• 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!

3 way DSP XO

Cosmik

Major Contributor
Joined
Apr 24, 2016
Messages
3,075
Likes
2,180
Location
UK
Hi everyne. BE718, I am going to buy MOTU 8A also, and go ahead with crossovers on PC side.

Pls, advise few questions:
1) What about ocasional freezing of OS, plugin crashes, etc. How you protect speakers?
2) You use accourate, and what do you use as soft player? What is the role/function of MOTU soft mixer in all the chain of your software setup Just share more detals about "soft" rooting in laptop itself.
3) Do you use sme in built DSP functions of MOTU? Are there any useful for speaker multiamping and correction?
4) Did you try AVB? Is AVB MOTU input compatible and accepting normal ethernet output from laptop, pc?

Thanks a lot for reply!
You raise a good point.
1) In theory the PC could go mad and send full scale output all the drivers. The mitigation against that is to use series protection capacitors with the tweeters (say 47uF..?). These will attenuate any damaging lower frequencies reaching the drivers and their effect can be compensated for in the DSP measurements and driver correction. Using the amplifier's volume control rather than software volume control means that even if the DACs send full output, the volume is limited to just being somewhat louder than your normal maximum listening level, so very unlikely that anything will be damaged.

The above scenario has only happened to me once, when I made a mistake with my own software - nothing was damaged.

If I were to develop a dedicated DSP box, I would certainly include intelligent failsafes to prevent potential damage. Kind of like the Sch**t amplifier that uses a microcontroller to monitor the output - but made to work properly.
 

Voldemaar

Member
Joined
Sep 21, 2017
Messages
5
Likes
0
Regarding cap for tweeters - that is exactly the way I am thinking to go, as most simple one...
 

watchnerd

Grand Contributor
Joined
Dec 8, 2016
Messages
12,449
Likes
10,408
Location
Seattle Area, USA
If one were to run a hybrid DSP/passive combo crossover on a 3-way system, would it be better to have the DSP crossover at the top or the bottom of the mid range?

Hypothetical example:

The Devialet Expert Pro, in dual stereo (or even quad mono) can do a high pass or low pass filter of 1-4th order, but it can't do both high and low pass on the same set of speakers.

If one applied that to a passive 3-way, would it be better to use the DSP 2-way crossover on the tweeter/mid portion (leaving the mid/woofer passive), or use the DSP 2-way crossover on the mid/woofer portion (leaving the mid/tweeter passive).

My first instinct is to think it would be better on the tweeter/mid as the ear is more sensitive there.
 

RayDunzl

Grand Contributor
Central Scrutinizer
Joined
Mar 9, 2016
Messages
13,202
Likes
16,982
Location
Riverview FL
My first instinct is to think it would be better on the tweeter/mid as the ear is more sensitive there.

If using a two-way cross on a three-way system, then you depend on the natural roll-off of the woofer-to-mid (your choice above) or the mid-to-tweeter to not interfere with the next higher range.

Not being fluent in that area, I'm just supposing, but it would seem to be problematic to me.

That assumes you don't use the analog crossover in the speaker for the cross not DSP'd, which now feels like a dumb assumption on my part.

What's your hypothetical plan again?
 
Last edited:

watchnerd

Grand Contributor
Joined
Dec 8, 2016
Messages
12,449
Likes
10,408
Location
Seattle Area, USA
If using a two-way cross on a three-way system, then you depend on the natural roll-off of the woofer-to-mid (your choice above) or the mid-to-tweeter to not interfere with the next higher range.

Not being fluent in that area, I'm just supposing, but it would seem to be problematic to me.

That assumes you don't use the analog crossover in the speaker for the cross not DSP'd, which now feels like a dumb assumption on my part.

What's your hypothetical plan again?

Take a 3 way passive speaker, ideally one already set up for tri-wiring.

Disconnect either the tweeter/mid passive and do it in DSP, leaving the mid/woofer passive, or vice versa.

I suspect it's not as easy as that because the passive network might need rejiggering if one leg or the other is disconnected.

(I'm a software / DSP applied physics guy, not an EE, so my passive electrical circuit knowledge doesn't go deeper than high school science).
 

RayDunzl

Grand Contributor
Central Scrutinizer
Joined
Mar 9, 2016
Messages
13,202
Likes
16,982
Location
Riverview FL
I'm a software / DSP applied physics

Can I combine two FIR filters - like one that does a good job on FR but lacks some low frequency phase repair, with one that does more LF phase correction only?

They would be text files - plotting an Impulse Response - easily manipulated in a spreadsheet - for the 6144 taps.

I haven't tried it yet...
 

DonH56

Master Contributor
Technical Expert
Forum Donor
Joined
Mar 15, 2016
Messages
7,835
Likes
16,497
Location
Monument, CO
I am not sure what you mean by "combine". You can convolve or cascade the responses depending upon how the system is set up and what you want to do. Or just design new tap weights that accomplish the end goal without having to use two filters. If it is for more than one speaker in an anechoic chamber, then cascading filters will change the phase and group delay which might alter the frequency response at the MLP.
 

RayDunzl

Grand Contributor
Central Scrutinizer
Joined
Mar 9, 2016
Messages
13,202
Likes
16,982
Location
Riverview FL
I am not sure what you mean by "combine". You can convolve or cascade the responses depending upon how the system is set up and what you want to do. Or just design new tap weights that accomplish the end goal without having to use two filters. If it is for more than one speaker in an anechoic chamber, then cascading filters will change the phase and group delay which might alter the frequency response at the MLP.

?

(Sorry)

What I mean:

My woofers go 180 degrees out of phase at the listening position, probably due to an asymmetrical room, around 47 Hz, making a measurably deep response dip (cancellation).

I have AcourateDRC (think Acourate Lite) which is highly automated (measure, draw a FR curve, and click "do it" to create IIR and FIR filters for the miniDSP) which does not give me any manual control of phase, but otherwise does a fine job of improving the sound here. It corrects phase (if you want it to), but apparently restricts its analysis to left and right and doesn't consider what happens when both channels play.

I have another filter generating software called rePhase, which is entirely manual, and permits direct manipulation of phase (and nothing else, if that is the intention).

Both tools can create an FIR centered on the same tap number (of 6144 available in the miniDSP).

I'm wondering if I can take the AcourateDRC FIR file, and apply a modifying phase correction to it in the 47hz area to my taste, from a rePhase FIR file.

My thought is to put both files (Impulse Response) into a spreadsheet (they are text files), add the coefficients, to create a single file with a taste of both of the input files.

When you suggest convolving or cascading or designing new tap weights, I'm unable to comprehend how to take any action given my knowledge/toolset.

---

Left and right and Both (stereo) response at the listening position. I've identified the 47 Hz dip to be a phase problem:

upload_2017-9-22_17-22-13.png


Phase: 180 degrees out at 47Hz (and problematic in the area around it)

upload_2017-9-22_17-24-24.png
 
OP
March Audio

March Audio

Master Contributor
Audio Company
Joined
Mar 1, 2016
Messages
6,378
Likes
9,317
Location
Albany Western Australia
Hi everyne. BE718, I am going to buy MOTU 8A also, and go ahead with crossovers on PC side.

Pls, advise few questions:
1) What about ocasional freezing of OS, plugin crashes, etc. How you protect speakers?
2) You use accourate, and what do you use as soft player? What is the role/function of MOTU soft mixer in all the chain of your software setup Just share more detals about "soft" rooting in laptop itself.
3) Do you use sme in built DSP functions of MOTU? Are there any useful for speaker multiamping and correction?
4) Did you try AVB? Is AVB MOTU input compatible and accepting normal ethernet output from laptop, pc?

Thanks a lot for reply!
Hi Voldemaar

1. Firstly my system PC is actually very stable, it doesn't crash or freeze, but obviously there is a potential for this to happen. As has already been mentioned, the only (simple) thing you can do is place a cap in series with the tweeter to stop low frequencies getting to it. High level digital noise has not got through to my speakers as of yet. I consider it low risk.

2. I use Roon as the media player. This plays to acourate convolver which does the filtering and in turn feeds the motu.
The motu has a software virtual signal routing panel and mixer. Essentially is is set to route 1:1 in this configuration. I will post pictures later which will make it clear.

3. I don't use any of the motu DSP functions but I suppose it would be capable of some basic speaker correction.

4. I haven't tried AVB. It does require a special AVB switch but that can then plug into a normal network. However I think this may only work with Mac at the moment.
Ignoring AVB functionality the motu can still plug into your normal network and you can access it over a web page to perform the setup firmware upgrades etc.

Hope this helps
 
Last edited:

jhaider

Major Contributor
Forum Donor
Joined
Jun 5, 2016
Messages
2,822
Likes
4,514
If one were to run a hybrid DSP/passive combo crossover on a 3-way system, would it be better to have the DSP crossover at the top or the bottom of the mid range?

The usual approach is to use one channel on the woofer, and one channel on the mid-tweeter. There are two main reasons. First, the woofer "needs" more power, because it is operating below the "baffle step." Second. the passive parts for a woofer-midrange crossover are usually larger/bulkier/pricier for a given quality.
 

RayDunzl

Grand Contributor
Central Scrutinizer
Joined
Mar 9, 2016
Messages
13,202
Likes
16,982
Location
Riverview FL
If you take all the DSP out of the picture, are they still 180 deg out of phase?

Yes.

upload_2017-9-22_23-19-33.png

(don't want to hijack the thread - I almost started one for this ("combining" FIR filter data), but didn't, yet, low audibility priority)
 
Last edited:
OP
March Audio

March Audio

Master Contributor
Audio Company
Joined
Mar 1, 2016
Messages
6,378
Likes
9,317
Location
Albany Western Australia
Motu Screen Shots

upload_2017-9-23_12-54-5.png


upload_2017-9-23_12-55-13.png


upload_2017-9-23_12-57-0.png



Acourate convolver set-up.

  • Takes "Virtual" inputs 1 and 2, which are simply left and right outputs from Roon.
  • Applies relevant filter ( sub, woofer, mid and tweeter for left and right channels)
  • Feeds that filter to relevant Motu Analogue output
Capture.PNG
 
Last edited:
OP
March Audio

March Audio

Master Contributor
Audio Company
Joined
Mar 1, 2016
Messages
6,378
Likes
9,317
Location
Albany Western Australia
Hi Voldemaar,

Just some measurements of the Motu 8A performed on a Keysight U8903B for your reference. It performs well.

IMG_20170923_154503.jpg



0 dB 1kHz. Please note scale on FFT is dBV. Full scale fundamental at 1 kHz is +20dBV. So subtract 20 dB off harmonic values for relative.

Standard_1.jpg
Graph_0.jpg

Graph_1.jpg



-60dB
Graph_2.jpg
Graph_5.jpg



12kHz jitter
Graph_8.jpg



-90.3dB sine wave
Graph_6.jpg


Intermod
Graph_9.jpg
 
Last edited:

dallasjustice

Major Contributor
Joined
Feb 28, 2016
Messages
1,270
Likes
907
Location
Dallas, Texas
Because the phase problem at 47hz is a non-minimum phase problem associated with Allison effect from the room, you will not be able to change the effect with any filters applied to R/L speakers. The only solution is to move the speaker or to use subs. Some folks have experimented with simply boosting amplitude at the dip. This is equally ineffective and probably counter-productive. If the speakers are moved more asymmetrically in relation to boundary causing the Allison effect, that problem will go away since all bass at that frequency is mono anyway.

If you like how your system sounds, I would ignore it unless you want to add a couple of subs.
?

(Sorry)

What I mean:

My woofers go 180 degrees out of phase at the listening position, probably due to an asymmetrical room, around 47 Hz, making a measurably deep response dip (cancellation).

I have AcourateDRC (think Acourate Lite) which is highly automated (measure, draw a FR curve, and click "do it" to create IIR and FIR filters for the miniDSP) which does not give me any manual control of phase, but otherwise does a fine job of improving the sound here. It corrects phase (if you want it to), but apparently restricts its analysis to left and right and doesn't consider what happens when both channels play.

I have another filter generating software called rePhase, which is entirely manual, and permits direct manipulation of phase (and nothing else, if that is the intention).

Both tools can create an FIR centered on the same tap number (of 6144 available in the miniDSP).

I'm wondering if I can take the AcourateDRC FIR file, and apply a modifying phase correction to it in the 47hz area to my taste, from a rePhase FIR file.

My thought is to put both files (Impulse Response) into a spreadsheet (they are text files), add the coefficients, to create a single file with a taste of both of the input files.

When you suggest convolving or cascading or designing new tap weights, I'm unable to comprehend how to take any action given my knowledge/toolset.

---

Left and right and Both (stereo) response at the listening position. I've identified the 47 Hz dip to be a phase problem:

View attachment 8824

Phase: 180 degrees out at 47Hz (and problematic in the area around it)

View attachment 8825
 

RayDunzl

Grand Contributor
Central Scrutinizer
Joined
Mar 9, 2016
Messages
13,202
Likes
16,982
Location
Riverview FL
Because the phase problem at 47hz is a non-minimum phase problem associated with Allison effect from the room, you will not be able to change the effect with any filters applied to R/L speakers.

Not to be argumentative, but here is an experiment, using 0, 2, and 4 ms delay applied to the right channel:

(Both speakers are playing back the test tone)

upload_2017-9-23_15-29-54.png


I'd like to focus a delay (phase) on the problematic range.

As an experiment.

With what I have available to me, for the experiment to be successful, would require combining the effects of two FIR filters, which was the question posed.

Combine an AcourateDRC filter (automated full range) with a rePhase (manually specified) filter.
 
Last edited:

dallasjustice

Major Contributor
Joined
Feb 28, 2016
Messages
1,270
Likes
907
Location
Dallas, Texas
So you are doing R+L logsweeps? Why not sweep each speaker at a time?
Not to be argumentative, but here is an experiment, using 0, 2, and 4 ms delay applied to the right channel:

(Both speakers are playing back the test tone)

View attachment 8844

I'd like to focus a delay (phase) on the problematic range.

As an experiment.

With what I have available to me, for the experiment to be successful, would require combining the effects of two FIR filters, which was the question posed.

Combine an AcourateDRC filter (automated full range) with a rePhase (manually specified) filter.
 

RayDunzl

Grand Contributor
Central Scrutinizer
Joined
Mar 9, 2016
Messages
13,202
Likes
16,982
Location
Riverview FL
So you are doing R+L logsweeps? Why not sweep each speaker at a time?

AcourateDRC does L/R/L for its measurement. It doesn't seem to consider the result that might occur when both are combined (my opinion)

The bass region looks good at the listening position with either L or R speaker alone, but the phase relationship between them at the 47Hz area with both speakers playing (normal condition) creates a cancellation in the 47Hz frequency range.

I do L/R/and Both when I measure.

Raw - no eq upload_2017-9-23_17-51-28.png

With AcourateDRC - upload_2017-9-23_17-53-14.png

The problem is not "obvious" when listening to music, but, as an experiment, I'd like to play with it. Using the tools at hand, it would seem I can make a difference by altering the timing (phase relationship) of the L/R in the problem area, as evidenced by the "brute force" delay experiment.

To put this into practice, I would want to focus on the problem area, and not delay the entire range by 4ms.

My thought is to "patch" the automated AcourateDRC filter with some additional phase correction created (manually) in a rePhase filter.

My question is "Can I can or can I can't combine two sets of FIR data into one?"
 
Last edited:

dallasjustice

Major Contributor
Joined
Feb 28, 2016
Messages
1,270
Likes
907
Location
Dallas, Texas
So your thesis is that the phase reversal is not caused by a room reflection. Instead, you believe the right and left speakers played at the same time cancel each other at 47hz?
 
Top Bottom