The Audyssey Pro Installer Kit Thread (FAQ in post #1) - Page 73

Great chart! So for say, 40 Hz you could simply half the phase numbers and distance values, correct?(Since wavelength twice as long?)

Yes, really interesting. If you look at the attachment, the green line represents the bass measurement with Audyssey default distances. The difference between the trough below 60 Hz and the peak right below 70 Hz is approximately 30 dB. The table suggests a sub distance adjustment of approx 6 ft.

The red line represents the bass measurement after adjusting the sub distance from 11.7 feet to 18.0 feet, or 6.3 feet, which is the adjustment that produced the smoothest response. This was done before I saw the table, obviously. Uncanny that the table predicted this!

that's ok, i trust your memory more than mine.

This is how I understand it.

Phase issues only happen between two speakers playing the same frequency range. This only happens at the crossover of the sub/satellites in room. Since Audyssey only measures speakers individually and doesn't measure the crossover it has no information regarding the type or slope of the crossover between the subs and mains, or the phase changes within that crossover. Is it possible to calculate the phase interaction at the sub/satellite crossover without that information? I don't know. But I'm fairly certain Audyssey doesn't attempt to do it. So its not just about compensating for arrival times. Its an issue of compensating for phase changes along the crossover slope. That's where the manual distance changes come in.

Hopefully someone with a better understanding of phase cuts in, but that's my current understanding of the problem.

Makes sense I suppose, but then again Pro does go so far as to recommend the "best" crossovers, which implies it must have some insight into the crossover circuitry built into each Pro-enabled AVR and/or a view of how the speakers will interact within the crossover range, else what is it basing this ranking upon? I agree with your certainty that Audyssey isn't attempting to optimize subwoofer distances to correct for phase issues at the crossover, as evidenced by AJ & Keith's experiences with needing to alter Audyssey-calculated subwoofer distances/polarity respectively. Just seems like it should be possible.

Fortunately I've never had to deal with this additional complexity as my subs are literally a foot or two from the mains in both of my rooms. Which brings up yet another point--could the adjustments being made by AJ and Keith, which are helpful for LF & RF, be causing even greater issues for the remaining surround channels. Not that one would care for stereo music, but perhaps Audyssey is doing a grand compromise for SW distances based on all channels.

Way beyond my ability to understand at this point. :-)

If the subs are more or less equidistant from the MLP, correcting both would be the first shot. With XT32 the distances for each are known and are probably both affected by whatever threw Audyssey off. But, hopefully the table will be a good start. Notice that getting within a foot or two is all that is necessary so it's not worth a lot of effort to optimize.

I still wonder what is throwing Audyssey off. Since Audyssey can measure distances to a tenth of a foot, it should get the sub and main distances very close. Maybe as suggested a few days ago, Audyssey isn't taking the phase shift caused by the receiver's bass management into account or that phase shift isn't what Audyssey is expecting.


I don't know how Audyssey lines up the distances to put all of the satellites and subs into alignment at their crossovers, but I'd guess that there isn't enough phase difference in the center between 80 and 90 Hz (group delay) to affect the results. Keep in mind that 20 or 30 degrees causes only a fraction of a dB of ripple at the crossover and there is surely far less than this between 80 and 90 Hz. I.e., if Audyssey aligns all subs and satellites at, say, 80 Hz, then all is well because only the sub has a phase change (group delay) and only at lower frequencies. Or, only the frequency chosen for the sub alignment is critical (but not very critical since considerable error makes little difference).


Thanks!
Stuart[/quote]

I would think so.


I suspect that the dip at the crossover that we've seen occur occasionally is an issue that the Audyssey software doesn't know about. The software thinks it has aligned the sub and satellites but in reality there is 1) an erroneous distance measurement or 2) a phase shift, maybe in the receiver's bass management (high and low pass filters), that the software does not know about. Without a problem like this I'm sure the alignment would be correct.

There could be a dip at the crossover frequency due to a room mode that is substantial and that Audyssey cannot fully correct. If that is the only problem, i.e., the sub alignment is correct, then changing the sub alignment won't help.

If both an alignment problem and an excessive dip exist, fixing the alignment will help but won't correct the share of the dip that is due to a room mode.


Maybe. Need to think more about what to include.


That is the point, and it seems that it does usually work. The question is why occasionally it does not work. A while back we realized that an erroneous -3db (F3) point can be identified when a sub/room response does not roll off until below 20 Hz. Maybe there's something like that going on here. We've tested the Audyssey software on many thousands of rooms and maybe found some that Audyssey did not anticipate. Ones that prevent an accurate sub distance reading.

Maybe this is one we should take to Chris again. Though I think he has commented on it before saying only that one can adjust the sub timing to improve the blend if necessary. Did I recall that correctly? In any event, I'm sure it has not escaped him that this should probably be addressed in the future.

I would like to think it does take into account the things you outline, but I would guess the most troublesome issues it addresses are in what the room does to the sound between the speakers and the listening area. There may be issues there that dominate the choice of crossover. Audyssey talks of a "blending" that is done in the receiver software and done with greater refinement in the pro software and with it's more accurate mic. I have no clue what all is involved but I can envision phase and magnitude issues that surely make some crossover points better than others.

The crossover circuitry isn't engaged when the chirps are being measured by Audyssey, but hopefully are taken into account somehow.

Well I certainly don't want to beat you up as you have always been very helpful to me, and I understand exactly the general point you are making about the way the after graphs tell you little or nothing about what is happening at the MLP.

But wrt to the mammarial considerations of the male of the bovine species, I don't fully agree. For example, in my own recent case, it was the after graph for the bass that alerted me to a problem and set me on the road to eliminating it. So in that sense at least, the after graph was useful

Ever heard the term "useful idiot"? I'll grant you the term to apply to the Audyssey graphs.

@harrison...

nice job there... pm sent...

while it might have the data (not really sure whether it does or not), don't forget that it's software, and it doesn't "learn"...

software will only do what the original developer told it to do... it is possible (likely) that the algorithms only go so far, and it is possible (likely) that it's not actually doing a full "compare and contrast" of all possible settings, it's merely a bounded mathmatical construct... keep in mind that all "results" that audyssey shows you in the much maligned after graphs are "predicted"...

translating the above: keith's results aren't surprising... given all the variables involved (and the logic path that would be required for a "full" analysis), i would expect that the end result isn't always what we might "want it to be"...

hth...

Looking at these charts, comparing the sub freq to the high freq it appears to me that the sub freq are a couple db hotter then the rest of the spectrum. Sub appears to be around 72.5db but all frequencies above 500hz are below that, by a couple db.

The reason I bring it up is my charts tend to be similar (using OmniMic) and I have always felt that using 80hz crossover gives too much bass or localized bass as compared to running the fronts large which sounds more correct to m. When I compare the graphs it does look like I get 2-3db more output (and flatter response) when I use a crossover, telling me that my sub level is a bit too hot.

So I'm curious what you think of your bass. And do you use DynEq and with any offset?

^^^

yea, they do... that's one of the things on my "mental list" to think about and decide whether or not i care... right now, it's trending towards the "not care" zone, at least for now... although i'm going to experiment a bit later today with turning the subs down a bit more before doing the pro measurements and see if that has any effect...

the bass is not lacking... in quality nor quantity...

yes, i use deq all the time (not for the measurements though)... offsets vary with material, but pretty much follow the "standard guidelines" (0 for bd's, 5 for orchestral music, 10 for tv/other music, 15 for compressed to hell modern music, and occasionally sports)...

Isn't Audyssey going to boost the levels back up as part of the calibration anyway?

^^^

you'd think it would... but this goes back to the "trust, but verify" from several pages ago...

again, it's not something i'm particularly concerned about at this point... in practice right now, there's actually a bit more of a boost on the true lower end, as i'm forcing a different "eq program" within the subs themselves AFTER running pro...

a "preference vs. reference" issue here, probably... a few db hot on the bass never hurts my ears...

I really wish I could see the graphs on my phone with the forum app

For me, I only have XT and no Pro so I chalked it up to the mic being +/-2db. But who knows, maybe it is the OmniMic. So not sure which to trust more. When I saw your curves similar to mine but using Pro it peaked my interest. For me dropping the subs by 2db got the bass back in line with the measured response of running the mains full range, with the benefit of flatter response and more headroom/dynamics.

^^^

dunno... we'll find out a bit more later on today... i'm curious now...

i doubt that dropping the sub 2db is making any material difference in headroom/dynamics... in either case, i have no worries there anyway... headroom, i got...

if anything, lowering the xover on your mains could potentially have a much larger effect there...

edit: to be honest, the curves are very close to being within the error tolerance of the equipment...

Use Safari.

agree, that is what I meant. I was running mains full range because bass was too much....dropping sub 2db and crossing mains at 80hz is now something I like and yields more headroom (less strain on receiver).

And yes, 2db could very well just be measurement tolerance, but to my ears a 2db difference in sub level is easy to sense.

^^^

ah, gotcha... like i said, we should find out this afternoon... i'll play a bit while i'm "working"...

dunno if i'd trust my ears to hear a 2 db difference between the lower octaves and higher octaves, especially with that mess in the diffraction/diffusion zone... not exactly a smooth transition there...

also, as noted, a bit hot* tends to be the "norm" for me, so it's entirely possible i'm so used to it, it's "right" (pref vs. ref again)...

also, it could just be the tolerance... or the ominimic being placed a few inches different than the audyssey mic (it wouldn't take much)...

who knows? i'm happy... if i stopped now, i'd certainly be adequately satisfied... the only reason i want to experiment with a few things is because the geek in me wants to know....

sometimes, it helps to take a look at your omnimic graphs without audyssey... it can help regain a bit of perspective... it's safe to say that if you don't have one of those "coveted rooms", your response is leagues better than it's ever been before...

* a BIT hot... as in "a few db, with a bit more on the really low end"...

designing and sharing a tool this nifty gets him a dog 120...

gonna download it and play with it a bit later on...

Thx, but I'm on the other side of the DMZ...Android

rright, i should have said use your browser..

They work for me using the app on Android...Samsung Galaxy S

They sometimes load, sometimes not. But even when they do I can't zoom on images which makes them impossible to read. I guess I meant to say I wish they would load into the gallery or somewhere they could be manipulated. I'm on the Galaxy Nexus.

As to using the browser, yes I can. But its far more combersome than the app - takes longer, have to login, blah, blah, blah. I'm just too lazy for that...

@keith...

since even though you earlier swore off any more measuring and tweaking, but subsequently showed interest again... here is some "stuff to think about" while you gird your loins for yet another battle with aud pro/omnimic...

warning: long and possibly boring post coming up... also, others should feel free to comment if they see something i am missing and/or interpreting wrong... it wouldn't be the first time i've made a mistake, nor would it be the last...

hopefully, this will tie together a lot of the posts we've made in the last several days, at least in my mind...

now that you've got the concept of room modes down, and ways to "make them better", you can use that along with more info from the bob gold's calculator and your omnimic graphs to gain a more complete understanding of "what is going on in your room", along with "what is audyssey really doing to make it better?"...

using my room in the calculator (roughly 16x14x9, and slightly oddly dimensioned, but the numbers it gives are close enough for gummint work) and my graphs, here is what i think:

if you look past the room modes from the calculator down to the bottom right hand corner, you'll see some verbiage that looks like this:

*- No modal boost: 1hz to 35hz
*- Room Modes dominate: 35hz to 114hz
*- Diffraction and Diffusion dominate: 114hz to 456hz
*- Specular reflections and ray accoustics prevail: 456hz to 20000hz

now take the 10-200hz graphs from this post...

edited to include pics of those two graphs... second edit: note that the db scale is slightly different, the graphs were not "locked"... so account for this when looking at them... as noted in a subsequent post to jeff, i'll lock them next time... sorry...

no aud:


aud:


for this post, only consider the 10-200 graphs (edit for clarity: and only up to about 115hz), the calculated room modes and the non-modal/modal bullet points... next post will look at the others...

from the no aud graph...

up to 30hz or so, there's no modal interaction, and it's pretty flat as a result... it also shows that i'm getting a pretty healthy amount of room gain.. note that although it's flat, it's also 5-8db hot...

immediately after getting into the domain of room modes, there's a peak... axial room mode at work there....

this is followed by a very sharp drop back to essentially flat from 40-65hz, and very close to the target spl... a good thing!

next comes a big rise peaking out almost 8db hot from 75-85hz... combination of modes at work here...

then that sharp almost 20db drop from 85-95hz... that 7-10db suckout stays consistent till we get out of the modal zone...

ok, so there's where the problems lie, and why they are there... now, what does audyssey do about it?

from the aud graph...

up to 30hz, it knocks the room gain off completely, and it remains flat... that's good... easy to build a house curve off of that, if desired...

it knocks off that monster peak caused by the first room mode issue, and gets it down to "close to flat"... that's good... side note: that ringing at around 35hz is where my old anti-mode made it's money... knocking off that peak alone was worth the expenditure...

it fixes the little dips just past 40hz and 60hz, and levels that area even more... that's good...

it completely controls the following peak up to the xover and knocks about 5db off of it above the xover... it then brings up the wide dip/null that by an average of 5db, and smooths it some...

that's both "good" and "ok", and illustrates what we've proven over the past several days, that eq'd/tweaked subwoofers can be made to deal with room modes better than main speakers... while it appears audyssey can pretty much kill anything (within reason, and with some user intervention) in the modal area by use of the subwoofers, it can't when using the mains, although it still does a more than respectable job... side note: this is why i wanted to try a "higher than suggested xover"... i have a small thought that i might get even better results with a 120hz one, handing the modal zone entirely over to the subs... as i have flanking subs, localization would likely not be an issue...

the previous two paragraphs are supported by the various subwoofer tweaks (distance/phase) having sometimes profound effects in the xover/just above it area... since the subs are still contributing a pretty healthy portion of the energy at and just above the xover, it would make sense that changing their parameters would affect the excitation of room modes in that area...

takeaway from all of the above is something we all intuitively "knew", which is not only does subeq "work", but it "works really well", although occasionally it needs a little help from the user... when taken with post 2, it also shows something else we also all pretty much knew, that subeq is pretty much the "most important part of audyssey processing"... not surprising, as it addresses the biggest issue most home users have, at least when it comes to getting pretty graphs...

whew... post 2 (shorter, i promise ) to come...

^^^

i'm trying to avoid yardwork, thus the reason for the brain dump... part 2 may have to wait until later today, swmbo has stuck her head in more than once to bestow the you are going to get up off your can and come help me, right? look upon me, most recently with meaning... but i'll get to it later, i want to do a complete brain dump of my thoughts on this before i forget some of it... like i alluded to in the first post, if nothing else, this is a useful exercise for me to see if "i really understand what i think i understand"... edit: plus i didn't want keith to get the mistaken impression that he was anywhere near "done learning and tweaking"...

yea, please do... you have a sharp eye for this stuff and know audyssey inside and out, so i trust you'll pick out any inconsistencies in my logic... while i did stay at a holiday inn express last night, i'm pretty much self taught* when it comes to this stuff...

edit: when you look at the graphs, note that the db scale is slightly different, so you have to account for that... i should have locked it rather than using auto... sorry... next measurements, i'll remember to lock it so it's more easily interpreted...

* self taught meaning "paid a lot of attention over the years to posts from people who know and have worn out my copy of the master handbook of acoustics"... without guys like you on avs teaching me (patiently, one might add), i'd still have no real clue how to interpret pictures..

Truly fascinating, Chris. Thanks for taking the time to post this. I shall study the first post in some detail and see how I can apply the new-found knowledge here. I am greatly encouraged by my first proper attempt at using the OmniMic to do something useful - and that gives me the motivation to carry on along the learning curve.