22 gauge wire for surround sound speakers - AVS Forum
Forum Jump: 
 
Thread Tools
post #1 of 26 Old 08-20-2012, 10:43 PM - Thread Starter
Newbie
 
ggpr3's Avatar
 
Join Date: Aug 2008
Posts: 3
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
Hello,

I have Klipsch RB61 and RS42 speakers for surrounds as part of my home theater. I believe they are 8 ohm and my longest run is approximately 45 ft. Based off the chart linked below i should need 16 gauge wire. I have a bunch of 22 gauge wire that I could use to wire the surrounds. What would i be sacrificing or risking by using the 22 gauge wire? Is it worth it for those specific speakers being used as surrounds to have the 16 gauge wire. If it will be noticeably better or prevent any risk of damaging the equipment i will definitely get the 16 gauge. Any thoughts?

Thanks

http://www.roger-russell.com/wire/wire.htm#wiretable
ggpr3 is offline  
Sponsored Links
Advertisement
 
post #2 of 26 Old 08-21-2012, 12:11 AM
AVS Addicted Member
 
arnyk's Avatar
 
Join Date: Oct 2002
Location: Grosse Pointe Woods, MI
Posts: 14,382
Mentioned: 2 Post(s)
Tagged: 0 Thread(s)
Quoted: 753 Post(s)
Liked: 1167
Quote:
Originally Posted by ggpr3 View Post

Hello,
I have Klipsch RB61 and RS42 speakers for surrounds as part of my home theater. I believe they are 8 ohm and my longest run is approximately 45 ft. Based off the chart linked below i should need 16 gauge wire. I have a bunch of 22 gauge wire that I could use to wire the surrounds. What would i be sacrificing or risking by using the 22 gauge wire? Is it worth it for those specific speakers being used as surrounds to have the 16 gauge wire. If it will be noticeably better or prevent any risk of damaging the equipment i will definitely get the 16 gauge. Any thoughts?
Thanks
http://www.roger-russell.com/wire/wire.htm#wiretable


According to a standard wire gauge table,
http://en.wikipedia.org/wiki/American_wire_gauge

22 gauge wire has a resistance of 16.14 ohms per thousand feet or 1.612 ohms per 100 feet. A 45 foot speaker cable has 90 feet of wire in it, so the 100 foot number is indicative of your situation.

The best info I can find about the impdance of these speakers is given at

http://www.hometheater.com/content/klipsch-reference-rb-61-ii-speaker-system-ht-labs-measures

"
The RB-61 II’s listening-window response (a five-point average of axial and +/–15-degree horizontal and vertical responses) measures +1.27/–2.40 decibels from 200 hertz to 10 kilohertz. The –3-dB point is at 50 Hz, and the –6-dB point is at 45 Hz. Impedance reaches a minimum of 4.82 ohms at 174 Hz and a phase angle of –49.43 degrees at 106 Hz.

The RS-62 II’s three-face averaged response measures +5.53/–2.90 dB from 200 Hz to 10 kHz. The –3-dB point is at 78 Hz, and the –6-dB point is at 66 Hz. Impedance reaches a minimum of 4.08 ohms at 203 Hz and a phase angle of –62.17 degrees at 120 Hz.
"

We know pretty surely that these speakers vary over a range from no less than 8 ohms to as little as about 4 ohms at one or more points well within the audio range. If you drive the speakers directly from a good power amp, the voltage at the speaker terminals will whatever the amplifier delivers at that instant. If we include your 22 gauge wire then the voltage at the speaker will include the loss in the speaker wire which varies strongly (at least several dB) due to the variable impedance of the speaker and the series impedance of at last 1.4 ohms due to the 90 feet of 22 gauge speaker wire.

Compared to what the room is doing to the response of the speakers, what the 22 gauge speaker wire does is fairly mild, but it is probably pretty clearly audible in a close blind test.

If you switch to the general recommendation of using 12 gauge wire, the wire's resistance become 1.6 ohms per 1000 feet or 0.16 ohms per 100 feet. This causes much less than 1 dB of additional frequency response variation due to the wire and makes the question of frequency-dependent losses in the wire moot.

Wire resistance doubles for every 3 gauge change, so 14 gauge wire would be pretty good as well.
arnyk is offline  
post #3 of 26 Old 08-21-2012, 07:47 AM
AVS Addicted Member
 
Ratman's Avatar
 
Join Date: May 2002
Location: Collingswood, N.J.
Posts: 14,539
Mentioned: 1 Post(s)
Tagged: 0 Thread(s)
Quoted: 210 Post(s)
Liked: 299
Using the KISS principle, use 16 AWG for a 45' run to speakers. No worries.
Save the 22AWG in case you need to install a doorbell.
Ratman is offline  
post #4 of 26 Old 08-21-2012, 10:27 AM
Senior Member
 
has7738's Avatar
 
Join Date: May 2012
Location: North America
Posts: 297
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 24
Perhaps I could distill Arnie's post (excellent, but might be gibberish to the unfamiliar) to this: Smaller wire means less flat frequency response. Larger wire results in flatter frequency response. In general, flatter response is more desirable. So you should use the larger wire. 12ga is my universal choice for all installations. It never misses. If you had to use 14ga or 16ga, that would be "acceptable". 22ga is too light for speakers, at any time, except for very very (very very) short runs. or where you really don't care about the result at all, our where you really have to hide those nasty wires and you have no choice. If that's the case, see the next 'graph.

Now, here's the outta-left-field part... If your AVR has Audyssey MultEQ calibration, and you use it properly, small variances in frequency response due to wire (and anything else) will be equalized out. That doesn't mean it's OK to use 22ga with Audyssey, it means that whatever impact wire may have will be taken care of. The caution is, if you used hundreds of feet of 22ga wire, even though Audyssey may give you flat response, you won't have the ability to deliver flat acoustic power at all frequencies.

That's it. Tweak-head mode off now.
has7738 is offline  
post #5 of 26 Old 08-21-2012, 10:56 AM
AVS Addicted Member
 
Ratman's Avatar
 
Join Date: May 2002
Location: Collingswood, N.J.
Posts: 14,539
Mentioned: 1 Post(s)
Tagged: 0 Thread(s)
Quoted: 210 Post(s)
Liked: 299
I wouldn't use or recommend 18 - 22 speaker wire for ANY speaker wire application. I'm familiar with the tech and gibberish. (Sometimes we must separate the wheat from the chaff).

16 will be fine at 45' and is "acceptable"
14 is a CYA and probably a better choice for "run once" and peace of mind
12 is overkill, but .... if you want to invest the extra money/effort, go for it.

No matter.... go for 16AWG or better.
Ratman is offline  
post #6 of 26 Old 08-21-2012, 12:43 PM
AVS Special Member
 
Colm's Avatar
 
Join Date: Aug 2002
Posts: 4,652
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 89
Quote:
Originally Posted by ggpr3 View Post

What would i be sacrificing or risking by using the 22 gauge wire?
Frequency response will not be quite as flat and you lose a little volume.
Quote:
Is it worth it for those specific speakers being used as surrounds to have the 16 gauge wire.
Yes, read the rest of the page the table is on.
Quote:
If it will be noticeably better...
Depends on your ears.
Quote:
...or prevent any risk of damaging the equipment...
No risk of damage from the smaller wire.
Colm is offline  
post #7 of 26 Old 08-21-2012, 01:36 PM
AVS Addicted Member
 
arnyk's Avatar
 
Join Date: Oct 2002
Location: Grosse Pointe Woods, MI
Posts: 14,382
Mentioned: 2 Post(s)
Tagged: 0 Thread(s)
Quoted: 753 Post(s)
Liked: 1167
Quote:
Originally Posted by has7738 View Post

If your AVR has Audyssey MultEQ calibration, and you use it properly, small variances in frequency response due to wire (and anything else) will be equalized out. That doesn't mean it's OK to use 22ga with Audyssey, it means that whatever impact wire may have will be taken care of. The caution is, if you used hundreds of feet of 22ga wire, even though Audyssey may give you flat response, you won't have the ability to deliver flat acoustic power at all frequencies.

I agree with all of the comments in this post, including the part about Audessey/MCACC/YPAO probably taking out any serious audible variations due to too-thin wire, except in extreme cases.

Not a lot seems to be published about exactly what adjustment range those facilities have, but other evidence suggests to me that their range might be as little as +/- 6 dB. If this is the case, you might still hurt oversall SQ even with them in play, if the wire is too thin and too long.
arnyk is offline  
post #8 of 26 Old 08-21-2012, 08:06 PM - Thread Starter
Newbie
 
ggpr3's Avatar
 
Join Date: Aug 2008
Posts: 3
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
Thanks everyone, your help is very appreciated.
ggpr3 is offline  
post #9 of 26 Old 08-22-2012, 12:34 AM
Senior Member
 
has7738's Avatar
 
Join Date: May 2012
Location: North America
Posts: 297
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 24
Quote:
Originally Posted by arnyk View Post

Not a lot seems to be published about exactly what adjustment range those facilities have, but other evidence suggests to me that their range might be as little as +/- 6 dB. If this is the case, you might still hurt oversall SQ even with them in play, if the wire is too thin and too long.
Audyssey's adjustment range is +/- 9dB. Their reasoning relates to over tasking power amps if more adjustment range was provided. Should take care of any average amount of under-sized wire with wildly complex speaker loads, though as I said before, using bigger wire is a far better solution. It's interesting, though, that something like Audyssey would eliminate any differences imposed by exotic speaker wire too. That expensive coat-hanger stuff? Or that $50/foot braided wire? Nope, whatever it does, it's EQ'ed out by Audyssey! Which I think is VERY cool!
has7738 is offline  
post #10 of 26 Old 08-22-2012, 04:17 AM
AVS Addicted Member
 
arnyk's Avatar
 
Join Date: Oct 2002
Location: Grosse Pointe Woods, MI
Posts: 14,382
Mentioned: 2 Post(s)
Tagged: 0 Thread(s)
Quoted: 753 Post(s)
Liked: 1167
Quote:
Originally Posted by has7738 View Post

Audyssey's adjustment range is +/- 9dB. Their reasoning relates to over tasking power amps if more adjustment range was provided. Should take care of any average amount of under-sized wire with wildly complex speaker loads, though as I said before, using bigger wire is a far better solution. It's interesting, though, that something like Audyssey would eliminate any differences imposed by exotic speaker wire too. That expensive coat-hanger stuff? Or that $50/foot braided wire? Nope, whatever it does, it's EQ'ed out by Audyssey! Which I think is VERY cool!

I appreciate your comments about how Audessy is implemented. What other details are known? Where are they publicized? What kind of equalizers (bands, Q, etc.?)
arnyk is offline  
post #11 of 26 Old 08-22-2012, 12:29 PM
AVS Special Member
 
JHAz's Avatar
 
Join Date: Mar 2009
Posts: 3,978
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 102 Post(s)
Liked: 157
Quote:
Originally Posted by arnyk View Post

I appreciate your comments about how Audessy is implemented. What other details are known? Where are they publicized? What kind of equalizers (bands, Q, etc.?)

In the past in the Audyssey thread Chris K from Audyssey provided a fair glimpse into how the system worked, but there's clearly a lot he would not get into, presumably because the company viewed the info as valuable trade secrets. He posted as Audyssey and a person could get a lot of info if they cared to search for that poster and scroll through a few hundred posts. I;'m pretty sure the 9 dB boost is from that particular horse's outh. I don't hink there's any limit on cut in Audyssey.

Graphic EQs and parametric EQs use infinite impulse response filters. Audyssey does not. It uses Finite Impule Response filters, which work far differently but I'll confess right here I don't really grok the differences. But you kind of can't talk about bands and Q with FIR filters, as far as my limited understanding goes, because the concepts don't apply, at lest to the individual filters. That's why there are hundreds of them for each channel. They just work differently. Given your background and level of understanding you might well be able to teach me a thing or 2 about the difference between IIR and FIR, but be warned that my dyslexia makes some concepts really hard to grasp . . . Here endeth the utterly uninformative non-lesson. Let's sing.
JHAz is offline  
post #12 of 26 Old 08-22-2012, 11:09 PM
Senior Member
 
has7738's Avatar
 
Join Date: May 2012
Location: North America
Posts: 297
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 24
Quote:
Originally Posted by arnyk View Post

I appreciate your comments about how Audessy is implemented. What other details are known? Where are they publicized? What kind of equalizers (bands, Q, etc.?)

There isn't a single document that details Audyssey, and getting info out of them is hard to do. They believe they have the definitive auto EQ system, and aren't about to tip their hand. I got the 9dB figure from someone on the inside, very inside. I'm an Audyssey Registered Installer.

Frankly the deep tech details are still veiled in a bit of fog, but I've been through their training, talked extensively with the key people, used the Pro version and software many times on installations, and seen the resulting response graphs (annoyingly shown without any X or Y scale so you can't even think of duplicating the resulting curve....!!!) I can say their strong points are using "fuzzy clustering" math to vastly improve the low frequency resolution of a swept "chirp" measurement.

A standard linear FFT has most of its data points above 1KHz, and taking the necessary spacial/temporal average with an FFT based RTA, no matter how tight the filters, reduces the resolution of the measurement. This integration is necessary for an FFT-RTA to make sense out of otherwise seemingly random results, but it really is reducing the total measurement resolution by averaging. By comparison, the Audyssey system places lots of data points in the LF region where it's most needed, and measuring chirps in multiple locations actually increases the measurement resolution, because of the way fuzzy clustering crunches the data. It's NOT an average. The resulting filters are highly precise, being basically not limited by constraints of Q, frequency or gain (except for the +9dB limit of course) and time. And, it is also my understanding that the system takes into account time-related issues, though honestly I had someone disagree with that once, but it makes sense if you take time domain data and generate time domain filters you deal with time and frequency domain issues both at once.

Because of their measurement and filter generation method, however, it's almost impossible to verify an Audyssey result with an RTA or conventional FFT. The only system that comes close is FuzzMeasure. But that's most likely why you'll hear some detractors state that what Audyssey claims to do is impossible, particularly when it comes to compensating for acoustic issues.

To be fair, I've seen Audyssey do some strange things too, but generally, at least when using the Pro software, a new measurement set fixes things. I've also had at least one situation where the result wasn't audible when switched on or off, but determined that was because the system was good to begin with, and the correction was subtle because that's all that was needed. It's more dramatic when dealing with bad systems and rooms than good ones.

I know that doesn't help much, but even for installers they don't give much info out. If the thing wasn't so darn good, I'd have a big problem with this, but since it does work, and works better than most others IMHO, they can go ahead and keep their secrets, I don't really care much.
has7738 is offline  
post #13 of 26 Old 08-23-2012, 11:02 AM
AVS Addicted Member
 
amirm's Avatar
 
Join Date: Jan 2002
Location: Washington State
Posts: 18,224
Mentioned: 3 Post(s)
Tagged: 0 Thread(s)
Quoted: 865 Post(s)
Liked: 516
Quote:
Originally Posted by has7738 View Post

Now, here's the outta-left-field part... If your AVR has Audyssey MultEQ calibration, and you use it properly, small variances in frequency response due to wire (and anything else) will be equalized out. That doesn't mean it's OK to use 22ga with Audyssey, it means that whatever impact wire may have will be taken care of.
While what you say can be generally true, in this specific instance it may not be a sure thing. A thin wire will have variable effect on the frequency response of the speaker as noted due to its dynamic impedance. Unfortunately this gets overlaid on the most complex part of the room measurement where room reflections already present a pretty wild looking response. Here is an example I have handy:

i-hNSV6pf-M.png
The auto eq system has to make sense of this highly varying data. It needs to separate what is correctable and what is not. The logic here is not deterministic (hence the reference to "fuzzy" analysis by Audyssey) and at any rate, not documented. It is for example thought that Audyssey attempts to correct what it perceives to be directivity issues in common consumer speakers (where the reflected sound has differing response than direct sound). It does so by boosting the mid-level frequencies. Such a logic may get triggered by insertion of that thin wire, and not without. Audyssey also has to be mindful of the variations in cheap mics provided by AVR manufacturers so I suspect it will not be anal about trying to achieve a "flat" response of any sort.

I took the above measurement for a similar application: how to compensate for the effect of video screen material in front of speakers ("2K" refers to the name of the screen there). The auto EQ system I am using, JBL Synthesis ARCOS/SDEC-4500, has the proper solution here in that it has a library of screen material that has been measured in the lab. Once you tell it that, it uses a set of "sweetening" filters that are overlaid on the room EQ results as global filters to compensate. As such, it does not attempt to auto detect them and be subject to vagaries of whether it gets them right. Going a significant step further, you can also tell it the speakers you have and if they are within the list of the ones in their library (which are limited to Harman brands such as Revel/JBL), it will take into account their measurements also in figuring out what to do with the above graph.

Probably way, way more than OP wanted to know smile.gif. My recommendation to him is to just use that 22 gauge wire. For movie sound, who is to say the modified version of the sound effects in the rear channel is less or more like what the spaceship produced??? biggrin.gif

Amir
Retired Technology Insider
Founder, Madrona Digital
"Insist on Quality Engineering"
amirm is offline  
post #14 of 26 Old 08-23-2012, 11:22 AM
AVS Addicted Member
 
amirm's Avatar
 
Join Date: Jan 2002
Location: Washington State
Posts: 18,224
Mentioned: 3 Post(s)
Tagged: 0 Thread(s)
Quoted: 865 Post(s)
Liked: 516
Quote:
Originally Posted by JHAz View Post

In the past in the Audyssey thread Chris K from Audyssey provided a fair glimpse into how the system worked, but there's clearly a lot he would not get into, presumably because the company viewed the info as valuable trade secrets. He posted as Audyssey and a person could get a lot of info if they cared to search for that poster and scroll through a few hundred posts. I;'m pretty sure the 9 dB boost is from that particular horse's outh. I don't hink there's any limit on cut in Audyssey.

Graphic EQs and parametric EQs use infinite impulse response filters. Audyssey does not. It uses Finite Impule Response filters, which work far differently but I'll confess right here I don't really grok the differences. But you kind of can't talk about bands and Q with FIR filters, as far as my limited understanding goes, because the concepts don't apply, at lest to the individual filters. That's why there are hundreds of them for each channel. They just work differently. Given your background and level of understanding you might well be able to teach me a thing or 2 about the difference between IIR and FIR, but be warned that my dyslexia makes some concepts really hard to grasp . . . Here endeth the utterly uninformative non-lesson. Let's sing.
I unfortunately missed that long thread when it was going on with Chris participating. But have read it after the fact. Unfortunately I don't think the right case was presented there. Much of what I recall Chris talking about with respect to IIR/FIR filter differences was really about Graphic EQs with their poor resolution independent of how they are implemented. Indeed you don't want to use graphic EQs because in low frequencies they don't have fine enough resolution to be useful. And at any rate, most of very cheaply implemented and don't do what their labels say they do on mass market gear.

IIR filters are cheaper to implement computationally so all else being equal, they can have higher power than FIR. The problem with them is that they are harder to design to be accurate. Good IIR implementations do exist and are used in top performing Auto EQ systems, two of which in published tests have beaten Audyssey in blind tests. Here is the summary results of that research:

Room%20Correction%20Preferences.png

Audyssey is the rightmost one, losing out to no EQ (i.e. degraded the response). #1, #2 are both from the same system using IIR. #5 is also IIR based. We see that they all outperform Audyssey in this test. The AES paper digs into potential reasons why ("The Subjective and Objective Evaluation of Room Correction Products", Audio Engineering Society 2009 Conference paper).

Roger Dressler put it best when we had this discussion on another forum: there is no data that points to choice of filter type being a deciding factor. But rather how the Auto EQ makes critical decisions relative to what should be corrected and what not per my other post.

The notable accomplishment of Audyssey is to use a multi-rate filter system where the computational burden is kept much lower enabling implementation at low cost in mass consumer products. As was just explained by another poster, in low frequencies one needs higher resolution and using traditional FIR filters would make the solution very expensive. So splitting the analysis into different frequency ranges and then fitting it in common DSPs in AVRs is clever and key to their success. From absolute fidelity point of view, I am not sure they have advanced the art there. Even in the Pro version which I used extensively I found the system to be lacking and very opaque in what it attempts to do (e.g. not providing high resolution post correction measurement).

My apology to OP for continuing this sidetrack smile.gif.

Amir
Retired Technology Insider
Founder, Madrona Digital
"Insist on Quality Engineering"
amirm is offline  
post #15 of 26 Old 08-23-2012, 11:52 PM
Senior Member
 
has7738's Avatar
 
Join Date: May 2012
Location: North America
Posts: 297
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 24
Quote:
Originally Posted by amirm View Post

While what you say can be generally true, in this specific instance it may not be a sure thing. A thin wire will have variable effect on the frequency response of the speaker as noted due to its dynamic impedance. Unfortunately this gets overlaid on the most complex part of the room measurement where room reflections already present a pretty wild looking response. Here is an example I have handy:
i-hNSV6pf-M.png
The auto eq system has to make sense of this highly varying data. It needs to separate what is correctable and what is not.
Actually, the system needs to first ignore response excursions that are specific to one measurement location and process measurements in multiple locations in a way that reveals response excursions that are not location specific.
Quote:
Originally Posted by amirm View Post


The logic here is not deterministic (hence the reference to "fuzzy" analysis by Audyssey) and at any rate, not documented.
Undocumented, yes. "Not deterministic"? Well, by definition, "deterministic" means "the doctrine that all facts and events exemplify natural laws". It would seem Audyssey, and all other attempts at room compensation would have to be deterministic. The "fuzzy" part you refer to is a portion of the term "fuzzy clustering". We are SOOO far off topic here, there's little point it attempting an original explanation. Google it. The end result is pretty deterministic.
Quote:
Originally Posted by amirm View Post


It is for example thought that Audyssey attempts to correct what it perceives to be directivity issues in common consumer speakers (where the reflected sound has differing response than direct sound). It does so by boosting the mid-level frequencies. Such a logic may get triggered by insertion of that thin wire, and not without.
And now we're into here-say, and theory of the unfamiliar. If the composite response, wire included, results in a response dip at a point of low impedance, Audyssey would sense and properly correct for it for several reasons. First, that dip would be over quite a significant area of the total response, fairly low Q as compared with other response anomalies. That dip would also be similarly present at every measurement position, and even if not equal at every position, fuzzy clustering would recognize the similarity and assign it significance quite easily. It's simply not going to cause the system to make an error.
Quote:
Originally Posted by amirm View Post


Audyssey also has to be mindful of the variations in cheap mics provided by AVR manufacturers so I suspect it will not be anal about trying to achieve a "flat" response of any sort.
If you've had a chance to compair even very inexpensive condenser capsules, you'd know that the response variation between samples is minuscule with respect to the response of speakers in rooms, or for that matter, small-wire induced response errors. Having visited the Audyssey lab, and seen a rig for testing measurement mics, I can say mic response is under scrutiny, though I cannot say that each mic is tested, or that the curve is compensated for somewhere. Regardless, mic variation is perhaps the least of the variables in any system, other than the purely electronic.
Quote:
Originally Posted by amirm View Post


I took the above measurement for a similar application: how to compensate for the effect of video screen material in front of speakers ("2K" refers to the name of the screen there). The auto EQ system I am using, JBL Synthesis ARCOS/SDEC-4500, has the proper solution here in that it has a library of screen material that has been measured in the lab. Once you tell it that, it uses a set of "sweetening" filters that are overlaid on the room EQ results as global filters to compensate. As such, it does not attempt to auto detect them and be subject to vagaries of whether it gets them right. Going a significant step further, you can also tell it the speakers you have and if they are within the list of the ones in their library (which are limited to Harman brands such as Revel/JBL), it will take into account their measurements also in figuring out what to do with the above graph.
The graph has no specific measurement condition data included with it, but it has the look of a single measurement location, FFT-based, swept-sine measurement, which is frankly useless on its own anyway unless done in an anechoic chamber. The graph is irrelevant to this discussion anyway, as it doesn't show the results of even the most rudimentary processing, like multi-location averaging.
[/quote]
Quote:
Originally Posted by amirm View Post


Probably way, way more than OP wanted to know smile.gif. My recommendation to him is to just use that 22 gauge wire. For movie sound, who is to say the modified version of the sound effects in the rear channel is less or more like what the spaceship produced??? biggrin.gif
Yes, we' are far from home here. But I do find your conclusion of "just use the 22 gauge wire" a bit shocking coming from someone with some rather authoritative articles published on his web site. And I doubt many here would agree with it.
has7738 is offline  
post #16 of 26 Old 08-24-2012, 08:23 AM
AVS Addicted Member
 
amirm's Avatar
 
Join Date: Jan 2002
Location: Washington State
Posts: 18,224
Mentioned: 3 Post(s)
Tagged: 0 Thread(s)
Quoted: 865 Post(s)
Liked: 516
Quote:
Originally Posted by has7738 View Post

Undocumented, yes. "Not deterministic"? Well, by definition, "deterministic" means "the doctrine that all facts and events exemplify natural laws". It would seem Audyssey, and all other attempts at room compensation would have to be deterministic. The "fuzzy" part you refer to is a portion of the term "fuzzy clustering". We are SOOO far off topic here, there's little point it attempting an original explanation. Google it. The end result is pretty deterministic.
While not relevant to the way I used the term, I do not consider Audyssey process deterministic. Using a single Mic location I have run both the Pro and Consumer versions and each time arrived at different results. In the case of the Pro version for example, it would complain about speaker phase reversal in one independent run, and not the next one. The unit was provided as part of the larger system from another manufacturer and when we asked them about this their answer was that "it does that!" And that we can ignore the message.

But again, that was not the context in which I used the term. I was explaining that relying on it reversing the effect of wire changing the system response is not deterministic. Let's say Audyssey determines there is a problem with power response if at 2000 Hz the response is 3 db below the average. You measure the system with a fat wire and it is at 2.5 db below average so no decision is taken by Audyssey to compensate. Now you change the wire to a thin one and for the sake of argument, the wire causes a 1 db drop at 2000 Hz and we are now at 3.5 db below average. As a result Audyssey applies a boost of a few db to compensate. This compensation is not proportional to change that the wire introduced since Audyssey has no idea that was the cause. It simply goes by the sum total of what it "hears." So when I say it is not deterministic, I mean in the context of one using his assumptions of the system and how it works. It just doesn't work as simply and predictably as we assume it does.
Quote:
And now we're into here-say, and theory of the unfamiliar. If the composite response, wire included, results in a response dip at a point of low impedance, Audyssey would sense and properly correct for it for several reasons. First, that dip would be over quite a significant area of the total response, fairly low Q as compared with other response anomalies. That dip would also be similarly present at every measurement position, and even if not equal at every position, fuzzy clustering would recognize the similarity and assign it significance quite easily. It's simply not going to cause the system to make an error.
You could very well be right. My point is that based on everything I know, there are enough heuristics that make the assumption subject to potential error. if such frequency response variations were large, then I would be with you. But with wire changes? I am not so sure smile.gif.
Quote:
If you've had a chance to compair even very inexpensive condenser capsules, you'd know that the response variation between samples is minuscule with respect to the response of speakers in rooms, or for that matter, small-wire induced response errors.
That is not relevant to my argument. I said that there can be no aspiration for a flat response since the measurement mic is too cheap and inaccurate to provide that correct view. And due to variations in mic, the system has to further assume that its targets for flatness cannot be relied upon. As you correctly state, for the most part this does not interfere with room equalization since the swings can be quite large and at any rate, we can make relative corrections and still be OK. A 15 db peak at 70 Hz needs to be corrected as an example regardless of how inaccurate the mic might be. As you know for the Pro systems they provide a much better mic and with calibration data so the problem with arriving at accurate and flatter response is an accepted one with the consumer mics provided with AVRs.
Quote:
The graph has no specific measurement condition data included with it, but it has the look of a single measurement location, FFT-based, swept-sine measurement, which is frankly useless on its own anyway unless done in an anechoic chamber. The graph is irrelevant to this discussion anyway, as it doesn't show the results of even the most rudimentary processing, like multi-location averaging.
The graph is what any single measurement would indicate to the system the overall response is. I post it to show that there is no pretty picture of a response there that makes it abundantly simple to see that there is an effect due to wire, or speaker, or room acoustics. It is a jungle of data which has to be made sense out of. There are different methods for trying to remove the "grass" from it and extract the true characteristics of the system. The best way to do that is not averaging by the way but to apply the ERB of the ear to it, taking into account that the ear resolution sharply reduces as frequencies go up. This is the subject of my upcoming article in Widescreen Review Magazine (should be on newsstands in September). Indeed, that graph is part of that article. In there I show further transformations of that measurement and how those could be more appropriate representations of "what we hear."
Quote:
Yes, we' are far from home here. But I do find your conclusion of "just use the 22 gauge wire" a bit shocking coming from someone with some rather authoritative articles published on his web site. And I doubt many here would agree with it.
I am a pragmatist. If I had run that wire behind the walls and had no other option than to use them for surrounds, I would. I am not recommending anyone go and deploy 22 gauge wire on purpose. Indeed, I am on record with the proper way to analyze the topic and determine what gauge should be used. See http://www.avsforum.com/t/1379661/does-expensive-speaker-and-interconnect-cables-really-make-a-difference#post_21347201, and then this follow up post: http://www.avsforum.com/t/1379661/does-expensive-speaker-and-interconnect-cables-really-make-a-difference/60#post_21354966

Anyway, thanks for the reply. I did not mean to turn this into a big argument smile.gif. Just thought since there was interest in how auto EQ systems work, that we use the current context to dig deeper.

Amir
Retired Technology Insider
Founder, Madrona Digital
"Insist on Quality Engineering"
amirm is offline  
post #17 of 26 Old 08-24-2012, 09:25 AM
AVS Special Member
 
commsysman's Avatar
 
Join Date: Dec 2007
Location: Southern California
Posts: 5,290
Mentioned: 2 Post(s)
Tagged: 0 Thread(s)
Quoted: 134 Post(s)
Liked: 257
If you use 22 gauge wire for a 40-foot run the wire has a higher resistance than the speaker, so more power will be delivered to the wire than the speaker; that is not good.

You need 14 gauge wire to ensure that at least 90% of the amplifier output reaches the speaker.

The idea that using smaller wire affects the frequency response is idiotic; wire has the same resistance at all frequencies.
commsysman is offline  
post #18 of 26 Old 08-24-2012, 09:29 AM
AVS Addicted Member
 
amirm's Avatar
 
Join Date: Jan 2002
Location: Washington State
Posts: 18,224
Mentioned: 3 Post(s)
Tagged: 0 Thread(s)
Quoted: 865 Post(s)
Liked: 516
Quote:
Originally Posted by commsysman View Post

The idea that using smaller wire affects the frequency response is idiotic; wire has the same resistance at all frequencies.
It does but your speaker does not. Since the total resistance (really impedance) as seen by the amp is the sum of speaker+wire, how high the wire resistance is as a percentage of the speaker wire causes variable dips. Here is the link to the theory again: http://www.avsforum.com/t/1379661/does-expensive-speaker-and-interconnect-cables-really-make-a-difference#post_21347201

Amir
Retired Technology Insider
Founder, Madrona Digital
"Insist on Quality Engineering"
amirm is offline  
post #19 of 26 Old 08-24-2012, 10:34 AM
AVS Addicted Member
 
Ratman's Avatar
 
Join Date: May 2002
Location: Collingswood, N.J.
Posts: 14,539
Mentioned: 1 Post(s)
Tagged: 0 Thread(s)
Quoted: 210 Post(s)
Liked: 299
No matter.... anyone that uses 22AWG speaker wire (for ANY length) needs a good spanking.
Ratman is offline  
post #20 of 26 Old 08-24-2012, 10:35 AM
AVS Addicted Member
 
arnyk's Avatar
 
Join Date: Oct 2002
Location: Grosse Pointe Woods, MI
Posts: 14,382
Mentioned: 2 Post(s)
Tagged: 0 Thread(s)
Quoted: 753 Post(s)
Liked: 1167
Quote:
Originally Posted by commsysman View Post

If you use 22 gauge wire for a 40-foot run the wire has a higher resistance than the speaker, so more power will be delivered to the wire than the speaker; that is not good.
You need 14 gauge wire to ensure that at least 90% of the amplifier output reaches the speaker.

Given that the speaker's impedance curve is only rarely below 8 ohms, the actual percentage is more like at least 95%
Quote:
The idea that using smaller wire affects the frequency response is idiotic; wire has the same resistance at all frequencies.

Shows ignorance of basic electrical circuits - how the series voltage divider works, and no understanding of the well known fact that loudspeaker impedance varies strongly with frequency.
arnyk is offline  
post #21 of 26 Old 08-24-2012, 11:08 AM
Senior Member
 
has7738's Avatar
 
Join Date: May 2012
Location: North America
Posts: 297
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 24
Quote:
Originally Posted by amirm View Post

While not relevant to the way I used the term, I do not consider Audyssey process deterministic. Using a single Mic location I have run both the Pro and Consumer versions and each time arrived at different results. In the case of the Pro version for example, it would complain about speaker phase reversal in one independent run, and not the next one. The unit was provided as part of the larger system from another manufacturer and when we asked them about this their answer was that "it does that!" And that we can ignore the message.
But, in no case has Audyssey ever claimed it could work using a single mic location! That's my point. There is no case were a single mic location gives you a meaningful measurement that any auto EQ system, or any person can respond to. Audyssey requires at least 6 positions, and more is better. Reading through the instructions would give you that perspective. Even your graph at a single point is meaningless. It's a mic at a single point in space, not a human with two ears moving around. Human ear/brain systems process and integrate multiple point data continuously, and thus can ignore point specific data, and focus on a result that represents a generalized, yet detailed sonic picture of the room and speakers. And as you well know, if you move a big chair and change the space, the result can be quite audible. That's what I mean by deterministic. Audyssey, when operated properly, will process multiple measurement points. Yet, as opposed to averaging FFT sweeps, or physically sweeping the measurement mic during an integration period with pink noise stimulus, fuzzy clustering post processing favors data that trends toward agreement between points, and rejects measurement noise. Averaging just smashes everything together with equal weighting.

The speaker that you saw come up out of phase is probably the subwoofer(s), and yes, this is an issue I've found frustrating as well. However, things do tend to right themselves once you get past the first measurement set. But I've found this more of a problem in rooms with a lot of internal reflections, less so with treated rooms. One system I set up used all omni speakers, and Audyssey was completely confused as to phase. Then again, so was I! I took some points very close to the speakers, and the phase ambiguity went away. Then just ignored it for the overall EQ.
Quote:
Originally Posted by amirm View Post


But again, that was not the context in which I used the term. I was explaining that relying on it reversing the effect of wire changing the system response is not deterministic. Let's say Audyssey determines there is a problem with power response if at 2000 Hz the response is 3 db below the average. You measure the system with a fat wire and it is at 2.5 db below average so no decision is taken by Audyssey to compensate.
Not sure why you would think Audyssey wouldn't attempt to fix a 2.5dB problem.
Quote:
Originally Posted by amirm View Post

Now you change the wire to a thin one and for the sake of argument, the wire causes a 1 db drop at 2000 Hz and we are now at 3.5 db below average.
Of course, the 22ga wire vs 12ga wire issue over a 40' run would be a much bigger change than 1dB....
Quote:
Originally Posted by amirm View Post

As a result Audyssey applies a boost of a few db to compensate. This compensation is not proportional to change that the wire introduced since Audyssey has no idea that was the cause. It simply goes by the sum total of what it "hears."
Again, not sure why you would say that the compensation would be wrong because Audyssey doesn't know the cause. How is knowing the cause relevant? Would any system ever know the specific cause of any measured response excursion? Would knowing that the left tweeter has a 2dB rolloff inherently, or that the rolloff is caused by the speaker grill change what you do to compensate?
Quote:
Originally Posted by amirm View Post

So when I say it is not deterministic, I mean in the context of one using his assumptions of the system and how it works. It just doesn't work as simply and predictably as we assume it does.
I would agree that Audyssey responds to the total response, and has no idea of what the specific cause is. You know what? That's pretty much what a set of human ears does too! Do you care or know what causes a response error? Or would you rather fix it? Also, Audyssey doesn't just "apply a boost of a few dB to compensate", it's far more than that. Again, a wire-caused response change would be present in every single cluster. That's the point, it would be one of the easiest to recognize, and apply a high precision fix to.
Quote:
Originally Posted by amirm View Post

I said that there can be no aspiration for a flat response since the measurement mic is too cheap and inaccurate to provide that correct view. And due to variations in mic, the system has to further assume that its targets for flatness cannot be relied upon. As you correctly state, for the most part this does not interfere with room equalization since the swings can be quite large and at any rate, we can make relative corrections and still be OK. A 15 db peak at 70 Hz needs to be corrected as an example regardless of how inaccurate the mic might be. As you know for the Pro systems they provide a much better mic and with calibration data so the problem with arriving at accurate and flatter response is an accepted one with the consumer mics provided with AVRs.
I agree that the consumer mics are a bit un-nerving, but unless you have actual test data that shows more than, say, a 1.5dB variance in response flatness, I'd have to say the comment is unfounded, and inappropriate to the discussion. And the key is variance...if they are all non-flat in the same way but have only 1dB of sample-to-sample variance, you build in a mic curve and the problem is over. Audyssey specifically cautions against using any other mic but the one that comes with your receiver, or the Pro kit. This would point to them having done considerable work to insure the mic is not an issue. The system couldn't work if the mics were so bad that they obscured valid data. Again, having been to their lab, I can assure you that mics response is under scrutiny to the extent that what is provided does not materially detract from the end result. They do test mics! How tight the mic specs are, I do not know, and I have no idea if each one is tested, or if there is an individual cal curve built into the receiver. I kind of doubt it, but it would be possible. More likely they have a spec for the mic.
Quote:
Originally Posted by amirm View Post

The graph is what any single measurement would indicate to the system the overall response is. I post it to show that there is no pretty picture of a response there that makes it abundantly simple to see that there is an effect due to wire, or speaker, or room acoustics. It is a jungle of data which has to be made sense out of. There are different methods for trying to remove the "grass" from it and extract the true characteristics of the system. The best way to do that is not averaging by the way but to apply the ERB of the ear to it, taking into account that the ear resolution sharply reduces as frequencies go up. This is the subject of my upcoming article in Widescreen Review Magazine (should be on newsstands in September). Indeed, that graph is part of that article. In there I show further transformations of that measurement and how those could be more appropriate representations of "what we hear."
I look forward to your article. But again, I point out, and I think we agree on this, a single point mic measurement isn't representative of anything but that specific single-point measurement. It doesn't correlate with what is heard, or what another point 1/4" away will look like. That makes a single point measurement simply unusable on it's own. As you allude to, post processing is necessary, but we could debate forever about what the "best way" is.
Quote:
Originally Posted by amirm View Post

Just thought since there was interest in how auto EQ systems work, that we use the current context to dig deeper.
Unfortunately, the AES paper you referenced may not be worth much in educating anyone on how (or if) auto EQ systems work. Without making specific references, the author(s) had an agenda, the science was bad, and the winning systems do not and did not ever exist in the real world. The presentation of a single point graph doesn't support or discredit auto EQ, but highlights the need for post processing, and should be presented as such. Part of post processing is collecting more data and integrating it. We agree that averaging isn't the best method (though I suspect your reference to averaging refers to averaging frequencies within the graph, or more correctly, smoothing. My reference was to averaging measurement points). My feeling is that both taking multiple measurement points and combining them in a way that reduces measurement noise and weights pertinent data is both valid and useful. Considering your suggesting of basically piping the measurements through something that impresses a human hearing transfer function is interesting, but has at least several obvious hurdles to overcome, not the least of which is that any human hearing transfer function, spacial and/or magnitude response, is at best a massive average over population. Perhaps you've worked that out, I look forward to reading the article.

My responses were stimulated by your claim that Audyssey would not properly recognize and respond with compensation to a response anomaly induced by undersized wire. I still say it can, and does, and if set up properly in accordance with their specified methods, can recognize that type of response anomaly quite easily.
has7738 is offline  
post #22 of 26 Old 08-24-2012, 01:18 PM
AVS Addicted Member
 
Ratman's Avatar
 
Join Date: May 2002
Location: Collingswood, N.J.
Posts: 14,539
Mentioned: 1 Post(s)
Tagged: 0 Thread(s)
Quoted: 210 Post(s)
Liked: 299
Come into my parlor said the spider to the fly.
Ratman is offline  
post #23 of 26 Old 08-24-2012, 04:40 PM
AVS Addicted Member
 
amirm's Avatar
 
Join Date: Jan 2002
Location: Washington State
Posts: 18,224
Mentioned: 3 Post(s)
Tagged: 0 Thread(s)
Quoted: 865 Post(s)
Liked: 516
For the sake of brevity I am not going to comment on the rest of your post but this bit:
Quote:
Originally Posted by has7738 View Post

Unfortunately, the AES paper you referenced may not be worth much in educating anyone on how (or if) auto EQ systems work.
That's puzzling to hear as the paper digs deep into both subjective listening tests results and objective measurements to correlate the two. As an example of the former, let's look at the listening test results arranged based on frequency of the source:

i-zCwpBvZ-M.png

Audyssey is the graph in pink (RC6). Ideal results would be zero (i.e. horizontal line in the middle). We see that Audyssey severely underperforms in both low and mid-frequencies compared to others. And it is not just the Harman/JBL system (RC1 and RC2) that does better (i.e. the employer of the authors) but also other competing systems. So it is hard to say they cooked the test to advantage themselves. There is a definite theme here as to what garners positive listener results which competing systems aim to optimize but Audyssey seems to have not to negative consequence in this study.

At my company we have deployed a lot of systems based on three of the tested brands above (Anthem, Audyssey and JBL Synthesis SDEC). Our collective experience absolutely agrees with that ranking. As an example, we have taken back a number of Anthem systems and replaced them with Audyssey based units and the customer always complains about taking a step backward in sound quality (sadly we had to do that due to HDMI problems with Anthem). I would be interested to know if you have also tested all of these systems and have read the paper and can explain how your view differs.
Quote:
Without making specific references, the author(s) had an agenda, the science was bad, and the winning systems do not and did not ever exist in the real world.
What science is bad? Systems were put in a reference listening room and listening tests performed. Measurements were made before and after and compared. Statistical analysis was performed to figure out what is going on. Granted the test matrix is huge here as system performance can be variable but to say there is bad science without saying what that is, doesn't give one anything to respond to smile.gif.

BTW, why has Audyssey been silent all this time and not publishing any listening tests of its own? This study came out in 2009. It is now three years later and still nothing. Why is that? Aren't their potential customers here least bit curious to have a proper objective evaluation of its merits? Do we not ask for blind tests of gear in general? Why not in the all-important category of room correction where there are ten different views of what one should do in a room? Surely we want to know if the manufacturer's assumptions are right. No?

Amir
Retired Technology Insider
Founder, Madrona Digital
"Insist on Quality Engineering"
amirm is offline  
post #24 of 26 Old 08-24-2012, 05:55 PM
Senior Member
 
has7738's Avatar
 
Join Date: May 2012
Location: North America
Posts: 297
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 24
Quote:
Originally Posted by amirm View Post

For the sake of brevity I am not going to comment on the rest of your post but this bit:
That's puzzling to hear as the paper digs deep into both subjective listening tests results and objective measurements to correlate the two.

Mea culpa, this one's my fault. I should never have taken a shot at the paper in this thread. In the interest of ending a now hopelessly meandering thread, let's either start a new thread or better, just end this one. It's of little use to go through a paragraph by paragraph critique of a 3 year old AES paper, especially in this thread. And it would take detail, time, and who's going to care at this point? The OP asked about 22ga wire, and look where we went.
Quote:
Originally Posted by amirm View Post

At my company we have deployed a lot of systems based on three of the tested brands above (Anthem, Audyssey and JBL Synthesis SDEC). Our collective experience absolutely agrees with that ranking. As an example, we have taken back a number of Anthem systems and replaced them with Audyssey based units and the customer always complains about taking a step backward in sound quality (sadly we had to do that due to HDMI problems with Anthem). I would be interested to know if you have also tested all of these systems and have read the paper and can explain how your view differs.
An interesting, if somewhat defensive stance, especially given the prior citation of a single point measurement. Based on your opinion and concept of how Audyssey works, should any of us place a high degree of confidence in your evaluation of other systems? I don't mean to keep taking shots at you here, clearly you do have your feed on the ground, and mean to be objective, at least I hope so. But we don't share the same experience at all. I've had excellent acceptance of Audyssey with clients. Other systems do often make a more dramatic difference, but after listening a while clients seem to want less. But I don't want to become anecdotal at this point.
Quote:
Originally Posted by amirm View Post

BTW, why has Audyssey been silent all this time and not publishing any listening tests of its own? This study came out in 2009. It is now three years later and still nothing. Why is that? Aren't their potential customers here least bit curious to have a proper objective evaluation of its merits? Do we not ask for blind tests of gear in general? Why not in the all-important category of room correction where there are ten different views of what one should do in a room? Surely we want to know if the manufacturer's assumptions are right. No?
Yes, that bothered me too. I don't know, perhaps they should respond directly. Perhaps its because there are so many issues to deal with, why bother? Certainly not worth doing in a forum, and if Audyssey has bigger fish to fry, I can't see the responsibility is mine either.

Let me just summarize, then leave it alone. The original statement I made was that Audyssey would EQ out the response anomalies caused by under-sized wire. I stand by that. You cited a single point graph, then highlighted the ambiguity of it, and the necessity of processing the data. You then made statements about what Audyssey would do, specifically, applying the wrong amount of correction because it doesn't "know the cause of the problem", and I claim that Audyssey wouldn't err in it's correction, and that not knowing the cause of a response error never precludes compensation of it. You cited ambiguity in the comparison of single point measurements, and I re-emphasize that single point measurements are by themselves useless. If that's what you're using to evaluate Audyssey and theorize about how it would correct for wire-caused response errors, consider this. Audyssey's method of extracting good measurement data from the chaotic is fuzzy clustering. To use fuzzy clustering you have to have data to cluster. You get that by taking data from more than one point. If you don't take more than one point, you have in effect completely bypassed Audyssey's data processing, and forcing it to create filters based on chaotic data. Yes, under those conditions, the result will be wrong every time. I theorize that someone who presents a negative opinion of a system that wasn't operated as intended may not understand the system fully. Sort of like saying "this light bulb is junk, it doesn't light up when I turn on the switch", then ignoring the fact that you have to screw the bulb into the socket first.

You also took a shot at their use of low-cost microphones without consideration that they may have taken the mic into account. If you were designing an auto EQ system and had to use cheap mics, would you just ignore the mic, or would you find a way to minimize or eliminate it's impact? Of course, you would do the latter. So did they.

The matter of the AES paper is a somewhat tangential attempt to discredit a system that is poorly understood. Without nit-picking the paper, we will just have to disagree on it.

Finally, your statement that as a pragmatist you'd use 22ga wire if that's all that would fit in an installation...have you ever actually seen a case where you couldn't get at least 16ga in? And, if for incredibly odd reason you absolutely had to use 22ga wire, would you actually use it for the entire run, or minimize it's impact by only using it for the bit that wouldn't fit any other way? Perhaps I'm still to green at this, but I have never found even the slightest need for 22ga speaker wire in a consumer hifi or HT installation.

Let me offer, for the sake of the thread, the suggestion that we just stop. Not much can be gained by further debate in this thread, and we've taken it way off topic anyway.

I look forward to reading your latest article with relish...perhaps on a hot dog. Perhaps we should shake hands at CEDIA and have a beer.
has7738 is offline  
post #25 of 26 Old 08-24-2012, 06:27 PM
AVS Addicted Member
 
amirm's Avatar
 
Join Date: Jan 2002
Location: Washington State
Posts: 18,224
Mentioned: 3 Post(s)
Tagged: 0 Thread(s)
Quoted: 865 Post(s)
Liked: 516
Quote:
Originally Posted by has7738 View Post

Finally, your statement that as a pragmatist you'd use 22ga wire if that's all that would fit in an installation..
I didn't say I would fit 22 gauge wire in a new installation if that is what you mean. Cost of wire is low compared to labor for the work that we do so we always use proper gauge wire. What I said is that if all you have is 22 gauge and the alternative of running a new wire is unpalatable from cost or aesthetic point of view, just go ahead and connect and use it for surround channels. If it were for front channels, then I would do everything in my power to remedy it. But for surround use in movies, I am just not going to lose sleep over it if you connect it and sounds fine to your ears. Plan to do it right the next time around smile.gif.
Quote:
have you ever actually seen a case where you couldn't get at least 16ga in?
No. Per above if we are running the wire it is not an issue. But from time to time one arrives at a situation on a job site where all that is there is the wrong kind of cable and the customer wants you to make it work cheaply without tearing up walls and such.
Quote:
I look forward to reading your latest article with relish...perhaps on a hot dog. Perhaps we should shake hands at CEDIA and have a beer.
Sounds good smile.gif.

Amir
Retired Technology Insider
Founder, Madrona Digital
"Insist on Quality Engineering"
amirm is offline  
post #26 of 26 Old 08-25-2012, 06:57 PM
Member
 
EEcle's Avatar
 
Join Date: Aug 2012
Posts: 28
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
Quote:
Originally Posted by has7738 View Post

Undocumented, yes. "Not deterministic"? Well, by definition, "deterministic" means "the doctrine that all facts and events exemplify natural laws".
Quote:
Originally Posted by has7738 View Post

And as you well know, if you move a big chair and change the space, the result can be quite audible. That's what I mean by deterministic.


Just to clear this up for any future readers, the word "deterministic" has a precise meaning in computer science which I believe is how amirm is using it. In that context a deterministic algorithm is one that, for a given set of inputs, will always produce the same output. Conversely a non-deterministic algorithm contains some randomness from the perspective of the consumer. So a routine that adds a list of numbers or paints a graphic on a display would have to be deterministic to be useful. A routine that deals cards or calculates video game enemy movement would be non-deterministic.

The Audyssey algorithm is almost certainly deterministic, which is to say if the microphone records exactly the same signals then the adjustments will be exactly the same every time. That is what you would expect for the system to be useful. Other components of the system though, especially the speakers and the microphone itself, are decidedly non-deterministic so the algorithm is operating on a slightly different data set every time its run.

Thus the process of tuning with Audyssey is non-deterministic (running it multiple times with the microphone in exactly the same positions may yield different results) even though the Audyssey algorithm is entirely deterministic. So you're both right biggrin.gif
EEcle is offline  
Reply Audio theory, Setup and Chat

User Tag List

Thread Tools
Show Printable Version Show Printable Version
Email this Page Email this Page


Forum Jump: 

Posting Rules  
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are Off
Pingbacks are Off
Refbacks are Off