Analog filter for room correction

[Neutrino]

Hi,

I was wondering whether anyone here has ever designed a EQ/Filter that would be suitable for room correction, i.e. for insertion into the monitoring chain. The requirements for such an EQ would be:

- no/minimum sound coloration
- fully parametric bell filters with high Q-values up to about 16
- possibly  a parametric low-shelf filter

Any hints to existing projects that would be appreciated.

Best

 

[ruairioflaherty]

Hi,

I was wondering whether anyone here has ever designed a EQ/Filter that would be suitable for room correction, i.e. for insertion into the monitoring chain. The requirements for such an EQ would be:

- no/minimum sound coloration
- fully parametric bell filters with high Q-values up to about 16
- possibly  a parametric low-shelf filter

Any hints to existing projects that would be appreciated.

Best

What is the goal of such high Q correction in a monitor path?  It will have severe issues with ringing.

If I was searching for a DIY EQ to EQ a monitor system the Barry Porter would be my first choice.

 

[gridcurrent]

I was wondering whether anyone here has ever designed a EQ/Filter that would be suitable for room correction,

in the last century.
read up on Boner's work.
the Altec filters were designed by Art Davis.

 

[Neutrino]

Thank you for the replies. I'll read up on your suggestions

What is the goal of such high Q correction in a monitor path?  It will have severe issues with ringing.

Room resonances are usually very narrow and require high Q-values. The standard approach is to use DSPs in the monitor path but that mostly requires an additional AD/DA conversion which I would like to avoid.  Hence my question.
Moreover, this only makes sense in a treated room, where the adjustments that have to be made are not more than a few dB and preferably attenuation. The ringing issue would also appear in a digital filter, wouldn't it?

 

[abbey road d enfer]

Hi,

I was wondering whether anyone here has ever designed a EQ/Filter that would be suitable for room correction, i.e. for insertion into the monitoring chain.

In the 1980's I designed a product marketed as the RE209 (Room Equalizer 2x9 bands), consisting in 9 fully parametric EQ sections (SVF filters) cascaded. It actually worked quite well, but operating it was very difficult because there were no FFT analysers available as today. Only a few acousticians were actually capable of owning and operating the necessary measurement equipment. Today, the measurement tools are available to anybody with a PC, a souncard and a measurement mic and the EQ itself can be done in the PC, in a much more convenient and repeatable manner..

The requirements for such an EQ would be:

- no/minimum sound coloration

"coloration" is due to two factors: incorrect frequency response and distortion. Distortion was typically in the 0.01% area, so not an issue. Frequency response is all about proper measurement/alignment of the monitoring system. Indeed, improper alignment results in some kind of "coloration".

- fully parametric bell filters with high Q-values up to about 16

Mine wasn't "fully" parametric, in the sense that each band was tunable in a ratio of 1:10, so there was no 20Hz-20kHz there. Most of the bands were concentrated in the low-to-lo-mid ranges, so room modes could be processed. As you may see on the attached picture, there is no mention of "Q", only BW (1/6th to 2 octaves), because expressing the selectivity of an EQ as "Q" is a misconceprion.

- possibly  a parametric low-shelf filter

RE209 had tunable High and Low-pass filters, as well as tunable Baxandalls.

 

[abbey road d enfer]

What is the goal of such high Q correction in a monitor path?  It will have severe issues with ringing.

If it is used to fight an already existing ringing (resonance), it actually makes it less severe. Perfect tuning is paramount though. A "Q" of 16 seems to correspond to a BW of about 1/6th octave. It is quite frequent to see room EQ filters with that sort of sharpness.

 

[Neutrino]

In the 1980's I designed a product marketed as the RE209 (Room Equalizer 2x9 bands), consisting in 9 fully parametric EQ sections (SVF filters) cascaded.

That sounds interesting. Is the schematic something that you could make available or do you still use it commercially?

I'm too inexperienced with circuit design to put together the bits of information myself.  But having a point to start with, I might be able to adapt it to my needs (maybe with some help here).

Best,

 

[abbey road d enfer]

That sounds interesting. Is the schematic something that you could make available or do you still use it commercially?

I'm too inexperienced with circuit design to put together the bits of information myself.  But having a point to start with, I might be able to adapt it to my needs (maybe with some help here).

Best,

Schemo is in Technical Documents under SCV Electronics.

 

[ruairioflaherty]

We've talked about this before Abbey and I believe we may have to agree to disagree.  My experience comes from the last 6 years of tuning rooms/speaker systems for many high profile clients here in L.A. (and 16 years before that of trying to make studios sound as good as possible). 

I've tried  analog EQs (everything from Whites to the Meyer and even high end Maselec), DSP based solutions (from low end to the nicest Lake/Hypex systems) and computer based solutions.  I've worked several times with a "master" in the field of room tuning who came in with an EQ in one hand and a hefty invoice in the other.

All of my experience tells me that high Q corrections do not work well, or at least do not make improvements in the kinds of rooms you'd actually make a record in.  They can be part of converting a bare office into something "better".

I spent several months a few years back really digging in and understanding a DSP room correction system called Dirac in my own mastering room, on the recommendation of a friend whose ears I trust.  No matter what I tried the system would introduce high Q corrections to "problems".  When playing a busy track of fiddle music I would hit the space bar and hear the filters ring on for almost a second, my old room was very controlled, this was all introduced by the corrections.

I truly wish there was a shortcut.  In the most recent room I built the bass trap is 14'x'14'x4' and almost 10' from the boundary, we had the luxury of space in that case and the results are stunning.  If i could have achieve that with an EQ I would have.  I did tweak the speaker response ever so slightly with a Maselec EQ, but not to correct low end room issues.

I'll finish by saying that you can indeed make FFTs and room sweeps look more correct with eq but in my experience hitting bypass often reveals that it was all for nothing.

For those still determined to EQ here's an article I found very helpful, for use with the free REW package.  It can help you understand why certain issues respond to EQ and others don't and how to determine which you have - https://www.roomeqwizard.com/help/help_en-GB/html/minimumphase.html 

 

[Neutrino]

Hi ruairioflaherty,

thanks for sharing your experience here. I can completely understand your opinion. I also tried several options including Dirac, which was a disappointment to me. Exactly for the reason you describe.  It made the response "more correct" in the measurement, but it just didn't sound right. I think because it was trying to correct too much. 

Right now, I have a DSP (minimum phase EQ) on my sub (but not on the satellites) and I must say that it works quite well. With small adjustments I don't get the feeling of unnaturalness (neither did I notice any ringing).  There is a real improvement in the sound. The disadvantage is that I have to use more EQ on the sub, because part of the low-end energy is also coming from the satellites which I don't want to high-pass. The bass frequency response in my room is better than +-5 dB in my room without EQ.

Yes, it would definitely be good to try this before spending a huge effort for nothing.

I read the article you posted a few months ago. Very useful information!

 

[abbey road d enfer]

All of my experience tells me that high Q corrections do not work well, or at least do not make improvements in the kinds of rooms you'd actually make a record in. 

You may have noted that I thoroughly avoid using "Q", and favoring BW. Is a 3dB notch at 1/6th octave a high Q? Tehnically, it is not, because "Q" does not apply to filters that don't go asymptotically to - infinity on both sides, but you may argue it's pedantic. More practically, a 3dB EQ involves a numerator and a denominator that more or less "moderate" each other, cancelling ringing. I would not advocate using more than 6dB boost or cut with such a narrow BW, even less at 1/12th octave.

 

[abbey road d enfer]

I think because it was trying to correct too much. 

That's a very common mistake. A large part of the art of tuning a monitoring system is to know how much of the imperfections to leave.

Right now, I have a DSP (minimum phase EQ) on my sub (but not on the satellites) and I must say that it works quite well.

Is it minimum-phase or linear-phase? There's a big difference here. Most analog EQ's are minimum-phase. Digital EQ's can be both. Being capable of producing linear-phase EQ is NOT a valid reason to use them. Room imperfection can be reduced (in the sense of reducing an equation or a polynomium) in MP terms and non-MP terms. EQ takes care of the former.Nobody can  fix the latter unless using impractical amounts of latency, with its cohort of artifacts (pre-echo).

The disadvantage is that I have to use more EQ on the sub, because part of the low-end energy is also coming from the satellites

I'm not sure I understand that. Most of the LF issues in rooms are due to room modes, which don't change with the number of sources. What changes with the number of sources is reflection issues, and it's been consistently demonstrated that reflection issues diminish when the number of sources increases; that is quite normal since frequency response alterations will tend to average.

...which I don't want to high-pass.

Are you sure about the pertinence of this choice? What slope and frequency is the x-over on the subs?

 

[Neutrino]

Is it minimum-phase or linear-phase? There's a big difference here. Most analog EQ's are minimum-phase. Digital EQ's can be both.

I'm familiar with the difference and I'm rather  sure my software is minimum phase only.:
https://www.alldsp.com/software.html

I'm not sure I understand that. Most of the LF issues in rooms are due to room modes, which don't change with the number of sources. What changes with the number of sources is reflection issues, and it's been consistently demonstrated that reflection issues diminish when the number of sources increases; that is quite normal since frequency response alterations will tend to average.

OK, that's what I observe in my measurements. The combined frequency response is better than that of the satellites alone. However,  the locations of the peaks and dips are the same. E.g. I have a peak at about 50 Hz which is a lot higher when I switch on the sub. If I'm only able to EQ the sub, I have to reduce that frequency by a larger amount than what I would have to do when I was EQing all sources simultaneously. Does this make sense?

Quote
...which I don't want to high-pass.
Are you sure about the pertinence of this choice?

My satellites are completely passive and they sound phenomenal as they are. I really like to hear them play full-range. Therefore, I'm reluctant to touch anything there.  Also, I don't even have a suitable filter to high-pass them for now. Do you think this would improve anything?

What slope and frequency is the x-over on the subs?

My sub is a dipole (i.e. one driver at the front one at the back, separate x-overs each). Filters are Linkwitz-Riley, 24dB:
Front: 21 - 90 Hz
Back: 21-59 Hz and 2 dB overall attenuation

 

[Neutrino]

Schemo is in Technical Documents under SCV Electronics.

Thanks for sharing!

 

[abbey road d enfer]

The combined frequency response is better than that of the satellites alone. However,  the locations of the peaks and dips are the same. E.g. I have a peak at about 50 Hz which is a lot higher when I switch on the sub. If I'm only able to EQ the sub, I have to reduce that frequency by a larger amount than what I would have to do when I was EQing all sources simultaneously. Does this make sense?

Yes, and no. If I understand correctly, you have an EQ on the sub and another on the satellites. If the EQ was on everything  you would need less EQ's.

My satellites are completely passive and they sound phenomenal as they are. I really like to hear them play full-range. Therefore, I'm reluctant to touch anything there.  Also, I don't even have a suitable filter to high-pass them for now. Do you think this would improve anything?

There is a problem with overlapping frequencies; in order to minimize problems, the phase difference should be less than 90° in the overlaps. With full-range satellites and two subs, there are areas where the phase difference is much larger, resulting in less efficient coupling between speakers and room. I would think in the 65-85 Hz range.

My sub is a dipole (i.e. one driver at the front one at the back, separate x-overs each). Filters are Linkwitz-Riley, 24dB:
Front: 21 - 90 Hz
Back: 21-59 Hz and 2 dB overall attenuation

  Is it a recommended setting for this particular sub?

 

[Neutrino]

Yes, and no. If I understand correctly, you have an EQ on the sub and another on the satellites. If the EQ was on everything  you would need less EQ's.

No, I only have EQ on the sub and none on the satellites. They are completely passive.

  Is it a recommended setting for this particular sub?

No, this is an outcome of a tuning process using  REW measurements to optimize the performance of the sub to my room. Note: both drivers are in the same box (front and back), but can be tuned separately. The front driver does the main work and the back driver can be used for a "controlled acoustic short circuit" to minimize the impact of some room modes. The back driver is phase-inverted w.r.t. to the front. There is also a delay on them that has also been tuned to my particular situation. I spent many, many hours with playing and trying to understand the impact of different setting and I think I'm close to the optimum of what can be achieved with this sub in my room.

 

[abbey road d enfer]

No, I only have EQ on the sub and none on the satellites. They are completely passive.

That means that, whatever EQ you apply to the sub(s), the satellites can trigger room modes. Let's say the satellites trigger a room mode at a given frequency, it will result in a response peak. In order to compensate, you may EQ this frequency out on the sub(s), but there may be a case where even completely kill (notch out) this frequency in the sub(s), there will still be a peak. And that would result in a variable distribution of sources, which in turn result in sonic aberrations. That may not be the case for you, but generally speaking, global EQ should be applied for dealing with room modes.

No, this is an outcome of a tuning process using  REW measurements to optimize the performance of the sub to my room. Note: both drivers are in the same box (front and back), but can be tuned separately. The front driver does the main work and the back driver can be used for a "controlled acoustic short circuit" to minimize the impact of some room modes. The back driver is phase-inverted w.r.t. to the front. There is also a delay on them that has also been tuned to my particular situation. I spent many, many hours with playing and trying to understand the impact of different setting and I think I'm close to the optimum of what can be achieved with this sub in my room.

OK; seems to me like a particular case of cardioid subwoofer. They are known to work well in free-field, but need extra TLC when used in small areas.

 

[Neutrino]

That means that, whatever EQ you apply to the sub(s), the satellites can trigger room modes. Let's say the satellites trigger a room mode at a given frequency, it will result in a response peak. In order to compensate, you may EQ this frequency out on the sub(s), but there may be a case where even completely kill (notch out) this frequency in the sub(s), there will still be a peak. And that would result in a variable distribution of sources, which in turn result in sonic aberrations. That may not be the case for you, but generally speaking, global EQ should be applied for dealing with room modes.

That's exactly what I meant.

Now as I think of it, it might even be advantageous to apply separate EQ to the satellites because you can take into account the location of the speakers in the room in the choice which frequencies to correct. E.g. in a typical small room the satellites are close to half-way between floor and ceiling, i.e. they hardly excite the first vertical mode. It's almost useless (or at least inefficient) to increase that frequency on the satellites, since you will force them to work harder, but the gain at the listening position is small. The sub on the other hand, being on the floor excites that room mode much stronger, therefore this is the place to correct it.

OK; seems to me like a particular case of cardioid subwoofer. They are known to work well in free-field, but need extra TLC when used in small areas.

I'm pretty happy with that one. It produces a very tight sound. The only constraint with this kind of subwoofer is that the listener shouldn't be located in the plane perpendicular to the main loudspeaker axis, because of the difference in phase between the two drivers (similar concept to fig-8 microphones). But that is not an issue in control rooms, since the sub is ususlly behind the desk.

 

[abbey road d enfer]

Now as I think of it, it might even be advantageous to apply separate EQ to the satellites because you can take into account the location of the speakers in the room in the choice which frequencies to correct. E.g. in a typical small room the satellites are close to half-way between floor and ceiling, i.e. they hardly excite the first vertical mode.

Room modes are independant on the location of the source; reflection modes are. Overall EQ for room modes, separate EQ's for reflection modes and speaker inaccuracies.

 

[Neutrino]

Room modes are independant on the location of the source;

I agree, but the extent to which the room mode is excited does depend on the location of the source: If the source is in the null of the mode, it doesn't excite it (or very litte).

What you call reflection modes, is it the same as SBIR (speaker boundary interference response)?

Coming back to the original topic: Would it be possible to use a completely passive circuit (for example notch filters) for room EQ? Of course, this would only allow for attenuation, but that's also what is needed in most of the cases. Is it possible to design a completely passive peak filter with variable Bandwidth, frequency and attenuation?