Frequency response in GS...AWFUL.
07-12-02, 12:57 PM
#16
Lexus Fanatic
Join Date: Jul 2001
Location: California
Posts: 6,084
Likes: 0
Received 0 Likes
on
0 Posts
I will play around a little more as that was a rather inelegant attempt but I cut some of the detail off of Percy's file in Paint and then attached it. Hopefully I can figure out how to do this a little better.
Ron
Ron
07-12-02, 04:45 PM
#17
Driver
Join Date: Feb 2002
Location: uk
Posts: 137
Likes: 0
Received 0 Likes
on
0 Posts
Percy,
You will never get a sensible response using sine waves unless you measure the speakers in an anechoic chamber. What you are actually measuring is the speakers and the room acoustics. This does not represent what you actually hear.
The only way you can measure using sine waves is by positioning the mic close to the speakers to cut out reflections. I bet that if you position the mic directly in front of the Dyn bass/mid you will record a fairly flat frequency response, at least in its passband.
To take out the acoustics of the car you have to gate the signal so that you measure the driver output before the reflections kick in. In practice, a room averaged response will provides a much more accurate representation of what your system sounds like.
Some comments on the curves. I'd ignore the dip between 300 and 700 Hz. It's probably inaudible. The peak at 800 Hz may be due to room acoutics but it may also be the Dynaudios, which have a peak in that area according to the published curves. The system is rolling off from 80 Hz and will therefore sound bass light. The dip between 800 Hz and 4 kHz is much too broad even for sine wave measurement. I would guess that you need to reverse the polarity of the tweeter. Also, the tweeter is too loud - everything should be balance to the 100 to 300 Hz region. Finally, you have what looks like an 18 dB per octave filter at 8 kHz, which is a definite problem not related to sinewave measurements. I'd try measuring each tweeter up close (1 or 2 inches away) to find out what's going on here.
Stevie
You will never get a sensible response using sine waves unless you measure the speakers in an anechoic chamber. What you are actually measuring is the speakers and the room acoustics. This does not represent what you actually hear.
The only way you can measure using sine waves is by positioning the mic close to the speakers to cut out reflections. I bet that if you position the mic directly in front of the Dyn bass/mid you will record a fairly flat frequency response, at least in its passband.
To take out the acoustics of the car you have to gate the signal so that you measure the driver output before the reflections kick in. In practice, a room averaged response will provides a much more accurate representation of what your system sounds like.
Some comments on the curves. I'd ignore the dip between 300 and 700 Hz. It's probably inaudible. The peak at 800 Hz may be due to room acoutics but it may also be the Dynaudios, which have a peak in that area according to the published curves. The system is rolling off from 80 Hz and will therefore sound bass light. The dip between 800 Hz and 4 kHz is much too broad even for sine wave measurement. I would guess that you need to reverse the polarity of the tweeter. Also, the tweeter is too loud - everything should be balance to the 100 to 300 Hz region. Finally, you have what looks like an 18 dB per octave filter at 8 kHz, which is a definite problem not related to sinewave measurements. I'd try measuring each tweeter up close (1 or 2 inches away) to find out what's going on here.
Stevie
07-12-02, 05:01 PM
#18
Driver
Join Date: Feb 2002
Location: uk
Posts: 137
Likes: 0
Received 0 Likes
on
0 Posts
sirsomo wrote:
>The GS stock system always did sound kind of high in the mid->high range to me, I just assumed that it was because my >previous car was all aftermarket and that was how factory >systems sounded?
That's absolutely right. They have a peak at around 1.5k - I've measured it (out of the car). It makes a saxophone sound like a kazoo.
My tone control settings with the stock drivers were 2 c'clock bass, 10 o'clock mid and HF. More bass when you're driving at speed, of course. With my replacement Vifas, the settings are not too different.
Stevie
>The GS stock system always did sound kind of high in the mid->high range to me, I just assumed that it was because my >previous car was all aftermarket and that was how factory >systems sounded?
That's absolutely right. They have a peak at around 1.5k - I've measured it (out of the car). It makes a saxophone sound like a kazoo.
My tone control settings with the stock drivers were 2 c'clock bass, 10 o'clock mid and HF. More bass when you're driving at speed, of course. With my replacement Vifas, the settings are not too different.
Stevie
07-12-02, 05:46 PM
#19
Lexus Fanatic
Join Date: Jul 2001
Location: California
Posts: 6,084
Likes: 0
Received 0 Likes
on
0 Posts
stevie - Now I am getting confused (don't worry, it happens a lot). I don't understand the rationale behind doing this measurement in any way that reduces the effect of the environment. By that I mean that you do not want to isolate the drivers from their interaction with this specific car environment. After all, Percy is trying to generate a curve of the response at the drivers head in the GS not simply plotting the response of the driver in an anechoic environment. The difference between sine waves of varying frequency and a so called "pink noise" generation would be dramatic if the human ear was listening but the microphone and spectrum analyzer is measuring amplitude at a certain frequency. Frequency meters will inevitably trigger on some portion of the curve of the signal for counting purposes and will in effect impose either a sine or square wave rationale on the signal in order to interpret it. If I put a certain frequency on a driver and at some location away from the driver measure what has happened in frequency or amplitude to the waveform, isn't this what Percy did? Maybe I don't understand exactly what Percy did but I can tell you that I am involved in electrical filter fabrication (up to 22 GHz right now) and optical multiplexers and we can send broad spectrum of signals at components and see what we get out on the output of the device or we can send specific frequencies (wavelengths) and accumulate data to construct a spectrum response. This has nothing to do with car audio but my understanding is that Percy took a certain set of components, the stock head unit and amplifer with the specified drivers, and generated a series of frequencies and measured what showed up at the drivers head to present an analysis of the acoustic environment, as described, in the GS. I am just not certain I see what the problem is with sine waves and as far as reflections, isn't this all part of the acoustic environment set up for this particular test? I imagine Percy will have additional comments and I hope other people here don't find this totally boring but I am very interested in understanding what hard data is available for improving the audio system in the GS. The difference with the dyns is amazing, the ADS's in the rear door is much more subtle. But my untrained, and old, ears are noticing an unusual bass response, even without replacing the sub yet. The dyns have really put in a lot of bottom end, too much for most music for my tastes, and I would like to smooth things out a bit as the bass control on the headunit goes from too much to not enough very quickly. Now, if I can just figure out how to post that doggoned graph directly I will feel like I made progress for the week.
07-12-02, 09:55 PM
#21
Moderator - Electronics Forum
Thread Starter
[QUOTE]Originally posted by stevie
[B]Percy,
You will never get a sensible response using sine waves unless you measure the speakers in an anechoic chamber.
***But the car was never an anechoic chamber and neither are most home audio systems. Just not possible. What I intended to do was to measure everything that had an effect on the final output, that of the electronics, speakers, reflections, etc, while NOT using pink noise. I've found pink noise to be off at times. Just seems like the mic wasn't responding to it.***
What you are actually measuring is the speakers and the room acoustics. This does not represent what you actually hear.
***Ok...you lost me on this one. GENERALLY speaking, if a mic "hears" it, you should hear it also. This doesn't account for notched or other hearing losses.***
The only way you can measure using sine waves is by positioning the mic close to the speakers to cut out reflections. I bet that if you position the mic directly in front of the Dyn bass/mid you will record a fairly flat frequency response, at least in its passband.
***What I wanted to do is to measure the whole ball of wax, reflections included. In an untuned vehicle or one with a stock system, there's pretty much no such thing as a flat frequency response. Plus, measuring directly in front of the speaker serves no real purpose since the drivers head is not directly in front, and at a close distance to the speaker. Note...mic directly in front of speaker would denote a near field measurement. I'm after free field, NOT near field. Plus the location of the mic is fairly critical. ***
To take out the acoustics of the car you have to gate the signal so that you measure the driver output before the reflections kick in. In practice, a room averaged response will provides a much more accurate representation of what your system sounds like.
***BUT the key here is EVERYTHING in the car. Electronics, interior acoustics, etc. Why measure the system without the reflections? Main point here is tuning and that accounts for the entire package.***
Some comments on the curves. I'd ignore the dip between 300 and 700 Hz. It's probably inaudible.
***Compared to the rest of the audio spectrum, I'd agree on this.***
The peak at 800 Hz may be due to room acoutics but it may also be the Dynaudios, which have a peak in that area according to the published curves.
***But at 15+ dBA more than the surrounding areas? The Dyn curves aren't THAT severe. I'd say both interior acoustics and electronics.***
The system is rolling off from 80 Hz and will therefore sound bass light.
***Yep.***
The dip between 800 Hz and 4 kHz is much too broad even for sine wave measurement.
***But still the ear would pick it up as something "missing".***
I would guess that you need to reverse the polarity of the tweeter. Also, the tweeter is too loud - everything should be balance to the 100 to 300 Hz region.
***Ideally, yes. But definitely not the case.***
Finally, you have what looks like an 18 dB per octave filter at 8 kHz, which is a definite problem not related to sinewave measurements. I'd try measuring each tweeter up close (1 or 2 inches away) to find out what's going on here.
***Better idea. Cut the mids out entirely (and sub), then measure from either that position or the driver position. Whole point to the measurement was to find out what was really going on. If they stuck an RTA over at the local IASCA event, they'd measure pretty much the same thing. ***
Percy
[B]Percy,
You will never get a sensible response using sine waves unless you measure the speakers in an anechoic chamber.
***But the car was never an anechoic chamber and neither are most home audio systems. Just not possible. What I intended to do was to measure everything that had an effect on the final output, that of the electronics, speakers, reflections, etc, while NOT using pink noise. I've found pink noise to be off at times. Just seems like the mic wasn't responding to it.***
What you are actually measuring is the speakers and the room acoustics. This does not represent what you actually hear.
***Ok...you lost me on this one. GENERALLY speaking, if a mic "hears" it, you should hear it also. This doesn't account for notched or other hearing losses.***
The only way you can measure using sine waves is by positioning the mic close to the speakers to cut out reflections. I bet that if you position the mic directly in front of the Dyn bass/mid you will record a fairly flat frequency response, at least in its passband.
***What I wanted to do is to measure the whole ball of wax, reflections included. In an untuned vehicle or one with a stock system, there's pretty much no such thing as a flat frequency response. Plus, measuring directly in front of the speaker serves no real purpose since the drivers head is not directly in front, and at a close distance to the speaker. Note...mic directly in front of speaker would denote a near field measurement. I'm after free field, NOT near field. Plus the location of the mic is fairly critical. ***
To take out the acoustics of the car you have to gate the signal so that you measure the driver output before the reflections kick in. In practice, a room averaged response will provides a much more accurate representation of what your system sounds like.
***BUT the key here is EVERYTHING in the car. Electronics, interior acoustics, etc. Why measure the system without the reflections? Main point here is tuning and that accounts for the entire package.***
Some comments on the curves. I'd ignore the dip between 300 and 700 Hz. It's probably inaudible.
***Compared to the rest of the audio spectrum, I'd agree on this.***
The peak at 800 Hz may be due to room acoutics but it may also be the Dynaudios, which have a peak in that area according to the published curves.
***But at 15+ dBA more than the surrounding areas? The Dyn curves aren't THAT severe. I'd say both interior acoustics and electronics.***
The system is rolling off from 80 Hz and will therefore sound bass light.
***Yep.***
The dip between 800 Hz and 4 kHz is much too broad even for sine wave measurement.
***But still the ear would pick it up as something "missing".***
I would guess that you need to reverse the polarity of the tweeter. Also, the tweeter is too loud - everything should be balance to the 100 to 300 Hz region.
***Ideally, yes. But definitely not the case.***
Finally, you have what looks like an 18 dB per octave filter at 8 kHz, which is a definite problem not related to sinewave measurements. I'd try measuring each tweeter up close (1 or 2 inches away) to find out what's going on here.
***Better idea. Cut the mids out entirely (and sub), then measure from either that position or the driver position. Whole point to the measurement was to find out what was really going on. If they stuck an RTA over at the local IASCA event, they'd measure pretty much the same thing. ***
Percy
07-13-02, 10:43 AM
#22
Driver
Join Date: Feb 2002
Location: uk
Posts: 137
Likes: 0
Received 0 Likes
on
0 Posts
Percy,
Well , you've opened a real can of worms here. I'll respond later today (I'm right in the middle of soundproofing my trunk).
I'd suggest, though, that you try measuring without the A weighting on your equipment. As far as I know, A weighting is used for sound pressure level measurements. It has no place in a frequency response measurement, as it will introduce a dip between 1 and 4k - which is in fact what your measurements are showing.
Stevie
Well , you've opened a real can of worms here. I'll respond later today (I'm right in the middle of soundproofing my trunk).
I'd suggest, though, that you try measuring without the A weighting on your equipment. As far as I know, A weighting is used for sound pressure level measurements. It has no place in a frequency response measurement, as it will introduce a dip between 1 and 4k - which is in fact what your measurements are showing.
Stevie
07-13-02, 11:05 AM
#23
Moderator - Electronics Forum
Thread Starter
Originally posted by stevie
Percy,
Well , you've opened a real can of worms here. I'll respond later today (I'm right in the middle of soundproofing my trunk).
I'd suggest, though, that you try measuring without the A weighting on your equipment. As far as I know, A weighting is used for sound pressure level measurements. It has no place in a frequency response measurement, as it will introduce a dip between 1 and 4k - which is in fact what your measurements are showing.
Stevie
Percy,
Well , you've opened a real can of worms here. I'll respond later today (I'm right in the middle of soundproofing my trunk).
I'd suggest, though, that you try measuring without the A weighting on your equipment. As far as I know, A weighting is used for sound pressure level measurements. It has no place in a frequency response measurement, as it will introduce a dip between 1 and 4k - which is in fact what your measurements are showing.
Stevie
Good luck with the trunk!
A weighted rolls off below 1khz so that the response will "simulate" that of a typical human ear. Since most ears aren't sensitive to low freqs (we feel them, not really hear them), this would be the most logical to use. From 1khz to 4khz, however, there is a slightly higher sensitivity, around the order of 1 to 2 dB at most. I'll be glad to send you a pic of the graph from Bruel & Kjaers publication.
To quote from B&K's manual...
"The A, B, C, weighting curves are internationally standardized and are used to make the sound level meter respond in a similar way to the human ear, at low, medium and high sound levels respectively. The differences between the three responses is most noticable at low frequencies."
If you use a weighting network, at low levels, of other than A weighted for tuning, then the bass will sound remarkably shallow. IME, I've found that A weighted produces the most accurate results and a good baseline from which to start off with. The peak measurement taken was only at 79 dBA, not really "loud".
Sidenote...Car and Driver also uses A weighted for their interior noise measurements.
Percy
07-13-02, 05:28 PM
#24
Lexus Fanatic
Join Date: Jul 2001
Location: California
Posts: 6,084
Likes: 0
Received 0 Likes
on
0 Posts
OK, I am going to watch and see how this discussion develops. Having never seen a curve such as this before I was wondering what can be learned from it. Percy - I may be on thin ice here but I view the curve as three separate regions, that covered by the sub, the mid, and finally the area that would be covered by the tweeter. My uninformed view is that the low end and top end don't look that bad, relatively speaking, it is obviosly never going to be a flat line and I am not sure our ears would like that anyway. Is the "less peaky" bottom and top response due to some acoustics in the car or is it speaker location relative to the mic (and a persons ear) location (or both, or something else)? The sub firing up through the rear deck and those marvelous dyn tweets up in the crotch at the A pillar are generating waves that will more directly impinge on the mic at the ear position. The mids are down in that ankle position in both the front and rear door and in order for the sound to reach our ears it will have to do some direct and reflection transmission. I may be ignoring for the moment any measurement peculiarities that you and Stevie will work out but I am trying to figure out what the graph means and how this relates to either improving things or just saying "that's the best we can do". So, am I off base looking at the three regions and the responses or is that the way to look at this curve? How do you improve acoustic response (and I assume I mean flattening the response curve out with as few peaks and troughs as possible) without moving speaker positions or is this the whole purpose of a good equalizer? I am slowly working up to the Mac outboard amp but an equalizer may need to be put in as well. Hope everyone else is enjoying this discussion as much as I am. Thanks for getting the starting data Percy. Not exactly being a car audiophile I was wondering do the car audio mags ever publish data like this for various vehicles? Seems like it would be a natural during a buildup to show the impact of various mods. Or do people just care mostly what color the speaker wires are and how many amps can dance on the head of a pin?
07-13-02, 06:48 PM
#25
Moderator - Electronics Forum
Thread Starter
Originally posted by RON430
OK, I am going to watch and see how this discussion develops. Having never seen a curve such as this before I was wondering what can be learned from it. Percy - I may be on thin ice here but I view the curve as three separate regions, that covered by the sub, the mid, and finally the area that would be covered by the tweeter. My uninformed view is that the low end and top end don't look that bad, relatively speaking, it is obviosly never going to be a flat line and I am not sure our ears would like that anyway. Is the "less peaky" bottom and top response due to some acoustics in the car or is it speaker location relative to the mic (and a persons ear) location (or both, or something else)?
***Ron, from what I'm seeing, it's mostly due to the tuning of the electronics. The easiest thing to say would be the acoustics, and I would agree. BUT, without an aftermarket system set COMPLETELY flat (and xover points noted as well as their levels), one would never really know the flat baseline. You're also right on the "region" of the weighting filters. I read further into the filters and a different company other than B&K noted that A weighted should be used for 40 phons (about 40dB...no weighting given) while C weighted should be used for 100 phons (about 100dB...no weighting curve given). One should also note that the weighting filters used are NOT as severe as the graph illustrates. I'll try to scan in a copy of the weighting filters for you to get an idea of what is going on.***
The sub firing up through the rear deck and those marvelous dyn tweets up in the crotch at the A pillar are generating waves that will more directly impinge on the mic at the ear position. The mids are down in that ankle position in both the front and rear door and in order for the sound to reach our ears it will have to do some direct and reflection transmission. I may be ignoring for the moment any measurement peculiarities that you and Stevie will work out but I am trying to figure out what the graph means and how this relates to either improving things or just saying "that's the best we can do".
***I'd say that for a stock system, it's the best we can do without an outboard EQ. I think I know what Stevie is trying to say about the anechoic chamber, etc. I believe he's thinking of speaker manufacturer specs, in ideal locations (anechoic) at near field or far field measurements. Unfortunately the "ideal" anechoic location will never occur in a car due to all the variables involved.***
So, am I off base looking at the three regions and the responses or is that the way to look at this curve?
***It's actually a fairly gentle sloping curve depending on where you look. Here's a description...
From 1khz (it's the reference point for all filter ing newtorks) to 500 hz, the slope goes down abour 4db. From 500 hz to 200, it's still on the same line but it's gone down another 6dB. 200 to 100, about 8 dB. 100 to 50, 12 dB. 50 to 20, 20 dB. Now for above 1khz. From 1k to 2k, the slope goes up 2dB. 2k to 5k, it's about the same curve but will shallow downward about 1.5 dB. Then from 5k to 20k, it drops about 10dB.
Compared to the graph, it's nowhere as smooth. The graph is ragged and it still doesn't really "cover the bases" of the weighting filter and it hardly even follows it's general shape. Even if I did switch over to C weighting, the lower area (below 1k) would be boosted anywhere from 42dB to 2dB, depending on where you're looking at the graph. This wouldn't be the appropriate filter to use unless you're constantly listening at over 100dB. I usuallly listen to the system on the average of anywhere from 60 to 70 dB. Most of the time it's around the low 50's.***
How do you improve acoustic response (and I assume I mean flattening the response curve out with as few peaks and troughs as possible) without moving speaker positions or is this the whole purpose of a good equalizer?
***Ron, you nailed it on the head. That's where the good old EQ comes in. There's NO WAY that you can hope to drop in a set of speakers and hope for an ideal "flat" response, or one of high end audiophile caliber. My guess is that the Dyns took out quite a bit of 630 hz energy (probably on the order of 10 to 20dB) and added to the 100hz and below frequencies. But, that's why they have EQ's. Every car is different (hmmm...just like every woman!) and there are always some nasties to tame.***
I am slowly working up to the Mac outboard amp but an equalizer may need to be put in as well.
***You'll need them!***
Hope everyone else is enjoying this discussion as much as I am. Thanks for getting the starting data Percy. Not exactly being a car audiophile I was wondering do the car audio mags ever publish data like this for various vehicles?
***I've wondered that also. It would be nice to actually get some real graphs/data on their response. You've got the McIntosh/Subaru systems (not really that good...highly detuned), Volvo/Dynaudio (needs more amplification and tuning), GM/Bose, GM/Monsoon, Lexus/Levinson, Lexus/Nak, etc. There's so many out there! One could make a killing in bucks just supplying the data if it's done right. Hmmmm....
***
Seems like it would be a natural during a buildup to show the impact of various mods. Or do people just care mostly what color the speaker wires are and how many amps can dance on the head of a pin?
OK, I am going to watch and see how this discussion develops. Having never seen a curve such as this before I was wondering what can be learned from it. Percy - I may be on thin ice here but I view the curve as three separate regions, that covered by the sub, the mid, and finally the area that would be covered by the tweeter. My uninformed view is that the low end and top end don't look that bad, relatively speaking, it is obviosly never going to be a flat line and I am not sure our ears would like that anyway. Is the "less peaky" bottom and top response due to some acoustics in the car or is it speaker location relative to the mic (and a persons ear) location (or both, or something else)?
***Ron, from what I'm seeing, it's mostly due to the tuning of the electronics. The easiest thing to say would be the acoustics, and I would agree. BUT, without an aftermarket system set COMPLETELY flat (and xover points noted as well as their levels), one would never really know the flat baseline. You're also right on the "region" of the weighting filters. I read further into the filters and a different company other than B&K noted that A weighted should be used for 40 phons (about 40dB...no weighting given) while C weighted should be used for 100 phons (about 100dB...no weighting curve given). One should also note that the weighting filters used are NOT as severe as the graph illustrates. I'll try to scan in a copy of the weighting filters for you to get an idea of what is going on.***
The sub firing up through the rear deck and those marvelous dyn tweets up in the crotch at the A pillar are generating waves that will more directly impinge on the mic at the ear position. The mids are down in that ankle position in both the front and rear door and in order for the sound to reach our ears it will have to do some direct and reflection transmission. I may be ignoring for the moment any measurement peculiarities that you and Stevie will work out but I am trying to figure out what the graph means and how this relates to either improving things or just saying "that's the best we can do".
***I'd say that for a stock system, it's the best we can do without an outboard EQ. I think I know what Stevie is trying to say about the anechoic chamber, etc. I believe he's thinking of speaker manufacturer specs, in ideal locations (anechoic) at near field or far field measurements. Unfortunately the "ideal" anechoic location will never occur in a car due to all the variables involved.***
So, am I off base looking at the three regions and the responses or is that the way to look at this curve?
***It's actually a fairly gentle sloping curve depending on where you look. Here's a description...
From 1khz (it's the reference point for all filter ing newtorks) to 500 hz, the slope goes down abour 4db. From 500 hz to 200, it's still on the same line but it's gone down another 6dB. 200 to 100, about 8 dB. 100 to 50, 12 dB. 50 to 20, 20 dB. Now for above 1khz. From 1k to 2k, the slope goes up 2dB. 2k to 5k, it's about the same curve but will shallow downward about 1.5 dB. Then from 5k to 20k, it drops about 10dB.
Compared to the graph, it's nowhere as smooth. The graph is ragged and it still doesn't really "cover the bases" of the weighting filter and it hardly even follows it's general shape. Even if I did switch over to C weighting, the lower area (below 1k) would be boosted anywhere from 42dB to 2dB, depending on where you're looking at the graph. This wouldn't be the appropriate filter to use unless you're constantly listening at over 100dB. I usuallly listen to the system on the average of anywhere from 60 to 70 dB. Most of the time it's around the low 50's.***
How do you improve acoustic response (and I assume I mean flattening the response curve out with as few peaks and troughs as possible) without moving speaker positions or is this the whole purpose of a good equalizer?
***Ron, you nailed it on the head. That's where the good old EQ comes in. There's NO WAY that you can hope to drop in a set of speakers and hope for an ideal "flat" response, or one of high end audiophile caliber. My guess is that the Dyns took out quite a bit of 630 hz energy (probably on the order of 10 to 20dB) and added to the 100hz and below frequencies. But, that's why they have EQ's. Every car is different (hmmm...just like every woman!) and there are always some nasties to tame.***
I am slowly working up to the Mac outboard amp but an equalizer may need to be put in as well.
***You'll need them!***
Hope everyone else is enjoying this discussion as much as I am. Thanks for getting the starting data Percy. Not exactly being a car audiophile I was wondering do the car audio mags ever publish data like this for various vehicles?
***I've wondered that also. It would be nice to actually get some real graphs/data on their response. You've got the McIntosh/Subaru systems (not really that good...highly detuned), Volvo/Dynaudio (needs more amplification and tuning), GM/Bose, GM/Monsoon, Lexus/Levinson, Lexus/Nak, etc. There's so many out there! One could make a killing in bucks just supplying the data if it's done right. Hmmmm....
***Seems like it would be a natural during a buildup to show the impact of various mods. Or do people just care mostly what color the speaker wires are and how many amps can dance on the head of a pin?
Percy
07-14-02, 01:44 PM
#26
Driver
Join Date: Feb 2002
Location: uk
Posts: 137
Likes: 0
Received 0 Likes
on
0 Posts
Hi Percy,
>What you are actually measuring is the speakers and the room acoustics. This does not represent what you actually hear.
***Ok...you lost me on this one. GENERALLY speaking, if a mic "hears" it, you should hear it also. This doesn't account for notched or other hearing losses.***
Yes, we'll hear it but not necessarily as it appears of the response curve. The difference is that our hearing has the capacity to discriminate. We can pick out one voice from a crowd, or listen to one person when other people are talking at the same time.
Your brain is able to distinguish between all those peaks and dips that are being picked up by the microphone. Your ear can tell when a system is properly balance because it recognizes the affect of the room and is able to separate that from what is being reproduced (broad generalization).
A mic measuring a sine wave sweep cannot distinguish between source signals and secondary, reverberant signals unless you use gating techniques. As you say, these filter out the room effects, but if you include the room effects you get a frequency response that is not very useful, because it's difficult to evaluate.
As an experiment, anyone with a Radio Shack SPL meter can try measuring a sweep or tones from a test CD and watching the meter. What you'll see is the meter jumping all over the place on tones that sound as if they are the same intensity.
***BUT the key here is EVERYTHING in the car. Electronics, interior acoustics, etc. Why measure the system without the reflections? Main point here is tuning and that accounts for the entire package.***
My experience with car audio is limited. I do know that in a normal room, third octave measurements using pink noise correlate closely to what we hear. There may be differences with the car environment that I don't know about. With third octave pink noise you are looking for a frequency response that is flat up to 2 or 3 k and which then rolls of at 6 dB per octave after that.
In a normal room using third octave pink noise, the position of the mic isn't very important, because most of the sound being measured is reflected rather than direct.
To your question: it is useful to be able to measure the system without reflections because then you know exactly what the system is doing. If there is a dip caused by poor crossover design, for example, you would want to fix that problems at source. You cannot correct for that kind of problem using a graphic equalizer.
What problems have you had using third octave pink noise measurements?
Does your system actually cut off at 9 kHz, i.e. can you hear this?
Stevie
>What you are actually measuring is the speakers and the room acoustics. This does not represent what you actually hear.
***Ok...you lost me on this one. GENERALLY speaking, if a mic "hears" it, you should hear it also. This doesn't account for notched or other hearing losses.***
Yes, we'll hear it but not necessarily as it appears of the response curve. The difference is that our hearing has the capacity to discriminate. We can pick out one voice from a crowd, or listen to one person when other people are talking at the same time.
Your brain is able to distinguish between all those peaks and dips that are being picked up by the microphone. Your ear can tell when a system is properly balance because it recognizes the affect of the room and is able to separate that from what is being reproduced (broad generalization).
A mic measuring a sine wave sweep cannot distinguish between source signals and secondary, reverberant signals unless you use gating techniques. As you say, these filter out the room effects, but if you include the room effects you get a frequency response that is not very useful, because it's difficult to evaluate.
As an experiment, anyone with a Radio Shack SPL meter can try measuring a sweep or tones from a test CD and watching the meter. What you'll see is the meter jumping all over the place on tones that sound as if they are the same intensity.
***BUT the key here is EVERYTHING in the car. Electronics, interior acoustics, etc. Why measure the system without the reflections? Main point here is tuning and that accounts for the entire package.***
My experience with car audio is limited. I do know that in a normal room, third octave measurements using pink noise correlate closely to what we hear. There may be differences with the car environment that I don't know about. With third octave pink noise you are looking for a frequency response that is flat up to 2 or 3 k and which then rolls of at 6 dB per octave after that.
In a normal room using third octave pink noise, the position of the mic isn't very important, because most of the sound being measured is reflected rather than direct.
To your question: it is useful to be able to measure the system without reflections because then you know exactly what the system is doing. If there is a dip caused by poor crossover design, for example, you would want to fix that problems at source. You cannot correct for that kind of problem using a graphic equalizer.
What problems have you had using third octave pink noise measurements?
Does your system actually cut off at 9 kHz, i.e. can you hear this?
Stevie
07-14-02, 02:36 PM
#27
Moderator - Electronics Forum
Thread Starter
[QUOTE]Originally posted by stevie
[B]Hi Percy,
>What you are actually measuring is the speakers and the room acoustics. This does not represent what you actually hear.
Yes, we'll hear it but not necessarily as it appears of the response curve. The difference is that our hearing has the capacity to discriminate. We can pick out one voice from a crowd, or listen to one person when other people are talking at the same time.
***Psychoacoustics.***
Your brain is able to distinguish between all those peaks and dips that are being picked up by the microphone. Your ear can tell when a system is properly balance because it recognizes the affect of the room and is able to separate that from what is being reproduced (broad generalization).
A mic measuring a sine wave sweep cannot distinguish between source signals and secondary, reverberant signals unless you use gating techniques. As you say, these filter out the room effects, but if you include the room effects you get a frequency response that is not very useful, because it's difficult to evaluate.
***That's where all of the processors come in. Ala Alpine F#1 Status. That's probably the only processor that will account for psyacoustics, interior anomalies, etc. It's measured and evaluated by it's own mic. Lots of processing power...the equivalent of two P4 1.3's in parallel.***
As an experiment, anyone with a Radio Shack SPL meter can try measuring a sweep or tones from a test CD and watching the meter. What you'll see is the meter jumping all over the place on tones that sound as if they are the same intensity.
***The Radio Shack meter is only good for 1khz. You have to add to different signal/frequency levels to get the proper readout. IOW...if you're measuring at 4khz and it gives 60dBA, it's NOT 60dBA. The meter (newer version) is not designed to be a 20-20khz measurement tool.***
My experience with car audio is limited. I do know that in a normal room, third octave measurements using pink noise correlate closely to what we hear. There may be differences with the car environment that I don't know about. With third octave pink noise you are looking for a frequency response that is flat up to 2 or 3 k and which then rolls of at 6 dB per octave after that.
***Usually with car audio, you get the RTA as flat as possible and then adjust the curve afterwords. The RTA is a good guideline, but not the final word. Most will roll off at 16k to 17khz, otherwise it will sound too bright.***
In a normal room using third octave pink noise, the position of the mic isn't very important, because most of the sound being measured is reflected rather than direct.
***In car audio, it's VERY critical. A reading from the head drivers position will be VASTLY different than a reading in the center console or the rear drivers position.***
To your question: it is useful to be able to measure the system without reflections because then you know exactly what the system is doing. If there is a dip caused by poor crossover design, for example, you would want to fix that problems at source. You cannot correct for that kind of problem using a graphic equalizer.
***That's where max flexibility comes in. Parametric with variable crossover settings and variable crossover slopes, as well as variable boost/cut levels for the xover will come in handy. There are alot of tricks you can do for the car, especially if you have the right equipment. With home audio, it's limited to a couple of bands, and they're not even parametric. Car audio is a different beast altogether. It's the only field outside of the professional studio where I know of that will utilize parametric eq's and delays. Most home setups won't utilize this.***
What problems have you had using third octave pink noise measurements?
***Sometimes during the pink noise generation, the frequencies are so short that the mic doesn't capture all of it. Other times it's overleveled (Alpine pink noise test disc has a +6dB over the rest of the frequencies at some parts of the disc) so that the reading is incorrect. This is where I have to go with sine wave (since the B&K is displaying all of the spectrum...I have to isolate it somehow) and single test tones. Effective and it works out nicely.***
Does your system actually cut off at 9 kHz, i.e. can you hear this?
***To me it sounds like a vast loss of detail and air. Dull sounding if you will. I'm used to SACD or HDCD sources with good speakers (Dynaudio ESOTAR series) or good headphones (Sennheiser HD580). To go back in the car after listening to these is painful! I've actually tested my hearing and on pure sine notes I can pick off 19khz. 16k to 17k (sound of TV tube being energized) is LOUD to me.***
Percy
[B]Hi Percy,
>What you are actually measuring is the speakers and the room acoustics. This does not represent what you actually hear.
Yes, we'll hear it but not necessarily as it appears of the response curve. The difference is that our hearing has the capacity to discriminate. We can pick out one voice from a crowd, or listen to one person when other people are talking at the same time.
***Psychoacoustics.***
Your brain is able to distinguish between all those peaks and dips that are being picked up by the microphone. Your ear can tell when a system is properly balance because it recognizes the affect of the room and is able to separate that from what is being reproduced (broad generalization).
A mic measuring a sine wave sweep cannot distinguish between source signals and secondary, reverberant signals unless you use gating techniques. As you say, these filter out the room effects, but if you include the room effects you get a frequency response that is not very useful, because it's difficult to evaluate.
***That's where all of the processors come in. Ala Alpine F#1 Status. That's probably the only processor that will account for psyacoustics, interior anomalies, etc. It's measured and evaluated by it's own mic. Lots of processing power...the equivalent of two P4 1.3's in parallel.***
As an experiment, anyone with a Radio Shack SPL meter can try measuring a sweep or tones from a test CD and watching the meter. What you'll see is the meter jumping all over the place on tones that sound as if they are the same intensity.
***The Radio Shack meter is only good for 1khz. You have to add to different signal/frequency levels to get the proper readout. IOW...if you're measuring at 4khz and it gives 60dBA, it's NOT 60dBA. The meter (newer version) is not designed to be a 20-20khz measurement tool.***
My experience with car audio is limited. I do know that in a normal room, third octave measurements using pink noise correlate closely to what we hear. There may be differences with the car environment that I don't know about. With third octave pink noise you are looking for a frequency response that is flat up to 2 or 3 k and which then rolls of at 6 dB per octave after that.
***Usually with car audio, you get the RTA as flat as possible and then adjust the curve afterwords. The RTA is a good guideline, but not the final word. Most will roll off at 16k to 17khz, otherwise it will sound too bright.***
In a normal room using third octave pink noise, the position of the mic isn't very important, because most of the sound being measured is reflected rather than direct.
***In car audio, it's VERY critical. A reading from the head drivers position will be VASTLY different than a reading in the center console or the rear drivers position.***
To your question: it is useful to be able to measure the system without reflections because then you know exactly what the system is doing. If there is a dip caused by poor crossover design, for example, you would want to fix that problems at source. You cannot correct for that kind of problem using a graphic equalizer.
***That's where max flexibility comes in. Parametric with variable crossover settings and variable crossover slopes, as well as variable boost/cut levels for the xover will come in handy. There are alot of tricks you can do for the car, especially if you have the right equipment. With home audio, it's limited to a couple of bands, and they're not even parametric. Car audio is a different beast altogether. It's the only field outside of the professional studio where I know of that will utilize parametric eq's and delays. Most home setups won't utilize this.***
What problems have you had using third octave pink noise measurements?
***Sometimes during the pink noise generation, the frequencies are so short that the mic doesn't capture all of it. Other times it's overleveled (Alpine pink noise test disc has a +6dB over the rest of the frequencies at some parts of the disc) so that the reading is incorrect. This is where I have to go with sine wave (since the B&K is displaying all of the spectrum...I have to isolate it somehow) and single test tones. Effective and it works out nicely.***
Does your system actually cut off at 9 kHz, i.e. can you hear this?
***To me it sounds like a vast loss of detail and air. Dull sounding if you will. I'm used to SACD or HDCD sources with good speakers (Dynaudio ESOTAR series) or good headphones (Sennheiser HD580). To go back in the car after listening to these is painful! I've actually tested my hearing and on pure sine notes I can pick off 19khz. 16k to 17k (sound of TV tube being energized) is LOUD to me.***
Percy
07-14-02, 04:10 PM
#28
Driver
Join Date: Feb 2002
Location: uk
Posts: 137
Likes: 0
Received 0 Likes
on
0 Posts
Percy wrote:
***That's where all of the processors come in. Ala Alpine F#1 Status. That's probably the only processor that will account for psyacoustics, interior anomalies, etc. It's measured and evaluated by it's own mic. Lots of processing power...the equivalent of two P4 1.3's in parallel.***
Does it work?
***That's where max flexibility comes in. Parametric with variable crossover settings and variable crossover slopes, as well as variable boost/cut levels for the xover will come in handy. There are alot of tricks you can do for the car, especially if you have the right equipment. With home audio, it's limited to a couple of bands, and they're not even parametric. Car audio is a different beast altogether. It's the only field outside of the professional studio where I know of that will utilize parametric eq's and delays. Most home setups won't utilize this.***
The car audio environment has a lot of problems that you don't get with home audio. Most audiophiles would be aghast at the though of a parametric equalizer - rightly so, because it's not needed.
***Sometimes during the pink noise generation, the frequencies are so short that the mic doesn't capture all of it. Other times it's overleveled (Alpine pink noise test disc has a +6dB over the rest of the frequencies at some parts of the disc) so that the reading is incorrect.
Is that normal for these devices? I've got an FFT system that works a treat, but it needs to average multiple measurements (50 to 100) to get an accurate reading. When I've finished upgrading the speakers in my car, I'll unplug all the leads on the computer and take it out to the garage to see what I've got. At the moment my system doesn't sound like it needs eq - just more definition at the bottom end.
***This is where I have to go with sine wave (since the B&K is displaying all of the spectrum...I have to isolate it somehow) and single test tones. Effective and it works out nicely.***
I'd take take those measurements with a liberal pinch of salt.
>Does your system actually cut off at 9 kHz, i.e. can you hear this?
***To me it sounds like a vast loss of detail and air. Dull sounding if you will. I'm used to SACD or HDCD sources with good speakers (Dynaudio ESOTAR series) or good headphones (Sennheiser HD580). To go back in the car after listening to these is painful! I've actually tested my hearing and on pure sine notes I can pick off 19khz. 16k to 17k (sound of TV tube being energized) is LOUD to me.***
I'd be very intested to learn what's causing this rolloff.
19 KHz is exceptional. That would make you around 16 years old, then, Percy-
Stevie
***That's where all of the processors come in. Ala Alpine F#1 Status. That's probably the only processor that will account for psyacoustics, interior anomalies, etc. It's measured and evaluated by it's own mic. Lots of processing power...the equivalent of two P4 1.3's in parallel.***
Does it work?
***That's where max flexibility comes in. Parametric with variable crossover settings and variable crossover slopes, as well as variable boost/cut levels for the xover will come in handy. There are alot of tricks you can do for the car, especially if you have the right equipment. With home audio, it's limited to a couple of bands, and they're not even parametric. Car audio is a different beast altogether. It's the only field outside of the professional studio where I know of that will utilize parametric eq's and delays. Most home setups won't utilize this.***
The car audio environment has a lot of problems that you don't get with home audio. Most audiophiles would be aghast at the though of a parametric equalizer - rightly so, because it's not needed.
***Sometimes during the pink noise generation, the frequencies are so short that the mic doesn't capture all of it. Other times it's overleveled (Alpine pink noise test disc has a +6dB over the rest of the frequencies at some parts of the disc) so that the reading is incorrect.
Is that normal for these devices? I've got an FFT system that works a treat, but it needs to average multiple measurements (50 to 100) to get an accurate reading. When I've finished upgrading the speakers in my car, I'll unplug all the leads on the computer and take it out to the garage to see what I've got. At the moment my system doesn't sound like it needs eq - just more definition at the bottom end.
***This is where I have to go with sine wave (since the B&K is displaying all of the spectrum...I have to isolate it somehow) and single test tones. Effective and it works out nicely.***
I'd take take those measurements with a liberal pinch of salt.
>Does your system actually cut off at 9 kHz, i.e. can you hear this?
***To me it sounds like a vast loss of detail and air. Dull sounding if you will. I'm used to SACD or HDCD sources with good speakers (Dynaudio ESOTAR series) or good headphones (Sennheiser HD580). To go back in the car after listening to these is painful! I've actually tested my hearing and on pure sine notes I can pick off 19khz. 16k to 17k (sound of TV tube being energized) is LOUD to me.***
I'd be very intested to learn what's causing this rolloff.
19 KHz is exceptional. That would make you around 16 years old, then, Percy-
Stevie
07-14-02, 05:19 PM
#29
Moderator - Electronics Forum
Thread Starter
[QUOTE]Originally posted by stevie
[B]
Does it work?
***Exceptionally well!***
The car audio environment has a lot of problems that you don't get with home audio. Most audiophiles would be aghast at the though of a parametric equalizer - rightly so, because it's not needed.
***Yep. Plus in home audio, the simpler the signal chain, the better it sounds. Plain and simple. A purists ideal.***
Is that normal for these devices? I've got an FFT system that works a treat, but it needs to average multiple measurements (50 to 100) to get an accurate reading. When I've finished upgrading the speakers in my car, I'll unplug all the leads on the computer and take it out to the garage to see what I've got. At the moment my system doesn't sound like it needs eq - just more definition at the bottom end.
***The pink noise that I'm running would be collected over a period of 5 minutes. But, there seems to be some wierd randomness in the levels itself. Now if I had a good B&K pink noise generator....****
I'd take take those measurements with a liberal pinch of salt.
***I'm betting that if I had a good pink noise generator and the 2260 B&K RTA observer (at nearly 7 grand) or even the PULSE B&K (at nearly 20 grand) that the results would come out close. Take a look at the B&K Pulse. http://www.bksv.com Very slick and it shows time decay as well as other goodies.***
I'd be very intested to learn what's causing this rolloff.
***So would I!***
19 KHz is exceptional. That would make you around 16 years old, then, Percy-
***31 and counting. Don't let the years fool you...I know lots of 16 year olds that are almost completely deaf.***
Percy
[B]
Does it work?
***Exceptionally well!***
The car audio environment has a lot of problems that you don't get with home audio. Most audiophiles would be aghast at the though of a parametric equalizer - rightly so, because it's not needed.
***Yep. Plus in home audio, the simpler the signal chain, the better it sounds. Plain and simple. A purists ideal.***
Is that normal for these devices? I've got an FFT system that works a treat, but it needs to average multiple measurements (50 to 100) to get an accurate reading. When I've finished upgrading the speakers in my car, I'll unplug all the leads on the computer and take it out to the garage to see what I've got. At the moment my system doesn't sound like it needs eq - just more definition at the bottom end.
***The pink noise that I'm running would be collected over a period of 5 minutes. But, there seems to be some wierd randomness in the levels itself. Now if I had a good B&K pink noise generator....****
I'd take take those measurements with a liberal pinch of salt.
***I'm betting that if I had a good pink noise generator and the 2260 B&K RTA observer (at nearly 7 grand) or even the PULSE B&K (at nearly 20 grand) that the results would come out close. Take a look at the B&K Pulse. http://www.bksv.com Very slick and it shows time decay as well as other goodies.***
I'd be very intested to learn what's causing this rolloff.
***So would I!***
19 KHz is exceptional. That would make you around 16 years old, then, Percy-
***31 and counting. Don't let the years fool you...I know lots of 16 year olds that are almost completely deaf.***
Percy
07-14-02, 08:23 PM
#30
Moderator - Electronics Forum
Thread Starter
Stevie,
Check out this link.
http://www.linearx.com/support/knowl...dgebaseTop.htm
Look for the IASCA test disc pdf. Read the last page...interesting and I knew something amiss was going on with the test disc.
Also, pink noise is A weighted. Yes, you must also use the A weighted filter or otherwise everything will look very "bottom heavy". The only noise that is linear is "white" noise. If you have the capability, use RANDOM generated pink noise, a good RTA and you should be all set. Roger Mussell of McIntosh (he holds the patent for the LDHP speakers design that McIntosh uses...for reducing 2nd and 3rd harmonic distortions) uses this method for room acoustics (and using Bruel & Kjaer's 3347 analyzer) with very good results. His analyzer costed him 20 grand...the mic and preamps were an additional 3 grand.
Percy
Check out this link.
http://www.linearx.com/support/knowl...dgebaseTop.htm
Look for the IASCA test disc pdf. Read the last page...interesting and I knew something amiss was going on with the test disc.
Also, pink noise is A weighted. Yes, you must also use the A weighted filter or otherwise everything will look very "bottom heavy". The only noise that is linear is "white" noise. If you have the capability, use RANDOM generated pink noise, a good RTA and you should be all set. Roger Mussell of McIntosh (he holds the patent for the LDHP speakers design that McIntosh uses...for reducing 2nd and 3rd harmonic distortions) uses this method for room acoustics (and using Bruel & Kjaer's 3347 analyzer) with very good results. His analyzer costed him 20 grand...the mic and preamps were an additional 3 grand.
Percy