Subwoofer level question. When to boost subwoofer output for dips?

[dasherzx]

I have some dips around 60-100hz for the left channel+subwoofers. I was wondering if its wise to boost the subwoofer output further because it doesnt seem to be a total null. Im using the minidsp flex and running the mains full range while the sub is crossed at 120hz with a 6db/octave butterworth filter, courtesy of a video i saw online recently. Anyways, even with a normal 80hz crossover the dip is still present whether i invert the phase or not.

 

[Chris A]

Create a plot for "excess group delay" (the portion of the group delay in excess of the minimum phase results only, i.e., that portion of the response in excess of what the Hilbert transform of the amplitude response predicts) from your in-room REW measurement(s). Smooth the plot using 1/3 octave smoothing to see the underlying trend.

In those areas of the plot that are not flat, you can't boost the input signal to compensate for apparent nulls in measured results. The only thing you can do is move the phase around a little.

You can thank D'Appolito for that little bit of wisdom.

Chris

 

[levimax]

I use REW's automatic filters. I have powerful subs that can play 15dB+ louder than the mains so I have plenty of headroom, if you don't have lots of headroom don't try to boost dips. First I try with no boost, then I try with 3 dB of boost and if the dips "come up 3 dB" that indicates they are more or less minimum phase so then I allow more boost to fill the dips "halfway" and then "all the way". If the dips don't come up with the first 3 dB of boost I leave them alone. I usually end up with 3 filters "no boost", "middle boost" , "full boost". I then listen to them on a range of music and pick the one that sounds best. I usually end up preferring the "middle boost" filter. There are no rules set in stone and a lot depends on your system (few systems have any LF headroom) and the room so you need to take time to listen to the results.

 

[Chris A]

Usually what occurs in-room is that you're taking measurements at the listening position (LP) or around the LP(s)--and then trying to boost room null positions.

That only thing that I've found that actually works, no matter how much "headroom" that you think you've got with the sub(s), is to add another low frequency source in another wall location--either another full-range loudspeaker or sub at the "antinode" room position on centerline between the left/right walls or between the front and rear walls to fill up the room's null mode positions as much as is possible. This is an effective method of dealing with a room null. But only so much can be filled using this approach,, especially if your LP is at a double or triple null room position.

Smoothing a room null mode then becomes a matter of working with the room's exact geometry and size--not amplifier or subwoofer power. Once you add another channel of low frequency output, you'll find this out for yourself.

It's like you've actually got two rooms that you're trying to fill up with low frequency output, It's very room and listening position-geometry specific.

Chris

 

[levimax]

Usually what occurs in-room is that you're taking measurements at the listening position (LP) or around the LP(s)--and then trying to boost room null positions.

That only thing that I've found that actually works, no matter how much "headroom" that you think you've got with the sub(s), is to add another low frequency source in another wall location--either another full-range loudspeaker or sub at the "antinode" room position on centerline between the left/right walls or between the front and rear walls to fill up the room's null mode positions as much as is possible. This is an effective method of dealing with a room null. But only so much can be filled using this approach,, especially if your LP is at a double or triple null room position.

Smoothing a room null mode then becomes a matter of working with the room's exact geometry and size--not amplifier or subwoofer power. Once you add another channel of low frequency output, you'll find this out for yourself.

It's like you've actually got two rooms that you're trying to fill up with low frequency output, It's very room and listening position-geometry specific.

Chris

I think it is an interesting subject with a lot of "dogma" and not so much settled science. If you are trying to fill dips "headroom" is the big elephant in the room. Assuming you have a 1,000 watt per channel stereo system "boosting" 6 dB effectively reduces it to a 250 watt per channel system and if "boosting" by 10 dB effectively reduces it to a 100 watt per channel system.

This is a real question, not trying to make a point. Outside of the obvious headroom issue why is "cutting" a minimum phase peak universally recommended but trying to fill a minimum phase null almost universally discouraged? It seems to me they are the same thing, the FR is only correct at the LP but that is the case with a cut or a boost but maybe I am missing something.

The other thing I question is that "adding a speaker" is always the best solution. I think for MC it is probably the best answer in most cases but for a stereo system I have found adding more speakers often causes more trouble that it solves (based on experimenting with 1, 2, 4, and 6 subwoofers in a stereo system.) One problem with "adding a speaker" is that while in theory you can find a "good place" for it to be helpful, the reality is this "good place" is seldom available in a domestic setting. Even if the spot is available there are other problems adding another sub causes, the biggest being giving up stereo bass and most importantly giving up all the advantages of a "point source transducer" (assuming full range speakers or tightly colocated subs and mains). Two point sources give you two in phase wave fronts which are relatively coherent and well behaved in the time domain which I find gives better slam / bass transient behavior. In addition 2 point sources gives a MUCH bigger sweet spot than adding more speakers. Finally an additional LF source in the room forces a low crossover or localization becomes and issue. In most cases this both reduces system headroom and limits DSP options.

As mentioned all of this is room and LP dependent but for me for stereo I have had consistently better results with 2 powerful full range speakers (or tightly collocated mains/ sub) optimized with the best practical locations available in the room and careful DSP and listening "mid-field". YMMV.

 

[Oddball]

We like analogies on this forum, just to have fun. Your question is almost like how do you like your steak done, if Argentina is better than local beef, should you best enjoy it in well lit room or with lights dimmed and what is the best company to enjoy the steak.

There are so many overriding factors that it will be yourself making decisions that measure well and sound well.

Stereo bass is really catching fire, so is "envelopment". Looks like if you are not attending to these you are still stuck in the stone age. But that is your decision to make. Stone age was at least simple. Also, having healthier interests in life that splitting hairs about audio helps mental health according to a good friend of mine that is a decent clinical psychologist. I obviously disregard his good advice as currently participating in this thread.

 

[YouDontHaveEnoughBass]

Out of curiosity could you link the video about running subs up to 120hz with the 6db roll-off please, I'm running mine 100hz 24db atm would be cool to see what such a shallow slope can do

 

[90sNintendoKid]

View attachment 538884
I have some dips around 60-100hz for the left channel+subwoofers. I was wondering if it’s wise to boost the subwoofer output further because it doesnt seem to be a total null. Im using the minidsp flex and running the mains full range while the sub is crossed at 120hz with a 6db/octave butterworth filter, courtesy of a video i saw online recently. Anyways, even with a normal 80hz crossover the dip is still present whether i invert the phase or not.

I would say that looking at the predicted response in the Dirac app, to me it looks like it’s already done something to that dip between 60hz and 100hz. I would use the moving mic method at the mlp and then use psycho acoustic smoothing and analyse the the frequency response in REW to see what it looks like.

I also concur with adding more subwoofers.

 

[90sNintendoKid]

Create a plot for "excess group delay" (the portion of the group delay in excess of the minimum phase results only, i.e., that portion of the response in excess of what the Hilbert transform of the amplitude response predicts) from your in-room REW measurement(s). Smooth the plot using 1/3 octave smoothing to see the underlying trend.

In those areas of the plot that are not flat, you can't boost the input signal to compensate for apparent nulls in measured results. The only thing you can do is move the phase around a little.

You can thank D'Appolito for that little bit of wisdom.

Chris

Aside from reading the book, may I please ask the best resources and other books you recommend for this type of speaker EQ knowledge?

 

[kemmler3D]

trying to fill a minimum phase null almost universally discouraged? It seems to me they are the same thing,

A peak is like 2+2 = 4, if you reduce that by 50% you get the intended outcome.

A null is like 2-2=0, so raising that by 50% still gets you 0, you can "go broke" on amp power or driver excursion trying to fill nulls.

the FR is only correct at the LP but that is the case with a cut or a boost but maybe I am missing something.

Yes, true, but in the case of bass peaks / nulls the physical size of the corrected area in the room is large (due to long wavelengths) so the correction at the LP is useful for listening.

Overall I think the caution against filling nulls in real rooms is a little exaggerated. You don't get perfect nulls when the room doesn't have perfectly reflective walls / ceilings with no furniture. Therefore when you boost a null the result isn't really just more null, you can fill them in practice, somewhat, if you have the headroom.

There are other bad side effects (making ringing worse) but it may be worth the trade-off in some rooms.

 

[dasherzx]

Out of curiosity could you link the video about running subs up to 120hz with the 6db roll-off please, I'm running mine 100hz 24db atm would be cool to see what such a shallow slope can do

This one, the 6db was just me trying out the lowest option on my minidsp since the video mentioned shallow roll off. Frankly I think 80 hz is still what most will do, I just don’t like having knowing my mains could go 35hz with no issue and having to cross them at 80

 

[levimax]

Frankly I think 80 hz is still what most will do, I just don’t like having knowing my mains could go 35hz with no issue and having to cross them at 80

Ideally you want flat (SPL and phase) driver response an octave on each side of the crossover frequency so for 80 Hz crossover having the mains with response to 35 Hz is a good thing.

 

[Chris A]

Aside from reading the book, may I please ask the best resources and other books you recommend for this type of speaker EQ knowledge?

The D'Appolito book (Testing Loudspeakers) is excellent, but its focus is on test, not dialing-in loudspeakers using DSP, etc.

I spent a few years experimenting with REW techniques and interpreting data, as well as using DSP crossovers in-room. That's the resource that I actually rely on.

One of the reasons why I started writing a book on the subject of home hi-fi audio, on how to practically achieve it, was that there was virtually nothing written on it. Unfortunately, I stopped writing when I saw what the GPT providers were doing to copyrighted materials.

I think Floyd Toole's book (any edition) is pretty balanced and insightful--but pretty short on the "how to". He also issued a multi-part white paper on subwoofers and low frequency audio in-room while he was at Harman. Geddes wrote on the subject of subwoofers in-room and some of that is good.

Chris

 

[Ultrasonic]

n those areas of the plot that are not flat, you can't boost the input signal to compensate for apparent nulls in measured results. The only thing you can do is move the phase around a little.

This is a common misconception. It would be true for a theoretical perfect null (complete cancellation) but is not true for pretty much any real-word dips in the frequency response in the bass region. It's trivial to prove otherwise by applying a boost and then measuring the result. Dirac does this as the OP's graph indicates.

The reason is that firstly you don't get complete cancelation in the real-word, and there are multiple reflections contributing and there are losses on reflection and transmission through objects in the room. Once you have incomplete cancellation you have a net signal that scales linearly with the signal level output by the speakers. So boosting does reduce the dips in the frequency response.

However, there are important issues to consider in terms of overall system headroom and distortion increasing if speakers/subs are pushed too hard. So just because dips can be boosted doesn't mean they should. Some boosting is appropriate in many systems though. Dirac will apply up to 10 dB boost in its stereo version for context which I do think is high, but in practice the way it bases a correction on a spatial average of multiple points means that typically the maximum boost applied is significantly smaller.

 

[ryanosaur]

You cannot boost your way out of a null. You are fighting physics here.

Also, adding that much trim to the signal will likely be in the digital domain. This is known to be one of the fastest ways to add distortion to your signal.

 

[Ultrasonic]

I have some dips around 60-100hz for the left channel+subwoofers. I was wondering if its wise to boost the subwoofer output further because it doesnt seem to be a total null. Im using the minidsp flex and running the mains full range while the sub is crossed at 120hz with a 6db/octave butterworth filter, courtesy of a video i saw online recently. Anyways, even with a normal 80hz crossover the dip is still present whether i invert the phase or not.

I wouldn't personally boost more than Dirac is already.

Have you experimented with your subwoofer and/or listening positions to reduce this dip? That's the first thing to do before using Dirac Live to try to optimise things further.

 

[Chris A]

You cannot boost your way out of a null. You are fighting physics here.

This is correct. If you want to change the amplitude response of the measured signal, simply change the position of the microphone. That's effectively what you're trying to do with boosting nulls via EQ in-room--a phase rotation of the direct arrivals and their reflections.

Also, adding that much trim to the signal will likely be in the digital domain. This is known to be one of the fastest ways to add distortion to your signal.

I'm not sure what the meaning of "in the digital domain" actually means here. If I understood it more, I might be able to agree with the statement.

Chris

 

[Chris A]

I wouldn't personally boost more than Dirac is already.

My experience with this app has been uniformly poor, and all that poor experience can be traced to the way in which it takes measurements--and usually centered in the transition band (Schroeder frequency in-room).

You need to pay attention to the direct arrivals (i.e., minimum phase) and not the interference of the direct arrivals + primary reflections. That means that measurements taken at the listening position are mostly problematic.

Virtually no one is paying attention to this issue, I've found. Dirac demonstrably cannot separate the direct arrivals from the reflections (non-minimum phase) at the listening position.

Chris

 

[Ultrasonic]

You cannot boost your way out of a null. You are fighting physics here.

I'm not sure if you were replying to me or happened to post just afterwards. True nulls can't be boosted but also true nulls don't exist in real rooms so this isn't the most relevant point. See my post above yours if you didn't before.

 

[Ultrasonic]

My experience with this app has been uniformly poor, and all that poor experience can be traced to the way in which it takes measurements--and usually centered in the transition band (Schroeder frequency in-room).

You need to pay attention to the direct arrivals (i.e., minimum phase) and not the interference of the direct arrivals + primary reflections. That means that measurements taken at the listening position are mostly problematic.

Virtually no one is paying attention to this issue, I've found. Dirac demonstrably cannot separate the direct arrivals from the reflections (non-minimum phase) at the listening position.

Chris

I only use Dirac Live in the bass region for this reason. Below 200 Hz currently.