[Spridle]

First let me say this is a great forum and the REW software is great.

Attached is my initial test using REW. 1/3 octave smoothing, RS Digital SPL meter, calibration files used, etc. Measurements are average of 3 positions across the couch in the main listening positions. When I set to 1/1 smoothing, graph results are almost identical (up to about 8kHz, then it rolls off a little) to my AudioControl RTA data of the same test environment, so I'm fairly confident on the calibration.

I have the SMS-1 parametric engaged and tone controls set to -1.5db on the treble. I have an Anthem D1 processor, ATI 3005 amps, and California Audio Technology 8.7 mains with s4 12" subs.

A picture of my room is also attached. Note the built-in nature does not allow any moving of speakers and my wife will not allow any treatments other that what you see (rugs, couches, irregular surfaces, etc.).

The dip you see at 50Hz is on purpose. I have a room mode at that frequency that really sticks out when you stand in certain places in the room so I attenuated it at the expense of the main listening position.

My main question is advice on adding parametric EQ (Rane PE 17s) on the mains. I would attenuate at about 245Hz (4.2Q) a collection of room modes, attenuate at about 780Hz (4.2Q) and boost at 1250Hz (6.0Q) and 3350Hz (7.0Q). With these filters, REW predicts there to then be a very smooth response. I will also try changing the capacitors on the crossover of mids/tweeter to smooth the dip at 3350Hz, since I think this developed after I padded the tweeter 4dB. This would be in lieu of the EQ boost at 3350Hz.

The Rane specs (120 dynamic range, 98db S/N and .005 THD) exceed digital recording and reproduction capabilities, so I should not be able to hear any ill effects.

Please give me your comments on my results and plans. Thanks!

 

[brucek]

up to about 8kHz, then it rolls off a little

Most people agree that the RS SPL meter isn't accurate or consistent above 5KHz. If you want to measure full range, best to get a better meter such as the Galaxy CM-140 or use a microphone such as the Behringer ECM8000 (and preamp).

I see very little wrong with your response. My advice would be to do nothing.

Adding parametric equalization to the mains and using it at frequencies above 100Hz has its problems. Room treatments are most effective at the frequencies above 100Hz up to about 500Hz.

The range that filter corrections are effective is much larger at low frequencies. At frequencies above subwoofer range, the effective listening area improvement of a corrective filter is quite tiny (smaller as the frequency increases) and so is of little value. Narrow filters at these frequencies could easily make things worse.

Narrow peaks or dips would not be audible, nor would they respond to equalization over any useful listening area.

Outside the low frequency subwoofer frequency range, equalization is quite ineffective in creating suitable results beyond a very small listening sweet spot (as a result of shorter wavelengths). The EQ at higher than subwoofer frequencies is really only valid at the point where the response was measured and a small region around that point (don't move your head).

The only time I see EQ being effective at higher frequencies is when wide bandwidth filters are applied to raise a large area such as would be accomplished with a tone control.

Beautiful room......

brucek

 

[Spridle]

Thanks brucek. I should have mentioned that 90% of my listening is music of all sorts (country, jazz, hip-hop, rock, and classical) mostly Music Choice through digital Comcast Cable, iTunes through digital Airtunes connection or CDs. The other 10% is movies on DVD or cable On Demand. CDs and Airtunes digital sound the best. So, my primary concerns are music.

Your comments on EQ on the mains seems to be the opinions of others as well. And, there are seemingly experienced others who say they EQ their mains with good results. When you say "narrow filters could easily make things worse," specifically what are you referring to here and what is considered "narrow." I've already ordered some used Ranes PE 17s so I will at least give it a try.

Can anyone comment who has EQed their mains? I'd like to here what you have to say. As I noted previously, the Ranes have very good specs and are analog.

One other question is the slope of my room response curve. I've read all sorts of things about this and house curves, but I'm totally confused. My room is about 33x22x10 or 7200sf and open to other rooms. What should I be looking for here.

Bottom line is that I want a sound that I like, which is impactful, realistic, clean, balanced and on the level of a live performance. Currently I get great sound sometimes with certain sources and volume levels, but generally I don't think it's quite there. Maybe it is source material, but sometimes the sound is thin lacking bass impact, sometimes it's muddy, and sometimes high female vocals and loud saxophone are screechy. Other times it sounds great. Don't quite know what is going on but I'm trying to find a good compromise, without additional room treatments.

One problem I know I have is that my speakers are not far enough apart. They are only 80" apart and my main seating position is about 144". Nothing I can do about it though.

If it helps, attached is LF waterfall plot.

Sorry for all the questions.

Thanks.

 

[brucek]

When you say "narrow filters could easily make things worse," specifically what are you referring to here and what is considered "narrow." I've already ordered some used Ranes PE 17s so I will at least give it a try.

Observe the peak at 700Hz or at 850Hz. You could easily apply a filter to those peaks and render them flat. That flat response will only be valid at that exact position where you measured because the bandwidth of those peaks is so narrow. A small positional change and those peaks may have been dips and so the filter you are applying to them will result in a greater dip. You're making things worse.

This is not the case with low frequencies where the dips and peaks are caused by room modes, and the wavelengths compared to the room dimensions are large enough that primary reflections from the walls, floor and ceilings of the room arrive at the listening position with a phase shift of much less than a cycle. These peaks respond very well to equalization, and over a larger listening area than higher frrequency peaks.

I know the Rane has good specs, but equalization at higher frequencies should be limited to broadly affecting large areas (like a tone control) and not for removing narrow peaks (i.e. the one at 700Hz or 850Hz in your plot).

I've read all sorts of things about this and house curves, but I'm totally confused.

In a nutshell, we hear poorly at low frequencies. A flat response may look nice on paper, but you don't hear 30Hz at the same level you hear 500Hz (and that response gets worse as the SPL level drops). Experience has shown that people like the bass boosted at lower frequencies. Different rooms require different boosts.

One problem I know I have is that my speakers are not far enough apart

If those are your speakers 'in the wall' in that picture, you couldn't have them in a worse spot. Mains speakers need room. They need space all around them. Several feet would be nice. I suspect that's not possible...

attached is LF waterfall plot.

Don't extend the horizontal axis past ~500Hz.

brucek

 

[terry j]

I 'eq' my mains, but it is a totally different procedure and with a totally different end result than what would be considered as 'eq'ing the bass'.

Admittedly I use a deqx, but we could be talking at cross purposes because of the phrase 'eq the mains'.

I agree totally with Bruce, although depending on the room and the listening results (you must audition the changes!) at times I will 'eq the room' up to 200, maybe at a stretch 300 hz but the last figure is pushing the boundary (for me at least).

The dips etc you see in the higher ranges (admittedly with an RS meter, so take on board Bruce's comments) are most definitely something you would not try to eq flat (or smooth, again terminology).

When I say I eq the mains, what i mean is that the native response of the speaker is made flat as it would be if measured in an anechoic chamber.

As an example, (and only because someone had linked to it elsewhere and it happened to be open for me to link to) see the native response of the speakers in the illustration http://www.soundstagenetwork.com/measurements/speakers/wilson_sophia_2/

I would set up the mic on axis (and gate out the lower frequencies if done in room) and then the speaker would be eq'd flat, which (to me at least) is what we want high fidelity speakers to do, reproduce all frequencies at an equal level.

However, once we place the speakers in the room, the room interactions take over and we would expect to see a graph along the lines of yours. It is totally acceptable to eq the bass flat at the LP, certainly to 100 hz as Bruce mentioned, but if you could imagine knowing you have a 'flat and perfect' pair of speakers what the result might be if you then started eq'ing the higher frequency dips and peaks, a recipe for disaster ha ha. To illustrate, lets assume your speakers are half way competent (and I'm NOT trying to suggest they are not!), what do you think the audible results would be of boosting the top end by 10 or 12 db (to account for room response) for example might be?

I'm not some sort of expert nor do i think I know much at all, but perhaps the explanation of how and what I do might clear up the terminology confusion of 'eq'ing the mains'?? Indeed, what other people mean by that term for all i know could be something totally different from what I mean, if you get my drift.

Not sure I would be too unhappy with a response like that!

 

[Wayne A. Pflughaupt]

Can anyone comment who has EQed their mains? I'd like to here what you have to say.

I have AudioControl 1/3-octave EQs in my system. The “it’s only good at the sweet spot” argument is the same one we hear from the guys who discount subwoofer equalization. I can’t say I’ve observed it to be an especially valid point, audibly speaking, for either the subs or mains. Sure, the equalization is typically optimized for a specific location, but there should be an audible improvement at other locations as well, even if not “perfect” at those locations. Just be sure to avoid ultra-narrow filters, as brucek recommended. In your situation, with the speakers stranded in the entertainment center, you’re undoubtedly getting some coloration that possibly can be at least partially addressed with EQ.

I agree with most of the recommendations of the other posters. You should be able to address room issues up to about 2-300 Hz, to some extent; you can’t use the same kind of drastic filtering above the subwoofer range that we use with subwoofers. Typically adjustments of a few dB are about all you can accomplish. One reason for this is because receivers are relatively low-powered (compared to good subwoofers) and run out of headroom in a hurry.

I would attenuate at about 245Hz (4.2Q) a collection of room modes, attenuate at about 780Hz (4.2Q) and boost at 1250Hz (6.0Q) and 3350Hz (7.0Q).

To repeat brucek’s recommendations, mains filtering should be more along the lines of broader tone control-type adjustments. Typically that means at least 1/3-octave (~ 4Q) or broader filtering. I’d be leery of the 6 and 7Q filters you’re considering (~ 1/4 - 1/5 octave). Up that high you’re getting close to notch-filter territory. Problems that narrow are typically not readily audible, especially when they’re dips in response, but feel free to experiment.

With these filters, REW predicts there to then be a very smooth response.

That’s a dangerous approach. Don’t get caught up trying to achieve a good-looking graph. The goal is an audible improvement. An audible improvement won’t necessary “look” the best.

For instance, the speakers I used to use, my RTA showed they had a spike at about 6 kHz, and it was audible. EQing that down to where it looked right on the RTA didn’t sound good. Adjusting it less delivered an audible improvement, even though it didn’t “look” correct on the RTA.

Looking at your graph, the areas I’d shoot for are the broad depressions between 4-500 Hz, and ~1200 Hz. (It looks to me like the latter requires a broader filter than the 6Q REW recommended. You might try shifting the frequency center upwards, to hit the center of the total width [~800 – 1800 Hz], rather than the deepest point.) Adjusting up a few dB in those areas might bring an audible improvement. Let your ears be your guide – if you don’t like what you hear, re-adjust the equalizer, or eliminate the filters entirely.

As I noted previously, the Ranes have very good specs and are analog.

Personally I’m concerned that the Ranes might not be up to the quality of your other equipment. I haven’t used the PE-17, but I have used its “first cousin,” the PE-15 in PA systems, and it introduced an audible high-frequency noise, even with cutting filters. When I was installing professional sound systems back in the early 90s, we had problems with Rane equalizers and their other electronics introducing hum into the system when they were situated close to things like amplifiers with large transformers. I’m not entirely positive, but I think the PE-17’s design dates back that far, or close to it. So be careful, if you get some undesirable artifacts, try putting it somewhere else in your rack.

But if nothing else, it should be able to let you see whether or not you can benefit from equalization, from an audibly-improved response perspective. If you get other undesirable artifacts, they can be alleviated by moving up to a higher-quality unit.

I've read all sorts of things about this and house curves, but I'm totally confused.

House curve: What it is, why you need it, how to do it

Maybe it is source material, but sometimes the sound is thin lacking bass impact, sometimes it's muddy, and sometimes high female vocals and loud saxophone are screechy. Other times it sounds great. Don't quite know what is going on but I'm trying to find a good compromise, without additional room treatments.

That’s a sad side effect of high performance systems: The better they get, the more they reveal any shortcomings that may be present in the source material.

Regards,
Wayne

 

[Spridle]

Thanks brucek, Terri and Wayne. Your comments are very helpful. My best course of action at this point will be to just try it and see how it sounds with EQ on the mains. I will be careful to use low Q, small cuts and focus only on a few areas. I think I will dial in and test with REW and then listen for about a week.

But the way Wayne, I researched the Rane PE 17 before I bought them and, as you note, I would expect them to be superior to the PE 15. Believe me, I could not tolerate anything that noticeably degrades the sound in any way. The specs on the 2 units are significantly different. For example, 98dB S/N on the PE 17 exceeds the theoretical maximum 96dB of 16-bit CDs. Here are the comparisons of the two units (maybe you could comment on these specs since you have experience in the recording industry):

PE 15
Freq Resp.: 15-33kHz +0/-3
THD+Noise: 0.02 ±.01
S/N Ratio Boosts/Cuts centered: 92±2dB at unity gain

PE 17
Dynamic Range: 120dB (20-bit equivilent)
Freq. Resp.: 20-20kHz ±0.5dB
THD+Noise: 0.005 ±0.001 @+4dBu, 20-20kHz, 80kHz BW
S/N Ratio Boosts/Cuts centered: 98±2dB at unity gain

 

[Wayne A. Pflughaupt]

I once (briefly) used a pro electronic crossover in my system that had a list price of close to $500 that had excellent specs, but it noticeably degraded and colored the sound.

Specs are nice, but they aren’t necessarily the be-all, end-all. For instance, specs won’t tell you if a component is prone to picking up noise from adjacent equipment in the rack. The cheap equalizers that the pro audio guys won’t even use as paperweights have decent specs on paper. I recently came across a user review of one complaining how noisy it was, even in bypass mode, despite its excellent published noise spec.

I trust Rane’s specs as being more reliable than the lower-end pro brands, and I don’t expect noise radiating in from adjacent gear to be a problem in a home installation. I’m just saying don’t perfunctorily expect a relatively inexpensive piece of pro gear to be as pristine as we’re used to seeing in home audio, as I did when I bought that crossover.

A good test you can perform when you get the Ranes, run pink noise through your system and switch the EQ in and out of bypass, with all single filters off (probably best to do this with a single channel, since you’re dealing with two equalizers). If you hear any timbre shift in the tone, then the equalizer is coloring the sound before any filters are added. That would be a deal breaker right there, IMO.

Another crude test, with no signal, turn the volume up high enough to hear the background hiss and switch the EQ in and out of bypass (again, with all filters switched off). This will let you know if the equalizer is as quiet as the rest of your gear. Of course, if any added noise isn’t audible at normal listening volume settings, then it’s not that big of a deal, but it’s good to know what you’re dealing with. Remember, any boosting filters you might use are going to add noise at the adjusted frequency, in the dB value of the boost, so it’s important for the EQ to have a quiet noise floor before equalizing. If the equalizer is quieter than your receiver, you can get away with some boosting and come away with no added noise to the system.

Good luck, and let us know how things turn out.

Regards,
Wayne

 

[Spridle]

OK, here are the initial results of inserting Rane PE 17 parametric EQs into my main channels.

First, I wanted to make sure that the Ranes did not add any noise or coloration. Here are the tests I ran:

1) Noise – Without the Ranes, I turned my system (600 watts bi-amped into each speaker) up all the way with no source input signal. There is no audible noise unless I put my ear a few inches from the tweeter and the noise is very faint. Next, I put the Ranes in the main lines and listened in bypass mode, then with all filters engaged and set to flat, and finally with the filters set to my intended positions. In all of these tests, the Ranes did not add any noticeable hum or additional noise.

2) Response – I used REW to measure the frequency response of the Ranes. In the flat setting with all filters on, the Ranes measures completely flat except for about a 0.25 db roll-off at the extreme ends. Then I played pink noise and switched back and forth to and from by-pass and noticed no change in sound or timber.

To create my filters, I used REW to create a filter file based on my prior room response measurements. Then I ran a sweep with REW through the Ranes and adjusted the frequency, level and bandwidth knobs until the response matched the virtual curves. This was important since the Ranes are analog and use knobs which can’t be adjusted with precision without using REW as a signal analyzer.

With the Ranes adjusted, I listened for a few days and was able to boost the sub a bit due and the sound is balanced. Sound is much improved.

Since I am just using the Radio Shack meter now, my next step is to remeasure with a Behringer ECM8000 (that came with my Velodyne SMS-1) and get a more accurate picture. Then I will tweak my lower end and experiment with house curves.

Any comments would be appreciated.

Here is the full-range response graph after using the Ranes and a small level boost to the subs.

 

[Wayne A. Pflughaupt]

Way to go, Spridle. :T Thoughtful and thorough analysis like yours can go along ways towards dispelling some of the myths floating around about how bad full-range equalization is.

Comparing your two graphs it looks like you took my advice and kept an ear out for audible improvement vs. picture-perfect results. I notice that the two areas I brought to your attention in the 4 - 500 Hz range and 1200 Hz are still visible in your second graph, but much less pronounced. Your experiment shows that proper equalizing technique with a quality product can have positive results.

Just curious, how many filters did you use, and at what settings? I see the notch you had at 3350 Hz is gone. Did you equalize it, or modify the crossover like you mentioned?

Regards,
Wayne

 

[Spridle]

I took the advice and went with broader filters with just enough gain/cut to get the overall smoothed response to ±3db. The four filters are:

250Hz -2.3dB 1.9Q
460Hz +2.5dB 2.3Q
1250Hz +3.5dB 2.3Q
3350Hz +3.1dB 4.0Q

I have not messed with the capacitors in the crossover yet.

I'm listening to some jazz right now and am amazed at how smooth, clear, detailed and balanced it is.

 

[Wayne A. Pflughaupt]

2.3dB; 1.9Q - how did you manage to get such precise setting indications from an analog EQ that gives bandwidth in octaves?

Taking a closer look at the Rane’s specs and comparing it to the equipment you have, it appears the Rane’s S/N ratio is actually on par with your other gear. Rane specs the PE17 S/N with an unweighted figure, which is technically superior to the A-weighted noise specs most home audio manufacturers use. As explained in Rane’s Audio Specifications paper, A-weighting can “improve” a component’s noise floor spec by up to 10 dB.

Regards,
Wayne

 

[terry j]

hi spridle

if you answer wayne, could you include your graphs again, but overlaid on each other?It is a lot of scrolling up and down to take in both.

 

[Spridle]

To get the precise filters out of the analog unit, I output the sweep tone from the headphone jack of my soundcard through the Rane and then from the output of the Rane directly back to the mic input on my soundcard. I loaded my REW filter file onto the screen for visual comparison. Using this method, I can dial in the exact filters. It's basically the same process you would use to calibrate a soundcard or mic preamp with REW. Hope this explains it.

 

[Wayne A. Pflughaupt]

Nice technique, Spridle. :T

Terry, here are the graphs.

Before EQ

After EQ​

Regards,
Wayne

 

[Spridle]

I will post some more graphs and data once I tweak a little more and measure with the Behringer ECM8000. Should be ready in a week or two.

 

[Rudy81]

I have just finshed my first run at this and have noticed the same high frequency roll off as your graphs show. Is there some reason the highs roll off like that? Some setting perhaps?

 

[brucek]

Is there some reason the highs roll off like that?

What mic/meter are you using?

brucek

 

[Rudy81]

I am using the ECM8000. Just got it yesterday. I noticed all my readings have a high frequency roll off above 5000Hz. I see many other plots posted on the forum with the same response.

I must be really stupid or something. I have read the help files many times, but I guess in my anticipation I overlooked the line that directs me to point the mic at the speaker, not at the ceiling, for full response testing. I will have to run everything again and see if that makes a difference.

BTW, this is a fantastic piece of software and this forum is really great. Thanks to all those that provided guidance and help.

Regards,
Rudy

 

[brucek]

I am using the ECM8000.

Hopefully you're using the calibration file with REW.

but I guess in my anticipation I overlooked the line that directs me to point the mic at the speaker, not at the ceiling, for full response testing.

Depends on whether you're measuring near-field or at your listening position. If you're specifically doing a near-field measure <12", then point at the driver(s), but normally the mic is pointed vertical with a 10-20 degree slant toward the front.

brucek