DiyMobileAudio.com Car Stereo Forum banner

1 - 20 of 217 Posts

·
Moderator
Joined
·
18,547 Posts
Discussion Starter #1 (Edited)
NOTE:
THIS THREAD IS REGARDING FREQUENCY RESPONSE AND DISTORTION MEASUREMENT METHODS. NOT KLIPPEL MEASUREMENTS. I need to have the discussions in separate threads to lower confusion.
All klippel info and build discussions can be found here:
http://www.diymobileaudio.com/forum/klippel-reviews-driver-specs/98579-moving-forward-one-step-time.html





As most of you know, I'm going to try taking the reigns on the klippel. To comlement that data, I also will be doing my best to provide frequency response and distortion measurements (much like Zaph does). I believe that having all these measurements can really help the purchase/use decision. And, imo, they are all equally important, while some aspects may be more important depending on driver/use.
With that said, this discussion will focus solely on the FR/Distortion measurement methods. Again, all klippel discussion will be kept separate as to alleviate confusion. See the link above.


In a nutshell what I'm about to post is:
  • Does anyone care to see data or are you here to post in OT and talk about your favorite subwoofers? I don't mean this as a stab; in fact, it's a legitimate question. If no one is here for the science anymore, then my efforts described below are really not warranted.
  • Can you offer insight or improvements to the below proposed test methods?
  • Can someone help us relate the data to what we hear (further information in the meat of the post below)?

This is your chance to speak up. All of those who say this place is in the shihole... NOW'S THE TIME TO SPEAK UP. Once I'm in build mode there will be no going back and if you have a problem with the test methods then, it will be too late. I really need people to speak up now if they have insight into how testing can improve this place.
I'm about 2-3 years too late, I fear, behind Nyugen to really "pick up the torch", so to speak. But, who knows, maybe we can get back to discussion of data and what it means and a lot of people will benefit from it as I benefited from Nyugen's testing.



Get ready for a looooooong thread....

As some of you may know, I'm currently slated to be the next klippel guy.
However, in addition to klippel measurements, I will also be doing FR and distortion measurements for this website (although, I'll be sharing my FR/distortion measurements with other sites and maybe even my own website *one day*).

Having said that, allow me to define my goals:
  1. Learn about driver response, distortion, and T/S parameters for myself. To do this, I feel it is most necessary for me to go through with testing drivers. It's really the only way I can really correlate data with what I hear in order to get a further understanding of what the data means itself. I've never been one to be satisfied (or even fully understand) textbook writings. I need the experiment to supplement the reading.
  2. Provide some set of data for the many car audio branded drivers out there and hopefully clear up some of the myth about them.
  3. Have a greater understanding of sound reproduction. To some the hobby ends with a pair of 6x9's. For me, the hobby doesn't end because there's endless knowledge. I guess I'm not in it so much for 'great sound' as I am in it for a greater understanding of what it means which has ironically allowed me to enjoy music even more. The more I learn, the more I want to learn and the more I really enjoy just jamming out. It seems contrary, I'm sure, for someone to say they're wanting to be intertwined in all the details while still not losing focus on the love of music itself, but that's exactly how I feel and what pushes me to learn more.
I'd like to either start or somehow contribute to a thread that will discuss what the data means and how it translates to what we hear.
A perfect example is the 2nd order distortion in some scan speak drivers; many attribute it to the 'warm' sound. The problem is, most people looking at the data will look for the "best" measurements, and in this case, the scan might easily fail in comparison to other lower distortion drivers. That would be one example of how the data taken at hand, without understanding of how it shapes the sound, can be counterproductive. If there's a way I can somehow start lumping correlations between data and sound then that would be great. I then fall into the issue of using "audiophile speak", though, where there's really no way to convey the sound without the risk of adverse translation in the meaning.
A real debacle, but this is where I see the *educate and test* portion some fellow forum mates urge, being useful. My question is... how. How do we explain the data? I personally can't do that until I get further analysis under my belt.

I am flat out being optimistic... almost to a fault. But, the goals listed above are what keep me pursuing this fantasy. Truthfully, I'm not really sure how it will transpire, but at the least I will have learned more about this hobby and have a greater understanding of measurements in general. Yesterday alone proved to be useful as, while studying up on baffle step and diffraction, I had a few *lightbulb* moments and now a couple new things to try out for the next revision of my install.

My biggest concern (hurdle, really, as the haters don't really bother me like they used to) is the naysayers that will come out of the woodwork after data begins to be posted. I've seen how many of the past threads have gone down (klippel Ultra LV-10 thread, anyone?). This is the one particular I'm not looking forward to having to deal with. That's why I'm posting this thread now.

Ironically, I can post multiple posts regarding the proposed test plan and methods and not have a single person chime in (pretty much what is going on already). But, you can be damn certain that once the data starts going up, there's going to be a lot of hand waving by people who have no idea how to tie their shoes.


(continued below)
 

·
Moderator
Joined
·
18,547 Posts
Discussion Starter #2 (Edited)
Now, with that said... Here is my current proposed test setup:
After some serious consideration, I've decided to go through with the build of the 10" IEC baffle. I'll have to figure out a way to clamp the 10"+ drivers for klippel/measruement testing, but I've absolutely rule out the possibiltiy of building an IEC baffle capable of supporting drivers of that size.

So, that's pretty much where I'm at right now.
The next steps for me are to:
  • decide how many layers will be sufficient for the baffle (ie: 1 or 2 3/4" mdf layes, or 1 or 2 1/2" or a mix, etc)
  • damping the baffle itself via a CLD method
Once I determine the above, I'll start building.


FR/Distortion testing:
This is currently what I'm considering:
Building the 10" IEC baffle to spec. This will support tweeters to small subwoofer (10") driver testing.
Drivers larger than 10" would possibly be measured free-air. I had considered having a separate IEC baffle for 12"+ driver testing, but the problem is that I simply don't have the space to do this with as the baffle size would exceed the geometry of the test room.
If I use the common 4911/d, where d=driver diameter, then the point at which I'd have to 'stitch' together the NF and FF measurements would be good with LF response in the NF up to 327hz, which is far beyond the desired usable range for most all of us in car audio (and probably audio, in general). For 10", I'm good until 491hz. Again, well within the desired range. Couple that with the non-need for HF response, and I see no reason at all that NF measurements of a subwoofer wouldn't provide useful and meaningful results sans baffle. The real problem would then be vibrations. That's the bug that's really biting me right now.
If this logic is fair, I may even step it down to an 8" IEC baffle, which would afford me the ability to have smaller baffle inserts. I know a sheet of MDF sounds cheap, but when you are cutting a large sheet down to 12" cutouts (required to fit a 10" driver), well, that 48x92" goes pretty quickly, and at nearly $30/sheet now, it's just not that cheap.

Regarding the klippel:
All klippel testing methods/info will be discussed here:
http://www.diymobileaudio.com/forum/klippel-reviews-driver-specs/98579-moving-forward-one-step-time.html


Resources:
Here is a link to the IEC baffle:
http://www.rjbaudio.com/Audiofiles/IEC baffle.jpg


Below is a plot of the baffle step/diffraction I can expect for a 10" IEC baffle supporting 3" to 10" drivers.
Curve #1 - 10" @ 1m
Curve #2 - 10 @ 0.01m
Curve #3 - 3" @ 1m
Curve #4 - 3" @ 0.01m

 

·
Registered
Joined
·
731 Posts
I obviously wont say any names, but I have a feeling that the individuals who are going to question the validity of the results are those that have some commercial or some other form of attachment to said driver or company. I think that if someone is truly skeptical of the results, they can go conduct the experiments themselves.

The goal of this exercise should be to provide an unbiased opinion on the measured transducers. There are other resources out there which provide similar measurements (Zaph, Mark K, etc), but lets face it; we can't expect them to have the time to measure the new hot thing as soon as its released.

I think you are on the right track. In regards to frequency response measurement and distortion, experimental apparatus isn't nearly as important as CONSISTENCY. All drivers that you measured should be done on an even playing field for direct comparison. Even if distortion measurements deviate from one scientist to another, the point remains that measurements within a population set are comparable. I know Zaph uses something like 95db/m for measuring distortion. This individual uses 100db @ 30 inches. I think in our case, we want to aim for the latter due to our higher noise floor environment.

That website, as well as Zaphs have a pretty good explanation on the measurement apparatus and method. Though I'm sure you have found your own resources for reference. Thanks for taking the time to do this, it really is an asset to our community.
 

·
Moderator
Joined
·
18,547 Posts
Discussion Starter #4
I think you are on the right track. In regards to frequency response measurement and distortion, experimental apparatus isn't nearly as important as CONSISTENCY. All drivers that you measured should be done on an even playing field for direct comparison. Even if distortion measurements deviate from one scientist to another, the point remains that measurements within a population set are comparable. I know Zaph uses something like 95db/m for measuring distortion. This individual uses 100db @ 30 inches. I think in our case, we want to aim for the latter due to our higher noise floor environment.
Agreed. My goal isn't so much absolute accuracy as it is consistency within my own methods.
As far as testing levels, I planned to mimic what Nyugen had done at one point, which was 95dB, at one point.
The hard part is when I 'stitch' the NF and FF results together. But, that'll be something I'll just have to do when I test. Obviously in the NF, the voltage output will be lower as to provide xdB levels, while at 1 m (FF), the driver will have to play louder to attain xdB. Although, I may find (upon further research) that simply keeping the levels the same and stitching will suffice.


That website, as well as Zaphs have a pretty good explanation on the measurement apparatus and method. Though I'm sure you have found your own resources for reference. Thanks for taking the time to do this, it really is an asset to our community.
I've got about 15 links currently and that's one source I used to come up with the baffle inserts design. I've also got D'Appolito's Testing Loudspeakers book on the way.



I appreciate your feedback.

- Erin
 

·
Moderator
Joined
·
18,547 Posts
Discussion Starter #5
The test baffle will be made of a single layer of 3/4" mdf until further testing and analysis proves otherwise.
This seems to best way for me to be able to test drivers. I'm thinking I'll have to add bracing here and there, but I'll have to see what happens once I get some drivers on the baffle.
Again, if anyone has suggestions, I'm open to them.
I will be buying the wood within the next couple days.

If any mfg's happen to see this, consider chiming in if you have input. Your drivers will be tested, even if I have to buy them myself. ;)
 

·
Registered
Joined
·
3,032 Posts
I don't see any "flaws" in your testing methods, but then again I'm still at the top of the tip of the iceberg when it comes to understanding the Klippel. :)

I, for one am not really concerned about the lack of testing of subwoofers. Although I think it could spark some interest from some of the newer & younger members to include a subwoofer test occassionally.

Sorry if I missed it above, but how often do you intend to test drivers?
 

·
Registered
Joined
·
307 Posts
"Does anyone care to see data or are you here to post in OT and talk about your favorite subwoofers? I don't mean this as a stab; in fact, it's a legitimate question. If no one is here for the science anymore, then my efforts described below are really not warranted. - yes
Can you offer insight or improvements to the below proposed test methods? - nope, well... extruded shapes ("i", "t", and "l" angles are good for this).
Can someone help us relate the data to what we hear (further information in the meat of the post below)?" nope, not me.
 

·
Moderator
Joined
·
18,547 Posts
Discussion Starter #9
I like that... flood the forum with so much data that people won't have time to complain. lol. ;)


I'm open to suggestions, which is why I'm having this thread now. Please provide specifics, though. I'm sure I could shoot holes in my testing method (I have a few times already), but I'm nearing the point where 'good enough' is taking precedent.

I've put in many, many hours of research in to the topic which lead me to the decision I've made.
If I allow people to dictate the testing apparatus after I post results, well, we'll never get meaningful results because there will always be something to complain about. Especially from those who have no insight into the topic itself; they'd rather just act like they have half a clue. Matter of fact, the gung-ho testers I've ran this method by (and probably a few who you have read results from on their own sites) have all given their "OK" when I've discussed it with them.
But, my goal is to give anyone who wants to add input to do so now... before it's built. I truthfully will not have time to do 3+ revisions of the same baffle (one that's an AES standard).

So, again, if anyone has suggestions, please give them and please do so with specifics.

Thanks, all.
 

·
Registered
Joined
·
1,486 Posts
Erin,

As I mentioned earlier, I'm glad to see the KDA found a good home. I have a few comments (keep in mind this is coming from someone with zero speaker testing experience).

Obviously in the NF, the voltage output will be lower as to provide xdB levels, while at 1 m (FF), the driver will have to play louder to attain xdB
Wouldn't you want to keep the NF and FF drive voltage constant and just adjust the plot as necessary to get things to align. The reason I suggest this is that you would expect more distortion with higher drive levels. If you increase the drive level between NF and FF measurement I would expect a discontinuity in the distortion plots. If you leave the drive level the same the distortion plot should be continuous but just reduced by the amount dictated by the change in measurement location. To me, this would be much easier to address (simple plot offset) and would provide more usable data.

On the subject of subwoofer measurement and vibration, I have a few ideas. Do you have any idea where you will set this up (upstairs?)? Depending upon your WAF I think we can put something together that will work very well. Think all thread and some bracing...
 

·
Registered
Joined
·
3,612 Posts
Erin,

I am VERY interested in this testing you will be doing and right now, I don't believe there is a better person out there conducting it. I will repeat what I was told before the midrange test and what I know you have already been told.....you are a crazy man and there is no way this can turn out good on your side. With that said....do it anyway. We need this info. Be prepared and when you think you have thick enough skin, double or triple that thickness. My experience is that you will have some people who either guide you or only guide you to a point so as to leave themselves some wiggle room if the results aren't flattering. I gave every opportunity with my test, but of course, after the entire thing was finished, then I was questioned. Since your measurements are technical in nature, you may have better results, but the Ultra thread still makes me worried. I like the fact of trying to decipher the results, however, because I don't think many know what the results even mean and your example of the scans is a great one. Good luck and if you need anything from me, please let me know!
 

·
Moderator
Joined
·
18,547 Posts
Discussion Starter #12
Wouldn't you want to keep the NF and FF drive voltage constant and just adjust the plot as necessary to get things to align. The reason I suggest this is that you would expect more distortion with higher drive levels. If you increase the drive level between NF and FF measurement I would expect a discontinuity in the distortion plots. If you leave the drive level the same the distortion plot should be continuous but just reduced by the amount dictated by the change in measurement location. To me, this would be much easier to address (simple plot offset) and would provide more usable data.
Totally gotcha there, man. I'm 110% with you.
I really should've mentioned that in my reply but I was in a hurry and didn't type up the difference in distortion measurements at various levels.
However, I'm not quite certain yet how the dayton software will permit me to stitch distortion measurements together.
I think you are absolutely right as far as stitching for distortion... I guess my reply was more in terms of FR. I really need to sit down with the dayton software to see how it will handle stitching for either measurement type.

I may be able to take the real data (ie: text file data) and adjust the offset to merge the results, but I'm not sure how the results are spit out (maybe an .frd file) for distortion. I know it is .frd for FR.



SSSnake said:
On the subject of subwoofer measurement and vibration, I have a few ideas. Do you have any idea where you will set this up (upstairs?)? Depending upon your WAF I think we can put something together that will work very well. Think all thread and some bracing...
Yes, the setup will be upstairs. I'm actually hoping it will be put together in time for the GTG so everyone can see the measurement rig and maybe we can test some drivers ourselves that day. :)
My problem isn't so much creativity as it is time. The next time I'll have free to build anything is next weekend... so, every other weekend (when my wife is off work). Otherwise, I'm pulling daddy duty about 90% of the time and Layla isn't taking long enough naps for me to build anything. :( ;)

But, if it's something you think you have time to help with, I'm definitely game. I'd welcome the help.
 

·
Moderator
Joined
·
18,547 Posts
Discussion Starter #15
Will do. I mentioned something to that effect in another thread elsewhere.



In other news, I finally broke down a couple nights ago and bought a good plunge/fixed based Bosch router set. got a steal on it, too.
I just picked up Jasper's 1-7" jig so I can flush mount small drivers such as tweeters.
I also purchased a couple caps for tweeter protection (many testers don't use these, but I will since I'll likely be testing other peole's drivers and some rare ones I own).
The crossover point will be around 700hz for both 4ohm and 8 ohm drivers.

So far I've got nearly $600 invested into the testing and I haven't even bought construction materials and I haven't even touched the topic of building a stand for LF driver testing.... :eek:
I may actually have to start a paypal donation fund. :(
 

·
Moderator
Joined
·
18,547 Posts
Discussion Starter #16
looking for some input here regarding driver testing levels, which Charles touched on above.

The typical driver testing is done at 2.83v @ 8 ohm. So, about 1.42v @ 4ohm. That's pretty much industry and hobbyist standard, it seems.
I'm more of the mindset that achieving a desired SPL is more pertinent because distortion increases with volume but moreso because, in the car, we're after a certain SPL threshold rather than X volts.
Though, testing at X volts also gives you a better understanding of the driver's sensitivity. But, surely someone who reads a driver's efficiency rating can summise the difference themselves. :)

I think that somewhere in the 90-95dB range would be about right, but am open to suggestions. It's hard for me to say what everyone else's listening levels are. I couldn't imagine that 95dB isn't enough for most anyone, though, unless they're driving in a 4x4 with mudders on and the roof off.
 

·
Registered
Joined
·
1,310 Posts
First off, I would like to say that I think selecting you/volunteering to do this is a wonderful thing. I think that your reviews are very well thought out and do a good job of presenting the facts without getting into too many personal superlatives. I also appreciate the open thread for input as I have some ideas I'd like to toss into the fold.

1.) I know that you are doing all of this in your spare time, and that the actual testing isn't usually the longest part of the process, it is building the baffle for the drivers. I think it would be a good idea to make your cutout 11 1/2x11 1/2. The reason for this is the USPS Flat Rate boxes. The large one is 12x12x5 1/2. Using that baffle cutout size would allow members with the materials and ability to make the baffle and mount the driver to it before sending it to you. The benefit then works both ways: you get something that is ready to go and the shipper gets VERY good protection for the speaker.

2.) I'm not sure if you're going to do subjective listening tests. I know npdang sometimes did them, but it seems to be a double-edged sword. Some people will always comment that it isn't how a speaker measures, but how it sounds. Others will say that personal opinion shouldn't be included. I personally think they are beneficial. If you chose to do them, I had a couple of suggestions.
-Pick a couple of standard tracks that you play on all of the speakers so the evaluation is consistent and only play the speaker being tested (no other drivers) so the evaluation is isolated.
-Write a review and make it a sticky on the drivers you know best (Scans?) and use that as your reference point. Again, this isn't to compare, but to give reference. If you give a good review of the baseline components, you can describe things as being heavier, darker, quicker, thinner, etc than said reference. Since people's taste are different, none of those comments make a driver better or worse than the reference drivers, but it does help a potential buyer know what they may be getting sound-wise.
-Do the subjective test after the physical performance tests. That way, you can help explain how the graphs CAN give you an idea of how a speaker sounds. Things like, "the warm sound of the speaker apprears to be from the higher even-order harmonics" or "though the testing shows a spike in response here, it doesn't seem to be very audible"

3.) I think it would be great if a single post could be stickied that compiled all of the testing data. Even if it is a "Cliff Note's" version, it would be great to have all of that data in one place. I'm not sure of if/how that can be done on a forum, but I think the first choice would be to do something like the comparisons are done on Zaph Audio. If not, maybe a spreadsheet type setup could be used. If could seperate the drivers by category vertically (columns), with each listed top to bottom alpabetically (so they aren't "ranked"). Then going across (rows), there could be data located. Maybe just T/S parameters, suggested high-pass, suggested low-pass, diamter, depth, MSRP, and a clickable link to the full-blown test/review. I know I have spent countless hours with 20-30 tabs open looking at offerings from Madisound, Parts-Express, and various actual car-audio websites trying to narrow down a driver I think would fit my needs. I think having something like this in one location would be very beneficial.

NOTE: I don't have any knowledge if my last suggestion can be accomplished on a forum. If anybody does know if it can be done and how to do it, I think your input would be greatly appreciated.
 

·
Moderator
Joined
·
18,547 Posts
Discussion Starter #18
1.) I think it would be a good idea to make your cutout 11 1/2x11 1/2. The reason for this is the USPS Flat Rate boxes. The large one is 12x12x5 1/2. Using that baffle cutout size would allow members with the materials and ability to make the baffle and mount the driver to it before sending it to you. The benefit then works both ways: you get something that is ready to go and the shipper gets VERY good protection for the speaker.
EXCELLENT idea!
I hadn't even considered that. Though, I was thinking earlier what size I'll need to have my baffle cuts be. That makes the decision for me. :)
Plus, like you said, it'll be easier for people to to send the driver mounted to the baffle piece and they'll have a bit of wiggle room in case their cut isn't just right and I need to flush trim it. Good call, man!


2.) I'm not sure if you're going to do subjective listening tests. I know npdang sometimes did them, but it seems to be a double-edged sword. Some people will always comment that it isn't how a speaker measures, but how it sounds. Others will say that personal opinion shouldn't be included. I personally think they are beneficial. If you chose to do them, I had a couple of suggestions.
That way, you can help explain how the graphs CAN give you an idea of how a speaker sounds. Things like, "the warm sound of the speaker appears to be from the higher even-order harmonics" or "though the testing shows a spike in response here, it doesn't seem to be very audible"
I'm inclined to agree with you here, but I need more time to consider it. It's a slippery slope for sure.


3.) I think it would be great if a single post could be stickied that compiled all of the testing data. Even if it is a "Cliff Note's" version, it would be great to have all of that data in one place. I'm not sure of if/how that can be done on a forum, but I think the first choice would be to do something like the comparisons are done on Zaph Audio. If not, maybe a spreadsheet type setup could be used. If could seperate the drivers by category vertically (columns), with each listed top to bottom alpabetically (so they aren't "ranked"). Then going across (rows), there could be data located. Maybe just T/S parameters, suggested high-pass, suggested low-pass, diamter, depth, MSRP, and a clickable link to the full-blown test/review. I know I have spent countless hours with 20-30 tabs open looking at offerings from Madisound, Parts-Express, and various actual car-audio websites trying to narrow down a driver I think would fit my needs. I think having something like this in one location would be very beneficial.
I'd really like that.
I'd really like to have some way where we can compare data on the same plot without me having to go through and do it each time. Maybe some java applet or something of the sort.
Zaph's page that allows you to switch between drivers and their measurements is something I'd love to do. Here's an example:
Zaph|Audio
I just don't know how I'd do it as I know zilch about java/web code... especially in the forum context. Not even sure if we could do that on a forum. Might need my own web page (which I don't have time for right now... maybe one day).
 

·
Moderator
Joined
·
18,547 Posts
Discussion Starter #19 (Edited)
the rabbit hole gets deeper....

I brought up the question of distortion correlating to what we hear earlier. I then posted the following on diyaudio forum. Got a nice nugget from Earl Geddes...
someone linked me to this thread as they know I am working on making a test apparatus for driver testing of my own.
Personally, I'm intrigued by the relationship of harmonic distortion and how a driver sounds. Can HD results give you an inclination of the sound a driver has? A specific example is one of scan speak revelator drivers, where it is often read that their odd order harmonics yield its "warm" sound (whether that is valid or not is not my concern). The notion is there.

I've been reading D'Appolito's book "Testing Loudspeakers" and in it, I found the following:
I have not found a quantitative or qualitative relationship between the various distortion types you can easily measure and loudspeaker preference. The audibility of nonlinear distortion is a complicated issue. It is relatively easy to detect a few percent distortion in simple signals such as a pair of sine waves. However, large levels of distortion can be tolerated in complex program material such as rock ‘n’ roll music. In my experience, the maximum sound pressure level a speaker can generate is dictated by the level of distortion the listener will tolerate. Distortion measurements do not directly predict how a speaker will sound, rather they help us judge driver linearity and by implication driver quality... Intermodulation distortion (IMD) produces output frequencies that are not harmonically related to the input. These frequencies are much more audible and annoying than harmonic distortion.

What does this tell me?
Well, if the above is valid, and there is little correlation between harmonic distortion and what you prefer to hear, and if a preference correlation isn't found, then it seems to say that you can't hear a difference... otherwise, subjectivity would leak in and you would then prefer a certain driver over another. I also have to think that measuring drivers at a fixed output in order to compare distortion levels at X volume is moot .
Of course, maybe I'm putting words into his mouth or I'm misunderstanding him altogether.

Any input on this?
I could go on for days on this topic, but to make a very long story short - No. There are several papers on my web site on this topic if you want to read them ( Papers).
me in return said:
Thank you VERY MUCH.

You wouldn't believe how hard a google search is on "correlation, sound, harmonic, distortion" is to obtain meaninful results. Seriously. I've been going in circles for days. :rolleyes:

I did find it interesting to see the large group of people who were extremely sensitive to distortion based on Klippel's method.
Configuration

However, the vast spread really makes it hard to say that X level of distortion is audble for the entire human race, and furthermore there's been no quantitative consensus on what each person who took the 'test' was running in regards to equipment.

It really makes me wonder if anything other than standard FR (and polars) and maybe CSDs (for resonance) supplemented by simple impedance sweeps is needed. IOW, it seems to me that HD levels may not be as important a test criteria as I thought. Especially since i've not been able to find any analysis or white papers strongly correlating THD to characteristics of a drivers "sound".

Additionally, I wonder just why we bother with THD at all? Is it simply a means of making sure the levels are below audibility threshold? Can we use the data to pinpoint specific 'problem areas'? IOW, since THD hasn't been shown to link 'sound' to the data, is the data still useful from a mechanical and ultimately acoustical POV (ie: resonance that becomes audible at x volume)?


Thanks again,
Erin

To summarize:
Two well known and well respected experts (I hate that term, btw) agree that there is no evidence of correlation in the distortion levels of a driver and what you hear/preference.

If that is indeed true, I can therefore only assume that distortion measurements are waste of time.
However, I want to believe that something can be gained from them...
 

·
Registered
Joined
·
1,486 Posts
I have read these papers before (about a year ago so forgive me if I mis-speak) and cannot argue that Gm seems to track more closely to what we hear (over THD and IMD - I would love to see additional test results on this subject). However, throwing out distortion measurements all together seems to provide little benefit as well. Has Geddes determined a practical test method for evaluating Gm (I'll review the papers when I have time to see if one was indicated)? If you convolved the orginal signal with the measured response you may be able to identify the distortion products but how do you correlate the distortion products to the orinigal signal to determine the order of the distortion products and other info required for Gm calculations? I can see how you might do this for simple sine waves but for complex signals I am not as confident...
 
1 - 20 of 217 Posts
Top