Larry Chanin
Stunt Coordinator
- Joined
- Jul 24, 2001
- Messages
- 218
This is a continuation of a fairly long thread started here:
Averaging Listening Positions when doing multi-channel equalization.
While there’s a lot of excellent discussion in the prior thread, it is seven pages long. So for those forum members who would prefer the “Reader’s Digest” version of the previous thread, I’ll start off with a Thread Recap before I continue reporting my results.
THREAD RECAP (Part I)
In anticipation of using an AudioControl Bijou multi-channel equalizer for the first time, I started to measure the characteristics of my room using Acoustisoft’s ETF5 software as a real-time acoustic analyzer. The original question of the thread was “Should I average the frequency response measurements from multiple room positions when performing a multi-channel room equalization?” This question was resolved fairly early. Despite what Mr. Lucas and the THX folks advise, most forum members seemed to think that special attention should be given to actual primary seating positions, not averaged results.
However, my first room measurements with ETF5 showed an impossibly large 55 dB drop off in frequency response going from a low frequency peak to a mid-range dip. It was this unusual result that turned out to be the main subject of the thread, and it really challenged everyone’s deductive reasoning skills as we tried to figure out what I did wrong.
A number of discoveries were made, each partially explaining the unusual room measurement.
The first most obvious problem “discovered” was I had located a 75 gallon fish tank near one of my front speakers. Duh!
The second discovery was a suspicious peak at exactly 60Hz where no room mode peak was predicted based on room dimensions. There were two main theories regarding the 60Hz peak:
1)acoustic noise, and
2)a sound card problem
It turned out that I had an intermittent bad connection to the Radio Shack sound pressure level meter, which accounted for a lot of the 55dB dip. In addition there was fish tank filter noise interfering with the measurements.
Once this was corrected I got frequency response curves that only had a ~13 to 18dB drop off in levels. Some members took the position that these measurement were reasonable, while others thought that it was still unreasonable. Part of the uncertainty in results was due to using the Radio Shack sound pressure level meter. Even with correction data there was a lot of doubt regarding its accuracy. Pete Mazz posted an excellent graph that showed that a tweaked Radio Shack meter with correction file was quite accurate up to about 10kHz.
The third discovery was that by opening various combinations of doors to my home theater I could actually improve the measured frequency response of the room. (In the interests of measurement consistency and to cut down on outside noise, I later decided to do all my measurements with the doors closed even though this didn’t necessarily yield the best room response.)
The fourth discovery was I had incorrectly used my surround processor's default DSP mode (Logic 7) and this was producing unusually attenuated results. It was agreed that depending on which speakers you are trying to send test signals to, there was a preference to which modes to use . In order of preference:
1.Bypass (With rewiring this mode is possible, but not practical for my Lexicon processor.)
2.Mono
3.Two-channel
4.Dolby Pro Logic
NOTE:I subsequently found that since I was trying to equalize all eight channels of my home theater, my Lexicon processor’s “Party” mode was the best for me because it sent unprocessed sounds to all speakers. This made it much easier for me to send test signals to all the speakers, particularly the surround back channels, without having to redo the wiring.
Although my measurements were starting to look better, there was still some reasonable doubt regarding the accuracy of the ETF5 results which still dropped off a fair amount from a low frequency peak. In trying to resolve the issue I used another, much simpler computer application, Frequency Plotter, that generates test tones and automatically records and plots the Radio Shack sound pressure level meter results. When replotted on an Excel spreadsheet with a logarithmic scale the results seemed to resemble the ETF5 log curve, but there were still some differences.
An other approach suggested was to buy a Behringer Ultra-Curve Pro Professional Equalizer and RTA. The object would be to use the Behringer to measure all channel’s frequency response and forget about using ETF5. I went ahead and bought the Ultra-Curve reasoning that in addition to providing an RTA, it could also be used to equalize the remaining two channels of my home theater that were not handled by the AudioControl equalizer.
As in the case of the Frequency Plotter comparison, I tried plotting the Behringer’s RTA results on an Excel spreadsheet to compare it with the ETF5 results. Again the results seemed to resemble the ETF5 curve, but there were some differences. Both of these comparisons were inconclusive because, I was still using the Radio Shack meter when using ETF5, and because I was not performing both pairs of measurements at the same sound pressure level.
The last discovery that I made was that the unique design of my main speakers was probably contributing to the usual frequency response. Mr. Matthew Polk, the co-founder of Polk Audio and the designer of the speakers informed me that these vintage Stereo Dimensional Array (SDA) type speakers required a different testing procedure. This was partially confirmed by the fact conventional speakers, such as my center channel speaker, did not exhibit the large dip in frequency response.
The fact that all three measurements (ETF5, Frequency Plotter, and the Behringer RTA) all showed a dip in frequency response for the SDA speakers, suggested to me that despite differences in measured results, perhaps there wasn’t anything wrong with ETF5. However, before I could investigate this further, my home theater computer that I was running ETF5 on became unstable, leaving me “dead in the water”, and that’s where the previous thread ended.
Part II
So, it’s now about three weeks since my computer difficulties. After reformatting my hard drive several times in unsuccessful attempts to solve my problem, it turned out that the problem was resolved by reseating my memory stick. Of course I had to reinstall and reconfigure all my software, including ETF5, not a lot of fun!
After getting up an running again I decided to give ETF5 an other chance. So to eliminate uncertainties regarding the Radio Shack sound pressure level meter I decided to buy a microphone preamplifier and use the measurement microphone that I used with the Behringer Ultra-Curve. (I should point out that I was beginning to suspect that the intermittent bad connection to the meter wasn’t due to bad cables, but rather the input connection to the Radio Shack meter itself.) It is not possible to adjust the Behringer Equalizer settings while in the RTA mode. On the other hand I can run ETF5 and the Ultra-Curve in equalizer mode simultaneously, adjusting equalizer settings and taking RTA measurements in almost real-time.
So I bought the EURORACK® MX602A 6-CHANNEL 2-BUS MIXING CONSOLE to use as a microphone preamplifier and began equalizing all eight channels of my home theater using both the Ultra-Curve, the AudioControl Bijou equalizers in conjunction with ETF5 as the RTA.
I decided to use the Ultra-Curve to equalize the main channels because it had more graphic equalizer bands as well as three parametric equalizer bands. This freed up the main channels on the AudioControl Bijou equalizer to be used to equalize the surround back channels.
The equalizer procedure recommended by Matthew Polk is logistically rather difficult, so on the main speakers I decided to start out doing serious equalizing at the low frequencies, (I used the parametric equalizer to cut certain low frequencies) but in general used only very modest equalization on the mid and high frequencies.
Here’s my first series of results measured from position #4. Listening Positions
Main Left Speaker
Main Right Speaker
Center Channel Speaker
Left Side Surround Speaker with Surround Subwoofer
Right Side Surround Speaker with Surround Subwoofer
Left Surround Back Speaker
Right Surround Back Speaker
LFE Subwoofer
The red curves are before equalization and the blue curves are after equalization.
I haven't done a lot of critical listening yet, but it definately sounds good to my untrained ears. My wife, who was working in an adjacent room, made an unsolicited comment that the Everly Brothers CD I was listening to sounded as if it were a live concert.
I'll continue to tweak the equalizer settings.
I'd like to thank everyone who provided advice and contributed to the prior thread. In case you're interested I have updated the photos on my homepage to show the equalizers and associated gear.
Thanks.
Larry
Averaging Listening Positions when doing multi-channel equalization.
While there’s a lot of excellent discussion in the prior thread, it is seven pages long. So for those forum members who would prefer the “Reader’s Digest” version of the previous thread, I’ll start off with a Thread Recap before I continue reporting my results.
THREAD RECAP (Part I)
In anticipation of using an AudioControl Bijou multi-channel equalizer for the first time, I started to measure the characteristics of my room using Acoustisoft’s ETF5 software as a real-time acoustic analyzer. The original question of the thread was “Should I average the frequency response measurements from multiple room positions when performing a multi-channel room equalization?” This question was resolved fairly early. Despite what Mr. Lucas and the THX folks advise, most forum members seemed to think that special attention should be given to actual primary seating positions, not averaged results.
However, my first room measurements with ETF5 showed an impossibly large 55 dB drop off in frequency response going from a low frequency peak to a mid-range dip. It was this unusual result that turned out to be the main subject of the thread, and it really challenged everyone’s deductive reasoning skills as we tried to figure out what I did wrong.
A number of discoveries were made, each partially explaining the unusual room measurement.
The first most obvious problem “discovered” was I had located a 75 gallon fish tank near one of my front speakers. Duh!
The second discovery was a suspicious peak at exactly 60Hz where no room mode peak was predicted based on room dimensions. There were two main theories regarding the 60Hz peak:
1)acoustic noise, and
2)a sound card problem
It turned out that I had an intermittent bad connection to the Radio Shack sound pressure level meter, which accounted for a lot of the 55dB dip. In addition there was fish tank filter noise interfering with the measurements.
Once this was corrected I got frequency response curves that only had a ~13 to 18dB drop off in levels. Some members took the position that these measurement were reasonable, while others thought that it was still unreasonable. Part of the uncertainty in results was due to using the Radio Shack sound pressure level meter. Even with correction data there was a lot of doubt regarding its accuracy. Pete Mazz posted an excellent graph that showed that a tweaked Radio Shack meter with correction file was quite accurate up to about 10kHz.
The third discovery was that by opening various combinations of doors to my home theater I could actually improve the measured frequency response of the room. (In the interests of measurement consistency and to cut down on outside noise, I later decided to do all my measurements with the doors closed even though this didn’t necessarily yield the best room response.)
The fourth discovery was I had incorrectly used my surround processor's default DSP mode (Logic 7) and this was producing unusually attenuated results. It was agreed that depending on which speakers you are trying to send test signals to, there was a preference to which modes to use . In order of preference:
1.Bypass (With rewiring this mode is possible, but not practical for my Lexicon processor.)
2.Mono
3.Two-channel
4.Dolby Pro Logic
NOTE:I subsequently found that since I was trying to equalize all eight channels of my home theater, my Lexicon processor’s “Party” mode was the best for me because it sent unprocessed sounds to all speakers. This made it much easier for me to send test signals to all the speakers, particularly the surround back channels, without having to redo the wiring.
Although my measurements were starting to look better, there was still some reasonable doubt regarding the accuracy of the ETF5 results which still dropped off a fair amount from a low frequency peak. In trying to resolve the issue I used another, much simpler computer application, Frequency Plotter, that generates test tones and automatically records and plots the Radio Shack sound pressure level meter results. When replotted on an Excel spreadsheet with a logarithmic scale the results seemed to resemble the ETF5 log curve, but there were still some differences.
An other approach suggested was to buy a Behringer Ultra-Curve Pro Professional Equalizer and RTA. The object would be to use the Behringer to measure all channel’s frequency response and forget about using ETF5. I went ahead and bought the Ultra-Curve reasoning that in addition to providing an RTA, it could also be used to equalize the remaining two channels of my home theater that were not handled by the AudioControl equalizer.
As in the case of the Frequency Plotter comparison, I tried plotting the Behringer’s RTA results on an Excel spreadsheet to compare it with the ETF5 results. Again the results seemed to resemble the ETF5 curve, but there were some differences. Both of these comparisons were inconclusive because, I was still using the Radio Shack meter when using ETF5, and because I was not performing both pairs of measurements at the same sound pressure level.
The last discovery that I made was that the unique design of my main speakers was probably contributing to the usual frequency response. Mr. Matthew Polk, the co-founder of Polk Audio and the designer of the speakers informed me that these vintage Stereo Dimensional Array (SDA) type speakers required a different testing procedure. This was partially confirmed by the fact conventional speakers, such as my center channel speaker, did not exhibit the large dip in frequency response.
The fact that all three measurements (ETF5, Frequency Plotter, and the Behringer RTA) all showed a dip in frequency response for the SDA speakers, suggested to me that despite differences in measured results, perhaps there wasn’t anything wrong with ETF5. However, before I could investigate this further, my home theater computer that I was running ETF5 on became unstable, leaving me “dead in the water”, and that’s where the previous thread ended.
Part II
So, it’s now about three weeks since my computer difficulties. After reformatting my hard drive several times in unsuccessful attempts to solve my problem, it turned out that the problem was resolved by reseating my memory stick. Of course I had to reinstall and reconfigure all my software, including ETF5, not a lot of fun!
After getting up an running again I decided to give ETF5 an other chance. So to eliminate uncertainties regarding the Radio Shack sound pressure level meter I decided to buy a microphone preamplifier and use the measurement microphone that I used with the Behringer Ultra-Curve. (I should point out that I was beginning to suspect that the intermittent bad connection to the meter wasn’t due to bad cables, but rather the input connection to the Radio Shack meter itself.) It is not possible to adjust the Behringer Equalizer settings while in the RTA mode. On the other hand I can run ETF5 and the Ultra-Curve in equalizer mode simultaneously, adjusting equalizer settings and taking RTA measurements in almost real-time.
So I bought the EURORACK® MX602A 6-CHANNEL 2-BUS MIXING CONSOLE to use as a microphone preamplifier and began equalizing all eight channels of my home theater using both the Ultra-Curve, the AudioControl Bijou equalizers in conjunction with ETF5 as the RTA.
I decided to use the Ultra-Curve to equalize the main channels because it had more graphic equalizer bands as well as three parametric equalizer bands. This freed up the main channels on the AudioControl Bijou equalizer to be used to equalize the surround back channels.
The equalizer procedure recommended by Matthew Polk is logistically rather difficult, so on the main speakers I decided to start out doing serious equalizing at the low frequencies, (I used the parametric equalizer to cut certain low frequencies) but in general used only very modest equalization on the mid and high frequencies.
Here’s my first series of results measured from position #4. Listening Positions
Main Left Speaker
Main Right Speaker
Center Channel Speaker
Left Side Surround Speaker with Surround Subwoofer
Right Side Surround Speaker with Surround Subwoofer
Left Surround Back Speaker
Right Surround Back Speaker
LFE Subwoofer
The red curves are before equalization and the blue curves are after equalization.
I haven't done a lot of critical listening yet, but it definately sounds good to my untrained ears. My wife, who was working in an adjacent room, made an unsolicited comment that the Everly Brothers CD I was listening to sounded as if it were a live concert.
I'll continue to tweak the equalizer settings.
I'd like to thank everyone who provided advice and contributed to the prior thread. In case you're interested I have updated the photos on my homepage to show the equalizers and associated gear.
Thanks.
Larry