Feature request - please help save our tweeters!

[lovingdvd]

Last night I had the master volume on my AVP set too high without realizing it. I ran a frequency sweep 20-20kHz with my UMIK-1 facing the ceiling and 90 degree cal file from the MLP, which is about 13.5 feet from the speakers. Because the MV was set way to high, REW showed a warning in red "0.0 dB headroom" and after the measure showed the warning that the results would be unreliable. That's understandable, as the mic was overpowered. Unfortunately, so was the tweeter...

As a result, I blew out the tweeter and I now have to repair the speaker. Normally I run the sweeps with the MV around -25dB to -20dB (IIRC). I had the MV up at -8.5dB because I was doing some Level adjustments. Then I forgot to lower it. Whoops! When I moved on to the frequency sweeps, the volume was still set at -8.5dB. In the measurement graph the FR nosedived starting right at the 2.5kHz crossover to the tweeter. I knew right away that the tweeter was toast.

Although the MV of -8.5dB is way too high for a frequency sweep, I do wonder why exactly it blew the speaker. These speakers can play at reference from 13.5 feet to the MLP. And it's not like the MV was at 0dB. -8.5dB uses far less amp power. One possible reason is that Audyssey XT32 has applied +8.5dB to the high frequencies in my calibration. So with the MV at -8.5dB, but Audyssey adding a boost of 8.5dB, it was more like closer to 0dB I guess? But even then, I would still think that the tweeters could handle this. Unless a speaker playing real movie/music content at 0dB reference is totally different than a speaker trying to handle a full spectrum frequency sweep at the same reference level, which perhaps is far more brutal on the speaker? I have read online that running a frequency sweep too loud can blow a tweeter, but I am wondering what level is considered "too loud". Well in my case I know what that level is but I am wondering if this is a normal level or if the speaker should have been fine at that level and instead by loud levels they are referring to MV of 0 or higher?

I'd love to hear your guys thoughts on this and whether it makes sense given the above that the tweeter blew out. Or whether you think that the tweeter should still have survived. Ultimately I am less concerned about blowing a tweeter when using REW (because I will be much more aware of the volume in the future) and more concerned about blowing a tweeter while playing real content at or near reference. Which I also did recently as a result of an Atmos music track, but that time I had the levels set +3dB too high on the speakers that blew, along with the Audyssey boost, so the speaker may have gotten peaks far above reference...

And now on to the feature request please.... It would be FANTASTIC if REW allowed us to set a threshold for the Headroom, so that if that threshold was crossed, REW would immediately stop the frequency sweep and display a warning. For instance I would likely set this threshold around 3dB - so that if at any point in the sweep it had less headroom then that, it would immediately abort the sweep and stop the signal. If REW had this feature, it would have saved my tweeter.

Its definitely not the fault of REW that the tweeter blew and certainly operator error here. I will definitely be a lot more careful to check the MV in the future before doing sweeps. But oversights do happen, and as much as I like to think I wouldn't do this again, it only takes one slip to lose a speaker.

What do you think?

Thanks!!

 

[Matthew J Poes]

Can you share your speaker model, microphone, and amplifier. That would help to figure out if this is normal.

Tweeters can't handle a lot of power but the crossover in most speakers are designed such that the amount of power that passes through a tweeter is very low even at very high volume levels.

The headroom in REW is a function of the digital signal it gets. The USB mics we use (and most inexpensive measurement mics) can handle 130db or so with low distortion. When you set the input and output gains too high the headroom problem is in the signal being passed through REW, not the mic itself. REW has no idea what the mics headroom actually is. A measurement that is at the signal limit is usesless so I could see maybe creating a safety limiter you could implement, but I guess the real answer is that it's important to be careful. Hard to include safeties in these softwares for this kind of thing. Similarly you could also blow a subwoofer trying to drive a 10hz signal through it too loud for too long. REW couldn't really protect against that.

Now are you sure the tweeters are fried? Can you share why you believe they are.

 

[lovingdvd]

Can you share your speaker model, microphone, and amplifier. That would help to figure out if this is normal.

Tweeters can't handle a lot of power but the crossover in most speakers are designed such that the amount of power that passes through a tweeter is very low even at very high volume levels.

The headroom in REW is a function of the digital signal it gets. The USB mics we use (and most inexpensive measurement mics) can handle 130db or so with low distortion. When you set the input and output gains too high the headroom problem is in the signal being passed through REW, not the mic itself. REW has no idea what the mics headroom actually is. A measurement that is at the signal limit is usesless so I could see maybe creating a safety limiter you could implement, but I guess the real answer is that it's important to be careful. Hard to include safeties in these softwares for this kind of thing. Similarly you could also blow a subwoofer trying to drive a 10hz signal through it too loud for too long. REW couldn't really protect against that.

Now are you sure the tweeters are fried? Can you share why you believe they are.

Sure - I'm happy to provide more information. I'm grateful for the help! I really need to better understand what is causing the tweeter to blow in my setup because this is not the first time (more on that below)... This post got pretty long, but all the answers to your questions are in here.

First let's go back in time to the fall 2016 when my dedicated room was completed. I have a 9.4.6 setup with only 7.4.4 active due to the current limitation in Atmos channels with my Marantz 8802A. Front speakers are KEF Ci5160RL-THX. Surround speakers are very similar but the 3160. Ceiling speakers are the Ci200RR. When I first got the room going back then, I was too exhausted from a very long build process and didn't have the time to focus on the audio calibration. So I just ran Audyssey XT32 and it sounded amazing afterward. I did not even bother to measure anything. All I knew is that I was thrilled with the sound, both for movies and music and didn't feel the need to look under the hood with measurements. Mainly because I wanted to go with the ignorance is bliss approach - once I saw a frequency response I would want to tweak and then its a big time sink.

After that initial setup everything was great. I would sometimes blast music and movies. Usually my comfort threshold is to listen around -5dB to -7dB for an Atmos movie. Occasionally as much as -3dB. I'm sure I do not need to tell you that this is loud, very loud. So I've watched dozens and dozens of movies (mostly action in Atmos) and never had a single problem with the speakers and really enjoyed it. These speakers really sound amazing. They are also all perfectly voiced with each other which is quite unique, or perhaps maybe not as unique now, but at least a few years ago when all this was planned that was pretty special. Also these speakers were chosen because they have very wide dispersion, which was particularly important to pull off Atmos in a room with a 7.5' ceiling.

So anyway, things have been great with the audio until two weeks ago where a number of factors came together to create a perfect storm that would ultimately knock out my tweeters in the Surround Back (SB) channels... It started when I decided to check the SPL levels of all my speakers. I noticed from the MLP the SB channels were about -3dB too low, because the seatback partially blocks the line of sight from the speakers to my ears. So I upped the Levels on the SB from -3dB to 0dB. Which hindsight this was a really poor ideal... To put this in context, most of the other speaker Levels were between -10dB and -5dB. So the SB at -3dB an outlier and even more so after I raised it to 0dB. Let's call this factor 1.

Then later that day I was watching the movie Chicago (the play) in Atmos. During a few musical numbers I decided to really blast it and took the MV to 0dB. I can't recall if I've ever had the system that loud. As mentioned above, normally about -3dB was the absolute max, and -5dB to -7dB was more "normal". Let's call this factor 2 (listening at the loudest level ever in a year). And while most movies use the SB channels mostly for ambiance and a few sound effects, this movie uses a LOT of full range discrete sound in the SB channels with instruments like horns and trumpets blaring out of the channels, and sound effects like a very loud "pop" as old-school flashes from cameras go off, etc. Let's call this factor 3.

Now let's add to the mix that Audyssey wants to try and flatten the natural speaker power HF rolloff it is measuring, which I don't think it should really do. In the process it is applying as much as +5dB to +9dB(!) boost to the HF end of the range. Let's call this factor 4.

The speakers are rated as 90dB efficiency at 2.83v/1m. Since these are 4 ohm speakers it means their efficiency is 87 dB at 1w/1m. They are rated for up to 180 watts, but as I understand that is very conservative and they can handle a lot more. These SB are mounted in corners giving them more dB boost. When using the SPL calculator with theses inputs and being 15.5 feet from the MLP, it takes 180 watts to reaches 105dB (giving it create for 2 channels in the pair playing together, otherwise if you count it as just one its 3dB lower), which is the top end of its "recommended amplifier power". Note that my amps are Emotiva XPA-5 Gen3 and into 4 ohm loads with 240v power deliver up to 550wpc.

OK so if we say it is delivering 180 watts to the SB when I have the MV at 0dB, then its probably sending something closer to 2x that or 360 watts due to Factor 2 above (raising the trim +3dB on the SB). Further, it we add in that Audyssey is boosting the signal by as much as +9dB more, the amp could easily be seeing all its got during peaks. And with these channels being very active in this movie, based on this math I think its easy to see how the SB tweeters got pushed over the edge.

So in summary, I think factors 1-4 all combined for the perfect storm to take out the tweeters in the SB. They sounded real flat (imaging no audio above about 3kHZ) and REW measures confirmed they were fried (more on that below). In the KEF the tweeter and midrange are in the same driver. After replacing that driver with a new one, they sounded great and REW confirmed my HF were back in action.

So that's the back story on how I had the two SB tweeters go on me. Of course now I am super paranoid about playing my system loud. I don't know where the breaking point is. Clearly I want to have a safe line in the sand. But I also don't want to be ultra conservative either. For instance I don't want to keep the volume lower than -5dB if in reality I am fine up to 0dB (with everything being EQ right and levels right). The challenge is I don't know how close to the edge of blowing tweeters I have been. Maybe for the past year I have been toeing the line and just a few dB away from the breaking point but never tripped it. Or maybe I am really no where close to it, if you take out a few of the 4 factors from the perfect storm. The challenge is I have no real way to know. But I MUST figure this out if I want to enjoy the system super loud while making sure I don't blow tweeters again. I should add that the room has acoustic stretched fabric for walls and an AT screen. All speakers are a major PITA to get to in order to service. So its not just the speaker repair but tearing things apart to get to them and then having to put them back together. I'd greatly appreciate any advice on how to better determine where the line in the sand is...

Now that brings us to yesterday... I was experimenting to find the best delay time to integrate the subs with the mains. But first I wanted to check all the Levels. To do this I had the MV at -8.5dB. When I was done I ran my first sweep on the front Main Left channel but forgot the volume was so high. My AVP is in another room so its not like I could see the volume displayed on the unit. I ran the sweep, and at that level it was too much and took out the tweeter in the front left main, as described in the opening post in this thread. The front main is a 5160 and recommended power is higher at "up to 250 watts". It however uses the same exact mid-range and tweeter as the 3160 used for the SB.

So I have now blown 3 tweeters in two separate cases in the past two weeks. This last case, with doing the frequency sweep with the MV so high, was really a fluke. I don't think this represents normal conditions the speaker are being operated on, as apparent these sweeps are brutal with the full signal going across the range. But perhaps there is some commonality here, so even tho I should have never run the sweep with the volume at -8.5dB, I'm not so sure that was high enough to take out the tweeters.

OK so how do I know the tweeters are blown? See the attached near-field measures from 16" from tweeter, on-axis, mic pointed straight at the tweeter, using UMIK-1 with 0 degree narrow band calibration file. (BTW I have tons of other measures and FR for the room, if anything else would help just let me know). You can see how the Center and Right channels go out to 20kHz. But look at how the left channel measures right after the sweep that blew the tweeter. The tweeter crosses over from the mid-range at 2.5kHz. You can clearly see how it drops right off starting in that region. Although these are the measures of the 5160, the 3160s with blown tweeters measured exactly the same way with this huge fall off.

What is really odd to me, however, is that the tweeter just seems completely dead. All 3 blown speakers measure the same exact way. Its either on and working, or completely off with no sound coming from it. Likewise with the frequency sweep that blew the tweeter, even that measure came out the same way. It's not like it measured out to say 8kHz or 10kHz and then died, and then in future sweeps measured nothing after 2.5kHz. So its almost like the power/current takes out the tweeter before it even starts playing. Almost like it has a fuse to protect itself. However upon examining it there appear to be no serviceable parts and no physical damage or melting etc. All wires are firmly attached to the coils and terminals and none are broken. Does this make any sense that the tweeter would just cutoff completely like this and identically on all 3 blown ones, as opposed to at least trying to play some sound, even if poorly?

My plan from here is to try and figure out what my real limits are in terms of MV, and also to stop Audyssey from adding so much boost to the HF. I will do that first by trying the Audyssey app, but that will have to wait until I replace my 8802A with the upcoming 8805 model in the spring. Then I can set a curve to get Audyssey to track the natural rolloff of the speaker better, perhaps adding up to +1dB to +3dB boost along the HF end to help with the acoustic fabric and AT screen the speakers are behind, but certainly nothing like the +5 to +9dB Audyssey is adding. That alone should buy me quite a bit more headroom. Tho until I am able to figure out what the real power handling and breaking point of these speakers, and whether these things were just flukes or areas I have to be very careful around, then a lot is still up in the air.

Thanks all for reading this far if you made it all the way through!

 

[Matthew J Poes]

Yow! I haven't heard of Audyssey providing so much boost it could fry tweeters like that. You may want to write Audyssey and/or Marantz to see what they say. Most correction systems prevent this. For example my speakers have a CD driver that breaks up at 19khz. 18khz is its upper limit. Audyssey does not try to fix that, it recognizes the problem. Usually there are DB boost limits in these systems to prevent it from doing what you are worried about. In addition, the tweeter would never see that power you are suggesting. Not only would it not draw that much to begin with for a given spl, but the boost you are suggesting would likely make the overall speaker sound louder (since our ears are pretty sensitive in the range it would be boosting I would think).

Now has it only blown after REW tests, or it blew from normal movies?

I will say this, the design and nature of the KEF speakers isn't of a type that is known to do full reference levels in large rooms well. I understand it met THX Ultra 2 standards, but a single dome tweeter has its limits. It is possible that your room size is large enough that you are just exceeding its capabilities. It is possible the Kef met the requirements barely and as such is being stressed passed its limits. Reference level equals 105db's if a movie is THX mastered. Lots of movies (even some THX movies) are exceeding those values during big dynamic peaks. It's a known trend and its certainly challenging a lot of speakers.

Honestly, if you are blowing tweeters during movies, I think you might just need louder speakers. something with more headroom. If you are only blowing them after taking measurements, it may be necessary to look at what you are doing during measurements. Remember that a measurement should be taken at around 75db's, no louder. As a first step, I think I would let Marantz, Audyssey, and Kef all know what is going on. This shouldn't be happening to you below reference levels with Audyssey. It implies something is wrong.

 

[lovingdvd]

Yow! I haven't heard of Audyssey providing so much boost it could fry tweeters like that. You may want to write Audyssey and/or Marantz to see what they say. Most correction systems prevent this. For example my speakers have a CD driver that breaks up at 19khz. 18khz is its upper limit. Audyssey does not try to fix that, it recognizes the problem. Usually there are DB boost limits in these systems to prevent it from doing what you are worried about. In addition, the tweeter would never see that power you are suggesting. Not only would it not draw that much to begin with for a given spl, but the boost you are suggesting would likely make the overall speaker sound louder (since our ears are pretty sensitive in the range it would be boosting I would think).

Now has it only blown after REW tests, or it blew from normal movies?

I will say this, the design and nature of the KEF speakers isn't of a type that is known to do full reference levels in large rooms well. I understand it met THX Ultra 2 standards, but a single dome tweeter has its limits. It is possible that your room size is large enough that you are just exceeding its capabilities. It is possible the Kef met the requirements barely and as such is being stressed passed its limits. Reference level equals 105db's if a movie is THX mastered. Lots of movies (even some THX movies) are exceeding those values during big dynamic peaks. It's a known trend and its certainly challenging a lot of speakers.

Honestly, if you are blowing tweeters during movies, I think you might just need louder speakers. something with more headroom. If you are only blowing them after taking measurements, it may be necessary to look at what you are doing during measurements. Remember that a measurement should be taken at around 75db's, no louder. As a first step, I think I would let Marantz, Audyssey, and Kef all know what is going on. This shouldn't be happening to you below reference levels with Audyssey. It implies something is wrong.

Thanks! I know it is a lot to read, but if you have a chance please go back through my last post, as I explain the history and details which answer your questions above (for example I explain under what conditions this has happened during movie playback (not just with REW) and what perfect-storm factors contributed to it).

I share your assessment that I am pushing the speakers near their limits for playing at reference in my size room. I was aware of this when I chose them. I do love their sound tho and would not consider changing them. Not to mention that my entire acoustic plan, room treatments and baffle wall etc are all entirely engineered around these speakers and their unique characteristics. I didn't think I'd want to play at reference when I planned the room, and generally I don't. If I knew that I should never go past -3dB and I would never have a problem, then I would be totally happy with that. But I wouldn't be happy drawing the line in the sand at say -7dB...

If these issues were happening from the beginning I'd be real concerned. But as I mentioned i the last post, I've gone a year with no issues despite doing a LOT of blasting. Just never quite as much as with that last time and all those perfect storm factors coming together.

Regarding Audyssey, I think the issue is that when it does its measurements it sees the farfield power response and that the mid and high frequencies taper off due to speaker directivity. As I understand it, that is perfectly normal and a lot of speakers and even horns do that. But Audyssey apparently tries to fllaten this power response to a flat line which results in excessive HF boost, at least in my case. See the attached aud-preout.gif. The red line is a measure my left channel pre-output from the AVP, as the baseline with Audyssey off. The green line line is the exact same measure except turning Audyssey on. You can see the boost and how it may be contributing (but not the sole cause) to the issue.

Also see aud-boost3.gif. The darker lines are the response with Audyssey off. The lighter color lines show that its boost has flattened that part of the curve. I understand that may not be favorable sound-wise, and certainly not dB output wise, especially when I am already close to pushing the limits of these speakers.

 

[Matthew J Poes]

I've never seen Audyssey do that before.

The roll off of the high is common for a number of reasons. The majority of speakers roll off to some degree off axis. Well done coaxial like Kef and waveguides do this in a smooth controlled manner. In addition most tweeters beam more at higher frequencies and most rooms are more absorbent at high frequencies. All of this conspires to roll off the highs.

The flattening by Audyssey is more extreme than I've experienced but I don't have your processor. I know it gets implemented a little differently in different products.

Could you take measurements with your mic pointed at the speaker and with a 0 degree file at 2 meters at different angles. I have a hunch but need to see it's response. Also take one with the mic in this position at your listening seat.

 

[Matthew J Poes]

Also please accept my apology for missing some of what you said. I did read it and thought I had read correctly. However I chose to ask the questions anyway so as not to provide useless feedback. I'm honestly shocked you are having this problem. Most modern speakers don't have issues hitting reference levels in the tweeter.

What if you used the tone control to bring the treble back down? Most work like a shelf filter. This would undo the excessive boost you mentioned.

 

[lovingdvd]

I've never seen Audyssey do that before.

The roll off of the high is common for a number of reasons. The majority of speakers roll off to some degree off axis. Well done coaxial like Kef and waveguides do this in a smooth controlled manner. In addition most tweeters beam more at higher frequencies and most rooms are more absorbent at high frequencies. All of this conspires to roll off the highs.

The flattening by Audyssey is more extreme than I've experienced but I don't have your processor. I know it gets implemented a little differently in different products.

Could you take measurements with your mic pointed at the speaker and with a 0 degree file at 2 meters at different angles. I have a hunch but need to see it's response. Also take one with the mic in this position at your listening seat.

You can imagine how adding this much boost can put the tweeter at risk, especially in a system where I am verging on pushing its limits even without the boost! As I understand it, I think Dirac may work differently and not try to flatten out the power response curve like Audyssey is doing. I have a question going in the Dirac thread on AVS about that, to see what those using Dirac would think I would see differently with it. I'm trying to stick with Audyssey tho because going with a dual MiniDSP 88A setup will add a lot of cost and complexity to my setup which I'd rather avoid. I'm thinking that the Audyssey app I'll have with the 8802A replacement in the spring may get me hopefully 80% or more of the way "there". For instance I know I can set a custom curve that should enable me to add some 1-3dB boost in the HF range but limit it to that.

My acoustic engineer designed the room and kept it more on the livelier side, relative to most of the rooms he does - because I like to use it for 2 channel music listening even tho that is lower priority that movie watching. It has a fair amount of absorption and diffusion and stretch fabric everywhere, and some mega bass traps. But perhaps these treatments and lack of lots of reflections are tripping up Audyssey and making it add so much boost. I'm not sure.

Regarding the measurements you are asking about - sure, I'm happy to take those measure and will post them tomorrow, err, i mean later today. I will take them with the mic pointed right at the tweeter and on eye-level with the tweeter. I know you want on-axis, but which approximate angles would you like to see (keeping in mind the angles will be approximate)?

In case I forget to mention it, keep in mind there is a Seymour Center Stage XD AT screen in front of the speaker. You can see its attenuation in the graph about 80% down this page http://www.seymourav.com/screens.asp . That screen is likely adding to this as well. Also keep in mind that ALL of the other speakers are playing through acoustically transparent stretch fabric. I have to find my measures on this but as I recall the fabric has a similar if not less rolloff than the screen. It would make sense that Audyssey should attempt to correct that SOME, perhaps about as much as the screen is attenuating, say up to +1 to +3dB, And I'd want it to, but no more than +3dB and certainly not +9dB!

Also please accept my apology for missing some of what you said. I did read it and thought I had read correctly. However I chose to ask the questions anyway so as not to provide useless feedback. I'm honestly shocked you are having this problem. Most modern speakers don't have issues hitting reference levels in the tweeter.

What if you used the tone control to bring the treble back down? Most work like a shelf filter. This would undo the excessive boost you mentioned.

No worries about the previous post. It was probably too long anyway. Just wanted to provide the full backstory.

Can you elaborate more on why you are shocked that my levels are too much for the tweeter? Is it possible that at higher levels the amps are sending "dirty" power and or otherwise at fault (or playing a role)? My amps are are Emotiva XPA-5 Gen3 amps (3 of them for the 15 channels, different amp (SpeakerPower) has subwoofer duties). It's not a specific faulty amp, because when the SB speakers that blew are on a different amp than the front L that blew. That said, perhaps there is something in their design that is not playing nice with my speakers or faulty in general (as opposed to a specific bad amp)?

Regarding the Treble control - yes I have played with it earlier as a possible workaround and measured the results from Treble 0 to Treble -6. It does a decent job undoing the excessive Audyssey boost. Attached here is aud-treble.gif which shows the HF response differences with treble 0 through treble -6. In all cases Audyssey is on. Looks like -3 or -4 could do nicely. HOWEVER there is a major catch. With the 8802A (and many other AVPs/AVRs I think), the Treble control only applies to the LEFT and RIGHT channels. I tested this and its the case for sure. So while it can be helpful for two channel music listening, its not a viable workaround because it leaves all other speakers vulnerable to the boost (not to mention I then would have a different HF curve for the mains than all the rest of the speakers). Great suggestion tho!

Thanks for all the thoughtful input. I really appreciate it!

 

[Matthew J Poes]

I'm shocked only because in recent years companies got good at designing speakers to be damage resistant. Blowing tweeters became a lot less common. Better crossovers, better tweeters, and the addition of protection circuits made it rare. Obviously not impossible though. Additionally, in my over 20 years in this hobby, I've never once blown a tweeter in a speaker, and I've done plenty of nutty things and listen plenty loud. I do watch movies at reference levels and have for some time with many different speakers.

I have Dirac as well. Can I answer your question? I've been using Dirac on and off for some time as I beta tested it when it first became available and had it for about a year and then recently reacquired it. I happen to prefer Dirac to Audyssey quite a bit. I think it sounds better and does a better job correcting. It too can apply too much boost though, that is up to you. You have to set the correction curve and could certainly do something a little silly. It had good protections to prevent that and I bet Flak (Flavio) could tell us if there are any kind of boost limits that minimize this. Certainly applying 6db boost could be a problem.

Do you run your speakers with the grilles on? Grilles also attenuate (and cause comb filtering). It may be good to remove the grilles to lessen the amount of total attenuation.

My theater uses spandex for the screen. After testing every material I could get my hands on, nothing was more acoustically transparent. The picture also looks great. I've owned the Seymour Screen you own and a number of others and previously used SI Black Diamond. In fact the frame I'm using is my old SI BD frame with the Spandex in place. Not that you should replace your screen, all materials attenuate to some degree. Not worth changing materials over 1 db or so difference. If your screen has a separate black layer that adds about 1.5 db average loss if I recall, right? Might be worth removing that and seeing if that helps (and how bad the effect on picture is).

So I will just tell you my theory. Speakers with a controlled directivity response fall off to the sides. It's intentional (and I know you know this). You actually want this behavior. It makes it so the sidewall reflections are lessened and the soundstage is more stable. However, to work as intended the speakers need to be angled so that their on axis response or close to it is at the listeners ears. With True CD you actually want to listen not on axis but slightly off-axis to the outside axis. For example, in the Geddes speakers I own, the listening axis is 22 degrees. Now he designed the speaker that way, but most designers actually design the speaker to be listened more on axis, including Kef. The proper listening axis depends on how directional it is, how quickly it rolls off. I am guessing that your speakers are in a baffle wall pointing straight ahead, correct? Not aimed at you or more. If you have a 15 foot listening distance to the speaker and the speakers are say 14 feet apart, then that is 25 degrees. If your 25 degree axis response is down 6db's from on-axis and that's your listening axis, then yeah, Audyssey is trying to correct that and shouldn't. My speaker is flat at 22 degrees so Audyssey wouldn't try to correct that. When you talked about the issue it hit me that this could be part of the problem.

Related to this is part of why I've never loved Audyssey. It's tonal balance is flat and bright. All the research shows that not to be the best response curve so why that is has always baffled me. I've told this story before around here, but I actually experienced Dirac before Audyssey. I made the wrong assumption that they operated in the same way (both using FIR filters above the transition zone) and thus would sound the same. I loved Dirac, I found it made everything better. I got Audyssey and actually sent back a number of processors because I was sure they were broken. I sent data to Audyssey telling them I was sure something was wrong. It turns out they just aren't the same and don't work the same, and as a result, don't sound the same. Not trying to sell you on anything, just letting you know my own opinion. I use Dirac with music, I don't use it for surround with movies as of right now because I use Dirac via the software. I would love to have a pre-pro in the future with built in Dirac so I can get the benefits in every scenario.

Too bad about the treble control, I didn't really think about how it would apply it. That is silly.

 

[lovingdvd]

I'm shocked only because in recent years companies got good at designing speakers to be damage resistant. Blowing tweeters became a lot less common. Better crossovers, better tweeters, and the addition of protection circuits made it rare. Obviously not impossible though. Additionally, in my over 20 years in this hobby, I've never once blown a tweeter in a speaker, and I've done plenty of nutty things and listen plenty loud. I do watch movies at reference levels and have for some time with many different speakers.

Thanks for that background. I think you will agree that KEF makes very high quality speakers so I do not think it is a problem with the speaker design. But rather something in my system that is pushing it over the edge. My feel at the moment is that it is a combination of pushing the system to play at near reference, which means the furthermost speakers (SB) need to play at reference from 15.5' from the MLP, combined with all the boost Audyssey is adding.

I have Dirac as well. Can I answer your question? I've been using Dirac on and off for some time as I beta tested it when it first became available and had it for about a year and then recently reacquired it. I happen to prefer Dirac to Audyssey quite a bit. I think it sounds better and does a better job correcting. It too can apply too much boost though, that is up to you. You have to set the correction curve and could certainly do something a little silly. It had good protections to prevent that and I bet Flak (Flavio) could tell us if there are any kind of boost limits that minimize this. Certainly applying 6db boost could be a problem.

The boost is +9dB. Yikes.

Do you run your speakers with the grilles on? Grilles also attenuate (and cause comb filtering). It may be good to remove the grilles to lessen the amount of total attenuation.

The grills are off. However, the LCRs are behind the XD screen and the rest of the listening bed speakers are being AT fabric. Both of these fabrics are adding to the slope. More on this below.

My theater uses spandex for the screen. After testing every material I could get my hands on, nothing was more acoustically transparent. The picture also looks great. I've owned the Seymour Screen you own and a number of others and previously used SI Black Diamond. In fact the frame I'm using is my old SI BD frame with the Spandex in place. Not that you should replace your screen, all materials attenuate to some degree. Not worth changing materials over 1 db or so difference. If your screen has a separate black layer that adds about 1.5 db average loss if I recall, right? Might be worth removing that and seeing if that helps (and how bad the effect on picture is).

I know what black backing material you are talking about and I am not using it. What gain do you get from the Spandex? The reason I picked XD is because it provided the highest gain for an AT screen I could find while also being the most AT-friendly. My screen is 140" 2.37 AR and with HDR I knew all the lumens I can get (and I'm still way short for proper HDR, but that's a separate discussion .

So I will just tell you my theory. Speakers with a controlled directivity response fall off to the sides. It's intentional (and I know you know this). You actually want this behavior. It makes it so the sidewall reflections are lessened and the soundstage is more stable. However, to work as intended the speakers need to be angled so that their on axis response or close to it is at the listeners ears. With True CD you actually want to listen not on axis but slightly off-axis to the outside axis. For example, in the Geddes speakers I own, the listening axis is 22 degrees. Now he designed the speaker that way, but most designers actually design the speaker to be listened more on axis, including Kef. The proper listening axis depends on how directional it is, how quickly it rolls off. I am guessing that your speakers are in a baffle wall pointing straight ahead, correct? Not aimed at you or more. If you have a 15 foot listening distance to the speaker and the speakers are say 14 feet apart, then that is 25 degrees. If your 25 degree axis response is down 6db's from on-axis and that's your listening axis, then yeah, Audyssey is trying to correct that and shouldn't. My speaker is flat at 22 degrees so Audyssey wouldn't try to correct that. When you talked about the issue it hit me that this could be part of the problem.

Yes, and your estimates and math is pretty close.

My baffle wall is straight so the speakers all point straight forward. My acoustic engineer and I considered a winged baffle wall to angle the L/R at the MLP. However it was deemed not to be worth it for a few reasons: 1) The KEF are very wide dispersion. At the MLP I am 22 degrees from the L/R. At this angle the FR is only down a max of -3dB. I think most speakers would be off considerably more than that at 22 degrees (for instance you mentioned -6dB which may be the case for many other speakers); 2) I would have needed to give up an extra foot or a bit more of room depth for the extra depth needed for the wings (baffle wall would have had to be deeper), and I was already running out of room to fit the 2nd row and back bar. So I wasn't willing to give that up; and 3) We preferred the extra rigidity and straight surface from one giant baffle wall rather than with the wings and the winged would have added to the build complexity and cost. Given the KEFs are only down about -3dB at 22 degrees we decided not to toe them in. Also I am not sure but I think KEF does not recommend toeing these speakers FWIW.

Related to this is part of why I've never loved Audyssey. It's tonal balance is flat and bright. All the research shows that not to be the best response curve so why that is has always baffled me.

If tonal balance of flat and bright is not ideal (and I agree with that in general), what type of curve is more ideal in your opinion? I know there has been a lot of talk about house curves for bass (I like mine about 7dB hot) but I haven't seem much about curve strategies for the HF. Do you have some info on that or links to some commonly recognized curves that are pleasing?

This is a little embarrassing, but I like the sound on the bright side. In my case it is too bright with some vocals.For instance at loud volumes if Prince screams in one of his songs it is so harsh I literally have to cover my ears or reach for the volume knob. So I think I will want to keep it a bit on the bright side, just not +9dB bright... Also I can only hear to about 14kHz, and I'm sure I am hearing less in the ranges leading up to that, so I think having things bright in terms of measurements helps things sound better to me. I've always wondered about speaker EQ calibration based on the listeners hearing, but never really came across anything about that. Just like I am saying in this case - if someone doesn't have great hearing due to growing older, can it make sense to use a brighter than normal curve if you only care about how things sound to that person in the room? Seems logical.

I've told this story before around here, but I actually experienced Dirac before Audyssey. I made the wrong assumption that they operated in the same way (both using FIR filters above the transition zone) and thus would sound the same. I loved Dirac, I found it made everything better. I got Audyssey and actually sent back a number of processors because I was sure they were broken. I sent data to Audyssey telling them I was sure something was wrong. It turns out they just aren't the same and don't work the same, and as a result, don't sound the same. Not trying to sell you on anything, just letting you know my own opinion. I use Dirac with music, I don't use it for surround with movies as of right now because I use Dirac via the software.

I was going to ask if you thought Dirac was really a night and day difference (I think most people say they like it better but harder to quantify how or why), but if you were sending back processors with Audyssey thinking they were defective, I guess that answers my question.

I would love to know just how much better Dirac would be for me. It would be a must-have due to the current issues with the HF boost XT32 is doing. However I will be getting the 8802A replacement when it comes out this spring, and that model supports the Audyssey app. The app now has a lot more customization and flexibility, closing the gap quite a bit on Dirac in terms of calibration options. For instance with the Audyssey app you can now set the point where it stops EQing, and can create custom curves for each speaker pair and the center separately. In my case I wonder if this will get me 80-90% there in terms compared to switching to Dirac.

Switching to Dirac is temping. But it adds a lot of complexity to my setup, compared to having Audyssey baked in, not to mention the cost, especially because I will need two MiniDSP 88A to do this due to my 9.4.6 setup. If it was purely a cost thing I may go for it. But the idea of adding two more units to my rack, all the XLR to Phoenix and vice-versa cabling on all those channels is a little daughtning. Especially because I can't be sure how much of a difference it would make. If Dirac in my setup would provide a noticeable and material benefit, then I would go for it. But if I can get 80-90% of the way there with the Audyssey app that would be good enough and favorable, to avoid all that complexity. I think I will wait and see what type of results I can get with the Audyssey app. If I'm not mostly satisfied at that point I will then likely get one 88A and play with it and see if I like it. Sell it if not, or buy a 2nd one for the ceiling speakers if so.

I would love to have a pre-pro in the future with built in Dirac so I can get the benefits in every scenario.

Well remember there is the in-between option - the MiniDSP 88A. Especially if you only need one (sorry I can't recall if you said you have an Atmos setup or not).

Too bad about the treble control, I didn't really think about how it would apply it. That is silly.

I will say this - and this is going back to what I said above about my preference for brighter sound - for two channel music, the treble control is a near-perfect anti-Audyssey Boost device. As shown in my last post, treble of around -6 with Audyssey on with 2 channels gives me the benefit of Audyssey in the low end, while completely undoing a good amount of its HF boost. But when I listen to music and movies like this at -6, it sounds SO FLAT I can hardly stand it. Maybe this is because I am used to it being around +6dB too bright, so comparatively it sounds so flat with treble at -6. Or maybe its because of my relatively poor hearing. When playing the with treble control, -3 sounds like a good compromise. This would mean for my room having a HF boost as much as +6dB (as opposed to Audyssey's +9dB). My engineer says he would never recommend more than +3dB for a soft dome tweeter, so +6dB still seems like it could put the tweeter at risk. Then again the difference between +6dB and +9dB is still twice the power, if I'm thinking about it right, so that is still a meaningful difference in terms of providing a little more safety and headroom for the tweeters, yes?

OK - circling back to why my HF rolloff may be more than usual and why Audyssey seems to struggle with it - I think its important to consider the affects of my XD screen and the acoustic fabirc. The XD screen creates about rolloff of about -5dB by the top end of the range. So if my off axis response from the L/R is already -3dB due to being 22 degrees off-axis, and then you drop another 5dB toward the high end of the range, that's -8dB rolloff right there. Interstingly I just checked and the max boost Audyssey applies to the Center channel is +5dB, hence about 3dB lower than the L/R HF boost, and why's that... because the Center channel does not have the same -3dB off-axis (since the MLP is on-axis to the center of course) as the L/R. That all seems to make sense and tie together well. Now for the rest of the speakers - they are all behind AT fabric. The measurements with/without the fabric in front of the speaker show it creates a -3dB to -4dB loss, plus most of these speakers are about 45 degrees off axis, that creates about a 4-6dB loss in the HF range - so that's a -7dB to -10dB combined loss for being behind fabric and off-axis. In this context it makes sense Audyssey trying to boost by +9dB. My only gripe with it is that it should allow you to specify a max because such boosts can put a tweeter at risk, and also that it is OK if the response is not perfectly flat - some rolloff is to be expected and perhaps desirable regardless of one's concern for protecting the tweeter.

Thoughts? Thanks!

 

[Matthew J Poes]

I agree with your conclusions as to why.

As to Kef and Toe, it would shock me if they didn't suggest toe. That wouldn't make a lot of sense. There are a ton of technical reasons for why they shouldn't think that, no good technical reasons not to toe. Sometimes people misinterpret a manufacturers advice to listen on axis as not toeing in. That isn't right though as we already pointed out.

Yes I think Kef makes great speakers and are well engineered.

I can't tell you the gain of my screen as I don't have an ability to test it. I would assume it's not as good as XD but have no idea. It's the same brightness from what I can tell as other similar screen fabrics but not as "shiny" as XD. It exceeds 16 ft Lamb of brightness on my 100" screen with an older JVC RS1 on low lamp. I've guessed it to be about .7-.8 maybe.

If you lived near by I would bring over Dirac and give you a demo for comparison.

Have you considered other processors? NAD or Arcam for example? They can do 7.2.4 and I imagine more is coming. They have dirac built in.

As for response curve, that's a whole ball of wax. Harman and BK did the best research on this and that's where I find the best suggestions. I like the Harman Curve or some variation of that. I'll try to find a reference and share. I like 6db of boost by 20hz and sloping up starting at 300hz or so. I like the treble from 1khz or -3db at 20khz relative to 1khz. Depends on the track and my mood.

 

[lovingdvd]

I agree with your conclusions as to why.

Thanks. This is all starting to make a lot more sense now. It seems that I have a system designed to play at or very close to reference just fine, but without consideration for EQ. Throw in a screen and AT fabric that creates up to -5dB rolloff and another 3-5dB of rolloff due to some speakers being off-axis, and I need as much as +9dB or so of EQ. Which then technically means I can play safely down to say -10dB (to leave room for the EQ boosts). That is not something I anticipated. Well, as I said I've played many movies as low as -3dB which is plenty loud and haven't had issues, so I think as long as I don't exceed that and generally stay in the -3 to -5dB MV range I should be OK.

As to Kef and Toe, it would shock me if they didn't suggest toe. That wouldn't make a lot of sense. There are a ton of technical reasons for why they shouldn't think that, no good technical reasons not to toe. Sometimes people misinterpret a manufacturers advice to listen on axis as not toeing in. That isn't right though as we already pointed out.

You could very well be right. It was a few years ago since I had these conversations and I could have mixed this up.

Yes I think Kef makes great speakers and are well engineered.

I can't tell you the gain of my screen as I don't have an ability to test it. I would assume it's not as good as XD but have no idea. It's the same brightness from what I can tell as other similar screen fabrics but not as "shiny" as XD. It exceeds 16 ft Lamb of brightness on my 100" screen with an older JVC RS1 on low lamp. I've guessed it to be about .7-.8 maybe.

Your estimate on gain seems about right. I tested gain with a light meter of many AT screens before picking XD. Most of them were in the 0.7 to 0.8 range, and typically advertised as around 1.0 gain. XD for references measures around 0.95 gain.

If you lived near by I would bring over Dirac and give you a demo for comparison.

Thanks - that is so kind of you! I see they have a demo, and I already have the UMIK-1 mic. But I'm clear how or if I could use this. I have a Windows 10 laptop that is pretty new but no blu-ray drive. Do they expect you to play back media through your laptop/PC in order to use the trial? I can't picture how to do that.

Have you considered other processors? NAD or Arcam for example? They can do 7.2.4 and I imagine more is coming. They have dirac built in.

Emotiva is working on a Dirac UMC-1 AVP that will do 13 channel Atmos and possibly 15 channel, plus the sub. But I'm a bit skeptical of a new product like that. I really like the Marantz line and features and would like to stay with it. It's possible it could make sense to consider another AVP tho.

As for response curve, that's a whole ball of wax. Harman and BK did the best research on this and that's where I find the best suggestions. I like the Harman Curve or some variation of that. I'll try to find a reference and share. I like 6db of boost by 20hz and sloping up starting at 300hz or so. I like the treble from 1khz or -3db at 20khz relative to 1khz. Depends on the track and my mood.

Thanks - would love some references / links to good discussions and example curves on the subject. Do you have some Dirac graphs or screenshots that show what your curve looks like as defined in Dirac, if convenient to post? It would be more intuitive I think for me to see the curve rather than trying to picture what you describe.

 

[Matthew J Poes]

I'll work on this tonight. I'm out with my family until later but try to post response measurements and references.

 

[lovingdvd]

I'll work on this tonight. I'm out with my family until later but try to post response measurements and references.

Great, thanks. Also I was do a little research on the Uni-Q driver which is the tweeter in my speakers. If you have a moment, please see this: http://us.kef.com/explore-kef/kef-innovation/uni-q . Note this in particular in the article:

The second advantage of Uni-Q is what we call 'matched directivity'. With the treble unit mounted at the centre of the bass driver's cone, its directivity (the spread of sound away from the main axis)... As a listener moves away from the main axis, the output of the treble unit falls off at approximately the same rate as that of the bass unit, thus improving the uniformity of tonal balance across the listening area, and improving the off-axis stereo imaging. The listener is not, therefore, as limited to a central 'sweet-spot' as with conventional speakers...

I think a point my engineer has tried to make to me is that by Audyssey blindly trying to flatten the natural RF rolloff, it is boosting the on-axis response of the speaker as well, which creates issues and perhaps is messing with the speakers natural design as far as the stereo imaging or other such aspects of the speaker's design...?

 

[Matthew J Poes]

Yeah the cone of a coaxial acts like a waveguide to help control directivity of the tweeter. As long as the crossover point is set so the directivity of both match at the crossover point you get a nice smooth roll-off to the sides. Usually a woofer cone is not a very good waveguide and it's throat is not able to match the tweeter transition smoothly. That is usually the case. Kef has spent a lot of time optimizing the cone profile and tweeter/cone transition to be as optimal as possible. They then added the tangerine phase plug and created an optimal coaxial. It's about as good as it gets. I still wouldn't say it's the equal of a true waveguide and CD but it's as close as a coaxial has ever been, while still being time aligned and point source.

 

[lovingdvd]

Yeah the cone of a coaxial acts like a waveguide to help control directivity of the tweeter. As long as the crossover point is set so the directivity of both match at the crossover point you get a nice smooth roll-off to the sides. Usually a woofer cone is not a very good waveguide and it's throat is not able to match the tweeter transition smoothly. That is usually the case. Kef has spent a lot of time optimizing the cone profile and tweeter/cone transition to be as optimal as possible. They then added the tangerine phase plug and created an optimal coaxial. It's about as good as it gets. I still wouldn't say it's the equal of a true waveguide and CD but it's as close as a coaxial has ever been, while still being time aligned and point source.

Thanks. Does it make sense boosting the on-axis frequency response, as Audyssey is doing, can "undo" the design of the speaker in terms of the balanced directivity that KEY so painstakingly designed and implemented?

 

[Matthew J Poes]

EQ can't change directivity. The change in response is the same at all angles. It can however change the intended tonal balance of the speaker.

What it can do is make it so the first reflections are boosted. This wouldn't generally be a good thing. Let's say normally the response is down 6 DB at the angle of incidence with the wall first reflection. If the direct signal is 74db then the first reflection is 68db. If you boost the highs by 6db then the on axis response is now 80db and the off axis is 74. The change with angle doesn't change but the absolute level does.

Since Audyssey is trying to flatten the response this wouldn't normally be a problem. It's only a problem now because you have a first reflection that may be as loud as your direct signal and you have a delayed rear wall reflection that's quite loud. It would dissipate with distance but it's still much louder than it was intended to be.

The point is, this isn't the right way to EQ the speaker and I can see it causing detriment to the sound in more ways than just tonal balance.

By the way, here is an article that explains this issue and why such speakers should be aggressively toed in for best. It's why I said I would be shocked for Kef to disagree. This phenomena is well known and science based.

http://www.libinst.com/PublicArticles/Setup of WG Speakers.pdf

 

[lovingdvd]

EQ can't change directivity. The change in response is the same at all angles. It can however change the intended tonal balance of the speaker.

Yes I guess that is what I was asking - if it can change the intended sound characteristics. Sounds like Yes.

What it can do is make it so the first reflections are boosted. This wouldn't generally be a good thing. Let's say normally the response is down 6 DB at the angle of incidence with the wall first reflection. If the direct signal is 74db then the first reflection is 68db. If you boost the highs by 6db then the on axis response is now 80db and the off axis is 74. The change with angle doesn't change but the absolute level does.

I'm not sure I totally follow that. Since both increase by 6dB why does it matter?

Also please note in case you think it changes anything - at the first point of reflection on both side walls there is a custom made (to my engineer's specs) absorption/diffusion panel. Also the ceiling has absorption/diffusion panels at the first point of reflection. So those are well treated areas. How much that helps with the situation I am not sure? The back wall also has some treatments.

Since Audyssey is trying to flatten the response this wouldn't normally be a problem. It's only a problem now because you have a first reflection that may be as loud as your direct signal and you have a delayed rear wall reflection that's quite loud. It would dissipate with distance but it's still much louder than it was intended to be.

The point is, this isn't the right way to EQ the speaker and I can see it causing detriment to the sound in more ways than just tonal balance.

Very interesting about the way Audyssey EQing things could be a detriment. With this in mind, what type of curve and boosts would you recommend (assuming that I had Dirac or the Audyssey app that supports custom curves) would be the right way to EQ the system in your opinion (what would a more ideal EQ curve look like)? I realize it is subjective and will require experimentation - but this will give me a better idea of what your approach would be.

By the way, here is an article that explains this issue and why such speakers should be aggressively toed in for best. It's why I said I would be shocked for Kef to disagree. This phenomena is well known and science based.

http://www.libinst.com/PublicArticles/Setup of WG Speakers.pdf

Very interesting, makes sense. Tho if I'm following it right, the primary benefit is increasing the sweet spot and or being able to sit off center but still enjoy great imaging. In my case I only care about one seat and its in the center. That said there may be other benefits, but it wouldn't have been worth not being able to fit my bar in the bar of the room (it was getting tight and any increase in the baffle wall depth, as would have been required to do the winged angles, would have required the extra depth).

 

[Matthew J Poes]

Ok so the increased highs with speakers set with no Toe causes the high frequency of the reflections to be raised too. Your absorbers ability to absorb is limited. It doesn't absorb 100% of the incident sound that hits it. It makes the reflection louder with that EQ which throws off the soundstage.

Here is a measurement Floyd Toole took of the reflection off an absorption panel. It shows what we would expect to see and what we actually see. If I recall the article correctly it's for a 2" OC703 panel.


If you think that is wrong, here is my custom designed panel. It's more than 6" thick in total including airgap and has a curved absorbing surface.

Keep in mind I don't have Harman resources so this measurement is only true love accurate down to about 2khz. Between 1 and 2 khz it's accurate but heavily smoothed. Below that I would largely ignore. Green line is the direct field and the yellow line is the reflection off the absorber. Amazing how much sound reflects off a wall with a thick absorber right? Gives you new perspective on the meaningfulness of 1.0 absorption coefficients.

Ok now for response curve
Look at Sean Olives experiment:
https://drive.google.com/file/d/0B9...cyLWEzZTAtMGJiODQ1ZTUxMGQ4/view?ddrp=1&hl=en#

And BK's preferred curves as examples:


And the JBL synthesis Curve here

 

[Matthew J Poes]

I just got a new laptop and haven't had time to do extensive measurements. I loaded Dirac and ran a test very quickly, just a single measurement point. Here it is so you can at least see the target curve I use. It's 5db's hot at 20hz and -3db's at 18khz (the upper limit of my speaker).



When I have a chance to do a full multi-point measurement and confirm with REW I can setup a separate thread and post so you can see.

Here is an example of what Dirac could do if you so choose, could be damaging to your tweeters potentially, and would sound terrible. It won't stop you from doing this (You can see the limits here, but you can still make this crazy curve. Now Dirac automatically reduces the output by a certain amount to give headroom. 6, 7, 8db's depending on the setting. That protects it from causing you to overload either the signal or having the typical problem of sending the excessive power we talked about earlier. However, if you set the volume such that its say 105db's at 1khz, then in this setting the treble is 10db's hot and Dirac can't protect against that.

 

[Matthew J Poes]

Here is another common curve you will see that I tend to like the sound of.

 

[lovingdvd]

Here is another common curve you will see that I tend to like the sound of.
View attachment 5274

Thanks! I'll dig more into those graphs and posts tomorrow. In the meantime I wanted to ask you an overarching question I have. With regards to getting rid of the flattening of the curve and using a custom curve that has some rolloff - I am concerned this is going to sound too dull when I drop the flat HF in favor of the more correct curve which allows for rolloff. Partially because of my HF hearing loss and partially because I tend to like things bright. Detailed comments about this concern are in post #10 above, here: https://www.avnirvana.com/threads/feature-request-please-help-save-our-tweeters.1843/#post-16839

Thanks again!

 

[Matthew J Poes]

Your concern is valid. I know a lot of guys who like a bright sounding system.

So first keep in mind that back to back bright often sounds better to people, over time a more neutral response with a slightly rolled off HF section tends to sound better. This is in general, your own preferences can vary.

Our ears adjust readily and so a more mellow system may sound just as good to you later on.

I mean normally if I was dealing with a client who had your situation I would probably be changing speakers for him. You can't and don't want to do that. If not for the blown tweeters I probably would have gone to custom EQ curves.

That leaves you with either leaving the EQ as is and keeping the volume down or removing the boosted highs and learning to like that.

Have you considered getting an Audyysey pro install with a custom curve. I have to imagine your processor supports the custom pro install.

Oh by the way I think I forgot to answer your dirac demo question. Yes they want you to use your computer as your source for the demo. Since you are just demoing the effect I would give it a try with music. Just play music through your laptop. If you must have a movie, maybe try Vudu or Netflix. Not the quality of Blu-ray but should give you the general idea. I use the software exclusively and I play all my music back through my computer. I use Tidal which gives me access to everything from 16 bit CD quality up to 24 bit 96khz HD audio quality.

While I agree that this correction helps with movies and music, some of the published studies have shown the audible benefit and perceived difference in this king of room correction is most evident with music and less so with movies. That has also been my experience. If the room is good and the speakers are good, room correction helps with the last 1%. With music it feels like it's more the last 10%. If the room is not good and the speakers have issues then I find all good room corrections make a much bigger difference. Especially Dirac.

 

[lovingdvd]

Your concern is valid. I know a lot of guys who like a bright sounding system.

Thanks for all the great graphs and additional information - its really helpful. Sorry for the long break since my last post - got busy with work.

So first keep in mind that back to back bright often sounds better to people, over time a more neutral response with a slightly rolled off HF section tends to sound better. This is in general, your own preferences can vary.

Our ears adjust readily and so a more mellow system may sound just as good to you later on.

Here's an interesting experiment I did over the weekend with two channel (plus subs) listening of Tidal MQA. I listened to various tracks with Audyssey on and the treble at -6. That combination gives me the benefit of Audyssey's sub and LF optimization, but with the HF boost completely remove (treble -6). It sounded fantastic. Perhaps a tad on the bright side, but just how I like it. Then I put the screen back on and left all settings/controls as they were. It sounded really poor. My screen attenuation grows from -1dB to -4dB or -5dB over the HF range according to my measurements (whereas the manufactuer's plot shows max -3dB). I guess then it is no surprise that the music sounded dull/flat with the screen in place and no HF EQ (treble still at -6, negating Audyssey).

Then I raised the treble control to 0. At this point it sounded great again, perhaps a little too bright. In looking at measurements from the MLP of with the screen removed and back in place, with Audyssey on and off, it seems that Audyssey is doing a great job actually correcting for the large HF rolloff that the screen is creating. If I'm being a bit picky, Audyssey is over-correcting by about 1-2dB. So if I was using a custom curve with Audyssey I would probably tell it to back off just a bit so that it is not adding that extra 1-2dB. I would also have it stop EQing at 15kHz since I can't hear that at all, so no point having it push +6dB or higher in that range.

Bottom line seems to be there - the screen is accounting for about 1-5dB loss across the HF range, and the AT stretch fabric in front of the rest of the speaker is doing about the same (maybe 1-4dB).

The main question that remains is whether I have to really reduce my MV by the same amount as in the top end boost. For instance if there is a +5dB boost at 5kHz and up, so I need to leave that headroom with the MV. I am pretty sure the answer is Yes. So if my SPL calculator shows I can hit 100dB with 180 watts (max recommended power) for my Surround Back channels, then really my max playback volume is 5dB lower, to account for the +5dB boost. So 95dB MV. Further, I can play it safe by giving up just a few more dB. For instance if I set it at 92dB instead, that only uses about 100 watts, which should leave me with good headroom to account for any content that may push the speakers too hard, or wiggle room in case some of my calculations are off, or if the dB efficiency is not quote what the spec says etc.

Ironically, 93dB is about the loudest level I am comfortable listening at when watching movies, and even a little lower for music. I've checked this recently with a SPL while watching movies and the peak in action scenes is low 90's (excluding the subs which are running +7dB hot). That is about my typical listening volume for the past year and I hadn't had tweeter issues until I went much louder and also includes the levels in the SB (which sent 4x power to the SB at already very high power levels - no wonder why the tweeters blew...). So the good part is that if I can listen "only" up to 93dB or so as my max SPL, I don't consider that much of a sacrifice or compromise, except perhaps in the rare event I really should want to crank things.

I mean normally if I was dealing with a client who had your situation I would probably be changing speakers for him. You can't and don't want to do that.

Right. Not to mention I really love the sound of the KEFs. I feel they are fantastic for movies and music. Maybe not as dynamic as some dedicated cinema speakers, but the KEFs sounded best to me. Plus they met my directivity requirements and were well suited for Atmos with my 7.5' ceiling. And at the time (not sure about now) they were about the only speakers that I could do in LCR, walls and ceiling and be perfectly voiced matches (same Uni-Q driver everywhere) and also completely hidden (behind screen, fabric etc).

If not for the blown tweeters I probably would have gone to custom EQ curves.

Can you please elaborate on that? What might such a custom curve for hearing loss look like? In my case my hearing loss seems to be mild below 2kHz, quite significant in the 2-7kHz range, and then less significant in the 8kHz range. But it seems clearly it would not be a good idea to add a boost too close to the 2.5kHz xo. One theory on why the tweeter blew is that having a boost above the xo still creates a lot of excess power requirements to the mid-range (to so the slope of the xo), which then carries over to the tweeter...

That leaves you with either leaving the EQ as is and keeping the volume down or removing the boosted highs and learning to like that.

Per the above, I THINK I am OK keeping the EQ and using a max SPL playback level in the low 90s - make sense?

Have you considered getting an Audyysey pro install with a custom curve. I have to imagine your processor supports the custom pro install.

Yes. However the Pro Kit is now obsolete. The newer Denon and Marantz receivers support the new Audyssey MultEQ app which allows for a lot more flexibility, like cutoffs and custom curves. With the 8802A replacement comes out in the spring it will have this feature and I will upgrade to it. Originally I was going to upgrade it mainly for the two additional Atmos channels it will support, but now having the Audyssey app is another benefit.

Oh by the way I think I forgot to answer your dirac demo question. Yes they want you to use your computer as your source for the demo. Since you are just demoing the effect I would give it a try with music. Just play music through your laptop. If you must have a movie, maybe try Vudu or Netflix. Not the quality of Blu-ray but should give you the general idea. I use the software exclusively and I play all my music back through my computer. I use Tidal which gives me access to everything from 16 bit CD quality up to 24 bit 96khz HD audio quality.

Yes, I'm not sure why I didn't think of that!

While I agree that this correction helps with movies and music, some of the published studies have shown the audible benefit and perceived difference in this king of room correction is most evident with music and less so with movies. That has also been my experience. If the room is good and the speakers are good, room correction helps with the last 1%. With music it feels like it's more the last 10%. If the room is not good and the speakers have issues then I find all good room corrections make a much bigger difference. Especially Dirac.

Yes I agree with that, with one cavet. If you have an AT screen and or speakers behind AT fabric, then the EQ becomes critical for correcting the attenuation caused by those materials.

Thanks!

 

[Matthew J Poes]

here are recent corrected measurements.


the outlier is no correction.


average of 6 measurements.

the bass needs a little more work but that's due to my three subs not being well integrated. I was playing with settings and lost my setup file for the sub amp. I have to find 4 hours to do it again. you get the idea.