Home Theater Forum and Systems banner

1 - 20 of 37 Posts

·
Registered
Joined
·
290 Posts
Discussion Starter #1
Guys,
Have you looked at the CEA2010 method of establishing objective data for subwoofer tests? The standard costs about $60 but you can get the gist of it here:
http://www.audiomatica.com/wp/wp-content/uploads/2012/03/appnote_007.pdf

Essentially this is a max spl test based on thresholds of nth order harmonics; starting from 2nd and going to 9th+. I like this method as it addresses Geddes' metric to some degree. And it's backed by some real logic from the study of distortion audibility. IOW, I dig it.

I plan to do my testing this way simply because it allows for a very simple way of providing members concrete data and the manufacturers will be held accountable for living up to an industry spec; not one that has propagated from forum mouths.


Now, this is only one aspect of woofer/driver measuring but I think it's something that we should all shoot for. The other obvious measurement is FR and compression. I have posted in Dale's thread regarding the way that compression testing has been done here in the past and how I disagree with the method. Not to sound like a jerk, but it's just fundamentally flawed. I'm not sure where this method came up. But, suffice it to say, we really need to revisit it because, as it stands, it's nothing more than a glorified magnitude response. I'm copying/pasting myself here so if you've seen it, you can blow over it:

I have a very fundamental issue with the talk around compression testing here. What's catching me off guard here is people saying to test compression by increasing the output to achieve 10dB steps. This is flawed. Compression shouldn't be solely focused on the FR. Compression is a measurement of input vs output. It is not a measurement of response as you increase the output itself. For example, if you increase the input voltage at a speaker's terminals then you should get that same relational value in dB output by the DUT. Anything less is due to the effect of compression.

Yes, you are looking to see how the response changes but what's the point if a driver is so inefficient at higher output that you're having to feed it twice as much as you should for the same output?

Consider this: What happens if you're increasing SPL by 10dB but your voltage ratio from your previous input voltage of 4v to your new voltage input, which should be 12.6 volts, is actually 14.6v? All you've done is increase the output but you've failed to acknowledge the 2v loss in your test, which equates to about 14watts or 6dB! Yikes! So, yea, the curve at the reference frequency increased by 10dB, but it's not illustrating the fact that you just had to make up more than 6dB by turning the amp gain up higher and higher.

This is what compression testing should tell you. Again, voltage in vs SPL out vs what should be there. Then you get the FR curve, but most importantly, you get to see how efficiently the driver is able to use the power provided to it.

Hope that makes sense. Maybe my assumption on how this kind of testing is being performed is wrong so please correct me if so.

- Erin


Below I've attached a picture of compression testing I did on a Seas w18nx driver. This is the 20-110hz band. As long as the lines are stacked on top of each other, what you're seeing is what is expected; no loss in output vs voltage input. Where the lines deviate, is an indication of how much output (in dB) is lost with the input vs the initial voltage vs frequency. As you go higher in frequency in the chart, you can see a loss of about 0.4dB from 1v to 8v input.






This is the same thing but from 400hz to 6khz. You can see a loss of about 0.8dB at 3800hz. Likely due to inductance issues (yet to verify).


Doing the compression testing in a way that illustrates the above isn't necessarily hard. It does require extra time and some math. The math simply consists of getting a base spl measurement. Fixing a frequency so it'd be reference (50hz seems logical) and measuring the voltage applied to the driver with a 50hz sine wave. Each subsequent sine sweep should be 5 to 10dB higher output in theory so you simply do the math to see what voltage that equates to. You now have a few different voltages that correlate to a few different theoretical output (dB). Then you sweep the driver with each of these voltages. Do this a so you've gone from (say) 80dB to 110/120dB and compare the real measurements vs a theoretical. You can do this all on one graph or break down the increments separately. The goal is to show how much output the subwoofer is losing due to compression; the only way to do this is to provide real measurement vs theoretical measurement. The delta is a loss in output due to efficiency. You'll still have your FR trend, but most importantly, you'll know that you gave up 2dB in output thanks to compression at 100dB and 4dB output at 110hz. This kind of information is excellent to have. ESPECIALLY for those who are going the diy route and need to have real numbers to base purchases on. With powered subwoofers, you get what you get unless you switch out the supplied amps.



In other news, I'm trying to figure out how I can obtain klippel measurements of the Rhymik sub. I can do standard FR/HD/IMD/etc... but getting the large signal parameters and curves isn't going to be easy because I want to measure the system with the servo coil active. I just don't think there's a way to do this with the klippel. Both systems (klippel and rhymik servo) are loopback systems; closed loop. trying to get one to integrate with the other is like trying to connect two hula-hoops together. :ponder:



Hope you're still awake, lol. I just wanted to post this here rather than out in public. Though, I do think the CEA2010 standard may need to be addressed... or I at least need to have a methodology thread of my own given I'm using a different set of measurement gear and doing things a bit differently than Dale. Thoughts?

- Erin
 

·
Banned
Joined
·
22,577 Posts
I think we checked in what it would take to get setup for that and Ed Mullen at SVS said it was $600 on the low side and $1200 on the high side.
 

·
Registered
Joined
·
290 Posts
Discussion Starter #5
oh, wow... no way. there's no reason you can't do that with what Dale already has. it's just a method. did you check the link? Dale's essentially doing the same thing. I'm just saying to take the distortion parameters prescribed by the standard and apply that to the measurements. done. :)

Edit: Well, actually, I'd have to check REW to see if it can be spit out. All else fails, we can figure something out. Bottom line: there's no reason you would need to spend that kind of money. Dale's already obtaining HD measurements. I'll have to look at REW again but if it's like most, the next step isn't a huge one. I just need to look in to the stimulus.
 

·
Banned
Joined
·
22,577 Posts
If we can do it for $60 then I see no reason not to.

If that is the case then I would like to know what Ed Mullen is suggesting we need that cost so much more.

Maybe Dale can chime in and let us know... or I can find out from Ed if Dale does not have more details.

OR... are you just referring to one particular test?
 

·
Registered
Joined
·
290 Posts
Discussion Starter #7
I'm just referring to one particular test. Check the link. It's a presentation on it. The CEA2010 spec costs $60 to buy from CEA but the link is essentially a tutorial on it. There's no reason to buy the spec. ;)

Essentially what Dale needs is:
A way to do a tone burst for the prescribed time of 6.5s for varying frequencies
A way to measure harmonic distortion and compare it to the fundamental either in dB or %

That's it. He would test until the distortion increased above the CEA allotted value (depending on which order of distortion it is). That would be the max spl for that frequency. You only do a handful of frequencies and that's it. Done. :)

The tone burst as stimulus might be tricky. I didn't see it offered in the signal sweep options. So, I can't say you can absolutely do it with REW because I'm not seeing the ability to do tone burst measurements, but maybe the author can look in to this for us and write up some code to make it so. The rest of it just would take some additional time but not much.

I'm not trying to push anyone in to anything. I just think that, as simple as it is, it's something we should consider and possibly pursue.


As far as compression testing goes, Dale already has that capability. That's just some math. If you're a fan of Excel, Dale, you can probably make all of the graphs pretty easily. And still import them in to REW if you save the data off as txt and convert to frd.
 

·
Banned
Joined
·
22,577 Posts
I will mention it to John and see what he says about the tone burst being added to REW.

If we can make it happen, I have no problem with it if Dale doesn't have any objections.
 

·
Registered
Joined
·
290 Posts
Discussion Starter #9
Cool.

Link John to that PDF. That should tell him all he needs to know. Maybe he can even provide a CEA stepped distortion plot so we don't have to do the math; we can just look at the results and see where we go over the distortion limit.
 

·
Registered
Joined
·
4,127 Posts
Sorry guys, I was sleeping at the time this conversation was going on. I have my own copy of the CEA 2010. Ed will give me the software but it requires a seperate OS which is about $600.
 

·
Registered
Joined
·
4,127 Posts
I missed one of Erin's posts in this thread. I was under the impression that I had to have a specific CEA software package in order for it to be legitimate. I can run these tests based on what you are saying using the testing method in the CEA 2010 document.
 

·
Registered
Joined
·
4,127 Posts
And I think having John add it to REW would be great as long as we do not run into any issues with CEA over providing it.
 

·
Registered
Joined
·
4,127 Posts
I'm just referring to one particular test. Check the link. It's a presentation on it. The CEA2010 spec costs $60 to buy from CEA but the link is essentially a tutorial on it. There's no reason to buy the spec. ;)

Essentially what Dale needs is:
A way to do a tone burst for the prescribed time of 6.5s for varying frequencies
A way to measure harmonic distortion and compare it to the fundamental either in dB or %

That's it. He would test until the distortion increased above the CEA allotted value (depending on which order of distortion it is). That would be the max spl for that frequency. You only do a handful of frequencies and that's it. Done. :)

The tone burst as stimulus might be tricky. I didn't see it offered in the signal sweep options. So, I can't say you can absolutely do it with REW because I'm not seeing the ability to do tone burst measurements, but maybe the author can look in to this for us and write up some code to make it so. The rest of it just would take some additional time but not much.

I'm not trying to push anyone in to anything. I just think that, as simple as it is, it's something we should consider and possibly pursue.


As far as compression testing goes, Dale already has that capability. That's just some math. If you're a fan of Excel, Dale, you can probably make all of the graphs pretty easily. And still import them in to REW if you save the data off as txt and convert to frd.
My ADD must seriously be kicking in today. I should probably combine into a single post. Math isn't a big deal for me and creating the excel sheet something I am infinitely familiar with. As far as the tone bursts go, could we use the RTA feature in REW? It allows for a single tone to be played and captured. That is what I use at the bottom for the HD. Could we not just us it for the same 1/3 octave spacing and apply the bursts? That being said it would be much better to have an auto burst that you did not have to time. It would take the human error risk out of the equation.
 

·
Registered
Joined
·
4,127 Posts
My biggest concern is around the legalities of using the test and publishing anything that refers to it. When I purchased mine earlier this year they were very adamant about not publishing it. I am sure audiomatica.com probably paid some fee in order to put it out there like that. I just want to make sure we are avoiding any type of pitfalls there.
 

·
Banned
Joined
·
22,577 Posts
If we simply follow the same testing method as they do, how can that be illegal?

Of course if it is a matter of $60, that ain't no big deal either.
 

·
Registered
Joined
·
4,127 Posts
The link that Erin provided is not to CEA and I am not sure what they may have worked out with CEA that would enable them to publish that information. It's probably nothing, but I would just rather be safe than sorry on that front.
 

·
Banned
Joined
·
22,577 Posts
I will dig in to it a little more. However, that are a lot of companies using it... and I can't see them paying a huge royalty for publishing if they buy the program.
 
1 - 20 of 37 Posts
Top