4-way with dual 21", dual 15" & dual 8" + TPL-150. An IBWWMTM...

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
wow another big drivers lovers :D big sound does not come from sissy driver

https://www.diyaudio.com/forums/sol...-active-minidsp-4-amplifer-4.html#post6030415

21" - 12" - 5" - small ribbon, all controlled by 2 of minidsp 2x4HD. plus dual-opposed subwoofer, and finishing 2nd subwoofer set.

my setup is not as big as yours but i'm sure you will get very decent result later. keep us posted

Thanks for sharing! Big drivers are for big boys...;)
How do you hook up your Two 2 x 4 HD's?
 
Thanks for sharing! Big drivers are for big boys...;)
How do you hook up your Two 2 x 4 HD's?

dcb1 is the center hub connection which drives : 2 minidsp, 1 subwoofers controller for behringer, later will put LXMini board. there is an additional liner supply using LT1085 to replace minidsp power supply. but my casing is too small, need bigger one to fit all boards

a few years ago, i build 2 way OB with 2x15" and fullrange. this will be revisited later utilizing LXmini board, i need to buy 4 Eminence beta 15a to accompany my existing AN classic 8". another on going project is Elsinore speaker

so there will be 3 sets of speakers and all of them will be driven by their own preamp and amplifier. my 6m wide wall will be fully occupied

in terms of 4 way amplifier, i'm still using 4 separate amplifier which is too hassle on rca and binding post connector. in the future i think that building 4way monoblock amplifier will be beneficial, especially when minidsp is put inside amplifier chassis. use 8pole neutrik speakon then you can change amplifier configuration from FIR to IIR

it is also beneficial for FIR to combine subwoofer & tweeter on 1 board and allocate more tabs to subwoofer. or when using IIR then set into 4ways. it can be set simply with speakon, just ensure it has enough cabling to switch between monoblock
 

Attachments

  • 20191215_100326.jpg
    20191215_100326.jpg
    805.8 KB · Views: 390
  • ACR_Alpair.jpg
    ACR_Alpair.jpg
    616.7 KB · Views: 396
Would 2 pcs MiniDSP 2 x 4 HD work?

Hi ZACC,
I believe the MiniDSP 2x4 HD machine is even less capable for FIR performance than the OpenDRC-DA8 machine you mentioned originally. It has fewer FIR taps available (1024 vs 1200) per equal channel, and only 4 channels (but you already knew that - hence your suggestion to buy two machines to give you the necessary 8 channels!)

In principle, buying multiple machines will not alter the inherent DSP power or FIR tap limits, it merely gives you more channels to play with, each having the same FIR correction ability as before.

However, since both those machines can be programmed to allocate more than a fair share of taps to a few particular channels (if willing to sacrifice the FIR on the other channels), it's possible to squeeze 2048 taps performance from them. Both are equally capable of achieving that, so whichever of these machines, MiniDSP 2x4 HD or OpenDRC-DA8 you choose, that maximum limit of 2048 taps per channel is the same absolute max FIR performance ceiling.
That is where you'll start to see the benefit of buying multiple machines because you'd then have enough channels to reach maximum DSP performance for every channel you need.

So for the OpenDRC-DA8...
ONE machine would work for starters, but you'd need TWO machines to run 8 channels with maximum 2048 tap performance.

And for the MiniDSP 2x4 HD...
You'd need TWO machines just to get 8 channels (running 1024 taps each), but you'd need FOUR machines if you want 8 channels with maximum 2048 tap performance each.

Yes, there's also other options of sub-dividing the taps, allocating perhaps max 2048 taps to the woofers (which really need most) and fewer 1024 taps to the tweeters and mids (which might be correctable with less taps / shorter impulse, and just adding delay instead to compensate and time-align with the longer length impulse offset on the woofers), so you could maybe get away with just three MiniDSP 2x4 HD machines instead of four.
Likewise you might choose a pair of OpenDRC 2x2 machines (with 6144 taps per channel each) for your 21 inch subwoofers and 15 inch woofers, but just one single more economical OpenDRC-DA8 running 2048 taps on 4 channels for the mids and highs. Also depends whether you prefer the unbalanced consumer RCA connections vs professional XLR setups.

Ultimately, any of these solutions can be made to basically work (if you mean basic functionality as a crossover / driver EQ / phase correction) but with perhaps 1024 taps will certainly be only a very loose approximation of the bass driver response at low frequencies. Depends whether you're an audiophile purist or just want a fun, loud system for watching action movies on home cinema... maybe 100% accuracy isn't the main goal.

I'll say again that I personally don't believe even 2048 taps is enough to do a 100% accurate job here for FIR correction of measured response, especially with a 21 inch subwoofer, and I'd recommend looking carefully at the specs below to compare the different hardware options for an 8 channel solution. All the MiniDSP stuff is solid, reliable and easy to use though.

MiniDSP 2x4 HD (Just one machine)
4 channels
4096 taps total available DSP power
equals 1024 taps each channel (equally shared between 4 channels)
maximum 2048 taps allocated per channel (if allocated to only 2 channels running FIR)

MiniDSP 2x4 HD (Two machines side-by-side)
8 channels
8192 taps total available DSP power
still equals 1024 taps each channel (equally shared between 8 channels)
same maximum 2048 taps allocated per channel (if allocated to only 4 channels running FIR)

MiniDSP 2x4 HD (Four machines side-by-side)
16 channels
16384 taps total available DSP power
still equals 1024 taps each channel (equally shared between 16 channels)
same maximum 2048 taps allocated per channel (if allocated to only 8 channels running FIR)

OpenDRC-DA8 (Just one machine)
8 channels
9600 taps total available DSP power
equals 1200 taps each channel (equally shared between 8 channels)
maximum 2048 taps allocated per channel (if allocated to only 4 channels running FIR)

OpenDRC-DA8 (Two machines side-by-side)
16 channels
19200 taps total available DSP power
still equals 1200 taps each channel (equally shared between 16 channels)
same maximum 2048 taps allocated per channel (if allocated to only 8 channels running FIR)

OpenDRC-DI 2x2 (Just one machine)
2 channels
12288 taps total available DSP power
equals 6144 taps each channel (equally shared between 2 channels)
maximum 6144 taps allocated per channel (all channels used)

OpenDRC-DI 2x2 (Four machines side-by-side)
8 channels
49152 taps total available DSP power
equals 6144 taps each channel (equally shared between 8 channels)
maximum 6144 taps allocated per channel (all channels used)

How does one hook them up for a 4-way stereo set-up? 1 per side?

You can hook up multiple machines / channels however you wish, as long as you feed them from a common digital source (via digital split), they all stay in sync anyway because their internal SRC are independent.

Myself, for safety, I've preferred to put a big sticky paper label on one OpenDRC machine "TWEETER" doing my left tweeter and right tweeter corrections, and a second machine labelled "MIDRANGE" for those 2 channels, and a third machine for "WOOFER" and so on, keeping the driver families separated and visibly labelled to avoid confusion, rather than just grouping LEFT signals together for instance, because if I accidentally plugged my left tweeter into OpenDRC's left midrange or woofer channel by mistake, I'd probably blow up the tweeters! Whereas if I accidentally plug my left tweeter into the right tweeter corrected output, no real harm done, but I'd notice the stereo image sounds wrong and can trace the fault back.

Also, grouped by driver family that way, you can easily switch between the 4x front panel OpenDRC user presets, ie. for comparing different FIR tweeter corrections (bracket-testing different EQ responses, etc.) within the same "Tweeter" OpenDRC machine box, without fiddling with other machine's front panels, and it makes more logical sense dealing with tweeter corrections all in one tweeter box, easier preset comparisons, and less likely to go wrong if all you've loaded into that particular machine are tweeter presets, nothing else.
Whatever happens, don't blow those expensive AMT tweeters!!!
 
Last edited:
OpenDRC-DI 2x2 (Just one machine)
2 channels
12288 taps total available DSP power
equals 6144 taps each channel (equally shared between 2 channels)
maximum 6144 taps allocated per channel (all channels used)

OpenDRC-DI 2x2 (Four machines side-by-side)
8 channels
49152 taps total available DSP power
equals 6144 taps each channel (equally shared between 8 channels)
maximum 6144 taps allocated per channel (all channels used)



You can hook up multiple machine channels however you wish, as long as you feed them from a common digital source (via digital split), they stay in sync anyway because their internal SRC are independent. Myself, for safety I've preferred to label one OpenDRC machine "TWEETER" doing my left tweeter and right tweeter corrections, and a second machine labelled "MIDRANGE" for those 2 channels, and a third machine for "WOOFER" and so on, keeping the driver families separated and visibly labelled, rather than just grouping LEFT signals together for instance, because if I accidentally plug my left tweeter into OpenDRC's left midrange or woofer channel by mistake, I'd probably blow up the tweeters! Whereas if I accidentally plug my left tweeter into the right tweeter corrected output, no real harm done, but I'd notice the stereo image sounds wrong and can trace the fault back.
Also, grouped this way, you can easily switch between the 4x OpenDRC user presets, ie. for comparing different FIR tweeter corrections (bracket-testing different EQ responses, etc.) within the same "Tweeter" OpenDRC machine box without fiddling with the other machine's front panels, and it makes more logical sense, faster comparison and less likely to go wrong if all you've loaded into that particular machine are tweeter presets, nothing else.
Whatever happens, don't blow those expensive AMT tweeters!!!

Nice breakdown of the minidsp alternatives.
I used four openDRC's for stereo 4-ways for quite a while too.
Worked super for me too!
In fact, it worked as well as my current setup, q-sys on a Core110f, but the current setup makes mobile use alot easier.

Anyway, shameless plug for 3 opendrc-di's i have for sale...i kinda quit wanting/trying to sell them...but this thread reminds me i need to clean some shelves...:)
FS: minidsp OpenDRC's DA8 and DI's
 
MiniDSP 2x4 HD (Just one machine)
4 channels
4096 taps total available DSP power
equals 1024 taps each channel (equally shared between 4 channels)
maximum 2048 taps allocated per channel (if allocated to only 2 channels running FIR)

MiniDSP 2x4 HD (Two machines side-by-side)
8 channels
8192 taps total available DSP power
still equals 1024 taps each channel (equally shared between 8 channels)
same maximum 2048 taps allocated per channel (if allocated to only 4 channels running FIR)

MiniDSP 2x4 HD (Four machines side-by-side)
16 channels
16384 taps total available DSP power
still equals 1024 taps each channel (equally shared between 16 channels)
same maximum 2048 taps allocated per channel (if allocated to only 8 channels running FIR)

OpenDRC-DA8 (Just one machine)
8 channels
9600 taps total available DSP power
equals 1200 taps each channel (equally shared between 8 channels)
maximum 2048 taps allocated per channel (if allocated to only 4 channels running FIR)

OpenDRC-DA8 (Two machines side-by-side)
16 channels
19200 taps total available DSP power
still equals 1200 taps each channel (equally shared between 16 channels)
same maximum 2048 taps allocated per channel (if allocated to only 8 channels running FIR)

OpenDRC-DI 2x2 (Just one machine)
2 channels
12288 taps total available DSP power
equals 6144 taps each channel (equally shared between 2 channels)
maximum 6144 taps allocated per channel (all channels used)

OpenDRC-DI 2x2 (Four machines side-by-side)
8 channels
49152 taps total available DSP power
equals 6144 taps each channel (equally shared between 8 channels)
maximum 6144 taps allocated per channel (all channels used)

Whatever happens, don't blow those expensive AMT tweeters!!!

Again, thanks for your input! This was very educational.
Question; I can only see OpenDRC-DI with digital out... I need analog out to my M2's and other coming amps. My source will be a Cambridge DAC/Streamer with digital inputs and digital and analog outputs.
OpenDRC : OpenDRC-DI
 
Anyway, shameless plug for 3 opendrc-di's i have for sale...i kinda quit wanting/trying to sell them...but this thread reminds me i need to clean some shelves...:)
FS: minidsp OpenDRC's DA8 and DI's

Nice shameless plug!
Tell me more about the "* Units #2, 3, & 4 are DRC-DI's. I originally purchased two of the units with DA-FPs (digital in and analog out) and converted them to DI's.
So the first two units purchased come with a DA-FP (no longer available), and currently installed DIGI-FP cards. I'll set those up as a DA-FP if requested."

Any idea of freight cost to Sweden?
 
Zacc, your frequency response in post #14 does look good overall. So I believe you when you write it sounds good.

The 2 dips between 100 and 300 Hz are probably due to SBIR, -reflections from nearby surfaces like floor, sidewalls and/or ceiling. If they are SBIRs they can’t be eq:ued, applies also for bothersome peaks due to SBIR. (Depending on destructive or constructive speaker boundary interference response.) Easiest way to check if SBIR or not is to move the mic, if the frequency moves that indicates it is SBIR. If it changes in amplitude but not in frequency it is more likely a mode. Easiest way to find out exactly where those reflection points should be, would be to check the ETC-diagram in REW and then apply the ”string method” for the reflection’s extra travel distance compared to direct sound. Measure only with 1 main speaker at a time if you are chasing reflection points! Those points can be treated with absorbtion, re-direction or blocked or with diffusion.

If you feel you get loud enough SPL above 30 Hz I think it could be a waste of of money with more large subs /woofers and beefy amplifiers. -Your room volume is huge and with a length of 14,8 m your lowest mode should be around 11,6 Hz and 2nd length mode about 23 Hz. The frequency curve in post #14 shows it starts to fall off at around 30 Hz. This may indicate you have a low ”mode support” from your end walls also for the 2nd length mode at 23 Hz and it will be impossible to reach the room’s pressure mode and reach down into infra frequencies. If the end walls are too ”weak” to efficiently reflect very low frequncies, the sound wave will more or less just pass through. (Of course it could also be that the mic was placed where the SPL of 23 Hz is low in your room in post #14.)

If you have a lot of time, you can check out how you frequency response, phase and waterfall diagrams varies depending on where you place the mic (sofa) in your room. A good description of it here: DOWNLOADS (Useful tools, documents, and other freebies) - Soundman2020 - Studio Design Forum "The Walking Mic Test Procedure". Look at how the response changes rapidly in the ”video clips” from the different measurements and mic positions.
 
Hi ZACC, thx for your interest.
All three are currently DRC-DI's.
But originally, two were DRC-DA-FP's. The DA-FP is digital input, balanced analog output.
MiniDSP used to sell various I/O modules for the DRC's, so you could easily change their configuration. I've attached a pict of a DA-FP panel/module to help visualize.

I have no idea about shipping to Sweden, but I could do some homework if you have a strong interest.
 

Attachments

  • DA-FP card.jpg
    DA-FP card.jpg
    69.6 KB · Views: 347
Yes it's true, sadly MiniDSP discontinued two previous versions ANALOG IN > ANALOG OUT and also DIGITAL IN > ANALOG OUT.
Nowadays they only offer the DIGITAL IN > DIGITAL OUT variety. Needless to say you can always feed those digital outputs into any external stereo DAC to get your preferred analog signals.

Like Mark100 as well, I have both the DA-FP output boards and the DIGI-FP output boards too for my OpenDRC boxes. They're quick and easy to swap if needed, 5 minutes with a screwdriver.
If you're looking for one FIR box that handles both analogue & digital I/O take a look at the Acoustic Power Lab APL1S although it only has 4096 taps per channel.
 
Last edited:
Zacc, your frequency response in post #14 does look good overall. So I believe you when you write it sounds good.

If you have a lot of time, you can check out how you frequency response, phase and waterfall diagrams varies depending on where you place the mic (sofa) in your room. A good description of it here: DOWNLOADS (Useful tools, documents, and other freebies) - Soundman2020 - Studio Design Forum "The Walking Mic Test Procedure". Look at how the response changes rapidly in the ”video clips” from the different measurements and mic positions.

Thanks for the tips. When the time comes I will have a closer look at this. Keep the good suggestions coming!

Now focus is on building test enclosures and start doing some measuring.
 
Hi ZACC, thx for your interest.
All three are currently DRC-DI's.
But originally, two were DRC-DA-FP's. The DA-FP is digital input, balanced analog output.
MiniDSP used to sell various I/O modules for the DRC's, so you could easily change their configuration. I've attached a pict of a DA-FP panel/module to help visualize.

I have no idea about shipping to Sweden, but I could do some homework if you have a strong interest.

Thx for info. For now, no real interest in these. :)
 
Yes it's true, sadly MiniDSP discontinued two previous versions ANALOG IN > ANALOG OUT and also DIGITAL IN > ANALOG OUT.
Nowadays they only offer the DIGITAL IN > DIGITAL OUT variety. Needless to say you can always feed those digital outputs into any external stereo DAC to get your preferred analog signals.

Like Mark100 as well, I have both the DA-FP output boards and the DIGI-FP output boards too for my OpenDRC boxes. They're quick and easy to swap if needed, 5 minutes with a screwdriver.
If you're looking for one FIR box that handles both analogue & digital I/O take a look at the Acoustic Power Lab APL1S although it only has 4096 taps per channel.

Let's see down the line what I end up with.

Is there a relation between amount of filters/corrections and quantity of taps needed? I mean, if the speakers are built with time aligned drivers, plenty of frequency overlap at x-over points etc. then filters should be easier to do. Less filters, less taps needed? Or have I not understood this at all...:confused:
 
Let's see down the line what I end up with.

Is there a relation between amount of filters/corrections and quantity of taps needed? I mean, if the speakers are built with time aligned drivers, plenty of frequency overlap at x-over points etc. then filters should be easier to do. Less filters, less taps needed? Or have I not understood this at all...:confused:

The number of taps a filter needs is based on how closely you want that filter to resemble it's theoretical ideal. You need more taps the lower you go in frequency to make the filter correctly. Below 100Hz the number of taps required to make accurate magnitude changes can be quite high.

The best way to see it for yourself is to use rephase. Add the EQ or correction and change the number of taps to see what filter you end up with. Rephase will show you the ideal filter and the one you created. You can then see how many taps you need. The filter will still work with a smaller number of taps but the slope and phase response may not be what you intended them to be.

Another way to get around the tap limitation of minidsp devices is to do the majority of your EQ and crossover tasks with IIR filters and then use an overall FIR filter for phase manipulation as only changing the phase needs a lot less taps. rehase will demonstrate that too :)

With a MinDSP I would start with delays and IIR filters first, you can do an awful to with them.
 
Is there a relation between amount of filters/corrections and number of taps needed?

Hi ZACC,
Yes, they are related like Fluid says in his post.
It's about accuracy, and more taps equals longer filter impulse duration, which equals better accuracy at lower frequencies.
As you increase the number of taps, firstly the treble accuracy improves drastically, then the midrange accuracy improves, and finally the bass starts to get accurate once you get lots of taps

Sorry - I am conscious of not wanting to de-rail your speaker DIY build thread about your monster size 4-ways, by going way off topic to talk too much about FIR correction in detail here! (Loads of other threads exist specifically about FIR, rePhrase, MiniDSP and all the math and theory stuff, etc.)

...but since you asked this question, let me just post a picture I created a while back on another forum, which tells the story better than a thousand words wall of text.
Blue line is the real measured response of a 15 inch woofer (target response.)
Red line shows how accurate an FIR curve matches with various numbers of taps.
If you can see the feint grey lines too, that's phase response, both the target phase and the FIR curve phase are grey.

Smaller tap sizes are pretty rough approximations!
1024 taps is barely enough for full-range audio - look at the giant errors in the bass missing the target by 13dB! (graph scale is +20dB to -30dB.)
Even at 2048 taps, you can see the treble's good but the bass and low midrange (below 300Hz) is still very coarse. By 6144 taps, all the midrange is lovely to 100Hz, while bass is okay-ish with reduction in error but not quite perfect. By about 16,834 taps you reach a quality plateau where it's so good you'd never hear the difference.

Observe the overall "trends" here... that's why more taps is a good thing!

attachment.php

attachment.php

attachment.php
 

Attachments

  • tapsaccuracy1.jpg
    tapsaccuracy1.jpg
    48.1 KB · Views: 463
  • tapsaccuracy2.jpg
    tapsaccuracy2.jpg
    49.2 KB · Views: 446
  • tapsaccuracy3.jpg
    tapsaccuracy3.jpg
    49.2 KB · Views: 446
Hi, ZACC

I like where you are going with your speaker system. I went big, but with horns. Nothing plays near live levels, with dynamics, like a big system. Good luck with your project.

YouTube

Joe, thanks for the link. I've seen this amazing system already and I bet it sounds great!

After the build I'm doing now, I will go all horns from 80Hz and up. Your system is an inspiration. Well done and thanks for sharing! :yes:
 
The number of taps a filter needs is based on how closely you want that filter to resemble it's theoretical ideal.

Another way to get around the tap limitation of minidsp devices is to do the majority of your EQ and crossover tasks with IIR filters and then use an overall FIR filter for phase manipulation as only changing the phase needs a lot less taps.

With a MinDSP I would start with delays and IIR filters first, you can do an awful to with them.

Thanks for this input fluid!

Sounds like this will suit me great as a start. If I later realize I need more taps, I will go shopping for more... ;)
 
Observe the overall "trends" here... that's why more taps is a good thing!


Dynaudio4way, again thanks for input!

Wow, that's a whole lotta taps! :D I'm starting to get the general idea now.
The most important question is, how will it sound...?:confused: Does the amount of filtering and use of taps degrade e.g. transients? Any other negative effects?

I believe in building speakers with good driver matching for easy filters. Drivers in bass, midbass and midrange should have similar characteristics. In this build I have chosen drivers with light cones, large motors and a lot of Sd in each frequency range. A measured frequency response says one thing but it says very little of how it sounds. In this build I'm trying to put a system together that will blow my socks off when listening to good music, recorded without maximum compression, with a lot of dynamics.
A kick drum should sound like a 5 kg (roughly 10 lbs for you 'mercans) piece of grade a beef thrown out of a second story window onto wet asphalt. :p
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.