Welcome, Guest
Username: Password: Remember me
  • Page:
  • 1
  • 2

TOPIC: The Darbari: a sealed 3-way 2-box active speaker

The Darbari: a sealed 3-way 2-box active speaker 5 years 1 month ago #15732

  • tcpip
  • tcpip's Avatar
  • Offline
  • Senior Boarder
  • Posts: 43
  • Thank you received: 3
  • Karma: 0
Hi,
I've been working on my first MiniDSP based project, also my first 3-way, and my first fully active speaker design. I call it the Darbari. See this page for full details (there are multiple Web pages) and photos.
I am using the MiniDSP 4x10HD, and it works as advertised. The live configuration is quite amazing, actually.
Am facing some challenges getting the final configuration of the crossover tuned.




The administrator has disabled public write access.
The following user(s) said Thank You: devteam

The Darbari: a sealed 3-way 2-box active speaker 5 years 1 month ago #15735

  • john.reekie
  • john.reekie's Avatar
  • Offline
  • Platinum Boarder
  • Posts: 3485
  • Thank you received: 1501
  • Karma: 137
Hi there, nice work!

Unibox does show the woofer to reach Xmax at 50 Hz with 100 W. There's enough volume to use a pair (and you can use a Linkwitz transform to make the Q whatever you want, within reason.) Alternatively you have the fourth pair of outputs on the 4x10 for a subwoofer ;)

What are the challenges you're still facing? (The link to "Crossover tuning" gives a not found error)
I am not miniDSP support.

"You must ask the right questions." - Dr. Alfred Lanning's hologram.
-> Have you read the User Manual??
-> Have you drawn and posted a diagram?
-> Have you posted a screenshot?
-> Have you posted your config file?
The administrator has disabled public write access.
The following user(s) said Thank You: tcpip

The Darbari: a sealed 3-way 2-box active speaker 5 years 1 month ago #15768

  • devteam
  • devteam's Avatar
  • Offline
  • Administrator
  • Posts: 8539
  • Thank you received: 1054
  • Karma: 157
That's a very serious looking speaker you've got here! Quite a lot of work on the woodworking I'm guessing. :-)
Glad to hear that the 4x10 is working as you expected! For filter design, I wonder if that dip for the midrange is some internal reflection. Have you looked into it? (e.g. more stuffing material). Just a wild thought not knowing much of your speaker.. You can easily see that with a calculation of inside backbox distance if I recall right.. Quite few years away though so might be wrong.. Someone like John is a lot more astute with speaker design!

DevTeam
MiniDSP, building a DIY DSP community one board at a time.
The administrator has disabled public write access.
The following user(s) said Thank You: tcpip

The Darbari: a sealed 3-way 2-box active speaker 5 years 1 month ago #15782

  • tcpip
  • tcpip's Avatar
  • Offline
  • Senior Boarder
  • Posts: 43
  • Thank you received: 3
  • Karma: 0
The Xmax with 100 W output is as per spec, and I am not complaining. I am surprised at how low (in SPL terms) this sounds. I was under the impression that your average driver fed 100 W of power will generate SPL to uncomfortable levels in-room. I am getting nowhere near that at 50 Hz with this woofer. A purely subjective complaint. I have never measured any other speaker at 50 Hz in-room for subjective assessment, so I guess this is my inexperience speaking. Does this mean that I need a 15-incher as a fourth channel?

Sorry about the crossover tuning page -- it was a placeholder link till now (I always keep that sort of link for my "next-to-come" page). It's up now. tcpip.dhandanought.org/node/58 You will see the challenges I am facing.

Right now, I wish I could clean out that "reverb" feeling I am getting with crisp clean male voices -- they are not crisp and clean at all, compared to what I've always gotten with my 2-way projects. :( This is quite upsetting. Setting a delay on the midrange (and a 0.06 msec additional delay on the tweeter, as per earlier tunings) has helped to reduce the problem quite sharply, but I want it as good as a good 2-way. Is that an impossible target with my speakers? And given the difficulty I am facing with this "reverb" thing, I'll never put another 10-inch in parallel with this one to increase SPL -- it will only make the midrange-woofer interaction worse, probably increasing the "reverb" feeling. :(

And the recessed midrange needs to be tackled. I am going to experiment with a few dB of boost using a parametric eq on the input channels, and I'll see if I can get something.

A very basic confession: I am not getting as much emotional enjoyment from the sound as I used to get from my earlier projects. The sound is probably too brassy or too sharp. It's not sharp enough to be an obvious problem -- no real severe listening fatigue. But it's something somewhere -- it is not allowing me to relax with the music. I am going to try cutting the hump at 6 KHz a bit, maybe steepen the cut in the existing notch (a notch is already there at 6 KHz) in the tweeter parametric eq. But this will only take care of any sharpness coming from the high freq. I am also unhappy with the quality of male vocals, due to the lack of clarity. Female vocals (think Joan Baez, "Diamonds and Rust") are just fine. A slightly lower-scale female vocal example (Diana Krall, "A Case of You" from her "Live in Paris" album) sounded a slight bit hard to decipher-the-words the last time I checked.
Last Edit: 5 years 1 month ago by tcpip.
The administrator has disabled public write access.

The Darbari: a sealed 3-way 2-box active speaker 5 years 1 month ago #15786

  • john.reekie
  • john.reekie's Avatar
  • Offline
  • Platinum Boarder
  • Posts: 3485
  • Thank you received: 1501
  • Karma: 137
Hi again, the term "reverb" is a bit of a worry, it suggests something seriously wrong... Would you mind posting your configuration file? (Add .txt to the file name.) I'm sure it's fine but it never hurts to have another pair of eyes have a look.

Please don't take the following the wrong way, I think I read that you have an engineering background so you understand the need to have a method of moving from a known starting point and isolating variables. Or just think of this as how I think I would proceed :)

While you sort out where the issues are, it might be helpful to directly alter the settings in the miniDSP rather than use ACD. Once you have a better understanding of what's going on, use ACD for the final tuning. The reason being that things like different crossover frequencies as I suggest below will be easier and quicker. The basic approach is to equalize the drivers flat past the crossover point and then insert the crossover. Here's my worked example: www.hifizine.com/2014/09/mini-convertible-crossover/

I would suggest getting the mid-tweeter right first, without the woofer. There will not be much low bass of course but f3 of something like 90 is not too bad. LIstening nearfield (eg 1m) will help remove the room effects from what you are hearing, when subjective auditioning.

I would consider using a passive notch filter on the RS150. At 9-10 kHz, the components will not be large and you may even have them leftover from previous projects,

I would definitely try higher mid-tweeter crossover points. A second order HPF at 1500 seems like it might be too little, too low, for that tweeter. Perhaps try 4th order (electrical) at 2500, and then maybe 2000 and 3000. I'd only be guessing that this is causing some of the strange perceptions but I did go through a similar experience with a different tweeter where I ended up moving the crossover from 1500 all the way up to 3500.....! A distortion measurement on the tweeter may help see if that is the cause. You can save different configurations for quick switching.

I think I read that you have only a small amount of stuffing in the mid enclosure. I would put in a lot, to absorb energy from the rear of the cone as much as possible.

Some off-axis measurements, both of individual drivers and combined with the crossover, will help I think. Even if you change nothing, it's better to see "past the edges."

Then add the woofer. I'd do measurements fairly close but halfway between the centers of the woofer and mid. Unless you can do the measurements outside.

For the BSC, it's a lot easier to handle if you put it in the input EQ block. I only realized this recently...

When doing subjective listening/assessment with the woofer included, beware of room effects. They can easily make +/- 10 dB difference at 50 Hz. I'd compare the nearfield measurement with the in-room measurement - if the latter has a considerable dip in the bass then moving the LP or the speaker is probably necessary to get a better idea of the speaker's real performance.

If by "Does this mean that I need a 15-incher as a fourth channel" you mean a 4-way, I would suggest considering 3-way plus sub/s rather than straight 4-way, for several reasons. Moving a driver at those frequencies can have a dramatic effect, and it's much easier with separate subs/s than if the driver is part of the speaker.

Rather than a single 15, I would suggest a pair of 12s in opposed configuration. That way mechanical vibration into the residence' structure is minimized.

OK, well, that's about everything I've learned over the least 4 years ! Hope this might help you.
I am not miniDSP support.

"You must ask the right questions." - Dr. Alfred Lanning's hologram.
-> Have you read the User Manual??
-> Have you drawn and posted a diagram?
-> Have you posted a screenshot?
-> Have you posted your config file?
The administrator has disabled public write access.

The Darbari: a sealed 3-way 2-box active speaker 5 years 1 month ago #15788

  • tcpip
  • tcpip's Avatar
  • Offline
  • Senior Boarder
  • Posts: 43
  • Thank you received: 3
  • Karma: 0
I will positively love to get any comments and feedback on my MiniDSP config file. Here it is.

Will respond to the rest of your email in the next post.
Attachments:
The administrator has disabled public write access.

The Darbari: a sealed 3-way 2-box active speaker 5 years 1 month ago #15789

  • tcpip
  • tcpip's Avatar
  • Offline
  • Senior Boarder
  • Posts: 43
  • Thank you received: 3
  • Karma: 0
john.reekie wrote:
Hi again, the term "reverb" is a bit of a worry, it suggests something seriously wrong... Would you mind posting your configuration file? (Add .txt to the file name.) I'm sure it's fine but it never hurts to have another pair of eyes have a look.
What I mean by reverb is by now clear to my mind, having done some explorations. It is basically two copies of the same audio output emerging from two drivers, time-misaligned. I am sure they are creating both phase mismatch problems and timing problems.

What is intriguing is the amount of time misalignment which must be happening. If the misalignment is very small (I don't know how to quantify it), the ear will just see the misalignment as phase misalignment and its effects. There needs to be a fairly large time misalignment for the ear to distinguish the two separate sounds sufficiently far apart to hear them as a kind of reverb or echo. I had read once that the ear hears a reverb if there is a gap of quite a few milliseconds -- maybe 10 msec or more. So, this thing intrigues me. The fact that I've licked it 80% by giving a time delay of 2 msec to the midrange makes me wonder... have I licked it at all or is there something bigger to fix?
Please don't take the following the wrong way, I think I read that you have an engineering background so you understand the need to have a method of moving from a known starting point and isolating variables. Or just think of this as how I think I would proceed :)
Oh absolutely agree. I will go insane if I try debugging by just floundering about blindly. :)
While you sort out where the issues are, it might be helpful to directly alter the settings in the miniDSP rather than use ACD. Once you have a better understanding of what's going on, use ACD for the final tuning.
This is exactly what I was planning to do. I wanted to fix things by sitting on the MiniDSP console and changing settings, but later go to ACD and generate a whole new set of biquads by feeding the stuff into the spreadsheet again.

I am facing one problem already -- I can't do certain changes sitting on the MiniDSP console, because some of the things I've configured till now are already in the form of biquads and can't be modified from the GUI. For instance, the stuff I have done on the midrange or tweeter crossover is all biquads. However, I can at least play with the parametric eq from the GUI and see if it fixes anything. Ditto, time alignment changes.
The basic approach is to equalize the drivers flat past the crossover point and then insert the crossover.
This is what I did for the midrange low-pass. Fixed the cone breakup with notch filters and then added the low-pass slope. But I haven't done that for everything else.
Will definitely take a look.
I would suggest getting the mid-tweeter right first, without the woofer. There will not be much low bass of course but f3 of something like 90 is not too bad. LIstening nearfield (eg 1m) will help remove the room effects from what you are hearing, when subjective auditioning.
This is what I tried doing actually, at least in the modelling of the crossover. I first modelled just a two-way crossover, played around to get the reverse-null right (by playing with the timing difference between mid and tweeter) and then added the woofer to the two as a separate step. I think I will now go back and do some listening after muting the woofer. Great idea.
I would consider using a passive notch filter on the RS150. At 9-10 kHz, the components will not be large and you may even have them leftover from previous projects,
Why is this required? I am really hoping I won't have to play with passive stuff right now.... :(
I would definitely try higher mid-tweeter crossover points. A second order HPF at 1500 seems like it might be too little, too low, for that tweeter. Perhaps try 4th order (electrical) at 2500, and then maybe 2000 and 3000.
Yes, I too was wondering whether these would help get the high-frequency "glare" or "hardness" out. I intend to flatten the current hump at 6KHz with a bit more eq, and then play with the tweeter slope and finally try different crossover points.
I think I read that you have only a small amount of stuffing in the mid enclosure. I would put in a lot, to absorb energy from the rear of the cone as much as possible.
No, actually, I've put in quite a bit of stuffing there. I've left some space for a few inches just behind the driver magnet, and from that point till the rear wall, I've put a lot of stuffing.
Some off-axis measurements, both of individual drivers and combined with the crossover, will help I think. Even if you change nothing, it's better to see "past the edges."
Ok, will add these to my to-do list.
Then add the woofer. I'd do measurements fairly close but halfway between the centers of the woofer and mid. Unless you can do the measurements outside.
Yes, will need to do the whole midrange-woofer thing again. I will need to get the time alignment spot on.
For the BSC, it's a lot easier to handle if you put it in the input EQ block. I only realized this recently...
Yes, I too have stumbled on the same conclusion after playing with the console. I will pull the stuff out of the midrange xo and into the input PEQ blocks.
When doing subjective listening/assessment with the woofer included, beware of room effects. They can easily make +/- 10 dB difference at 50 Hz. I'd compare the nearfield measurement with the in-room measurement - if the latter has a considerable dip in the bass then moving the LP or the speaker is probably necessary to get a better idea of the speaker's real performance.
I have not even reached the point where I compare near-field with in-room. Currently, whether I am eight feet away from the speakers or three feet away, the "reverb" like effect is audible.
If by "Does this mean that I need a 15-incher as a fourth channel" you mean a 4-way, I would suggest considering 3-way plus sub/s rather than straight 4-way, for several reasons.
Actually, I will not be embarking on any such modifications to these speakers (or extensions) right now... I need to get the current configuration to give me the best that it can, and I need to be convinced about my own ability to tackle such a project before I add one more driver. :)

Looking forward to hearing more from you. :)
The administrator has disabled public write access.

The Darbari: a sealed 3-way 2-box active speaker 5 years 1 month ago #15790

  • john.reekie
  • john.reekie's Avatar
  • Offline
  • Platinum Boarder
  • Posts: 3485
  • Thank you received: 1501
  • Karma: 137
Hello again, gosh, this impression of "reverb" seems to be the biggest worry. Perhaps first eliminate electrical issues by say using the same source, 4x10 (crossovers disabled), and an amp with an existing (full range) speaker? If that's sounds fine, then flatten the mid say 100 Hz -10 khz and have a listen to that alone? Then add tweeter etc. I'd be grasping at straws to try and explain what you're hearing, so hopefully there's a way to find a configuration that doesn't have that effect, and then reintroduce things until it reappears.

I took a quick look at the config file. Sorry I'm a bit out of time and will look again tomorrow but I didn't see any real issues, other than the levels and crossover woofer-mid (based on assumptions and specs). I'm not really sure what I was looking for, but it doesn't hurt to look...

Sorry for misunderstanding about the stuffing in the mid enclosure.

I wouldn't say the passive notch is "necessary" ... it's just something I'd try?
I am not miniDSP support.

"You must ask the right questions." - Dr. Alfred Lanning's hologram.
-> Have you read the User Manual??
-> Have you drawn and posted a diagram?
-> Have you posted a screenshot?
-> Have you posted your config file?
Last Edit: 5 years 1 month ago by john.reekie.
The administrator has disabled public write access.

The Darbari: a sealed 3-way 2-box active speaker 5 years 1 month ago #15791

  • tcpip
  • tcpip's Avatar
  • Offline
  • Senior Boarder
  • Posts: 43
  • Thank you received: 3
  • Karma: 0
Thanks a ton for the amount of time you are putting in. Really makes a difference to my confidence level.

I have decided to create a new xo config v2 from scratch. I will start with the raw FRD files. I will change the following:
  • I will linearise tweeter and midrange more carefully, then put a fairly steep xo between them, and just listen to the configuration keeping the woofer out of circuit.
  • I will then linearise the mid and woofer more carefully than I did last time. In particular, I will use notches to cut the woofer's cone breakup region. I will then put them together using fairly steep crossovers.
  • I'll listen to just mid+woofer and see what it sounds like.
  • I will put BSC and overall tonal balance things in the input PEQ phase.
  • I will take fresh gated farfield measurements of woofer and mid, to see what I can get in terms of their time alignment.
Let's see what I get. I think I will only be able to do all this over the weekend. The day job takes away all free time, and my seven-year-old claims the rest. I will definitely post all results here and on my website.
Last Edit: 5 years 1 month ago by tcpip.
The administrator has disabled public write access.

The Darbari: a sealed 3-way 2-box active speaker 5 years 1 month ago #15849

  • tcpip
  • tcpip's Avatar
  • Offline
  • Senior Boarder
  • Posts: 43
  • Thank you received: 3
  • Karma: 0
Hi John (and others),

I seem to have solved my problems. I re-did the crossover, and I can listen happily now. Full details are here:

tcpip.dhandanought.org/node/58

Thanks a lot for help and encouragement. :)

My observations about MiniDSP instability are quite interesting and unexpected. I wonder whether others have seen such things too?

And any suggestions to help me reproduce hard rock drums realistically will be a great help. I seem to have goofed up in that department.
Last Edit: 5 years 1 month ago by tcpip.
The administrator has disabled public write access.

The Darbari: a sealed 3-way 2-box active speaker 5 years 1 month ago #15862

  • john.reekie
  • john.reekie's Avatar
  • Offline
  • Platinum Boarder
  • Posts: 3485
  • Thank you received: 1501
  • Karma: 137
Hi again, glad to hear you're making progress :)

Have you done in-room measurements? The bass region is all (well, a lot) about the room.
I am not miniDSP support.

"You must ask the right questions." - Dr. Alfred Lanning's hologram.
-> Have you read the User Manual??
-> Have you drawn and posted a diagram?
-> Have you posted a screenshot?
-> Have you posted your config file?
The administrator has disabled public write access.

The Darbari: a sealed 3-way 2-box active speaker 5 years 1 month ago #15875

  • tcpip
  • tcpip's Avatar
  • Offline
  • Senior Boarder
  • Posts: 43
  • Thank you received: 3
  • Karma: 0
The bass measurements are the next phase. First I will sit back and listen for a few weeks, and tweak the rest of the response curve to get the sound to my liking. Am lazy. :)

I found a bit of a glare in the sound, so I put in a treble shelf from about 2000 Hz upwards, gain of -2 dB, Q of 0.7. Within a few hours, I realised that the sound had become "uninteresting", and in some tracks, "muddy". So I went back and shifted the frequency from 2000 Hz to 5000 Hz. Everything else is unchanged. Immediately, the sound became better -- it's the right mix of details and smoothness which I keep striving for. This is the kind of tweaking I aim to do in the next couple weeks.

BTW, can you tell me something about the shelving filters in the PEQ section of the Input blocks? These have a "frequency" setting, but this frequency does not seem to be the turnover frequency -- they seem to be the middle-frequency of the sloping portion. Am I right? Also, what's the slope of these shelving filters? Are they 12 dB / oct?

Another question -- about room equalisation this time. Have you used Speaker Workshop? I guess I can't do in-room far-field measurements easily using Speaker Workshop, right? Ideally, I should use REW, am I right? One reason I am asking this is because Speaker Workshop does not seem to have a facility to get a frequency response curve from a swept-sinewave signal. And I am not at all sure, but I am getting the feeling that a swept-sine signal is better than an impulse based measurement for really low frequencies far into the room. Any thoughts?
Last Edit: 5 years 1 month ago by tcpip. Reason: Added question about SW vs REW for DRC
The administrator has disabled public write access.

The Darbari: a sealed 3-way 2-box active speaker 5 years 4 weeks ago #15881

  • john.reekie
  • john.reekie's Avatar
  • Offline
  • Platinum Boarder
  • Posts: 3485
  • Thank you received: 1501
  • Karma: 137
Hi again, not lazy, sensible!

I have no experience with Speaker Workshop. Perhaps someone else does. Swept sine should be more immune to noise but at low frequencies you can get ambient showing up anyway if conditions are not good.

The frequency of the shelving filters is halfway. The slope depends on the Q - I couldn't tell you what the mapping is exactly, I just adjust the Q to get the compensation I want. I think Q of 0.5 might be 6dB and 12dB would be... 0.7?
I am not miniDSP support.

"You must ask the right questions." - Dr. Alfred Lanning's hologram.
-> Have you read the User Manual??
-> Have you drawn and posted a diagram?
-> Have you posted a screenshot?
-> Have you posted your config file?
The administrator has disabled public write access.

The Darbari: a sealed 3-way 2-box active speaker 5 years 4 weeks ago #15885

  • tcpip
  • tcpip's Avatar
  • Offline
  • Senior Boarder
  • Posts: 43
  • Thank you received: 3
  • Karma: 0
Thanks for the details about the shelving filters. But I think it is highly likely that the Q will affect the slope of the filters. Wish someone from their product team would enlighten us, and more important, would update their manuals. :(

Try giving a massive boost or cut, of maybe 30 dB (that's about the max you can do from their GUI. See what the slope looks like, visually, in the region in the middle, away from either end. Then change the Q and see if the slope in the middle region changes. I doubt that it does.
Last Edit: 5 years 4 weeks ago by tcpip.
The administrator has disabled public write access.

The Darbari: a sealed 3-way 2-box active speaker 5 years 4 weeks ago #15886

  • john.reekie
  • john.reekie's Avatar
  • Offline
  • Platinum Boarder
  • Posts: 3485
  • Thank you received: 1501
  • Karma: 137
Hi again, by "their GUI" do you mean the miniDSP 2x8 plugin?
I am not miniDSP support.

"You must ask the right questions." - Dr. Alfred Lanning's hologram.
-> Have you read the User Manual??
-> Have you drawn and posted a diagram?
-> Have you posted a screenshot?
-> Have you posted your config file?
The administrator has disabled public write access.
  • Page:
  • 1
  • 2
Moderators: devteam