to all 4 adjacent sides to break up any possible standing waves or cancelling waves. When I design cabinets I make sure to take advantage of every precaution and allowing for any changes to be easily made even after they are in use.
Make em big and you can always cut them down if you want.

IMG_20220430_085631.jpg


This will all be 1 speaker, the blue panels will be mounted along the outer edge forming a line array next to the woofers. The 10 larger panels
will be removed from cabinets and mounted above the woofers held in place with a wire frame that can be moved to aim panels. I have some 8" mirage subs I might use instead but I'll need 2 more if they test good, they will reduce cab size dramatically or I can just use the original BPS150i cabs and modify them so both drivers are front mounted. I'll scrap the plate amp and mount the dual mono QSC's on the back of each cab.

Ive been testing combinations of drivers for 2 years now and reached a sonic perfection that I have not heard before, even when there used to be high end stores still in business, I never heard anything close even at prices over $16k just for speakers, mine have more volume and retain every last bit of sound on every recording. I can add that I'm able to play any music and even poorly recorded songs come alive with the exception of possibly an original cassette tape source and 1 Queen greatest hits with live tracks that sounds like it was recorded on a portable cassette tape recorder. My neighbors can corroborate these facts, since they now listen to whatever I'm playing, without choice.

i guess what I'm saying is try your speakers without any DSP before you proceed, I'm sure you will come to the same conclusion, also most subs I tested tend to be near flat up to 3KHz, running a 2 way greatly improved my overall flat response and removed a phase issue that was impossible to fix. No matter what I did I had 1 driver that was always out of phase. I can't explain it but I could fix 1 side but the other ended up with 1 driver unable to match the rest or match the opposite side. Here's the other big issue you addressed. Time alignment, 2 speakers is simple to correct without the need for any delay, just move the tweeter back 4 inches (if you actually think you can tell the difference). Try testing it, 2 way system, mount a tweeter on top on piece of wood, put a piece of cloth even with the motorboard of the woofer. Have a friend move it back and forth untill you think it's in 1
perfect sync. If he moves it back and front and you can't tell, you'll be like 90% of us. If it ends up even with the woofers motor then you have super hearing of you just told him to stop by luck. I moved mine 1 foot and heard no difference. Set DSP for a 25" difference and it sounded the same even reversing the drivers delay it sounded in sync, but 1 foot delay sounded off, 4 inches or zero had no effect. Automatic delay setting had set mine to 50" delay on 1 side! DSP was disconnected and sold yesterday.

I wish you good luck with you speakers and hope you find your sonic perfection, I know it's not easy and it's not always going to be found by working out long formulas or special programs, I have found that hands on is the only way to be sure you covered all bases. A good test mic that gives exact repeatable results and at least 2 programs that show exact same results using that same mic is the only way to know you're at a good starting point. Only problem is I haven't found 2 test programs that even come close to the same results, my faith in test programs is non existent. Untill they all read the exact same results, I won't be working with digital and it's flawed design. Analog shows me consistent results calibrated sound meters seem to be the only true results than can be trusted, repeatedly.
 

Christoph-ASR

Member
Joined
Jun 20, 2022
Messages
43
Likes
18
Location
NYC
Top Bottom