-
Welcome to ASR. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!
Subwoofer level question. When to boost subwoofer output for dips?
- Thread starter dasherzx
- Start date
Ultrasonic
Major Contributor
I've not tried to measure or calculate this any time recently as I believe it's not trivial and there is debate over applying bass correction into the transition frequency range anyway. Rather the range I am currently using is judged partly on the basis of how it sounds.
My room is not huge though at something like 4.6 m x 3.3 m x 2.3 m (going from memory as I type). Putting these numbers in the amroc calculator gives a Schroeder frequency of 155 Hz for what it's worth. The reverberation time in my room is probably lower than assumed due to acoustic treatment though, which I believe will change this.
amroc - THE Room Mode Calculator
The number one room mode calculator for rectangular rooms. Read and hear the tune of axial, tangential and oblique modes. Scientific sources included.
amcoustics.com
Your mains have a 4th-order (24 dB/oct) native acoustic slope at the bottom of their range (if they are ported) and your sub has a 1st-order electrical slope at the top of its range. Asymmetric slopes inherently have completely different phase trajectories. Trying to lock them in phase over a very wide bandwidth is impossible.
I would abandon the full range experiment. Implement a proper, symmetric crossover, crossing over as low as possible while remaining safely clear of the mains' port tuning frequency and excursion limits. Once the slopes match, use a combination of subwoofer time delay and allpass filters to cleanly lock the phase responses together.
Torin Krell
Active Member
- Joined
- Jun 28, 2025
- Messages
- 192
- Likes
- 254
Interesting thread! Even experts disagree on sub integration and overall setup configuration.
My approach (IMHO) is to first optimize speaker and sub location, room treatment (and if possible room choice) before depending on DSP like Dirac. Optimizing speaker and sub placement/phase is more effective than some may realize, however it does take time and patience to iterate through several small adjustments.
I also found that boosting my subs in overall level slightly higher allows me to smoothly attenuate peaks/resonances with PEQ rather than needing to boost dips.
My approach (IMHO) is to first optimize speaker and sub location, room treatment (and if possible room choice) before depending on DSP like Dirac. Optimizing speaker and sub placement/phase is more effective than some may realize, however it does take time and patience to iterate through several small adjustments.
I also found that boosting my subs in overall level slightly higher allows me to smoothly attenuate peaks/resonances with PEQ rather than needing to boost dips.
Chris A
Senior Member
The reason I ask is that the region at and just above the Schroeder frequency in my room is precisely that which Dirac cannot seem to handle correctly. So my my room, that means that 77 Hz --> ~300 Hz are the problem frequencies. Dirac tries to correct that band by over-attenuating--resulting in a very thin and lifeless sound.
Of course, Dirac apparently doesn't know or change its pro-forma corrections that my loudspeakers have directivity down to ~100 Hz...
Chris
Keith_W
Master Contributor
I too was writing a book. But I stopped writing and decided to release parts of it for free, like that REW book in my sig.
Anyway, to answer the OP's question - "can you boost subwoofer output", this is the answer: think of a null as a hole which is infinitely deep that looks like this:
As you approach the centre of the hole, the slope gets steeper and steeper. When you are at the centre of the hole, you will need an infinite amount of energy to fill an infinitely deep hole. But that's the rub, not all nulls are infinitely deep. Some of them are close to the centre, but not at the centre. Which means that you don't require infinite energy, merely a lot of energy.
The question is "how much energy do you need"? Well, that depends on how close you are to that hole, and how much headroom you have with your subwoofer. None of us can answer that for you. The only way to find out is to try. Make sure you pay attention to the test result. In particular, look for distortion, test for multiple locations, and look for ringing. You will need before and after measurements.
My understanding has been that most room correction works on cutting the signal, reducing peaks, rather than boosting from the start. Thus, the original signal is usually preserved and in negative trim. Once you start boosting trim levels past 0, the signal is now being boosted purely in the digital domain. The way it was explained to me is that at this point, DSP is "amplifying" the information and can be error-prone which translates to distortion.
I haven't ever seen anything that refutes the person who said this to me. If erroneous, I would gladly welcome and appreciate correction.
Ultrasonic
Major Contributor
Yes, this is exactly what I've done.
Whilst this idea is appealing (and I did this myself years ago for this reason) I'd say it's not generally a good idea for two reasons.
- Whether you increase sub level and attenuate, or leave it where it is and boost more, if the resultant frequency response is the same then the subwoofer is outputting exactly the same thing. So it's not somehow being pushed less hard to reduce the dip in the frequency response if that makes any sense?
- To optimise the transition from main speakers to subwoofer you want them to be level matched at the crossover frequency. That way above the crossover frequency the main speakers will dominate and below the crossover he subwoofer will. If instead the subwoofer is at a significantly higher level at the crossover frequency, it will still dominate for a range of frequencies above the crossover frequency and you're more likely to run into problems with the sub being localisable. And just the whole transition being less seamless.
Ultrasonic
Major Contributor
How many measurements points are you using and how spread out are they? Too few points or them being too close together could lead to worse performance for what you describe.
Ultimately everyone should experiment with Dirac Live and use whatever they enjoy the most (assuming listening pleasure is the objective), so if you've figure this out, just stick with it.
Indeed it appears we hit "post" likely within seconds of each other.
I do agree with you for the most part, though I don't see how boosting a signal by any amount can help fix a 15-20dB suck-out, for example. You do say that some boost is acceptable, and this I agree with. Yet it still is unlikely that you can overcome every such instance by brute force alone. For myself, this would speak to positional EQ and seeing if better gains can be had by adjusting placement of Speakers and LP rather than tweaking room correction/PEQ settings.
To the image in OP, the -5db dip is pretty minor and may well not even require further adjustment other than to satisfy a perfectionist trait.
Ultrasonic
Major Contributor
My understanding has been that most room correction works on cutting the signal, reducing peaks, rather than boosting from the start. Thus, the original signal is usually preserved and in negative trim. Once you start boosting trim levels past 0, the signal is now being boosted purely in the digital domain. The way it was explained to me is that at this point, DSP is "amplifying" the information and can be error-prone which translates to distortion.
I haven't ever seen anything that refutes the person who said this to me. If erroneous, I would gladly welcome and appreciate correction.
As per my post above, ultimately boosting or cutting to achieve the same frequency response makes no actual difference.
What I think you make be thinking of though is that there needs to be enough headroom in the digital domain such that any boosts don't lead to digital clipping (which would sound awful!) So many DSP solutions will initially reduce signal level to create the required headroom to then be able to apply boosts. Having said this I'll note that the miniDSP products that I'm most familiar with don't do this. Well, I don't know about the latest Tide model I suppose but the others don't anyway.
I don't think there's inherently anything "error prone" to amplifying signals in the digital domain that is relevant here though.
Last edited:
Well on this forum we have few official reviews of the room correction systems, and they seem overall positive. They are mostly focused on FQ response though. Tearing apart the room correction systems with more or all REW screens happens occasionally and at least my take on it is there is no free lunch - it will be good and bad and how to prioritize is difficult to tell upfront. I just use my ear is such cases.
We don't have reviews of pre-amps, AVRs or AVPs that would compare the raw 2 channel specs to what happens if the room EQ is engaged. There is speculation that with room EQ engaged, SINAD would take a hit. There are also speculations as to various reasons why this is to be expected. I have asked @amirm to make an exception and test Denon 3900H which might poised to be a best selling AVR on this basis. It is a difficult and time consuming test, but at that point we would know what is at least the SINAD price to pay for the room EQ. This could be different with different systems, and not sure if Amir will have the ART unit.
We don't have reviews of pre-amps, AVRs or AVPs that would compare the raw 2 channel specs to what happens if the room EQ is engaged. There is speculation that with room EQ engaged, SINAD would take a hit. There are also speculations as to various reasons why this is to be expected. I have asked @amirm to make an exception and test Denon 3900H which might poised to be a best selling AVR on this basis. It is a difficult and time consuming test, but at that point we would know what is at least the SINAD price to pay for the room EQ. This could be different with different systems, and not sure if Amir will have the ART unit.
Ultrasonic
Major Contributor
I wouldn't suggest anyone tries to 'fill in' a 20 dB dip in the frequency response with EQ but my point above is that it's not that EQ can't. Apply a 5 dB boost to this hypothetical 20 dB dip and it will become a 15 dB dip.
Anyone with a measurement microphone can easily demonstrate this for themselves even without applying any EQ. Measure a frequency sweep and one output level and then another at a level 5 dB louder (or whatever) and they'll see bass response dips will increase by the same amount. This is exactly the same as applying a boost at the dip frequency. Occasionally there will be near perfect nulls where this won't be seen but this is very much the exception rather than the rule in my experience. And will be more so for anyone who is using effective bass traps.
Keith_W
Master Contributor
My understanding has been that most room correction works on cutting the signal, reducing peaks, rather than boosting from the start. Thus, the original signal is usually preserved and in negative trim. Once you start boosting trim levels past 0, the signal is now being boosted purely in the digital domain. The way it was explained to me is that at this point, DSP is "amplifying" the information and can be error-prone which translates to distortion.
I haven't ever seen anything that refutes the person who said this to me. If erroneous, I would gladly welcome and appreciate correction.
I have created a simulation for you.
First, here is a set of filters. A low-pass filter for the sub (red) and a high-pass filter for the main speakers (green). Both are minimum-phase 4th order Linkwitz-Riley with an XO point of 80Hz. Pay attention to the vertical scale on the left. You will see that the filters are normalized to 0dB. This is what will be sent to the DAC.
Now we add a PEQ. This one is at 50Hz, Q of 2.8, and gain of 20dB to simulate boosting a dip at 50Hz by 20dB.
After we convolve the PEQ to the LPF, we normalize the gain to 0dB (since you can't send >0dB to the DAC). Look at what happens to the rest of the spectrum - it has dropped by 20dB to accommodate that single +20dB PEQ.
Now you load the filters and start listening. The music has dropped by 20dB. To compensate, you turn up the volume knob by 20dB. You have now applied a 20dB boost at 50Hz.
The question is: can you afford to do this? The answer is: it depends. Maybe you have a system that is already struggling with volume and high distortion at the low end (e.g. speakers that are too small for your living space and listening distance). Maybe you have giant speakers with a tonne of amplifier power in a small room.
But that's not the end of the story. Check out how much ringing you get in the LPF with the PEQ (blue) compared to the straight LPF (red)!
Ultrasonic
Major Contributor
My response to this is always that the distortion to the bass response introduced by room modes is HUGE in comparison to the sort of potential SINAD changes you're alluding too. Quite how much will depend on the room but we're talking many orders of magnitude different here.
Edit: you'd also need to somehow modify the SINAD definition so as to excluded the intended/desired EQ as being 'distortion'.
Last edited:
My comment was to overall EQ distortion - if we see a number will be 1Khz or so.
You don't need distortion from the room - even for great subs distortion (like THD or SINAD) towards the low end of their response is huge. But apparently as you approach 20-30hz band that is acceptable as we can tolerate such numbers in this band.
If you are talking about the impact of the room to performance of the subs, that is a bit different, and even more serious problem. But many tools to deal with that.
Chris A
Senior Member
This is related to gain structure. Each implementation on each preamp or external box must be optimized based on total gain available and noise floor considerations (i.e., hardware-related limitations). If a preamp (usually AVPs and AVRs) or post-preamp correction box (for instance, like miniDSP) has limited gain relative to noise floor, then this can be an issue--for instance the original miniDSP "2x4", which apparently has total gain limitations/problems in the DSP device.
Other miniDSP boxes (i.e., those that have been used by the guys that I've helped dial-in their systems remotely) all seem to have noise floor or total gain problems, especially the ones where they have elected to load Dirac firmware--which effectively replaces the ability of the end user to do their own PEQs, i.e., you can either have Dirac or you can have adjustable PEQs, but not both. For guys having loudspeakers that are in the 105-110 dB/2.83v sensitivity range per channel, I don't recommend any miniDSP products except perhaps the "2x4 HD"...without Dirac.
As far as the commercial-grade DSP crossover boxes (outside of the lowest priced miniDSPs, DriveRack, Behringer or automotive-grade DSP boxes), I have not had any issues with total gain availability and/or noise floor. It's like the commercial-grade boxes are designed to handle a wider dynamic range without distortion or noise.
[Note: I'm assuming higher efficiency loudspeakers above, not those having less than ~95 dB/2.83v-m). For those loudspeakers having much lower efficiencies, you've always got the transient thermal and "Xmax issues" associated with the loudspeaker, as well as getting enough good clean electrical power to cover all the PEQs and source material dynamic range. These loudspeaker types always have 20+ dB higher modulation distortion in addition to the problems just stated. It's a "runaway problem".]
Chris
Last edited:
Ultrasonic
Major Contributor
No, I wasn't talking about distortion on bass output of subwoofers or loudspeakers. Rather considering the peaks and troughs in the bass response caused by room modes to be distortion compared to the smooth bass response we'd want. This is clearly audible, and much more so than let's say the difference between 1 kHz SINAD of 80 dB vs 90 dB (totally hypothetical numbers but you get the idea).
Chris A
Senior Member
And I very much thank you for that, since my book starts on measurement after inserting your book into the learning process beforehand. That saved me a ton of time and work. I'm sure that effort was very time consuming for you to put that together.
If you had instead charged for that book, I still would have referenced it before the reader started on the material that I've provided (i.e., in the acoustic measurements section of the book).
My problem is that the book I outlined was getting way too large to cover the subjects in the sort of depths that I believe need to be covered. So I de-scoped my outline to keep the book under ~500 pages total. It's a real challenge to give enough detail to do it, but still have something that can be easily carried around in paperback format. (I think that Toole has the same issue keeping his book editions under a certain page limit.) My plan was to wait to see if there were questions coming in--in each section of the book, then plan to write another book to fill out those areas in greater detail.
It's amazing to me how many pages can go by just to get the text to be tractable by non-engineering users: it requires a lot of explanation and examples to get it to the point where the average hi-fi enthusiast (i.e., those without engineering background) can use it to achieve hi-fi results.
Chris
Last edited:
Ultrasonic
Major Contributor
The part in bold is not true for every miniDSP I can think of. The ability to apply FIR filters is lost on units that run Dirac Live, but not PEQ. I use both PEQ and Dirac Live on my miniDSP SHD. The PEQ I use for high-frequency adjustments based on time-gated measurements, which compliment using Dirac Live in the bass region.
I did though run into the issue of the noise floor of my SHD being audible when pairing Meadowlark Shearwater loudspeakers and a Bel Canto Evo 4 Gen 2 power amp run at 29 dB gain. It's one reason I switched to using the SMSL DAC in my signature. With my current lower sensitivity ATC speakers and lower gain used on my Hypex Nilai power amps (22 dB) the noise floor of the SHD is no longer an issue though. So more an issue of my previous gain structure really.
Similar threads
