Acceptable Group Delay - Subwoofers

[NTK]

The test you refer was done with headphones so if was incompetent for evaluating all effects and significance of group delay at L.F.

View attachment 422799

Regarding the suitability of the test method to loudspeakers, you can post that question to Aki Mäkivirta, who is the second author listed on the paper.

Aki Mäkivirta - Genelec.com

Aki Mäkivirta came to Genelec in 1995 from the Nokia Research Centre. He had gotten to know Ari Varla in a joint venture between the companies, within which they designed a speaker with digital signal processing. Before that, active monitors were based on analogue filtering and signal...

www.genelec.com

 

[kimmosto]

Regarding the suitability of the test method to loudspeakers, you can post that question...

Probably not. That study is really important and useful, and could be one of the reasons why Genelec has improved significantly especially in midrange GD since the first Ones and last analogue Masters. Much cannot be done anymore with their existing speaker concepts, latency limits and reliability/durability policy for pro audio.

Ears and auditory weighting has been almost standard in group delay testing. For example ref [8] Krauss, G. J., “On the audibility of group distortion at low frequencies,” in Proc. Audio Eng. Soc. 88th Conv., Montreux, Switzerland, Mar. 1990 was using Stax headphones: Stax Lambda Pro and Stax Monitor SRM.
Older study ref [7] Greenfield, R. and Hawksford, M. J., “The audibility of loudspeaker phase distortion,” in Proc. Audio Eng. Soc. 88th Conv., Montreux, Switzerland, Mar. 1990 was using Celestion SL700 loudspeakers. That wasn't very "deep" science with L.F. either.

Fortunately we can make our own applications and conclusions so possible poorness of audio science does not prevent anything.

 

[DaveBoswell]

One thing I have not seen discussed much is whether we can sense group delay distortion in low frequencies not by hearing the delay but by feeling it via a delay in the tactile sensation. I have read numerous reports over the years that we can sense bass directionality via tactile sensation, even at low frequencies where we would not be able to do this with our ears by interpreting the relative levels and phases between each ear (i.e. we can sense the chest slam from bass drums or nearby fireworks hitting the front of our bodies first and as such perceive the sound as coming from the front).

As such, one would think that there is a point where an appreciable delay between the physical sensation of the base for short impulses and the higher frequencies would be perceived.

The folks who setup tactile transducers in their setups have known for some time that getting the delay right between the tactile transducers and the audio is very important for the experience to feel seamless and natural. I have a sneaking suspicion that we might be more sensitive to group delay distortion a low frequencies than these headphone only studies would suggest - would be interested to know if anyone has come across any literature along these lines?

 

[TheZebraKilledDarwin]

Hey community,

I’m resetting my strategy for deploying multiple subwoofers in my setup. Information sources include fellow members' testing (@Nuyes I measured 10 Subwoofers) and spreadsheets (@sweetchaos Subwoofer Comparison). Also using Audioholics, Buttersworth, and Erin’s to identify those in my price and output ranges. Nuyes testing brought to my attention group delay and how varied it can be below 80hz. I really don't know anything about it and will certainly not be attempting to explain its significance. But it got me wondering at what point would measurements show "this is now a problem"? I have (tried) to understand posts and papers group delay that have been linked on other threads and I simply confidence in drawing my own conclusions. I did see one post on DIYAudio stating the ‘ideal’ is sub-one cycle at a given frequency. If true, the ideal group delays should measure less than:

20hz — 50ms
30hz — 33.33ms
40hz — 25ms
50hz — 20ms
60hz — 16.67ms
70hz — 14.29
80hz — 12.5ms

Ok, this is may not be the best rule of thumb and responders may quickly call out a 'general rule' error. If so, let's find something agreed upon to the best of our ability. Right now it's what I have found that I can run with. I want to learn more.

So why am I asking for help with this subject and possibly making knowledgable people on the forum tired of beating a dead horse? I want to sell my current subwoofer to find a more affordable unit to be able to scale up (2-4 subs). My current SVS 3000 micro nails the form factor, but there are less expensive options - namely the Kali WS-6.2. If I could afford them, I would be looking at the Paradigm MilleniaSub, Sigberg Inkognito, or continue with more 3000 micros. I'm not looking for crazy output or sub-20hz output. I just want to get the most for my money and be done with it with as much peace of mind as I can knowing I made the most educated decision I could.

The Kali looks viable. But what about group delay for a ported subwoofer like this one? Looks like there is a direct correlation between ported subwoofers and higher group delays. I don't mind the look of the WS-6.2 because my intention is to conceal them. Output looks respectable based on Erin's video review I stumbled upon through thread Anyone Tried Kali Audio WS-6.2 Subwoofer?. True it may not pass CEA-2010 below 25hz, and I'm ok with that.

20hz ≈ 45ms
30hz ≈ 42ms
40hz ≈ 26ms
50hz ≈ 16ms
60hz ≈ 12ms
70hz ≈ 10ms
80hz ≈ 9ms

Good, you just proved your own point, Justin. The Kali should fit the bill, so why post this at all? Glad you asked...

Presuming one cycle is the ideal, at what point would the group delay become audibly unacceptable? This isn't meant to put anything to bed, but to take my next subwoofer search and have a thread people can look back to in layman's terms, see how we got there, and what is reasonable and within an acceptable margin. Go easy on me.

Thanks!

How long are the decay times in your room?

 

[just1n]

How long are the decay times in your room?

I installed a second subwoofer since this thread started. I hope to take some measurements over the weekend if I can. I have never taken post-calibration measurements.

Anyone have a guide I can reference for MLP or MMM?

 

[kyuu]

Anyone have a guide I can reference for MLP or MMM?

For MMM you might try this thread.

 

[ExPerfectionist]

I installed a second subwoofer since this thread started. I hope to take some measurements over the weekend if I can. I have never taken post-calibration measurements.

Anyone have a guide I can reference for MLP or MMM?

Be sure you set up the polarity / time alignment properly with the sub distance settings. REW is a big help with that. Here's a few videos.

 

[just1n]

Be sure you set up the polarity / time alignment properly with the sub distance settings. REW is a big help with that. Here's a few videos.

Thank you. I am hoping that Dirac has correctly phased both subwoofers. I will find out soon enough.

 

[dualazmak]

This isn't meant to put anything to bed, but to take my next subwoofer search and have a thread people can look back to in layman's terms, see how we got there, and what is reasonable and within an acceptable margin. Go easy on me.

OK...
Just for your possible interest and reference, let me share my rather primitive approach of "using time-shifted multiple-Fq rectangular tune-burst signals", "energy peak matching using single/simultaneous tone-burst consists of multiple -Fq component" and "precise wave shape matching for 0.1 msec precision time-alignment".
- Precision measurement and adjustment of time alignment for speaker (SP) units: Part-1_ Precision pulse wave matching method: #493
- Precision measurement and adjustment of time alignment for speaker (SP) units: Part-2_ Energy peak matching method: #494
- Precision measurement and adjustment of time alignment for speaker (SP) units: Part-3_ Precision single sine wave matching method in 0.1 msec accuracy: #504, #507
- Measurement of transient characteristics of Yamaha 30 cm woofer JA-3058 in sealed cabinet and Yamaha active sub-woofer YST-SW1000: #495, #497, #503, #507

In case if you would be interested in using/applying the test tone-burst signals I prepared/used in these measurements and tunings, please simply let me know through PM communication writing your wish.

After these measurements and tunings in my listening environments, I have been enjoying really nice tight clear non-distorted bass sound with my multicannel audio rig.

Spoiler

Just for example:
- Reproduction and listening/hearing/feeling sensations to 16 Hz (organ) sound with my DSP-based multichannel multi-SP-driver multi-amplifier fully active stereo audio system having big-heavy active L&R sub-woofers: #782
A nice smooth-jazz album for bass (low Fq) and higher Fq tonality check and tuning: #910, #63(remote thread)

Just for your reference, please find the latest Fq-SPL spectrum (at listening position) of my audio setup under the below spoiler cover.

Spoiler

As for "calibration" of my ECM8000 measurement microphone, you would please refer to my post here.
- Frequency response of my BEHRINGER ECM8000 measurement microphone (specially selected unit in 2008): #831

I dare did-not/do-not go into further precise delay compensation tunings within the L&R subwoofer-covered Fq-zone, in my case ca. 15 - 90 Hz; my L&R heavy large subwoofer YAMAHA YST-SW1000 works in Helmholtz resonator (so-called air woofer) mechanism (ref. here) which cannot be precisely delay-controlled even applying upstream fine multiple DSP filters with group delay settings within 15 - 90 Hz, I believe.

Furthermore, please note, at least for me, I have no interest at all on "absolute group delay", but I have been being only interested in "relative group delay between the multiple SP drivers", i.e. "time-alignment all over the SP drivers", measured at my listening position.

 

[kimmosto]

One thing I have not seen discussed much is whether we can sense group delay distortion in low frequencies not by hearing the delay but by feeling it via a delay in the tactile sensation.
...
As such, one would think that there is a point where an appreciable delay between the physical sensation of the base for short impulses and the higher frequencies would be perceived.

Ears are poor for detecting magnitude differences. Tactile sensation does not seem to have that limit, or tactile sensation could magnify differences due to some thresholds. Excess group delay could drop peak energy and sound pressure quite radically though is does not have much effect to tone. Therefore it's easy to be convinced why excess delay is very difficult to hear by ears only (with headphones), but multiway speakers with too steep IIR XO slopes or subwoofer can be unacceptably lame due to relatively weak...missing tactile sensation.
Compression of the drivers and wavefront (spherical, cylindrical, planar) are also significant so radiating area and location of drivers matter. Rear and side woofers etc. are no no.

 

[just1n]

I tried doing a MLP measurement of my system earlier in the week and realized I did not disable Auromatic3D. I will need to remeasure.

One thing that confused me is that it seems the LFE signal did not test both subwoofers and maybe just subwoofer 1’s output.

Anyone know a what to measure subwoofers individually, or two measure them together as one LFE channel?

 

[ExPerfectionist]

You can measure both, so individually so you can see how they act on their own, and as a pair. That will also help you adjust "sub distance" / polarity to fine tune and make sure they're playing in unison.

 

[just1n]

Thank you. I’m wondering if I need separate subwoofer sweeps or if there is a setting in AVR to enable.

 

[levimax]

I have a pair of "mains" speakers that are old school large volume infinite baffle with a 12 " woofer. They have very little group delay as expected. I am trying to integrate them with two modern SVS 3000 subs which are 13" in relatively small sealed enclosures using lots of DSP. Group delay is ~85 ms at 20 Hz. It has not been easy to get the subs to integrate as I can set the delays so they are the same at an 80 Hz or 60 Hz crossover point but I start to get cancelling instead of blending at lower frequencies which makes sense looking at the group delay graphs. Is there anyway around this?

See below Green is Sub GD and Red is mains GD.

 

[terryforsythe]

Is there anyway around this?

It depends on how deep down the rabbit hole you want to go.

If you use something like CamillaDSP or a miniDSP unit, you could cross them over at, say, 80-90Hz with very steep crossover slopes. I use an 8th order for my subwoofer/speaker crossover, which helped me to get a smoother response by narrowing the frequency range at which they are overlapping.

Another option you could try is to flatten the group delay of the subwoofer with a FIR filter and adding an appropriate amount of time delay to your speakers. CamillaDSP will work for that. Some miniDSP units let you add FIR filters and some don't.

 

[TheZebraKilledDarwin]

A GD discussion without a single measurement of the decay sound in the room?

Contrary to group delay from speakers, the acoustic decay behaviour (resonances) in rooms is so strongly audible, that it is always annoying in untreated living rooms of normal sizes, it always degrades the sound quality immensely. While the group delay of a subwoofer may, or may not be heard by experienced and trained ears in highly optimized rooms...

 

[dualazmak]

Well, I believe we need suitably calibrated microphone measurement at your listening-position in your own listening-room in your listening-acoustic-environment using independent (second separate) PC or Mac (or other suitable independent digital sound recorders) for completely objective XO/EQ/Delay/Phase/Gain tuning of your L&R subwoofers for optimal matching with your other SP drivers.

Please refer to my above post #29 for example efforts on such primitive but reliable objective approaches and tunings.

 

[Keith_W]

I have a pair of "mains" speakers that are old school large volume infinite baffle with a 12 " woofer. They have very little group delay as expected. I am trying to integrate them with two modern SVS 3000 subs which are 13" in relatively small sealed enclosures using lots of DSP. Group delay is ~85 ms at 20 Hz. It has not been easy to get the subs to integrate as I can set the delays so they are the same at an 80 Hz or 60 Hz crossover point but I start to get cancelling instead of blending at lower frequencies which makes sense looking at the group delay graphs. Is there anyway around this?

See below Green is Sub GD and Red is mains GD.



View attachment 430592

That measurement is completely free of room artefacts. How did you take the measurement? Where was the mic?

Anyway it may be possible to design an all-pass filter to nudge the two curves closer together. You can use the green curve as a guideslope and time-reverse it and convolve it to green, but this will create a lot of pre-ringing. The alternative is to not time reverse it, and put that AP filter on the subwoofer (red).

Another approach is to not worry about it. Cross your subs over to your mains at say 50-60Hz with a steep filter and your group delay problem will go away

 

[levimax]

That measurement is completely free of room artefacts. How did you take the measurement? Where was the mic?

Anyway it may be possible to design an all-pass filter to nudge the two curves closer together. You can use the green curve as a guideslope and time-reverse it and convolve it to green, but this will create a lot of pre-ringing. The alternative is to not time reverse it, and put that AP filter on the subwoofer (red).

Another approach is to not worry about it. Cross your subs over to your mains at say 50-60Hz with a steep filter and your group delay problem will go away

Those measurements were done near field as I just wanted to see the group delay at the speaker as I was trying to figure out why adding the subs REDUCED LF output below the crossover point despite being time aligned at the crossover point.

I have played around with all-pass filters but am worried about ringing as it is a pretty strong filter to get rid of all the group delay.

I also read the "stereo bass" thread and am worried about summing the bass to mono

For now I settled on a solution of crossing to the subs at 30 Hz as there is a 30 Hz room mode so the mains play strong down to 30Hz in stereo and then the subs cover below that. I don't like this solution as it seems like a waste of the 2 SVS 3000 subs but with some FIR room correction filters I get fairly smooth response below Schroeder down to 20 Hz and the imaging and the balance sound very good to me.

I am sure this is not the final solution

 

[dan_remaincalm]

i've been playing around with this stuff too, to room-correct and integrate a set of neumann kh-120 ii with a pair of svs sb-1000 subs with a minidsp flex, REW and rephase in a heavily treated small room. mostly done.

from what i can tell, the svs subs introduce a ton of GD - they go pretty laggy under 50Hz. these ones also have a little bit of rumble that the measurements pick up on, which isn't ideal either.

i'm struggling with a few things:

* i can use massive FIRs (in equalizer APO) to make the step response look amazing, but the pre-ringing is bonkers. i believe this is just a fundamental constraint of this sort of correction?
* ... music sounded pretty great though - really prominent visceral "chest kick" with bassdrums, in particular, and a sense of extra clarity, but i suspect i was just feeling the pre-delay bass come through before the real kick attack. it was a pretty impressive sensation though.

anyway, i settled with using a 1024-tap FIR on just the mains, to line them up better with the subs (works better than a flat delay). went from +880deg to +560deg at 20Hz, +250deg to +120deg at 100Hz.

however, after a ton of unscientific A/B testing, I think I can only hear it on one very odd techno track, the change was basically inaudible on everything else despite the measurements looking a lot better.

FIR off vs 1024 tap (current) vs massive FIR:

Step responses: off vs current 1024 tap FIR vs abandoned massive FIR

1024 tap step response:

The one that I gave up on, step response:

Here's the whole system, 1024 tap - most of that weird crap in the low end is the SB-1000 I think, plus some of it is environmental rumble (I live near a busy road)

anyway it's sounding ok i guess.