Group Delay Questions and Analysis

[wesayso]

That's for you to answer Jim. In your shoes I'd buy a soundcard with both optical and coax and find out. The software package doing the playback is also often a point of discussion. I trust JRiver in it's bit-perfect output, I'm messing up those bits on purpose anyway .
My Asus Xonar Essence ST and my DAC do sound different though. Not sure which I like most.

 

[gmad]

Greg, Can you help me understand the process a bit better?

Q1) OK, you took my system mdats. Somehow, those mdats became a blueprint of sorts for the DRC. Can you explain simply what that involved?

Q2) If I am understanding this correctly, the room measured mdats are essential to the process?

Q3) So basically, if I change the room response in any significant way, one would have to remeasure (make new mdats), and plug them into the DRc all over again to establish a new correction blueprint, yes?

A1) The mdat measurements were simply exported from REW as impulse responses in wav format, then converted to raw (using Audacity) to feed DRC.

A2) Yes. The window size (4 cycles in this case) determines the area over which the corrected sound will be valid (or artifact free). With a small window like this, you'll probably notice only improvements from wherever you listen.

A3) Yes. This is one of the reasons I would advise against offline correction, and encourage you instead to make the leap to computer based playback.

 

[xrk971]

Jim,
That "no photo" track from Conqueror sounds really nice on my 10F/RS225 FAST boxes. Great track for testing speakers, lots of textures and dynamic range.

 

[jim1961]

Jim,
That "no photo" track from Conqueror sounds really nice on my 10F/RS225 FAST boxes. Great track for testing speakers, lots of textures and dynamic range.

Yea, there are 20-30 tracks I rotate around when doing system evaluations. This is one of them.

 

[jim1961]

Is there a way to make the DSP only work on part of the frequency range?

Say 300hz and below?

 

[gmad]

It's possible, but I think it would be better to first try a slightly shorter window (3 cycles) if you feel there was too much of a change. Like I said earlier, this is powerful stuff and I think it's best to ease into it.

By the way, I'm guessing that with the 4 cycles window length, the correction is transitioning from speaker/room correction to speaker only correction at just a few hundred hz.

 

[wesayso]

I'd say first try a few other target curves to know what does what.
But it's difficult to judge after hearing only one song. The difference in target curves does a lot to the perceived sound, even if it's a little change.
You could also do 4 cycles below and 2 up top...

 

[Barleywater]

The computer only has 1/8" audio outs and USB period. The DAC USB inputs are inferior (use a different internal chipset).

Believe me, there is no way to do this correctly without adding a coax out to the computer.

To best of my knowledge Rega DAC USB connection is asynchronous; USB connection is only supplying digital data that is handed to the DAC circuit.

If real audible difference exists, problem is most likely software.

Do you have documentation about your Rega DAC having two different DACs built in?

 

[jim1961]

Left Ch. https://www.dropbox.com/s/yvychnovgjcscll/05 23 15 REW MM Left 1M.mdat?dl=0

Right Ch. https://www.dropbox.com/s/t9eczhz3gtcawtt/05 23 15 REW MM Right 1M.mdat?dl=0

Here are some new mdats. I changed things a bit.

Lets try the 3 cycles, and can we knee the response at 5K to be flat to 20k?

 

[gmad]

I'd say first try a few other target curves to know what does what.
But it's difficult to judge after hearing only one song. The difference in target curves does a lot to the perceived sound, even if it's a little change.
You could also do 4 cycles below and 2 up top...

Good point. Small changes in FR target make a big difference especially when the IR is clean. Also, I think our choice of FR target is probably affected by the time domain behavior of the system; once the IR is very clean we may no longer be happy with a particular FR target.

 

[jim1961]

To best of my knowledge Rega DAC USB connection is asynchronous; USB connection is only supplying digital data that is handed to the DAC circuit.

If real audible difference exists, problem is most likely software.

Do you have documentation about your Rega DAC having two different DACs built in?

Rega DAC D/A processor Page 2 | Stereophile.com

In the case of direct USB connection, the data go to that Burr-Brown PCM2707 receiver mentioned above. With S/PDIF inputs, the data go to a Wolfson Microelectronics WM8805 receiver chip.

 

[gmad]

Left Ch. https://www.dropbox.com/s/yvychnovgjcscll/05 23 15 REW MM Left 1M.mdat?dl=0

Right Ch. https://www.dropbox.com/s/t9eczhz3gtcawtt/05 23 15 REW MM Right 1M.mdat?dl=0

Here are some new mdats. I changed things a bit.

Lets try the 3 cycles, and can we knee the response at 5K to be flat to 20k?

ok, no problem as long as you're ok with having multiple variables.

 

[jim1961]

ok, no problem as long as you're ok with having multiple variables.

If you want to tinker, and give me back different versions, that is ok also as long as you label what each is.

Since I have to burn a whole CD to listen, may as well fill it up with more than one version.

 

[gmad]

If you want to tinker, and give me back different versions, that is ok also as long as you label what each is.

Since I have to burn a whole CD to listen, may as well fill it up with more than one song or one version.

Sounds good. I'll take my time and give you a few things to try.

 

[jim1961]

Here is what I changed.



This is probably as far as I can go diminishing the 40hz GD. It is starting to push up around 65hz.

 

[jim1961]

Here is my present GD (red) compared to the last correction curve (orange)

Now my problem seems more at 55hz than 40hz.

 

[wesayso]

If you remember this from before:

The answer is a ported speaker will usually have a phase delay of 180 degrees (but you are using the rear of the driver wave which is already 180 degrees out of phase) so the resulting sound (at the tuning frequency) is delayed but in phase.

The problem arises at higher frequencies - that's why you don't want any higher frequencies coming out of the port ideally - they will dilute the 'punch' significantly as they won't be exactly in phase with the driver output (the port provides the correct delay only for the tuning frequency).

Yes you do want to match distances of the sub and main driver to maintain phase unless you have a phase adjuster built in so you can change it manually.

Figure out the port frequency and the sound will be one cycle delayed at that frequency (but in phase). Nothing you can change about that with the ported box(es) there...

 

[jim1961]

If you remember this from before:



Figure out the port frequency and the sound will be one cycle delayed at that frequency (but in phase). Nothing you can change about that with the ported box(es) there...

I remember. I have also read that one octave above the tuning frequency, The GD of a ported speaker is about the same as a sealed one.

I have retuned my main sub to 16hz, and my mains to 26hz.

 

[wesayso]

So at 26 Hz it will be 38.5 ms delayed. Don't look at the wiggles, Smooth it a bit to see you're below 5ms at 55 Hz. The wiggles aren't static, they move with time.
I'm sure Greg will eek every ounce of Group Delay out of it.
Couldn't you seal off the mains? Let the 3 subs do their job there?

 

[jim1961]

So at 26 Hz it will be 38.5 ms delayed. Don't look at the wiggles, Smooth it a bit to see you're below 5ms at 55 Hz. The wiggles aren't static, they move with time.
I'm sure Greg will eek every ounce of Group Delay out of it.
Couldn't you seal off the mains? Let the 3 subs do their job there?

No. Lose too much SPL, and creates cone excursion problems.