Well just been messing around with miniDSP today and learning about adujsting PEQ and its effects trying to get a reasonable flat/curve.
When in my mind so far it is absoultley pointless
WHY?
The measurement at the listening position goes tits up
green line first measurement 2.0m left speaker
black line both speakers @ listening posistion
blue line right speaker @ listening position
red line left speaker @ listening position
Now I expected a difference between L/R due to room boundaries.
But from hump at 15hz--100hz throws out any nearfield measurement thats worth doing
thoughts please
As this is DIY forum then the end result should suit the end listener
When in my mind so far it is absoultley pointless
WHY?
The measurement at the listening position goes tits up
green line first measurement 2.0m left speaker
black line both speakers @ listening posistion
blue line right speaker @ listening position
red line left speaker @ listening position
Now I expected a difference between L/R due to room boundaries.
But from hump at 15hz--100hz throws out any nearfield measurement thats worth doing
thoughts please
As this is DIY forum then the end result should suit the end listener
Last edited:
hi sorry I was under the impression 1m is nearfield industry standard
due to my set up 1m is not practical
1.4m high and lower cabinet 1m wide
even an 2mm nearfield as you say would be far worse at 3.6m listening position
measurements taken at 35" from floor level (ear height) slightly off 10 degrees off axis
Last edited:
I see. You can read the paper at this link to see the details about the near field and why you probably should not be calling your measurements (any of them) a "nearfield" measurement:
http://www.xlrtechs.com/dbkeele.com/PDF/Keele (1974-04 AES Published) - Nearfield Paper.pdf
A nearfield measurement is not meant to be a "system" measurement. It's performed on one driver VERY close to radiating surface (e.g. the dustcap or center of cone) and is valid for that driver only.
Your measurements are picking up reflections off of the back wall, floor, ceiling as well as the usual cabinet edge baffle diffraction. There is likely no good way to avoid this with your setup because the boundaries are very close (acoustically speaking) to the loudspeakers.
http://www.xlrtechs.com/dbkeele.com/PDF/Keele (1974-04 AES Published) - Nearfield Paper.pdf
A nearfield measurement is not meant to be a "system" measurement. It's performed on one driver VERY close to radiating surface (e.g. the dustcap or center of cone) and is valid for that driver only.
Your measurements are picking up reflections off of the back wall, floor, ceiling as well as the usual cabinet edge baffle diffraction. There is likely no good way to avoid this with your setup because the boundaries are very close (acoustically speaking) to the loudspeakers.
are the measurements not exactly what i would be hearing?
but can these not not be EQd out? not totaly but to a good degree shirley
this is no different to buying the JBL M2 that measures flat and plonking it in my room?
the measurements on that too would be inperfect at the listening posistion
cheers
Are the measurements exactly what you are hearing? Probably not, and it has a lot to do with how you make the measurement, how long you gather the time record, how you process that into a frequency response, etc. etc. etc. A microphone is NOT an "ear+brain" substitute. The brain does a lot of very interesting things with the stimuli it receives from the ears...
Understanding what is going on acoustically and psycoacoustically is a deep subject.
How you understand the measurements you make is also not a trivial subject either.
And yes, if you put the worlds "best loudspeaker" in the same position that your speakers are in now it would likely have some major flaws to the sound as well. There is no magic to acoustics, and no getting around the physics of it either. Generally, speakers placed against walls will sound bad because of peaks and dips introduced into the response by the sound reflected off the wall. Speakers placed in a corner couple most strongly to the room modes and the bass can sound way off, etc. One exception is a Klipschorn, which is designed for that location. You would get better sound with the monitors on stands moved out into the room and away from all walls by at least 1.5m. Subs can go in a corner, but you will likely need to EQ the low end a lot. That is likely what you are facing now. Try moving the subs around. One trick uses the principle of reciprocity: put the sub in your listening position (drivers where your head goes) and turn on some music. Move around the room, placing your head where the sub might/could be located. You will hear the bass as if you were sitting in your listening location and the sub was where your head is located. Once you find a good spot, move the subs there and return to your chair.
thankyou again for replying charlie, but moving the speakers and cabs around is not possible due to size and weight.
the sound i heard was horrible i like to think i can hear a half decent sound without looking at a graph
so here is to a bit of REW auto EQ not exactly majic but i think i am getting their.
measurement taken 3.6m away 10 degrees off axis @ listening position
the sound i heard was horrible i like to think i can hear a half decent sound without looking at a graph
so here is to a bit of REW auto EQ not exactly majic but i think i am getting their.
measurement taken 3.6m away 10 degrees off axis @ listening position
Charlie2, I think you should also use RTA measurement with very long cycle and move your mic around your head when measuring (MMM technique) RTA can not be use to set parameters for eq, but it is closer to what you hear.
Another thing, it is a big difference when measuring one speaker at time and stereo. With stereo there is always summing at low frequencies. Just try to make one speaker's response nice, don't measure at all both!
Don't try to equalize room modes too much! Means sharp bumps and dips 30Hz-300Hz. Our ear/brain system gets adapted to moderate room modes. Room acoustic treatment is the way to go.
Your speaker system is diy multiway. Have you measured each unit's response and matched crossover slopes and delays? Can you equalize each unit with dsp?
Another thing, it is a big difference when measuring one speaker at time and stereo. With stereo there is always summing at low frequencies. Just try to make one speaker's response nice, don't measure at all both!
Don't try to equalize room modes too much! Means sharp bumps and dips 30Hz-300Hz. Our ear/brain system gets adapted to moderate room modes. Room acoustic treatment is the way to go.
Your speaker system is diy multiway. Have you measured each unit's response and matched crossover slopes and delays? Can you equalize each unit with dsp?
+1 on the measure one speaker at a time. Try a measurement at 0.5m away and on axis with HF unit (the "butt cheeks" should be aligned top/bottom for wide horizontal directivity, as shown, it is wide in vertical direction). Then apply a gate in the "IR Windows" button and set to 1ms before and 4ms after to look at response unaffected by room reflections. You won't see bass measurements but that needs to be done separately with a "near field" measurement about 2 in away from cone. There are ways to "stitch" or combine the near field and far field (in this case 0.5m) together for a measurement that is indicative of what you can get in an anechoic room. Measurements at listening position will generally be full of dips and peaks due to room modes unless you have a sound treated room and take care not to put speakers close to nearby walls.
Don't give up - it takes some time and patience but getting good measurements at closer range and nearfield is very useful for designing XO's and for tuning general balance of speaker, independent of the room.
There are some good items here that may help blend nearfield and farfield frd files to be useful for developing XO's:
http://audio.claub.net/software/jbabgy/jbagby.html
Don't give up - it takes some time and patience but getting good measurements at closer range and nearfield is very useful for designing XO's and for tuning general balance of speaker, independent of the room.
There are some good items here that may help blend nearfield and farfield frd files to be useful for developing XO's:
http://audio.claub.net/software/jbabgy/jbagby.html
Might as well weigh in.
What I usually do when setting up a new set of speakers is this:
- measure each driver (or pair of drivers if they're playing the same signal), nearfield. Usually within 1 driver diameter
- use REW to average some on-and off-axis responses, and auto-EQ the results, with plenty of spare frequency range at each end (eg, compression driver running down to 1kHz, would EQ it flat to 700Hz).
- apply desired crossover slopes. Since your drivers are forced flat (even outside their frequency range), the acoustic crossovers ought to be the same as whatever electrical crossover you choose.
- time-align
The result will be a speaker that's producing a flat response. After that, its a question of acoustics. Since acoustic issues are largely reflection-based, I don't worry so much - the initial wavefront is good, and IIRC our brains can usually filter out longer reflections.
Chris
What I usually do when setting up a new set of speakers is this:
- measure each driver (or pair of drivers if they're playing the same signal), nearfield. Usually within 1 driver diameter
- use REW to average some on-and off-axis responses, and auto-EQ the results, with plenty of spare frequency range at each end (eg, compression driver running down to 1kHz, would EQ it flat to 700Hz).
- apply desired crossover slopes. Since your drivers are forced flat (even outside their frequency range), the acoustic crossovers ought to be the same as whatever electrical crossover you choose.
- time-align
The result will be a speaker that's producing a flat response. After that, its a question of acoustics. Since acoustic issues are largely reflection-based, I don't worry so much - the initial wavefront is good, and IIRC our brains can usually filter out longer reflections.
Chris
I'd try the suggestions above, centered on the tweeter-midrange junction. I'd also try the gating mentioned within REW. This has worked well for me, too.
Here is a link to a short excerpt on the subject of corner-located loudspeakers written by PWK: https://community.klipsch.com/index.php?app=core&module=attach§ion=attach&attach_id=53973
The real issue with corner loading (as you effectively have as shown in the picture) is early reflections at midrange frequencies from side walls, front wall, and equipment between the speakers--not bass frequencies. You can re-EQ the bass to flatten the response to corner loading to significantly reduce modulation distortion, as I see you have already done in your last REW plot. The corner loading will actually smooth the LF response--as shown in the linked article, above. Also see http://www.diyaudio.com/wiki/Corner_Horn_Imaging_FAQ
Some fuzzy absorption tiles on the side walls and front wall between the speakers will work to control early midrange reflections at less than 1 metre from the midrange driver cabinet that impair stereo imaging performance. I'd use a generous amount of them. Hanging thick tapestries or quilts will also work, if the material is very soft and fuzzy.
Room modes will produce peaks and dips in response that are "non-minimum phase", meaning that EQ can't solve those issues. Boosting dips doesn't work - they're there for the room, unless you're using metre-thick absorption panels or the newer resonant bass panels (from D'Antonio) around the room for LF control. However, I've found that pulling the peak responses down to nominal works well.
I'd also recommend overall EQ with rising bass response toward the LF end, and flat response within 1-2 dB at all frequencies higher than about 300 Hz. The 120-240 Hz dip shown in your in response curve probably needs to be boosted a bit after you find a bass trap or two for the corners to eat the midbass boom.
YMMV.
Chris
Last edited:
In the post above yours, there is a link to software that will do the same thing: http://audio.claub.net/software/jbabgy/BDBS.html
Charlie2, nice system.
As others have mentioned, psychoacoustics plays a large role when measuring speakers at the listening position. The first 31 pages of this PowerPoint presentation from JJ Johnston details what our ear/brain hears at the listening position: http://www.thewelltemperedcomputer.com/Lib/room_correction.ppt
A key point in the presentation is frequency dependent windowing (FDW) when taking measurement at the listening position as this is more representative of what our ears hear. Meaning the window varies with frequency. Typically long window at low frequencies (200 to 500ms) and progressively shorter above the Schroeder frequency to basically measuring the direct sound at higher frequencies. One can get a visual of what an FDW window looks like from Denis's excellent documentation here:DRC: Digital Room Correction
REW has FDW capability and there is a long thread on how it works at: Feature Request: Frequency Dependent Windowing - Home Theater Forum and Systems - HomeTheaterShack.com
If measuring at the listening position, I would recommend using the FDW functionality of REW and apply the Psy smoothing to the graph as well.
A fully treated room with proper absorption, diffusion and bass trapping is a lot of $$$'s. A reasonable compromise is to treat early reflections from the side walls (absorption or diffusion to break up the specular reflections) and floor (thick carpet/underlay is good) will help. Use REW's ETC display to look at the first 20ms or so to determine reflection points. Rule of thumb is to have an early reflection down -15 to -20 dB from the direct sound.
REW's waterfall display will show you how your room decays over a 300ms window in time slices. This will point out any low frequency room modes.
As others have mentioned, time aligning the system will help with the impact of the sound since the full frequency waveform from the individual drivers will arrive at your ears at the same time.
If using a computer based audio system, one can use modern DSP software to employ linear phase digital XO's, time align the drivers, and apply an overall room correction (eq) to suite your preferred listening target. Some of the more advanced DSP software will also linearize each individual driver. Meaning, taking near field measurements of each driver and correcting it's frequency response. Then the overall correction at the listening position will be less, but result in an even smoother response.
You mentioned the JBL M2 earlier. Note that the full package of these speakers do come with a DSP for tuning and room integration.
Hope that helps.
As others have mentioned, psychoacoustics plays a large role when measuring speakers at the listening position. The first 31 pages of this PowerPoint presentation from JJ Johnston details what our ear/brain hears at the listening position: http://www.thewelltemperedcomputer.com/Lib/room_correction.ppt
A key point in the presentation is frequency dependent windowing (FDW) when taking measurement at the listening position as this is more representative of what our ears hear. Meaning the window varies with frequency. Typically long window at low frequencies (200 to 500ms) and progressively shorter above the Schroeder frequency to basically measuring the direct sound at higher frequencies. One can get a visual of what an FDW window looks like from Denis's excellent documentation here:DRC: Digital Room Correction
REW has FDW capability and there is a long thread on how it works at: Feature Request: Frequency Dependent Windowing - Home Theater Forum and Systems - HomeTheaterShack.com
If measuring at the listening position, I would recommend using the FDW functionality of REW and apply the Psy smoothing to the graph as well.
A fully treated room with proper absorption, diffusion and bass trapping is a lot of $$$'s. A reasonable compromise is to treat early reflections from the side walls (absorption or diffusion to break up the specular reflections) and floor (thick carpet/underlay is good) will help. Use REW's ETC display to look at the first 20ms or so to determine reflection points. Rule of thumb is to have an early reflection down -15 to -20 dB from the direct sound.
REW's waterfall display will show you how your room decays over a 300ms window in time slices. This will point out any low frequency room modes.
As others have mentioned, time aligning the system will help with the impact of the sound since the full frequency waveform from the individual drivers will arrive at your ears at the same time.
If using a computer based audio system, one can use modern DSP software to employ linear phase digital XO's, time align the drivers, and apply an overall room correction (eq) to suite your preferred listening target. Some of the more advanced DSP software will also linearize each individual driver. Meaning, taking near field measurements of each driver and correcting it's frequency response. Then the overall correction at the listening position will be less, but result in an even smoother response.
You mentioned the JBL M2 earlier. Note that the full package of these speakers do come with a DSP for tuning and room integration.
Hope that helps.
Whenever I want to have a "final" advice on my mixes/masters in the studio, I set up a linear-phase plugin (from Waves) on the DAW, that "smoothes" the room modes for my listening position. But curiously, I like it with corrections only half the way, for some weird psycho acoustic reason... I don't like it all the way flat. It makes sense now.
I've often found the half way mark to sound more reasonable, too.
That's news to me. 1 meter is usually considered nearfield for most speakers. 2mm? That's special. Perhaps we are confusing drivers with speakers?
Have you read the Don Keele paper on near field measurement?
http://www.xlrtechs.com/dbkeele.com/PDF/Keele (1974-04 AES Published) - Nearfield Paper.pdf
Regarding "nearfield", there is no consensus on where this is. It is different for drivers than systems and it varies with frequency, so in reality it is all over the place. Keele's "nearfield" was for a particular purpose and should not be considered a global definition.
I want to point out that the "room boundary" model that most people use in their calculations is mostly wrong. That is because it is based on an empty room with completely rigid boundaries. It is possible to do the calcs with absorbent boundaries - my software does that - but of course real rooms with real furnishings is not possible. The simple models are a decent start, but as one adds more and more LF damping they diverge from reality quite a bit. If I ever get bored with retirement I plan to write a paper on these differences.
The best reference given is the one to JJ's paper. One wants a long time bass for LFs and a short one for HFs because that the way the ear processes. Basically the HF is a simple direct field and the LFs are the reverberant field.
The correct way to deal with LF modal problems is with multiple subs. This can be found discussed in several places.
I want to point out that the "room boundary" model that most people use in their calculations is mostly wrong. That is because it is based on an empty room with completely rigid boundaries. It is possible to do the calcs with absorbent boundaries - my software does that - but of course real rooms with real furnishings is not possible. The simple models are a decent start, but as one adds more and more LF damping they diverge from reality quite a bit. If I ever get bored with retirement I plan to write a paper on these differences.
The best reference given is the one to JJ's paper. One wants a long time bass for LFs and a short one for HFs because that the way the ear processes. Basically the HF is a simple direct field and the LFs are the reverberant field.
The correct way to deal with LF modal problems is with multiple subs. This can be found discussed in several places.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.