miniDSP OpenDRC-DI - AVS Forum
Forum Jump: 
 
Thread Tools
Old 08-08-2014, 03:31 AM - Thread Starter
AVS Special Member
 
RayDunzl's Avatar
 
Join Date: Sep 2013
Location: Tampa, Florida
Posts: 1,399
Mentioned: 2 Post(s)
Tagged: 0 Thread(s)
Quoted: 478 Post(s)
Liked: 671
miniDSP OpenDRC-DI

New toy inserted.

This one:

http://www.minidsp.com/products/open...ies/opendrc-di

I'll be back later...


links::: 1.5RQ > digits > 1177a > OpenDRC-DI > DEQ2496 > DAC2 > KCT > FPB 350mcx > reQuest > Sweetspot
RayDunzl is online now  
Sponsored Links
Advertisement
 
Old 08-08-2014, 05:56 AM - Thread Starter
AVS Special Member
 
RayDunzl's Avatar
 
Join Date: Sep 2013
Location: Tampa, Florida
Posts: 1,399
Mentioned: 2 Post(s)
Tagged: 0 Thread(s)
Quoted: 478 Post(s)
Liked: 671
Here's a graph of the phase (delay?) by frequency for my system at the sweet spot.



It looks like the lows become progressively late, though from 800hz on up the output is already phase coherent.

This will be the first thing I experiment with, using the tool rePhase.

I'll be back later...


links::: 1.5RQ > digits > 1177a > OpenDRC-DI > DEQ2496 > DAC2 > KCT > FPB 350mcx > reQuest > Sweetspot
RayDunzl is online now  
Old 08-08-2014, 09:14 PM
AVS Special Member
 
DonH50's Avatar
 
Join Date: Feb 2010
Location: Monument CO
Posts: 6,343
Mentioned: 1 Post(s)
Tagged: 0 Thread(s)
Quoted: 258 Post(s)
Liked: 295
Ideally the phase would be a linear line, which in turns leads to constant group delay. That means all frequencies are delayed equally, no time distortion.

"After silence, that which best expresses the inexpressible, is music" - Aldous Huxley
DonH50 is offline  
Old 08-09-2014, 02:17 AM - Thread Starter
AVS Special Member
 
RayDunzl's Avatar
 
Join Date: Sep 2013
Location: Tampa, Florida
Posts: 1,399
Mentioned: 2 Post(s)
Tagged: 0 Thread(s)
Quoted: 478 Post(s)
Liked: 671
First try at phase correction gives (using the same scale as original above):



Group Delay is decreased, but not so flat, so have to think about that part a little bit more.


I'll be back later...


links::: 1.5RQ > digits > 1177a > OpenDRC-DI > DEQ2496 > DAC2 > KCT > FPB 350mcx > reQuest > Sweetspot
RayDunzl is online now  
Old 08-12-2014, 02:08 PM - Thread Starter
AVS Special Member
 
RayDunzl's Avatar
 
Join Date: Sep 2013
Location: Tampa, Florida
Posts: 1,399
Mentioned: 2 Post(s)
Tagged: 0 Thread(s)
Quoted: 478 Post(s)
Liked: 671
Well, having played for a while, a little surprise.

Most of my investigation so far is with phase correction. To do that the DSP delays or advances frequency bands in time relative to each other, but does not aterially change the frequency response.

So, by doing that, waves are relocated in time, and when summed with the non-relocated waves, produces a clip. The results are a bit unpredictable when trying to guess how much attenuation is enough to prevent clips.

I've had to attenuate the input to the DSP by 9dB to prevent clippage showing up at the input to the Behringer. That may or may not be enough for all source material.

I've thought to use white noise input at 99% full scale, but the no-clip input setting determined with that source was about -3dB.

Unfortunately, the miniDSP hardware does not have a level indication, and the software plugin has what appears to be a very rudimentary level meter.

I'm liking the unit though, overall.

I'll be back later...


links::: 1.5RQ > digits > 1177a > OpenDRC-DI > DEQ2496 > DAC2 > KCT > FPB 350mcx > reQuest > Sweetspot
RayDunzl is online now  
Old 08-14-2014, 12:42 AM - Thread Starter
AVS Special Member
 
RayDunzl's Avatar
 
Join Date: Sep 2013
Location: Tampa, Florida
Posts: 1,399
Mentioned: 2 Post(s)
Tagged: 0 Thread(s)
Quoted: 478 Post(s)
Liked: 671
Weekend Update...


I've brutally twisted the phase into alignment from about 700hz on down.

Increasing torque was applied to the lower frequencies, correcting a maximum of 1477 degrees of retardation at 30hz.

The bottom end is all cleaned up.


Any questions?


Yes, why didn't the Group Delay flatten out more?

Yes, uh, well, the, uh... Are there any other questions?

no.

Post adjourned.

I'll be back later...


links::: 1.5RQ > digits > 1177a > OpenDRC-DI > DEQ2496 > DAC2 > KCT > FPB 350mcx > reQuest > Sweetspot
RayDunzl is online now  
Old 08-16-2014, 10:47 AM - Thread Starter
AVS Special Member
 
RayDunzl's Avatar
 
Join Date: Sep 2013
Location: Tampa, Florida
Posts: 1,399
Mentioned: 2 Post(s)
Tagged: 0 Thread(s)
Quoted: 478 Post(s)
Liked: 671
Effect of time displacement on REW's logarthmic sweep frequency response

Very small to none.

(Yes, I know I have a big suckout in the lows due to the room)


I'll be back later...


links::: 1.5RQ > digits > 1177a > OpenDRC-DI > DEQ2496 > DAC2 > KCT > FPB 350mcx > reQuest > Sweetspot
RayDunzl is online now  
Old 08-16-2014, 10:58 AM - Thread Starter
AVS Special Member
 
RayDunzl's Avatar
 
Join Date: Sep 2013
Location: Tampa, Florida
Posts: 1,399
Mentioned: 2 Post(s)
Tagged: 0 Thread(s)
Quoted: 478 Post(s)
Liked: 671
Group Delay reduction - green trace is with the time filter, red is without


I'll be back later...


links::: 1.5RQ > digits > 1177a > OpenDRC-DI > DEQ2496 > DAC2 > KCT > FPB 350mcx > reQuest > Sweetspot
RayDunzl is online now  
Old 08-16-2014, 01:09 PM - Thread Starter
AVS Special Member
 
RayDunzl's Avatar
 
Join Date: Sep 2013
Location: Tampa, Florida
Posts: 1,399
Mentioned: 2 Post(s)
Tagged: 0 Thread(s)
Quoted: 478 Post(s)
Liked: 671
Since the miniDSP and FIR filter is messing with my signal, let's look at the output of the preamp.

(This measurement includes a 25 foot RCA to 1/8" miniplug from preamp to PC onboard sound device, and any defects caused by them.)




Blue is without the filter, and green is with. Basically no difference in frequency response is seen, results within .1dB or so. (Flat dark blue line is just the measurement cursor)






The FIR filter is changing phase, so here is that measurement at the preamp:

Blue is the preamp output without the filter, green is preamp output with the phase correction filter applied.







And finally, for right now...

Top green is the in-room measurement of phase with the FIR filter, red is the unfiltered room measurement of phase, and the lower green is the phase correction applied at the preamp by the filter (as measured at the preamp out).

This shows how the FIR filter creates* a reasonable inverse to the phase error detected in the room, and applied that to the signal, and corrected the sound in the room.




*rePhase is a manual tool, so the inverse isn't perfect, I have to move a lot of sliders and mess with things a while to have gotten it that close

I'll be back later...


links::: 1.5RQ > digits > 1177a > OpenDRC-DI > DEQ2496 > DAC2 > KCT > FPB 350mcx > reQuest > Sweetspot
RayDunzl is online now  
Old 08-21-2014, 12:35 PM - Thread Starter
AVS Special Member
 
RayDunzl's Avatar
 
Join Date: Sep 2013
Location: Tampa, Florida
Posts: 1,399
Mentioned: 2 Post(s)
Tagged: 0 Thread(s)
Quoted: 478 Post(s)
Liked: 671
A tip on using rePhase:

At first I used a lot of travel on a lot of sliders to get that major phase/delay out of the low end.

Then experimented on another page of the tool.

Filters Linearization tab, Minimum-phase Filters Phase Linearization

One setting there can twist the big curve in the phase nearly flat.

Then, go to the Paragraphic Phase Eq sliders to clean up, after the major demolition work is done.

I'll be back later...


links::: 1.5RQ > digits > 1177a > OpenDRC-DI > DEQ2496 > DAC2 > KCT > FPB 350mcx > reQuest > Sweetspot
RayDunzl is online now  
Old 08-29-2014, 01:38 PM - Thread Starter
AVS Special Member
 
RayDunzl's Avatar
 
Join Date: Sep 2013
Location: Tampa, Florida
Posts: 1,399
Mentioned: 2 Post(s)
Tagged: 0 Thread(s)
Quoted: 478 Post(s)
Liked: 671
Here are some current measurements.

Red = no DSP

Green = FIR filters applied

1/3 octave smoothing applied
Attached Thumbnails
Click image for larger version

Name:	spl.png
Views:	37
Size:	76.0 KB
ID:	235529   Click image for larger version

Name:	phase.png
Views:	34
Size:	63.9 KB
ID:	235537   Click image for larger version

Name:	gd.png
Views:	31
Size:	63.8 KB
ID:	235545   Click image for larger version

Name:	impulse.png
Views:	35
Size:	66.9 KB
ID:	235553   Click image for larger version

Name:	step.png
Views:	37
Size:	69.2 KB
ID:	235561  


I'll be back later...


links::: 1.5RQ > digits > 1177a > OpenDRC-DI > DEQ2496 > DAC2 > KCT > FPB 350mcx > reQuest > Sweetspot
RayDunzl is online now  
Old 08-29-2014, 03:12 PM
Advanced Member
 
3ll3d00d's Avatar
 
Join Date: Sep 2007
Location: London, UK
Posts: 639
Mentioned: 1 Post(s)
Tagged: 0 Thread(s)
Quoted: 289 Post(s)
Liked: 85
what's the ripple in the IR (and SR) before the impulse? a preringing artefact of the filter?
3ll3d00d is online now  
Old 08-29-2014, 03:41 PM - Thread Starter
AVS Special Member
 
RayDunzl's Avatar
 
Join Date: Sep 2013
Location: Tampa, Florida
Posts: 1,399
Mentioned: 2 Post(s)
Tagged: 0 Thread(s)
Quoted: 478 Post(s)
Liked: 671
It's the filter, yes.

The FIR filters here are 'acausal'...

There is time-shifting of frequency bands in my filter, to align the time of the peaks.

There is also a time delay - about 50ms in this filter, allowing plenty of time for the filter to grab something that would have occurred later and make it occur sooner.

JJ has looked in on this thread, and asked a few questions in PM, I'm sure he could give us a totally accurate yet incomprehensible reply.

I'll be back later...


links::: 1.5RQ > digits > 1177a > OpenDRC-DI > DEQ2496 > DAC2 > KCT > FPB 350mcx > reQuest > Sweetspot
RayDunzl is online now  
Old 08-29-2014, 03:53 PM
Advanced Member
 
3ll3d00d's Avatar
 
Join Date: Sep 2007
Location: London, UK
Posts: 639
Mentioned: 1 Post(s)
Tagged: 0 Thread(s)
Quoted: 289 Post(s)
Liked: 85
do you detect any audible trace of that?

I use FIR filters as well, in my case generated by acourate which has a preringing compensation function. I thought the goal was to have a gently rising phase response at low frequencies as a completely flat phase response is one of the thing that introduces excessive preringing.
3ll3d00d is online now  
Old 08-29-2014, 06:41 PM - Thread Starter
AVS Special Member
 
RayDunzl's Avatar
 
Join Date: Sep 2013
Location: Tampa, Florida
Posts: 1,399
Mentioned: 2 Post(s)
Tagged: 0 Thread(s)
Quoted: 478 Post(s)
Liked: 671
Quote:
Originally Posted by 3ll3d00d View Post
do you detect any audible trace of that?
Audible? Shhhh!!!! Can't talk about what you hear here!

Quote:
Originally Posted by 3ll3d00d
I use FIR filters as well, in my case generated by acourate which has a preringing compensation function. I thought the goal was to have a gently rising phase response at low frequencies as a completely flat phase response is one of the thing that introduces excessive preringing.

Maybe you have something to teach me, because I don't know. I'm reading things, but it takes a while to soak in.

Post a picture if you have one.


I looked at Dirac (automation) and decided against it, since their implementation locks up the hardware and prevents the use of the Open source tools (and costs another $500)...

And I looked at Acourate but am still having enough fun with rePhase that I haven't looked back at it yet.


Here is the other impulse display, first attachement below.

The filter pre-ringing is reportedly 40-60dB down. That's in the generally accepted hard-to-hear area when other things are going on.

Second attachment, filtered impulse response - 1/3 octave 200Hz range, with pre-ringing on the un-filtered trace that isn't evident in the full-range impulse response. The unfiltered peak there is 18.1ms "late".
Attached Thumbnails
Click image for larger version

Name:	impulse2.png
Views:	25
Size:	146.4 KB
ID:	236129   Click image for larger version

Name:	impulse 200 1-3 octave filtered.png
Views:	25
Size:	169.4 KB
ID:	236137  

I'll be back later...


links::: 1.5RQ > digits > 1177a > OpenDRC-DI > DEQ2496 > DAC2 > KCT > FPB 350mcx > reQuest > Sweetspot
RayDunzl is online now  
Old 08-29-2014, 07:14 PM
Senior Member
 
Join Date: Jul 2014
Location: Canada
Posts: 464
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 205 Post(s)
Liked: 69
Great thread Ray!

Nice new mug shot as well!
Garidy is offline  
Old 08-30-2014, 04:53 AM
Member
 
GeekWave's Avatar
 
Join Date: Dec 2013
Location: Wellington, New Zealand
Posts: 51
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 15 Post(s)
Liked: 16
woah! ...cool thread - I don't know what half of it means, but Id like to learn... Subscribing

openELEC HTPC - Marantz SR6003 - Emotiva UPA1's - JBL Studio 590's - Dual SVS-PB2000's - JBL Studio 520C - JBL ES10's - miniDSP 2x4
FYI - I have NO affiliation with the media player 'Geek Wave'
GeekWave is offline  
Old 08-30-2014, 07:57 AM
Advanced Member
 
3ll3d00d's Avatar
 
Join Date: Sep 2007
Location: London, UK
Posts: 639
Mentioned: 1 Post(s)
Tagged: 0 Thread(s)
Quoted: 289 Post(s)
Liked: 85
I think acourate aims to balance a few things with its phase correction; no major discontinuities in the phase response, as similar a response as possible across the 2 channels (summarised by calculating the IACC of the 2 step responses), minimal preringing.

The details, of how these features are implemented, are proprietary though so I don't know how much help I can give you. I just control the strength of its various parameters (excess phase correction, preringing compensation & the window size for its frequency dependent windowing) without knowing exactly what they do. Perhaps that gives you some food for thought though on the parameters you can tweak, for example to see if a different balance of adjustment reduces the ringing further but maintains the quality of the correction.

There is also a good guide on using rephase over on the jriver forum - http://yabb.jriver.com/interact/index.php?topic=87538.0 - which might be worth a read.
3ll3d00d is online now  
Old 08-30-2014, 10:26 AM - Thread Starter
AVS Special Member
 
RayDunzl's Avatar
 
Join Date: Sep 2013
Location: Tampa, Florida
Posts: 1,399
Mentioned: 2 Post(s)
Tagged: 0 Thread(s)
Quoted: 478 Post(s)
Liked: 671
What I can't figure out yet is which knobs to twist to predictably affect Group Delay...

Here's an old study.

I'll be back later...


links::: 1.5RQ > digits > 1177a > OpenDRC-DI > DEQ2496 > DAC2 > KCT > FPB 350mcx > reQuest > Sweetspot
RayDunzl is online now  
Old 08-30-2014, 07:08 PM
Senior Member
 
Join Date: Jul 2014
Location: Canada
Posts: 464
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 205 Post(s)
Liked: 69
Ray

Would you say that thus far, that there has or hasn't been a sonic improvement, in your opinion?

Thanks
Garidy is offline  
Old 08-30-2014, 09:07 PM - Thread Starter
AVS Special Member
 
RayDunzl's Avatar
 
Join Date: Sep 2013
Location: Tampa, Florida
Posts: 1,399
Mentioned: 2 Post(s)
Tagged: 0 Thread(s)
Quoted: 478 Post(s)
Liked: 671
There have been measureable changes.

I'll be back later...


links::: 1.5RQ > digits > 1177a > OpenDRC-DI > DEQ2496 > DAC2 > KCT > FPB 350mcx > reQuest > Sweetspot
RayDunzl is online now  
Old 08-30-2014, 09:12 PM
Senior Member
 
Join Date: Jul 2014
Location: Canada
Posts: 464
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 205 Post(s)
Liked: 69
Quote:
Originally Posted by RayDunzl View Post
There have been measureable changes.
Measurable by human machines or electronic machines?

Last edited by Garidy; 08-30-2014 at 10:42 PM.
Garidy is offline  
Old 09-06-2014, 02:38 AM - Thread Starter
AVS Special Member
 
RayDunzl's Avatar
 
Join Date: Sep 2013
Location: Tampa, Florida
Posts: 1,399
Mentioned: 2 Post(s)
Tagged: 0 Thread(s)
Quoted: 478 Post(s)
Liked: 671
Tonight's activity:

Listening to piano then realizing I could fix the couple of notes that were offending my ear.

So I did.

I'll be back later...


links::: 1.5RQ > digits > 1177a > OpenDRC-DI > DEQ2496 > DAC2 > KCT > FPB 350mcx > reQuest > Sweetspot
RayDunzl is online now  
Old 09-07-2014, 11:45 PM - Thread Starter
AVS Special Member
 
RayDunzl's Avatar
 
Join Date: Sep 2013
Location: Tampa, Florida
Posts: 1,399
Mentioned: 2 Post(s)
Tagged: 0 Thread(s)
Quoted: 478 Post(s)
Liked: 671
I have added frequency amplitude EQ to the original phase corrections, in case I didn't mention that.

REW has a automatic filter generator function, tried that. It creates overlapping filters of various Q and amplitude.


So... Not always being one to follow convention, I created a 128 band 1/12th octave equalizer with the DSP.

Every "musical" note from 13.75Hz to 21096Hz has it's own slider.

The filter type is "raised cosine", and with a Q of 18, it appears to be able to adjust any note without molesting its neighbors.

I'll be back later...


links::: 1.5RQ > digits > 1177a > OpenDRC-DI > DEQ2496 > DAC2 > KCT > FPB 350mcx > reQuest > Sweetspot
RayDunzl is online now  
Old 09-10-2014, 10:07 AM - Thread Starter
AVS Special Member
 
RayDunzl's Avatar
 
Join Date: Sep 2013
Location: Tampa, Florida
Posts: 1,399
Mentioned: 2 Post(s)
Tagged: 0 Thread(s)
Quoted: 478 Post(s)
Liked: 671
1/3 octave frequency response today.

Unequalized, gray, equalized, green


I'll be back later...


links::: 1.5RQ > digits > 1177a > OpenDRC-DI > DEQ2496 > DAC2 > KCT > FPB 350mcx > reQuest > Sweetspot
RayDunzl is online now  
Old 09-10-2014, 10:18 AM
Advanced Member
 
3ll3d00d's Avatar
 
Join Date: Sep 2007
Location: London, UK
Posts: 639
Mentioned: 1 Post(s)
Tagged: 0 Thread(s)
Quoted: 289 Post(s)
Liked: 85
Quote:
Originally Posted by RayDunzl View Post
1/3 octave frequency response today.

Unequalized, gray, equalized, green

does it sound on the bright side?
3ll3d00d is online now  
Old 09-10-2014, 11:01 AM - Thread Starter
AVS Special Member
 
RayDunzl's Avatar
 
Join Date: Sep 2013
Location: Tampa, Florida
Posts: 1,399
Mentioned: 2 Post(s)
Tagged: 0 Thread(s)
Quoted: 478 Post(s)
Liked: 671
As in screechy, no.

REW's measurements assume Pink (-3dB/octave going up) is flat.

I think... looking at that now...

I'll be back later...


links::: 1.5RQ > digits > 1177a > OpenDRC-DI > DEQ2496 > DAC2 > KCT > FPB 350mcx > reQuest > Sweetspot

Last edited by RayDunzl; 09-10-2014 at 11:29 AM.
RayDunzl is online now  
Old 09-10-2014, 11:35 AM
Advanced Member
 
3ll3d00d's Avatar
 
Join Date: Sep 2007
Location: London, UK
Posts: 639
Mentioned: 1 Post(s)
Tagged: 0 Thread(s)
Quoted: 289 Post(s)
Liked: 85
I am not sure what you mean by it "assumes pink is flat". Can you clarify? (NB: I know what pink noise is just not what you mean in that statement)
3ll3d00d is online now  
Old 09-10-2014, 11:38 AM - Thread Starter
AVS Special Member
 
RayDunzl's Avatar
 
Join Date: Sep 2013
Location: Tampa, Florida
Posts: 1,399
Mentioned: 2 Post(s)
Tagged: 0 Thread(s)
Quoted: 478 Post(s)
Liked: 671
Using Pink and White noise sources, measured by REW at the preamp, with no EQ

Red is Pink noise - expect downward slope, but it is displayed as flat.

Green is White noise -expect flat slope, but it is displayed as rising.


I'll be back later...


links::: 1.5RQ > digits > 1177a > OpenDRC-DI > DEQ2496 > DAC2 > KCT > FPB 350mcx > reQuest > Sweetspot
RayDunzl is online now  
Old 09-10-2014, 11:54 AM - Thread Starter
AVS Special Member
 
RayDunzl's Avatar
 
Join Date: Sep 2013
Location: Tampa, Florida
Posts: 1,399
Mentioned: 2 Post(s)
Tagged: 0 Thread(s)
Quoted: 478 Post(s)
Liked: 671
Here's a different measuring tool, which displays without fudging...

Pink



White


I'll be back later...


links::: 1.5RQ > digits > 1177a > OpenDRC-DI > DEQ2496 > DAC2 > KCT > FPB 350mcx > reQuest > Sweetspot
RayDunzl is online now  
 
Thread Tools


Forum Jump: 

Posting Rules  
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are Off
Pingbacks are Off
Refbacks are Off