Share your in-room measurements?

[rdenney]

Since my last measurements, I have changed my preamp and altered the gain structure setup for my equalizer. And I am trying to do things a bit more carefully.

Here's the setup: Dayton Audio calibration mic into Presonus Studio 24c (both mic and interface have been calibrated, the latter with a loopback test) into a line input on my B&K MC-101 Sonata preamp. On the preamp, the line amp is engaged, the tone controls are defeated, and the EQ loop is engaged. The PEQ is a Yamaha YDP2006 digital parametric equalizer from a couple of decades ago. The Yamaha equalizer provides six parametric filters for each channel, with +/- 12 dB of gain, and a Q range of 1-10, plus low and high shelf filters. Each channel also includes six notch filters, of which two can be reconfigured as low and high-pass filters. It also include pre-emphasis to hot-rod the high frequencies going into the processor, which I have turned off. The DSP portion includes an internal in-and-out converters comprising 20-bit Burr-Brown PCM1760P ADC and a PCM63P DAC clocked at 44.1 KHz. Analog gain amps are provided by common 5532 op-amps. This equalizer has good specs for its day--distortion better than -80 dB, noise likewise, and a dynamic range of 106 dB. In a playback chain, this isn't terrible, though the specs won't look as good as one of the better miniDSP models, and I have sort-of poked around what those units would require (in dollars) to make a real improvement.

The output of the B&K preamp is fed to a pair of B&K Reference 125.2 amplifiers driving Revel Concerta F12 speakers in a biwire arrangement that does nothing but let me use both of the amps I own. (The amps will be replaced with an incoming Buckeye NC502MP amp in the search for more power and also just because.

Obviously, REW isn't going to attempt to support that Yamaha commercial PEQ directly. I do wish that they would provide a configurable equalizer, rather than a (long) list of more current models. Being able to program the number and type of filters would at least minimize the back-and-forth required to get REW not to suggest more filters than I have.

So, to the measurements. I have used the "psychoacoustic" smoothing, because my purpose isn't looking at comb filtering but rather identifying what PEQ filters I need. And with 1/12 smoothing, the EQ algorithm in REW asked for many more filters than the Yamaha provides.

Before:

I used the equalizer feature to attempt to match a target with a low-frequency cutoff of 25 Hz and a 0.6 dB/octave downward tilt from there up. REW distinguishes the bass range from everything else, so I had to define around the 200-Hz boundary it defines. I have to say that my room is not very intense on first reflections, so I'm not thinking the stronger downward tilt of the Harman model would be exactly right in this room. So I've gone a bit flatter, which fits my preference in any case.

I gave REW the range of up to 1000 Hz, recognizing that the changes I wanted to see at 900 Hz were probably not minimum phase.

The big hole at 200-400 Hz troubles me. It's well below the crossover between the woofers and mid-range driver, which is at 575 Hz, so I don't think it's a suckout. I see big spikes in excess group delay at 240, 325, 353, and 425 Hz, but I'm still trying to sort out in my head what all that actually means. I think it means not trying to fix the dip in that region with narrow filters. But there's something happening in the room to cause that issue--the anechoic response of these speakers doesn't show it.

Here are the PEQ corrections I made:

Note that Filter 1 on the right channel is switched out, as is Filter 6 on the left channel. The Q range goes to 10, but I have no idea what the numbers mean, or how they compared to the Q values used in REW for the "Generic" equalizer. I kept the numbers fairly low to keep the filter shapes broad, in the (ignorant) hope that it will work around those frequencies that see excess group delay.

Filling that gap requires a LOT of gain in the equalizer. I cannot for a moment consider dropping everything else in the band. But I did substantially attenuate the signal so that with the equalizer looped in by the preamp. the overall loudness noticeably decreases. I should measure some levels with my SSVM to determine where I can set those values to avoid clipping the return input of the preamp (which is upstream of the line stage). I did notice more distortion in the REW plot than I have seen before, with lots of regions above 1% coming out of the speakers. I can't hear a problem, but that doesn't mean it isn't there. So, there's more to do. I'm not worried about starving the amp--I'm already at full input capability of the amp halfway up the volume pot of the preamp.

Here's the response with the EQ corrections:

I cannot explain the rolloff in the top octave. I always see it, but it's not the microphone, which was provided with a calibration file that shows little correction up there. I know the speakers are capable up to 20 KHz, but I don't know what in my room might be sucking them up. That's on the list to fully understand. Not that I can hear much up there.

And there's a bump at 3-4 KHz that I did not mess with--a trim at that point should make the whole graph acquire that obvious downward spectral tilt. I still have an open filter on each channel if I decide I can mess with it.

Rick "still on the steep part of the learning curve, it seems" Denney

 

[storing]

but it's not the microphone, which was provided with a calibration file that shows little correction up there. I know the speakers are capable up to 20 KHz

None of these are 100% proof though; ok I know in this case it's extremely likely these aren't the culprit, but still, the 'question everything' principle has showed so many measurement/assumption errors in my day-job measurements that I've come to default to it whenever possible. In this case you could for instance measure with the microphone directly in front (as in almost touching) of the tweeter to make sure those frequencies measure without attenuation there.

 

[flipflop]

Rick "still on the steep part of the learning curve, it seems" Denney

It seems like you've regressed on the learning curve since the last time you posted in this thread.

 

[rdenney]

It seems like you've regressed on the learning curve since the last time you posted in this thread.

Not unlikely, but saying so without explanation isn’t helpful.

Rick “one thing that has changed since prior entries is an expanded vertical scale” Denney

 

[rdenney]

None of these are 100% proof though; ok I know in this case it's extremely likely these aren't the culprit, but still, the 'question everything' principle has showed so many measurement/assumption errors in my day-job measurements that I've come to default to it whenever possible. In this case you could for instance measure with the microphone directly in front (as in almost touching) of the tweeter to make sure those frequencies measure without attenuation there.

Good idea—I’ll add it to the further exploration list.

Rick “left the stuff set up just for that” Denney

 

[MakeMineVinyl]

Following up with my previous MiniDSP 4x10HD adventures, I've gone back to the purely analog crossovers / EQ for the LF, HF and VHF horns, but I have routed the signal for the subwoofer through the MiniDSP to perform the crossover and PEQ. I tried using REW to work out the PEQ filters necessary but these never yielded a good response, so I just went the manual route working with the RTA to adjust the filters and checking the results in REW. What I ended up with was a crossover point for the subwoofer of 46Hz (previously was 60Hz) and I ended up only needing to use 3 of the 5 available PEQ filters. The region between 10Hz and 20Hz is a bit hot, but I think I'll leave it as-is since I'm not anywhere near stressing the four 18" subs at my maximum listening levels.

The graph below is of the left channel at the listening position 11 feet from the speakers using a MiniDSP UMK-1 microphone. 1/6th octave smoothing.

 

[flipflop]

Not unlikely, but saying so without explanation isn’t helpful.

You don't see the stark contrast between what you said in post #430 + #436 and the content of post #481?

 

[rdenney]

You don't see the stark contrast between what you said in post #430 + #436 and the content of post #481?

Not really. Check the scales. Even in the older post, there was about a 10-dB difference (or more) between the levels in the low hundreds of Hz and the level around 300, without EQ.

If anything, I've learned not to hide response errors with a compressed vertical scale.

I do note that the peak around 4-5K was observed in anechoic testing done by Sound Stage Network at the Canadian NRC:

And the peak there was a little more pronounced off-axis:

(It occurs to me, looking at these, that my measurements are looking more like the severely off-axis plots, making me wonder if there are more reflections than I had thought. But when I set at the listening point and look for first reflections, I simply don't see very many candidates in that space. But maybe I need to do some toeing experiments. Another thing to try: some absorptive material behind the listening chair.)

But my measurement mic, which was at the same vertical level as used in that testing, was no more than 10 degrees off-axis from the slightly toed-in speakers.

I set REW's test levels (which use pink noise) at 85 dBC SPL at the microphone position, which is about two meters or a bit more from the speakers--the same as what the NRC used in the anechoic chamber.

But my response after equalization is not quite as good as their spin measurements without EQ, which leads me to think the room is doing something funky. I've changed some things since March:

1. I'm using the -20 dBm input and output gain levels on the Yamaha PEQ instead of the +4dBm gain levels. The signal path goes from the input to one stage of a 5532 op-amp, through the second stage of the 5532, to the input gain pot, to yet another 5532 stage, to the pre-emphasis filter (which is switched in physically using a relay), to a fourth 5532 stage, and then to the PCM1760 ADC. The -20/+4 switch wraps around the second gain stage. When at -20, the output of that stage is fed back through to the input signal negative through a 68K resistor. The +4 position bypasses that resistor. The -20 is supposedly for lower-voltage sources, and my source is the record bus of the preamp, which is unbuffered and at input source level. The level display for the input is handled in the digital realm of the Yamaha unit, which indicates to me that clipping there would indicated digital clipping in the processor.

The output gain circuit has a separate switch, also for +4/-20 dBm. After comping back from DSP, the signal goes through the PCM63 DAC, in series through two 5238 op-amps. The second of these is has a filter wrapping around it to provide de-emphasis, again switched the the pre-emphasis relay. The output of that second op-amp is shunted to signal ground through a 7.5K resistor and a 510-ohm resistor in series. At +4, both resistors are used; -20 bypasses the 7.5K resistor. So, the -20 setting on the input side is padded down with a resistor, and the +4 setting on the output side is padded down with a resistor. I'm using -20 for both, which lowers the signal level into the DSP unit, but for some reason results in a higher indicated level on the "input level" LED array that is controlled by the DSP processor. To keep that from showing clipping, I turned down the input level pot considerably--maybe -10 dB on its scale--and the output of the unit is noticeably lower than when it is bypassed. But that provides room for the amplification in that 300-Hz hole, it seems to me. I just turn up the preamp a bit to compensate.

When I was using the +4 switch positions, the gain level was always well below full scale unless I added positive gain by turning the input pot above 0 dB. ("0" is at about 2 o'clock on that input gain pot.)

I need to put in a reference signal and measure some voltages to really know what's happening.

2. The preamp is now a B&K MC-101 Sonata instead of the previous Adcom GFP-565. In both cases, the equalizer rides in the external processor loop, so that I can bypass it entirely in the preamp. (I am not using my dbx 400T to provide that processor loop--that sits in the tape loop of the preamp, upstream from the processor loop, which is upstream from the line amp.

Rick "gotta get the signal generator and SSVM out of the closet" Denney

 

[rdenney]

Well, I just had an epiphany, of the sort that anyone who has 1.) tuned complex antennas, and 2.) studied the acoustics of brass instruments should have experienced a long time ago.

One of the weaknesses of my current listening space is a partial wall close to the back of my head. THERE IS MY DAMAGING FIRST REFLECTION! Well, D'UH!

(The view of my listening chair from the left-channel speaker.)

The wavelength of 250 Hz (the deepest part of my left-channel dip) is 56 inches, close enough.

The microphone was 14 inches from the wall.

...wait for it...

If a wave is at peak pressure when it passes the microphone, it will be a quarter wavelength down 14 inches later when it hits the wall, which is at zero pressure differential. 14 inches later, it will be at a half wavelength, which is at peak negative pressure, negating the peak pressure from the direct signal.

So, a quarter wavelength from the wall should be a deep null. I feel kinda dumb.

The good news is that only a small part of the wall needs to be treated to diminish the first reflection.

Rick "looking into absorption panels" Denney

 

[rdenney]

Okay, much screwball stuff going on here.

Discovery 1: Yes, the reflection from the back wall is causing problems. But except at a few sharp frequencies, I could EQ around it, up to a point. Once I discovered...

Discovery 2 (Edit: see later post): The Yamaha PEQ is, for some reason, adjusting frequencies at half the frequency being displayed. Some divider somewhere ain't dividing. I'm expecting it to correct itself at some point, but if'n it does, I've at least made notes.

Discovery 3: It's not the measurement mic--I measured each driver from very close, and got readings I fully expected to get.

Discovery 4 (Nope! See later posts): The top octave reappeared when I toed the speakers in to point directly to the listening position. The bottom octave weakened considerably, probably because I killed some modes behind the speaker by turning it at a sharper angle.

Curiosity 1: Or, could it be that REW is halving every frequency all of a sudden? (That was a joke. Sorta.) (Edit: Not a joke. See later posts.)

Discovery 5: Nothing I used to improvise either damping or redirection of the reflections behind the microphone made more than a subtle difference in the readings. The dip was there even when I moved the mic three feet into the room. I stopped worrying about it, and I'll let the acoustic panels I put on my ebay watch list go to someone else.

There are still lots of comb effects going on with reflections, so maybe I should keep those panels on my watch list. I just don't know where I'd put them to have any real effect. I get not ceiling first reflection--it slopes up and the mirror point is behind the speaker. There is no side reflection on either side. The reflection from behind the speaker won't be much, simply because the mid and treble sound doesn't go to the rear strongly.

But now I'm roughly +/- 3 dB around my target, and that seems to be as good as it gets.

Here are the left and right channels against my EQ target:

And here's the final EQ, back to 1/12-octave smoothing, and still with an expanded vertical scale.

Rick "done for now, but giving myself a couple of days to receive advice and think of stuff before putting the apparatus away" Denney

 

[rdenney]

Okay, now I'm really confused. Attempting to figure out why the frequencies were doubled, I started comparing REW traces from the same speaker on different days. The levels were different--not important. The frequencies are doubled--it looks like a linear lateral translation on the log display. Here are two traces of the same speaker, made on different days and with somewhat different conditions. But notice how the shape of one is similar to the other but shifted laterally on the response graph.

I'm a little PO'd, to be honest. I work on this stuff instead of getting sleep, and this has seriously ruined a precious evening for me. Now, I have no idea where ground truth is. Am I getting great bass, or the top octave? I'm going to have to start over from scratch, if I can figure out how to make REW read consistently.

No wonder nothing made sense.

Rick "whose last few posts represented hours of wasted effort" Denney

 

[storing]

started comparing REW traces from the same speaker on different days

Weird. And interesting. To me only the part up to 120Hz looks shifted. The rest looks almost inverted. (peak at 210, 350, 950 vs roughly corresponding valleys in red graph). In any case no obvious effect across complete sweep. Can be pure coincidence, slightly different mic placement, you standing in a different place, ... (since you mention somewhat different conditions). Or could be REW messing up delays (as in thinking it's recording a 100Hz since but actually the 200 Hz one is playing already) but I don't think it typcially does that?

 

[flipflop]

I do note that the peak around 4-5K was observed in anechoic testing done by Sound Stage Network at the Canadian NRC:

There is no 4-5 kHz peak in that frequency response.

And the peak there was a little more pronounced off-axis:

It shows that 3-4 kHz doesn't lose as much energy, when going off-axis, as the surrounding frequency ranges. Still no peak to be found.

It occurs to me, looking at these, that my measurements are looking more like the severely off-axis plots, making me wonder if there are more reflections than I had thought.

It's completely normal. In-room measurements are dominated by reflected sound. Unlike measurement mics, humans hear both the direct and reflected sound above the transition frequency.

But my response after equalization is not quite as good as their spin measurements without EQ

Why would it be? How are the two things even comparable?

 

[rdenney]

Weird. And interesting. To me only the part up to 120Hz looks shifted. The rest looks almost inverted. (peak at 210, 350, 950 vs roughly corresponding valleys in red graph). In any case no obvious effect across complete sweep. Can be pure coincidence, slightly different mic placement, you standing in a different place, ... (since you mention somewhat different conditions). Or could be REW messing up delays (as in thinking it's recording a 100Hz since but actually the 200 Hz one is playing already) but I don't think it typcially does that?

Sure, there are differences, overall level and speaker toe being two of them. But mic placement is similar.

Being delayed by a constant time would not cause such a straight visual translation on a log scale. If the sweep takes one second, and 40 Hz is being assumed by REW when 20 Hz is playing, the sweep would have to ascend exponentially to create what we are seeing. But that seems to me like REW would analyze that as 100% distortion of the second harmonic--the second harmonic would be at signal level and the assumed fundamental would be zero. Not seeing that.

Here's my arm-waving in the hopes that someone with real expertise will set me straight:

I'm really thinking it has something to do with sampling rate. For a while when I first got REW, I was using the OS (Java) interface, with all the Microsoft features turned off, at 48KHz, because that's what I could get to work. I made my first loopback calibration at that sampling rate. Then, when I used REW to measure an open-reel tape deck, I switched everything to 96 KHz sampling, including making a new loopback calibration file. I also switched to the ASIO driver that came with the Presonus sound interface I'm using.

But then I noticed that my measurements of that Teac deck seemed bloated in the bottom octave. I thought it was something to do with the calibration, and didn't really worry about it much for that purpose. But I dug up that calibration file and looked at the measurement that made it--the bottom of the calibration rolled off substantially starting at 80 Hz, being 5 dB down by 20 Hz. Applying that calibration would bloat the bottom octave, for sure. But that roll-off sure looked like the calibration measurement was shifted to the right by an octave.

So, I decided to measure again a loopback calibration. Who needs sleep? The Presonus Studio 24C is intended for commercial use so it has Neurik combination sockets on the front for inputs to accommodate balanced and unbalanced microphone (XLR), instrument (1/4" phone, two-conductor), and line inputs (also 1/4"). The XLR connector goes through a fixed mic preamp that has phantom power and a very low input impedance for condenser mics, and the instrument/line input skips that and has an input impedance of the usual thousands of ohms. Both have a gain stage after that with up to 50 dB of gain. My measurement mic is a Dayton Audio model that uses a microphone interface with phantom power, and I plug it into the Presonus with a conventional balanced microphone cable. When I created the 96 KHz calibration measurement for testing the Teac, I incorrectly used an XLR-RCA adapter plug to connect the line out on the Presonus to (yes) the microphone input. I repeated that mistake (at first) last night, resulting in another rolled-off calibration measurement. I'm sure that it is the low microphone input impedance causing the problem. Presonus has nearly identical performance specs for both mic-in and line-in, but a line source with higher output impedance will react with a mic input with low impedance, even if the level is sufficiently attenuated. Sure enough, the loopback measurement showed a 5-dB rolloff by 20 Hz (but was surprisingly acceptable above that).

I switched to a 1/4"phone/RCA adapter, and re-recorded the loopback measurement. That fixed it--20Hz to 20KHz at +0/-0.1 dB, with a noise level measured by REW at -85 dB, pretty much across the spectrum.

But that calibration assumes a line input, and I'm taking it on faith that the Presonus specs for their mic input are valid. I'm sure it's very close for room measurements, where that level of precision is unnecessary in any case.

So, here's the question: Is there any way that me fiddling around with a 48 KHz-sampled calibration file could confuse REW 5.1 into thinking the files were recorded at 48KHz sampling rates, such that switching the calibration file back to a 96K calibration wouldn't fix it? REW is subject to needing to be restarted from time to time, as I have observed. In any case, I need to do all the measurements again to settle it in my mind.

Rick "learning about the equipment as much as the process" Denney

 

[rdenney]

I hope this is the final episode in the continuing saga...

...at least until the next time.

After building a new calibration file last night (really this morning), I decided to do a bit of laptop maintenance this evening, which included a fresh boot. Mostly, I was getting rid of Pro Tools One, the cheapie version of Pro Tools, but still a machine hog of the first water. Plus, I never got it to recognize my audio interface, so screw it. I have the (excellent) PreSonus Studio One (the license for which came with my interface), but for the most part Audacity does the music recording I need with no fuss. I also updated the control software for the PreSonus interface.

And this evening, the ASIO driver is working fine, frequencies are being recorded correctly, and everything is behaving.

Here are fresh measurements from the listening position. I'm using large pillows to simulate my body sitting there, and the speakers are now toed in so that they are point right at my face while sitting here. The Yamaha equalizer is back to using the +4 dBm input and output settings, which prevents accidentally overloading the inputs, and I showed no signs of clipping the input even with some substantial elevations of some frequencies.

I am of the mind that the downward spectral tilt is an outcome rather than an objective--the product of speakers with good directivity in a typical room. This is not a typical room--the only first reflections are from the front and back walls, and the room opens into another room, providing an axial modal distance of 40-odd feet with pretty good absorption, plus a sloped ceiling that has to be modeled at four feet high to predict the nulls that I’m seeing. There is interaction among competing long dimensions in the space, not the result of anything I can really fix.

But I noted a real drop in higher frequencies off-axis both in the room (the results were affected by toe-in) and in the anechoic measurements from the Canadian NRC shown upthread. So, there will be some downward tilt if there is any reflection at all contributing to the integrated sound at the listening position. I prefer not to exaggerate that, so I set the target to roll off at 25 Hz and slope down at a rate of 0.8 dB per octave.

I figured out how to constrain REW's equalizer optimizer to just the number of PEQ channels I had available (six) and let it fly. I entered these making a guestimate on the needed Q, given that REW's default Q scale covers a different range, and perhaps with a different profile, than the Q adjustment on my equalizer. The values I used are:

Channel	Left f (Hz)	Left gain (dB)	Left Q	Right f (Hz)	Right gain (dB)	Right Q
1	106	-5.5.2	2.5	133	-7	4
2	240	+3.5	4	257	+8	2
3	300	+4	2	355	-8	9
4	520	-7.5	8	1.16K	+3	5
5	2.9K	-3.5	1	1.54K	-3	5
6	40 (not suggested by REW)	+3.5	2	3.86K	-4	1.2

Everything above 400 Hz is probably well above the Schroeder frequency. But looking at those NRC spins, and despite them not being visible to all observers, there is about a 3-4 dB dip in the axial response between 1 and 2 KHz compared to the little bump between 3 and 4 KHz. This becomes more pronounce off-axis, but smoothly, which Toole tells us is ripe for electronic equalization. So, I didn't mind too much the low-Q adjustments I made in the 3K range, or the bit of fill I added to the right channel just above 1K. The one departure from Toole would probably be the left-channel adjustment at 520 Hz, which is a high-Q cliff on the right edge of the adjustment at 300 Hz. I bet I won't notice whatever problem it causes.

REW did not suggest the boost at 40 Hz in the left channel. The rear port of that speaker isn't in quite the same shape of space as with the right speaker. I boosted it to match the right channel, but didn't notice really all that much resonance.

Here are the two channels relative to the targets, with the above EQ added:

And here's the final result:

The only resonances I have noted are either below 40 Hz, or at 120 Hz, where that frequency rings after the others have decayed. But it doesn't stop ringing at several dozen dB below the signal, making me think it's either power hum in the system or some residual noise in the house. I think it's the latter--my RTA sees it in the room at maybe 10 dB above the noise floor. But I don't see any harmonics of it, so it's not buzzy. 60Hz doesn't stand out in the room. It could be the air handler. So, the big problems are not resonant modes so much as hard nulls, at 53 and 90 Hz in the right channel, and at 72 Hz in the left channel. I think only subs will help address those. I think moving the speakers around will likely move the problem around without solving it.

The big moves in the equalizer did not add any distortion or overloading. The distortion graphs in REW showed that above 100 Hz, distortion rose above 1% (-40 dB) briefly at around 250 Hz, and briefly again at around 2.5 KHz, whether or not the EQ was engaged. The NRC measurements showed the 2.5 KHz distortion spike, which is assuredly a crossover issue—the 5-1/2” mids are maybe just starting to break up a bit before the crossover sends the signal to the tweeters. Crossover frequencies are at 575 Hz and 3 KHz. REW is apparently interpreting the interference with reflections from the back wall that causes the null at around 250-300 Hz as distortion, as it should.

Finally, I am considering adding maybe 0.7 ms of delay on the right channel to time-align it with the left channel. My listening positing is not exactly centered, and because of that the phantom image leans right a bit (meaning, right in front of me, because I'm a bit right of center). I stretched a tape and the difference in distance is 8 inches. I have also added a fraction of a dB to the left channel to balance the loudness. The Yamaha equalizer has very fine delay, measured in inches if preferred, as part of its DSP. A listening test confirms that the phantom image sits right between the two speakers, even though I'm slightly offset from center. That's firmly in the category of doing it because I can--I never noticed any lack of stereo imaging without it.

(If I do add subs, I have a another Yamaha PEQ that I can use to cross them over and adjust their time delay, assuming the subs don't do it.)

Rick "time for some sleep" Denney

 

[Le Concombre]

I can't think of a good reason for more than a couple of db difference between MMM and point sample above the room transition frequency, apart from SBIR which should not be a huge issue at 10KHz and above.

Did you use a timing reference signal with the sweeps? If there is any delay in the chain, there could be impact on the HF curve. Long time ago I measure an AVR via its Airplay input and say -20db at 8KHz, which was clearly wrong. Adding the timing reference sorted that out.

I have taken the time to redo MMM with UMIK's 0°calibration file and with the mike fixed to a stick, thus 1m from by body but my arm. The HF roll off is now quite OK. I have created 2 sets though ; one has a -2.5 dB 2 octaves eQ point @8000K to match Bob Katz's HF roll-off and also because Steve Hoffman often complains that 3 db @ 8K should be subtracted from this or that mastering he hasn't done ; maybe due to the influence of Bob Katz there's a link ! Without that eQ point, the roll-off is quite consistent with Toole's ideal steady state. As of the bass I have tweaked it to follow Toole's ideal down to 64 Hz and below that I let the resonance be : most recordings cut before and the vast majority of those who don't seem to benefit from the bass lift. Please also note the difference when correlated PN rather than uncorrelated PN is used to measure both channels at the same time. I indulged myself to use a dozen eQ points per channel in those latest iterations but please note how the essential is reached with only 5 eQ options per channel with my Meyer analog parametric equaliser that I use to correct my analog rig.

 

[Le Concombre]

Addendum : a quick word about "target" choices.

yesterday alone I came across 2 Bernie Grundman masterings, Joe JAckson's Body &Soul SACD and H Hancock's Speak like a Child 24/192 ; both exhibited round 8 K bumps. A pure compensation would be more Hz precise, less than 2 octaves, and more than 3 dB knocked out. But it justifies having a set with the Bob Katz's HF roll-off for quick shift

Speaking of whom, H Hancock's Speak like a Child benefited from a set done after Katz's target : better delineation of layers, less clogged upper bass (double bass has no LF extension in this recording ; same with the original Blue Note LP) ; so, yes, maybe I will end up fiddling with bass options too. I also made a set lowering the 39 peak to Toole's level.

But I strongly advocate that the bass raise down to 64 (reading Toole's ideal steady state from right to left) shouldn't be any higher than Toole's (so very very tiny raise). I gave Trained listeners another try, could demo that the general drums and bass interaction, being more prominent, is "better" ; but on my system it was at the expense of vocals and of the impacts of drums strokes that are much more life like with the leaner in the 80+-30 Hz region Toole's (or Katz for that matter)

 

[Madjalapeno]

Have the house to myself, so had time for some initial REW measurements with my LX521's.

Measurements are via a MOTU M-2, and UMIK-1.

So far only one measurement for each speaker, and one for the L & R together. Listening position is approx 3.0m from L, and 2.9m from R speaker, and they are pulled 1m away from the rear wall. No EQ or room treatments at all.

 

[Madjalapeno]

Tried again with the moving microphone method. 2 scans, with 30+ samples.

Having played with various house curves and convolution filters (light blue, bottom plot), I think I prefer the default sound.

Not sure if REW has a problem with di-pole designs, but it sounds really flat with the EQ. Or maybe I just prefer the default sound.

 

[Le Concombre]

Tried again with the moving microphone method. 2 scans, with 30+ samples.

View attachment 175663


Having played with various house curves and convolution filters (light blue, bottom plot), I think I prefer the default sound.

Not sure if REW has a problem with di-pole designs, but it sounds really flat with the EQ. Or maybe I just prefer the default sound.

I'm not surprised but the main issue comes from full range correction. You could try taking a few dB off the 220 broad peak and boosting the peak inside the dip around 320. I would also try to take a few dB off the 500 peak but every move above 300 is to be considered risky