Upgrade from miniDSP 2X4 HD to SHD Studio in a DEQX based system?

[davewantsmoore]

9 hours ago, almikel said:

if REW uses Impulse or Step to calculate its Acoustic Timing Reference for aligning a mains woofer with a sub,

It seems like you're confused about how the accoustic timing reference works.

 

It's a "blip" in the audio, which is the "zero reference".   So it doesn't use anything to "calculate it".   It just looks at where the "blips are" and aligns them in time.

 

9 hours ago, almikel said:

From the above graphs it's very clear that neither the Step or Impulse Responses are particularly useful for achieving time alignment, only phase

In practise, phase also has "issues" .... due to the inaccuracies of in room measurments.

 

In short... consider that the amplitute response you measure at 10Hz, 20Hz, 30Hz, whatever....... dramatically affects the phase chart of a woofer in the 50 to 150Hz range (where you might be crossing over your sub/woofer to a mid/woofer).

 

The trick (in that case) is looking at your LF response, and asking "is this really correct"  (ie. is my data a good represnentation of reality)... and if not, doing something about it.     Eg. something like deleting the ampliture response below where it is accurate, and replacing it with a simulation.

 

 

Taking the measured in-room response "warts and all" and correcting it flat .... is OK, for a well designed system....  but this is based on the proviso that the crossovers and big picture EQ for the drivers were designed using something better than flawed / in-room data  (ie. the "well designed system" bit).

 

I hope that makes sense.   These things are difficult to explain behind a keyboard.

 

9 hours ago, almikel said:

The author of MSO (I'm again paraphrasing - funky stuff going on with SNA tonight) has found good success with tweaking the amplitude (EQ) and delay of multiple subs in the frequency domain only when integrating multiple subs with main speakers

Yes.

If you correct the amplitude response of speakers.... the phase response fall into place.    If it doesn't... then you are measuring a non-minimum-phase effect ....which is probably something you should ignore (at least in the design of the speaker itself, eg. the EQ you apply to the speaker).

 

9 hours ago, almikel said:

...his approach aligns with @davewantsmoore's  of aligning phase as the priority

I don't really do it like that.

 

I design speakers with flat amplitde response (and crossover which rolloff the way I want them)....  Now I know they are flat and will sum correctly (when placed where physics says they are timealigned).     If I now measure something differerent "in room" - I must ask "why?"  (it's an error in the data).

 

At this point, I now have the basics sorted...... from there, using techniques similar to MSO.... I can tweak the response a bit here and there to get a better sum from multiple sources...... and then last I can transform the entire summed response to be something which aligns with my preference and hearing, and the power response of the speakers/room  (eg. a straight line drop of 10dB from 10Hz to 10Khz... or something).

 

9 hours ago, almikel said:

The goal of perfect time alignment between mains and sub/s is possibly/likely one of those things that audiophiles try to finesse too much

It's important.... in so far as if woofers are out of phase by more than 1/4WL .... then it's akin to a car driving around the racetrack with not all wheels spinning the same speed (or even in the same direction).

 

.... but looking at the measured phase curve..... relies on an enormous assumption that the amplitude response is a sensible represenation of the overall picture.

 

9 hours ago, almikel said:

- IMHO a smooth/flat FR at the listening position across the region between subs/mains is more important...obviously time alignment is part of that - but smooth phase and FR throughout the crossover region are the ultimate arbiters - not aligning any peaks in time domain measurements.

If you have a woofer and a subwoofer ..... that both have well designed flat resposne, and 24db high/low pass filters at the same frequency  (eg. design in anechoic conditions.... eg. simulated or outside).

 

Then just put them in room and they are alomst completely fine (aside from delay based on distance).... trying to deisgn these things with in-room data is fraught with issues.

 

This is why you'll recall seeing in many threads people suggesting "take it outside"  (or, use a simulator).

 

9 hours ago, almikel said:

(edit) if your sub ended up multiple cycles in front or behind your mains, but your room's frequency and phase response was smooth at the listening position - this would sound fine.

Given the wavelngths involved (eg. ~4m @ 80Hz) .... then this doesn't happen in practise.

 

I don't think it sounds fine either.   I think we are partiuclarly sensitve to timing at LFs ......  but it's just at other issues (eg. wild swings in levels.... eg. the typical +/-10dB of a good system in the bass - which is gargantuan) swamp the issue.....  and it is heavily related to the recording as well.

 

It's hard to explain in general terms..... but I think we all know some recording where the bass is extremely "spatial" and "dimensional".

 

 

 

I've said it before.... but all of these sorts of issues.... are why "EQ gets a bad name".    It is so easy to "measure -> fix"the wrong things .... and the paradox is that your "result" ...... looks perfect.

[davewantsmoore]

Heh. Sorry.... I suspect some of my posts aren't as helpful as I want them to be... and pose more quesitons than answers, or some such.

[Satanica]

13 hours ago, almikel said:

just sanity checking here after @davewantsmoore's post above, which was one of those "light bulb moment" posts for me...what does REW use/how does REW calculate its Acoustic Timing Reference for aligning a mains woofer with a sub?

as per above, if REW uses Impulse or Step to calculate its Acoustic Timing Reference for aligning a mains woofer with a sub, then it would be wrong...I'm not saying how REW does it is incorrect...I jut don't know how REW calculates its Acoustic Timing Reference.

From the above graphs it's very clear that neither the Step or Impulse Responses are particularly useful for achieving time alignment, only phase - and the above are theoretical/nearly perfect simulations - real room measurements would be much messier.

 

In a previous thread on a similar topic, the creator of Multi Sub Optimizer (MSO) chimed in - I can't seem to open another tab/window in SNA to check the details or even his name...but he discussed (and I'm paraphrasing here as I can't open another SNA tab to check) that theoretically impulse response could be used to time align perfect linear phase crossover filters, but based on current technology and the room's response, the actual acoustic crossover will never be close to a theoretical linear phase crossover - hence time alignment using impulse response was not a valid approach.

 

The author of MSO (I'm again paraphrasing - funky stuff going on with SNA tonight) has found good success with tweaking the amplitude (EQ) and delay of multiple subs in the frequency domain only when integrating multiple subs with main speakers - ignoring time domain measurements such as Step and Impulse Response...he's a smart guy...much smarter than I am...

...his approach aligns with @davewantsmoore's  of aligning phase as the priority - and seeking a smooth FR at the listening position.

 

The goal of perfect time alignment between mains and sub/s is possibly/likely one of those things that audiophiles try to finesse too much - IMHO a smooth/flat FR at the listening position across the region between subs/mains is more important...obviously time alignment is part of that - but smooth phase and FR throughout the crossover region are the ultimate arbiters - not aligning any peaks in time domain measurements.

(edit) if your sub ended up multiple cycles in front or behind your mains, but your room's frequency and phase response was smooth at the listening position - this would sound fine.

If your sub was much closer to "proper" time alignment, but not quite there, resulting in dips/peaks in the FR and big swings in phase - this would sound much worse (end edit)

 

cheers

Mike

Hi Mike, I honestly don't know what it does to in regards in terms of it's calculations. I was really just trying to follow a bouncing ball which was the miniDSP guide. I'm glad I did because my subs were measured to be about 58 an 68ms ahead of my left speaker presumably because of the processing delay of the DEQX and I don't think these measurements are too little to not "finesse" about. For me it was out of curiosity and getting them in the right ballpark.

 

With Dirac Live through the miniDSP SHD's there isn't anything such as subs. The left and right speakers are treated as a whole with its measurements. It actually doesn't know if there are subwoofers attached to the system. Well at least it doesn't appear to because only a Left and Right speaker show up in the Dirac Live configuration and there are no subwoofer sweeps; just full range room measurements Left and Right sweeps. The latter is something DEQX really should do but doesn't and I think I can remember that you used REW to confirm that the crossover region(s) of your multi DEQX system were OK.

Edited September 12, 2020 by Satanica

[almikel]

On 12/09/2020 at 8:36 AM, davewantsmoore said:

It seems like you're confused about how the accoustic timing reference works.

clearly a little (or maybe still a lot) - and I greatly appreciate the time you spend explaining stuff... 

 

On 12/09/2020 at 8:36 AM, davewantsmoore said:

(Acoustic timing reference) It's a "blip" in the audio, which is the "zero reference".   So it doesn't use anything to "calculate it".   It just looks at where the "blips are" and aligns them in time.

back in the day, guys like Wallace Sabine would use a starting gun to measure the impulse response of a room - which is sort of equivalent to your "blip" - which may be adding to my confusion...

 

I understand how the timing reference can time align 2 identical subs to each other (or identical woofers in main speakers)

Based on your earlier "ideal" graphs of time aligned impulse, step, phase of a low pass and high pass combining, I don't get how you could use the blip to align a sub with a mains woofer...

Below is the time aligned impulse response from your post

...maybe the key is to ignore the pre-ringing of the high pass (as noise), and the peak of the low pass as irrelevant - the only thing important is when the low pass driver starts moving (ie red line moves from zero), compared to when the high pass driver starts moving (ignoring pre-ringing somehow)

But looking at the Step response of the time aligned example

The low pass response starts to move after the high pass response.

 

I'm still confused on the mathematical and practical differences between:

• Acoustic Timing Reference using a "blip" in the audio, which is the "zero reference".   So it doesn't use anything to "calculate it".   It just looks at where the "blips are" and aligns them in time.
• Impulse and Step responses in relation to the above

I get that mathematically the Impulse Response is the derivative of the Step Response - which is why the red impulse response drops back to zero when the red step response stabilises....

 

No scales on the X axis are shown - but assuming the vertical graduations are equivalent between Impulse and Step (??? big assumption), the peak in Step seems way behind the peak in Impulse...

 

I'm wondering if this is because the Impulse is the derivative of Step, and the transfer function for Step had a constant at the end that disappeared on differentiation??

Or is this gibberish?...thinking further I think it is - any constant that got lost during differentiation would impact the Y axis, not the X axis...

 

(edit) - often we see graphs of perfect Impulse Responses stacked on one another (which as per dave's graphs above does not achieve time alignment), but never Step Responses - With Impulse aligned (which isn't time alignment) Step remains offset  -  what causes this? (end edit)

 

...apologies to @Satanica for the off topic - I'm just trying to get my head around the maths and apply it to reality

 

cheers

Mike

 

 

Edited September 13, 2020 by almikel

[davewantsmoore]

14 hours ago, almikel said:

.maybe the key is to ignore the pre-ringing of the high pass (as noise), and the peak of the low pass as irrelevant - the only thing important is when the low pass driver starts moving (ie red line moves from zero), compared to when the high pass driver starts moving (ignoring pre-ringing somehow)

 

Well.... it knows exactly what the "blip" looks like.... so it juts finds the (identical) blips in each (which will occur well before any other sound) .... and aligns them both.

[sir sanders zingmore]

2 hours ago, davewantsmoore said:

Well.... it knows exactly what the "blip" looks like.... so it juts finds the (identical) blips in each (which will occur well before any other sound) .... and aligns them both.

Is that a good (ie the "right") way to do it?

[Satanica]

2 hours ago, davewantsmoore said:

Well.... it knows exactly what the "blip" looks like.... so it juts finds the (identical) blips in each (which will occur well before any other sound) .... and aligns them both.

Just to be clear for others who have not used the functionality; it doesn't actually align anything but tells you how much you could align them if you wanted to.

[Satanica]

17 hours ago, almikel said:

...apologies to @Satanica for the off topic - I'm just trying to get my head around the maths and apply it to reality

No worries Mike,  I think this is on topic and I wouldn't care if it wasn't.

I don't like the thought of a thread just being about me just because I started it.

[Satanica]

From: https://www.roomeqwizard.com/help/help_en-GB/html/makingmeasurements.html

 

 

Measuring with a timing reference

REW can make use of a timing reference when it measures, according to the setting on the measurement panel or in the Analysis preferences. The timing reference selection controls whether REW uses a loopback on the soundcard as a timing reference, or an acoustic timing reference, or no reference. Using a timing reference allows REW to eliminate the variable propagation delays within the computer and soundcard so that separate measurements have the same absolute timing.

If a loopback is selected the reference channel signal must be looped back from output to input on the soundcard and measurements will be relative to the loopback timing. Usually this means measurements will have a time delay that corresponds to the time it takes sound to travel from the speaker being measured to the microphone.

If an acoustic timing reference is used REW will generate a timing signal on the output that has been selected to act as the reference before it generates measurement sweeps on the channels being measured. The level of the timing reference is set separately from the measurement level using the control at the top of the measurement dialog. The timing signal is a high frequency sweep to allow accurate timing, a subwoofer cannot be used as the reference channel. Measurements will have a time delay that corresponds to the difference in their distance from the microphone compared to the distance of the reference speaker - if the reference speaker is further away the delay would be negative. When an acoustic timing reference is used individual measurements taken from the same mic position will have the same relative timing, allowing trace arithmetic to be carried out on the traces in the All SPL graph. Note that multiple sweeps cannot be used when using an acoustic timing reference.

If using a timing reference REW can calculate the delay through the system being measured relative to the reference and show it in the measurement Info panel as "System Delay" in milliseconds, with the equivalent distance in feet and metres shown in brackets. For speakers the delay estimate is based on the location of the peak of the impulse response. Subwoofers have a broad peak and a delayed response due to their limited bandwidth so the delay is instead measured relative to the start of the impulse response. The start of the impulse response cannot be located as precisely as the peak, however, so delay values are less accurate for subwoofer measurements.

 

 

 

Interesting is the very last sentence which at least raises one question for me. How did it know whether I was measuring a speaker or a subwoofer against the timing of my Left speaker? I can't remember a setting to signify as such; so I wonder if it has the smarts to know if it measures the response of a speaker as full-range then it applies its calculations on the peak of the impulse and if it measures the response of a speaker as bass-range-limited then it assumes its a subwoofer and applies its calculations on the start of the impulse. There seems to be that implication above although I'm not absolutely sure. And the very last sentence that subwoofers are less accurate to measure for the given reason is perhaps what I observed. I measured my subwoofers numerous times and there was a variance of about 3 milliseconds with each one without me physically moving anything whereas measuring my Right speaker resulted in virtually the same measurement compared to my Left speaker every time.

Edited September 14, 2020 by Satanica

[davewantsmoore]

6 hours ago, Satanica said:

Just to be clear for others who have not used the functionality; it doesn't actually align anything but tells you how much you could align them if you wanted to.

Are you talking about REW?! 

[Satanica]

Just now, davewantsmoore said:

Are you talking about REW?! 

Yes sir.

[davewantsmoore]

7 hours ago, sir sanders zingmore said:

Is that a good (ie the "right") way to do it?

I use loopback timing reference

[gibbo9000]

4 hours ago, Satanica said:

How did it know whether I was measuring a speaker or a subwoofer against the timing of my Left speaker?

From my experience it doesn't have the capability to distinguish.  If you are using the acoustic timing reference it expects you to turn off everything but the Left Main (if chosen as timing reference) and the speaker you are testing - so just Right Main or Sub.  It took me a few attempts to work out that is what it wanted (and how to do it!).  That way it simply compares the timing of the reference 'ping' on Left to the other speaker you are comparing it too. 

Will defer to others if I have that wrong - but worked for me.

[Satanica]

23 hours ago, gibbo9000 said:

From my experience it doesn't have the capability to distinguish.

Not before the measurement, no. But from the documentation it seems to imply that REW will determine what is a speaker and what is a subwoofer from the bandwidth detected from a measurement. And for its acoustic timing calculation it will use a different method for a speaker versus a subwoofer. To quote the documentation again.

 

For speakers the delay estimate is based on the location of the peak of the impulse response.

 

So for regular speakers REW will use peak of the impulse for comparison against the peak of the impulse of the timing reference speaker.

 

Subwoofers have a broad peak and a delayed response due to their limited bandwidth so the delay is instead measured relative to the start of the impulse response.

 

And for subwoofers REW will use the start of the impulse for comparison against the peak of the impulse of the timing reference speaker.

 

The start of the impulse response cannot be located as precisely as the peak, however, so delay values are less accurate for subwoofer measurements.

 

As I mentioned before this is consistent with my experimentation that from measurement to measurement there was a slight difference with each of my two subwoofers. Seemingly, it is harder for the fuzzy logic of the REW program to determine the start of an impulse from a subwoofer and therefore less accurate when compared to determining the peak of an impulse from a speaker. I found there was a variance of about 3ms from measurement to measurement when measuring my subwoofers. If this inaccuracy presents any consequence then I trust that Dirac Live compensated for it.

Edited September 15, 2020 by Satanica

[almikel]

On 14/09/2020 at 5:29 PM, davewantsmoore said:

I use loopback timing reference

I agree this is the best timing reference to use with REW - and assuming that's the timing reference used:

On 12/09/2020 at 8:36 AM, davewantsmoore said:

It seems like you're confused about how the accoustic timing reference works.

 

It's a "blip" in the audio, which is the "zero reference".   So it doesn't use anything to "calculate it".   It just looks at where the "blips are" and aligns them in time.

From Satanica's review of how REW calculates "time alignment",

12 hours ago, Satanica said:

For speakers the delay estimate is based on the location of the peak of the impulse response...

...Subwoofers have a broad peak and a delayed response due to their limited bandwidth so the delay is instead measured relative to the start of the impulse response....

...The start of the impulse response cannot be located as precisely as the peak, however, so delay values are less accurate for subwoofer measurements...

 

we're back to Impulse Response - but at least the start of the Impulse Response for the sub, which is what I understood from your posts

On 14/09/2020 at 7:54 AM, davewantsmoore said:

Well.... it knows exactly what the "blip" looks like.... so it just finds the (identical) blips in each (which will occur well before any other sound) .... and aligns them both.

But they won't be identical blips in each after one has gone through a high pass filter and the other through a low pass filter (for the mains/sub alignment case) - but I'm happy to keep things simple for this discussion - let's assume a perfect acoustic LR4 crossover has been achieved...

...would REWs method of aligning the peak of the high pass Impulse Response with the start of the low pass Impulse Response work?

 

Having done my head in contemplating how to achieve "ideal" time alignment between subs/mains "in room", I think REW's approach is probably as good is it gets as a starting point, then tweak from there looking at both phase and amplitude throughout the crossover region at the listening position - seeking a flat FR and consistent changes in phase (ie no dips/peaks/swings in either)...

...which is hard to achieve "in room" at the "listening position"

 

 

mike

 

[Satanica]

10 hours ago, almikel said:

But they won't be identical blips in each after one has gone through a high pass filter and the other through a low pass filter (for the mains/sub alignment case)...

As per the miniDSP\REW instructions all bass management should be disabled before the acoustic time alignment and I did remember to do this.

Setting things up[Top]

First of all, disable bass management in the nanoAVR-BM plugin. The easiest way to do this may be to switch to an unused configuration i.e. one that is at the default settings. Or, if you are connecting your computer directly to an AVR for speaker timing measurement, disable bass management in your AVR (often accomplished by setting all speakers to "large".).

 

https://www.minidsp.com/applications/auto-eq-with-rew/measuring-time-delay

 

[davewantsmoore]

11 hours ago, almikel said:

But they won't be identical blips in each after one has gone through a high pass filter and the other through a low pass filter

 

REW Manual.

 

Quote

If an acoustic timing reference is used REW will generate a timing signal on the output that has been selected to act as the reference before it generates measurement sweeps on the channels being measured. The level of the timing reference is set separately from the measurement level using the control at the top of the measurement dialog. The timing signal is a high frequency sweep to allow accurate timing, a subwoofer cannot be used as the reference channel. Measurements will have a time delay that corresponds to the difference in their distance from the microphone compared to the distance of the reference speaker - if the reference speaker is further away the delay would be negative. 

 

The timing reference is played through a different speaker then the one(s) being measured.

Edited September 16, 2020 by davewantsmoore

[frednork]

Have only just caught up on this and have found (after some probing questions from @davewantsmoore ) that even  with an acoustic timing reference on the same speaker being tested for a full range sweep played off a file,  that rew rarely will align the time the same way for all sweeps done. so I need to do an "estimate IR delay" to get phase aligned and sometimes that is not that successful either.

Edited October 8, 2020 by frednork

[almikel]

12 hours ago, frednork said:

Have only just caught up on this and have found (after some probing questions from @davewantsmoore ) that even  with an acoustic timing reference on the same speaker being tested for a full range sweep played off a file,  that rew rarely will align the time the same way for all sweeps done. so I need to do an "estimate IR delay" to get phase aligned and sometimes that is not that successful either.

lol - what a rabbit hole "time alignment" can be...

...it's been 12 months or more since I made a measurement of my room...likely a good thing...

...system sounds fine, bass is great...thinking I won't F$%#@ck with it...

[frednork]

33 minutes ago, almikel said:

lol - what a rabbit hole "time alignment" can be...

...it's been 12 months or more since I made a measurement of my room...likely a good thing...

...system sounds fine, bass is great...thinking I won't F$%#@ck with it...

Yep, maybe wait till you get itchy

[davewantsmoore]

23 hours ago, frednork said:

even  with an acoustic timing reference on the same speaker being tested for a full range sweep played off a file

?

The use of an accoustic timing reference (if done properly) should result in good data.

23 hours ago, frednork said:

so I need to do an "estimate IR delay"

This is rarely reliable.... it will often get "close", but not for lumpy and/or LF data.

[frednork]

6 minutes ago, davewantsmoore said:

?

The use of an accoustic timing reference (if done properly) should result in good data.

This is rarely reliable.... it will often get "close", but not for lumpy and/or LF data.

Hmmm, might try another laptop and see if it is any better

[almikel]

On 10/10/2020 at 9:17 AM, frednork said:

Hmmm, might try another laptop and see if it is any better

I would doubt changing laptops would make any difference...but I use an external USB sound card as my mic requires phantom power...and when I bought my USB sound card 10 years ago I didn't understand enough to get one that supports proper 2 channel to provide loopback ...I need a new USB sound card

 

Mike