I tried to copy your post to this thread as well, no luck to copy it with the pictures without having to copy each picture back into the post. 🙁
(a moderator could move the post for you, and even my reply to that post if you ask them nicely enough 😉)
Yes, if you have an REW file with a few measurements you can view those IR's simultaneously using the "Overlays" button (top center) and select the "Impulse" tab within the Overlays window:
Example overlay window:
Here's the REW file you posted a couple of days ago (fundamental only29th november), comparing left and right IR with Overlays.
(a moderator could move the post for you, and even my reply to that post if you ask them nicely enough 😉)
Yes, if you have an REW file with a few measurements you can view those IR's simultaneously using the "Overlays" button (top center) and select the "Impulse" tab within the Overlays window:
Example overlay window:
Here's the REW file you posted a couple of days ago (fundamental only29th november), comparing left and right IR with Overlays.
Something like this, but a massive spike introduced?
It will be down to incompetence over here some place!
This is supposed to be the comparison of right speaker with and without the RHS panel in place...........
It will be down to incompetence over here some place!
This is supposed to be the comparison of right speaker with and without the RHS panel in place...........
I think it's correct, that large spike probably is from the back wall. If we look at what happens early on, at about 700us there's a peak missing in the measured result with panel. A reduction from 22% to about 8% at that peak. An IR like this mostly shows high frequency material. That's just the nature of these graphs. But it does help to identify things like wall reflections, if you know where to look. The wall is nearby, so the time travel distance is shorter than that peak at ~4.7 ms.
Show the original (no treatment) and the back wall cushions if you like, that should show us something identifiable.
How big can i attach?
Here is most of the measurements, i hope the individual titles are self explanatory, i trimmed out enough to get down to less than 10Mb
Here is most of the measurements, i hope the individual titles are self explanatory, i trimmed out enough to get down to less than 10Mb
14 is no treatment, 12 and 11 are differing front and back wall as described. 4 panels in total distributed 3 on front wall and 1 on back or 2 on front, 2 on back.
I only look at SPL, because i don't understand better/more
If one looks at 7,11, 12 and 14 it seems that treating the front wall and a small back wall panel may smooth things significantly, if the front wall could do better at less than 100Hz, even better. and i like the front wall ideas as i feel they may enhance the room!
M
I only look at SPL, because i don't understand better/more
If one looks at 7,11, 12 and 14 it seems that treating the front wall and a small back wall panel may smooth things significantly, if the front wall could do better at less than 100Hz, even better. and i like the front wall ideas as i feel they may enhance the room!
M
Here's a little sample of how the panels in the back operate:
Both 11 and 12 have a panel at the back, right? Compared to the orange "no panel" we see the peaks at 5 and ~5.8 ms disappear. The high peak at 1.3 probably is either the left or right speaker in a stereo measurement? (and not being in the exact sweet spot?) I usually make sure to be in the exact sweet spot for these type of comparisons.
I'm not exactly sure what the p3 mod is, I'm guessing panels on front wall?
Comparing that to one rear panel it is quite obvious which one makes the most difference 😉.
Yet again a slightly different spot in the room used for this measurement judging from the peak shift at start. This way REW couldn't auto adjust position of the peak zero and I had to use the cursor on the first peak to set that to zero.
Both 11 and 12 have a panel at the back, right? Compared to the orange "no panel" we see the peaks at 5 and ~5.8 ms disappear. The high peak at 1.3 probably is either the left or right speaker in a stereo measurement? (and not being in the exact sweet spot?) I usually make sure to be in the exact sweet spot for these type of comparisons.
I'm not exactly sure what the p3 mod is, I'm guessing panels on front wall?
Comparing that to one rear panel it is quite obvious which one makes the most difference 😉.
Yet again a slightly different spot in the room used for this measurement judging from the peak shift at start. This way REW couldn't auto adjust position of the peak zero and I had to use the cursor on the first peak to set that to zero.
When chasing wall reflections it is best to do it one channel at a time. For instance, we get a reflection from the right wall. Measure the right channel and you'll most probably have a clear peak at that reflection. Play a stereo pair and the left speaker wont have the right wall reflection at the same spot due to different distances to that wall. Now it will look like there's only halve as strong a reflection from that wall. In a similar way arrays average away the floor and ceiling reflections. Yes, there still are floor and ceiling reflections but each speaker is at it's own height, different from the one above or below, so they all have a different point of reflection and 24 others that average that spot out. 😉
Don't ever think there are no floor and ceiling reflections with arrays 😀.
Don't ever think there are no floor and ceiling reflections with arrays 😀.
Great insight thanks.
As usual my inexperience leads me to inconsistencies and misrepresentation, but i am learning!
Another interesting experience, sat at a different listening position yesterday evening i preferred that result, that was just TV voice.
So, P3, was an EQ modification.
The Flex has 4 presets available that each remember the setup of routing, set up and EQ. The P3 mod was actually a right channel bass EQ modification (cut) correcting an over-boost modification i had made, but will not be dissimilar otherwise to any of the no panels options.
I will now set up Preset 4 with the same speaker EQ on the input side for both channels, trying to match the Harman curve, and an individual speaker+room EQ following that on the output side of each channel. This will all be done in IIR with REW but will allow me eventually to compare using a different preset an output side FIR calculated filter on each output with re Phase and so a simple comparison.
Having done this experiment and seen your comments, i shall do it again more rigorously and see what i learn, when there is a suitable gap in seasonal celebrations, maybe January......!
However, seeing that last graph, is removing peaks always a good thing, it may damp out some "good reflection ambience" as well?
As usual my inexperience leads me to inconsistencies and misrepresentation, but i am learning!
Another interesting experience, sat at a different listening position yesterday evening i preferred that result, that was just TV voice.
So, P3, was an EQ modification.
The Flex has 4 presets available that each remember the setup of routing, set up and EQ. The P3 mod was actually a right channel bass EQ modification (cut) correcting an over-boost modification i had made, but will not be dissimilar otherwise to any of the no panels options.
I will now set up Preset 4 with the same speaker EQ on the input side for both channels, trying to match the Harman curve, and an individual speaker+room EQ following that on the output side of each channel. This will all be done in IIR with REW but will allow me eventually to compare using a different preset an output side FIR calculated filter on each output with re Phase and so a simple comparison.
Having done this experiment and seen your comments, i shall do it again more rigorously and see what i learn, when there is a suitable gap in seasonal celebrations, maybe January......!
However, seeing that last graph, is removing peaks always a good thing, it may damp out some "good reflection ambience" as well?
Maybe it is, maybe it isn't related to the Stereo cross-talk but once you get lower levels of reflection one might notice that a voice is easier to follow/understand off axis instead of perfectly centered equal distance to both speakers. What happens is that both ears suffer from cross-talk, not only hearing the right, but also hearing the left speaker with the right ear and vice versa. If perfectly centered, this effect is the same at both ears. Basically creating a comb filter pattern at both ears that has nulls in the speech frequencies. Once you move to either side, both ears get a slightly different comb pattern which allows us to easier understand the spoken words.
That's why I have so many tweaks and trickery in my setup, basically targeted to battle this comb filtering to increase intelligibility. Works perfect for movies too where I don't use an actual center channel.
That's the loaded question. I'd say chances are rather big you'll notice having a panel behind the listener. You may notice it as a reduction in excitement. Instead of using an absorbing panel you could opt to have diffraction there, or maybe a combination. Diffraction would work well as a keeper of ambience while not intruding with the stereo image you perceive. Basically you're 'decoupling' the reflections because they get scattered into different smaller ones.
Maybe something along these lines:
A Do It Yourself version of course, see my post here and the reaction from @jzagaja : https://www.diyaudio.com/community/threads/a-3-way-design-study.376620/page-107#post-7874350
So templates are available to create something like that. This may be a very useful addition to any setup that wants some ambience from the back.
I took a look at the link for that panel, its rather expensive!
But yes, we are a DIY community.
I made some panels, bass traps etc before but wasn't totally convinced by any of it...........
I have carried out my two stage EQ now, and in doing that find the predicted and actual aren't matching very close, this may be mic position (which i don't repeat very exactly) or something else, but i shall potter along becoming gradually familiar and learning!
But yes, we are a DIY community.
I made some panels, bass traps etc before but wasn't totally convinced by any of it...........
I have carried out my two stage EQ now, and in doing that find the predicted and actual aren't matching very close, this may be mic position (which i don't repeat very exactly) or something else, but i shall potter along becoming gradually familiar and learning!
Yes, I would never suggest buying a panel 😀. I'm convinced by the positive effect a room can have on music, but what's coming from behind should be decorrelated to achieve that.
A diffusor creates that decorrelation. Absorbing panels are easy enough to create but may not always be the proper solution.
A diffusor creates that decorrelation. Absorbing panels are easy enough to create but may not always be the proper solution.
Having been inactive over the holiday period, i now find i have evolved in my mind a strategy for a new EQ, based on some learning and useful inputs on the threads.
This is based on the fact that i can use two stages of EQ on each channel in my Flex, i can use Biquads on the input and either Biquads or FIR generated taps on the output of each channel.
Consequently i intend to measure the FR in REW of on of the speakers with the mic at 1m standing in the middle of the room with a measurement window so that reflections are eliminated, this would be at least 2m from any wall so a window to allow for that. then i will implement EQ to match the speakers (both the same) to a desirable curve (currently Harman).
I don't know how to do that measurement yet!
After loading that for both speakers (the same) i will re position the speakers in their normal position and measure again, and implement EQ with REW etc on the output side of the Flex to allow for in speaker in room effects.
That way i see the speaker and the room being separated.
I haven't started to try that yet, so would benefit from any thoughts or ideas anyone can offer to get me started!
Happy new year
Mike
By the way, listeners have commented (un-asked) on two specific aspects, first the separation of instruments and voices from each other and reproduction of previously unheard content, second how the sound spreads evenly both in the listening room and on out into other rooms.
This is based on the fact that i can use two stages of EQ on each channel in my Flex, i can use Biquads on the input and either Biquads or FIR generated taps on the output of each channel.
Consequently i intend to measure the FR in REW of on of the speakers with the mic at 1m standing in the middle of the room with a measurement window so that reflections are eliminated, this would be at least 2m from any wall so a window to allow for that. then i will implement EQ to match the speakers (both the same) to a desirable curve (currently Harman).
I don't know how to do that measurement yet!
After loading that for both speakers (the same) i will re position the speakers in their normal position and measure again, and implement EQ with REW etc on the output side of the Flex to allow for in speaker in room effects.
That way i see the speaker and the room being separated.
I haven't started to try that yet, so would benefit from any thoughts or ideas anyone can offer to get me started!
Happy new year
Mike
By the way, listeners have commented (un-asked) on two specific aspects, first the separation of instruments and voices from each other and reproduction of previously unheard content, second how the sound spreads evenly both in the listening room and on out into other rooms.
Before you proceed with the above mentioned strategy, I'll give you the following picture to "consume".
It's the in-room prediction of an unshaded array of 25x TC9 FD18-08 EQ-ed flat at a listening angle of 10 degree and 2.7 meter.
Negligible floor and ceiling reflections were used (20 dB suppressed), just the position/distance change is shown here.
Take from it what you want, just wanted this effect to be clear. It might influence your current thinking 😉.
It's the in-room prediction of an unshaded array of 25x TC9 FD18-08 EQ-ed flat at a listening angle of 10 degree and 2.7 meter.
Negligible floor and ceiling reflections were used (20 dB suppressed), just the position/distance change is shown here.
Take from it what you want, just wanted this effect to be clear. It might influence your current thinking 😉.
Intriguing reading and looking, currently wondering what effect i am looking at, I see 3
1, bass rolls down from 40Hz, is this the EQ as applied?
2. The levels are reduced as distance increases, not surprisingly
3. The response goes wild in upper frequencies at lower frequencies when measured closer.
Now i have read about and failed to understand (or observe) comb filtering at close distances.
Might this be what its about?
Anyway, a gated measurement at 1m would be irrelevant below its low frequency limitation at ~200Hz and all over the place like this above say 2K, so perhaps not much use!
Further thinking is required, i would like to EQ the speakers in one tranche of EQ and the room in the second, how might one do that simply. i could do a rough attempt at the published curve? - However maybe i should attempt a different strategy and aproach broad trends only in the first stage and local in the second.
NB, i still haven't tried a FIR based EQ at all..........
M
1, bass rolls down from 40Hz, is this the EQ as applied?
2. The levels are reduced as distance increases, not surprisingly
3. The response goes wild in upper frequencies at lower frequencies when measured closer.
Now i have read about and failed to understand (or observe) comb filtering at close distances.
Might this be what its about?
Anyway, a gated measurement at 1m would be irrelevant below its low frequency limitation at ~200Hz and all over the place like this above say 2K, so perhaps not much use!
Further thinking is required, i would like to EQ the speakers in one tranche of EQ and the room in the second, how might one do that simply. i could do a rough attempt at the published curve? - However maybe i should attempt a different strategy and aproach broad trends only in the first stage and local in the second.
NB, i still haven't tried a FIR based EQ at all..........
M
Yup, that's just the EQ
That's a given 😉
I just wanted to show that moving closer to a line array just messes up all that's good about them. Once you get past a certain point things level out much more. You need to be at that certain distance to have a good outcome. Starting at about 2.5+ meters would be a good step forward (but not moving the mic forward 😀). The closer you go, the worse the frequency response is going to be, yes that is where comb filtering gets worse and there will be more uneven 'spray patterns' (or lobes) hitting floor and ceiling. See a little write up here. So I would not recommend using a close up measurement for anything with arrays. You'd be better off taking multiple measurements at a few different spots and average those (which would also serve a great deal to remove room effects that are local). I would not recommend using the average of a multi spot measurement in a FIR generating tool but for general EQ work it would work quite well. You could even average the total measurements of the left and right speaker (measured separately mind you!) to average out the room even more.
Next you can measure at the sweet spot, one speaker at a time and do either fine tune EQ or some FIR correction. Whatever you may wish. Sounds like a plan? 🙂
To show the results of moving further away I've made a similar sim to the one I posted earlier:
It isn't perfect but it isn't bad either. We optimize at listening spot and get some deviation if we move further out. Even the one at 4 meter is within +/- 3 dB up to 11 KHz. Mind you, this is looking at it without adding the room effects, if we add those, that's where most other type of speakers would show more/higher deviations than line arrays in frequency response evenness.
Here's where my frequency shaded array shows an improvement:
The result at 4.0 meter varies +/- 1.5 dB in this case. That's where a lot of my optimizations were aimed at, both on and off axis. Trying to get there without losing the array advantages.
Is it worth it? In real world rooms I'd say not really. The room will be a bigger factor in what we hear than the mods I've made. You need to be somewhat fanatical to get more out of it.
It isn't perfect but it isn't bad either. We optimize at listening spot and get some deviation if we move further out. Even the one at 4 meter is within +/- 3 dB up to 11 KHz. Mind you, this is looking at it without adding the room effects, if we add those, that's where most other type of speakers would show more/higher deviations than line arrays in frequency response evenness.
Here's where my frequency shaded array shows an improvement:
The result at 4.0 meter varies +/- 1.5 dB in this case. That's where a lot of my optimizations were aimed at, both on and off axis. Trying to get there without losing the array advantages.
Is it worth it? In real world rooms I'd say not really. The room will be a bigger factor in what we hear than the mods I've made. You need to be somewhat fanatical to get more out of it.
Last edited:
That must have been what i meant to suggest all along!!!!
Actually it is similar in some ways to what i have done by chance and coaching, but more extreme, in that i have done an average of 9 at a listening position, i shall now take a look at a wider position variance to see what sort of consistency i can get for the average perhaps removing as much room effect as possible for the first stage EQ.
M
Anyone have some thoughtful views on what to smooth and when please.
Following advice above i revised my (current!) EQ strategy as follows
1. make a series of measurements around the whole listening area at different heights, both left and right channel. Average them. Smooth quite significantly and note the level difference on average, and set the Flex channels at that difference. I found 1.2Db and set that. This way the filters are not used to match levels and i find there is effectively more capacity for later on.
2. Measure as above again and average as before. This will be to create what should be mostly the Speaker EQ which goes in the input side of the Flex and is limited to IIR. Smooth extensively and ask REW to create the filters to match the target curve from 35-20K, first off it may not use all 10 bi quads available. reduce the smoothing until it first uses all 10. Install this same filter on the input side of both channels. NB, i could also create that filter set in REW manually if i thought i knew better?
3Measure again and average each channel separately and smooth with Vari or Phsyco, ask REW to make a filter set for each channel and load those.
4. Measure and compare, it wont be as good as the predicted because its based on an average, and no chance i get the mic back in the same positions
5. Have a look at phase and wonder if i will get my head around FIR in Rephase.
6. By the way, i do feel i am getting a gradually more coherent sound as i play around.
The whole REW file exceeds attachment sizes so i made some screenshots
Please be Honest!
M
Following advice above i revised my (current!) EQ strategy as follows
1. make a series of measurements around the whole listening area at different heights, both left and right channel. Average them. Smooth quite significantly and note the level difference on average, and set the Flex channels at that difference. I found 1.2Db and set that. This way the filters are not used to match levels and i find there is effectively more capacity for later on.
2. Measure as above again and average as before. This will be to create what should be mostly the Speaker EQ which goes in the input side of the Flex and is limited to IIR. Smooth extensively and ask REW to create the filters to match the target curve from 35-20K, first off it may not use all 10 bi quads available. reduce the smoothing until it first uses all 10. Install this same filter on the input side of both channels. NB, i could also create that filter set in REW manually if i thought i knew better?
3Measure again and average each channel separately and smooth with Vari or Phsyco, ask REW to make a filter set for each channel and load those.
4. Measure and compare, it wont be as good as the predicted because its based on an average, and no chance i get the mic back in the same positions
5. Have a look at phase and wonder if i will get my head around FIR in Rephase.
6. By the way, i do feel i am getting a gradually more coherent sound as i play around.
The whole REW file exceeds attachment sizes so i made some screenshots
Please be Honest!
M
