Anyone ever make copies of a big name sub?

[WmAx]

So what your saying is that two subs that have identical THD and frequency response graphs will sound identical on musical material?

Well, I should point out that the THD levels on these subs is *far* below audibility, so you could raise THD considerably before it become audible. But yes, if the frequency response is the same, and distortion is below audibility, and no other mechanical noises are a factor(vent turbulence or noisy driver mechanics, for example), then yes, they will sound identical(assuming that the sound output source of the subwoofers is located exactly in the same spot in a room).

-Chris

 

[MACCA350]

Well, I should point out that the THD levels on these subs is *far* below audibility, so you could raise THD considerably before it become audible. But yes, if the frequency response is the same, and distortion is below audibility, and no other mechanical noises are a factor(vent turbulence or noisy driver mechanics, for example), then yes, they will sound identical(assuming that the sound output source of the subwoofers is located exactly in the same spot in a room).

-Chris

So impulse response is not a factor?

 

[WmAx]

So impulse response is not a factor?

Refer to post # 17.

-Chris

 

[MACCA350]

So a sub with flat frequency response and low THD will always have good impulse response? I am having a hard time believing that.

As I understand it, Frequency Response is the measurement of a varying frequency at a constant amplitude and the THD is the Harmonic Distortion nothing else, whereas Impulse response is the measurement of sound pressure versus time, showing how a device responds to an impulse. IMHO the sweeping tones of a frequency sweep(and the fact that THD charts only refer to THD) cannot adequately show the full story of how a loudspeaker reacts to an impulse(like beats in music).

One of the guy's over at AVTalk was researching the ability to record the actual waveform produced by a short dynamic music passage of different subs to see how well they actually reproduced the passage by comparing them to the input signal. Unfortunately My browser wont connect to AVTalk anymore(it's really annoying) but I believe it was under the sub woofer section.

cheers

 

[emorphien]

MACCA350 that has always been my rough understanding of it as well.

 

[WmAx]

So a sub with flat frequency response and low THD will always have good impulse response? I am having a hard time believing that.

For your subwooofer example, the rate of frequency roll-off and corner shape of the roll-off dictate the degree of 'ringing', as seen in an impulse response.

One of the guy's over at AVTalk was researching the ability to record the actual waveform produced by a short dynamic music passage of different subs to see how well they actually reproduced the passage by comparing them to the input signal. Unfortunately My browser wont connect to AVTalk anymore(it's really annoying) but I believe it was under the sub woofer section.

No need to try to record and compare a dynamic part of a music signal, though if one wanted to, it's relatively easy to do. But a music signal is not useful. One needs to derive the impulse response. No music signal will equal the broad instance spectrum of an impulse. The useful thing about having the impulse response, is that you also get precise phase data as well as frequency response data, and this allows you to observe various narrow band resonance(s) of a driver. Of course, there will be no such resonances in any competent subwoofer, since the bandwidth is so narrow and so low in frequency that the cone nor any part of the assembly should resonate in that narrow bandwidth. It's an entirely different story with mid-range drivers, however.

-Chris

 

[MACCA350]

For your subwooofer example, the rate of frequency roll-off and corner shape of the roll-off dictate the degree of 'ringing', as seen in an impulse response.

How is 'the rate of frequency roll-off and corner shape of the roll-off' shown on a frequency response graph. I mean, isn't a frequency response graph just showing the amplitude at those frequencies?

No need to try to record and compare a dynamic part of a music signal, though if one wanted to, it's relatively easy to do. But a music signal is not useful. One needs to derive the impulse response. No music signal will equal the broad instance spectrum of an impulse. The useful thing about having the impulse response, is that you also get precise phase data as well as frequency response data, and this allows you to observe various narrow band resonance(s) of a driver. Of course, there will be no such resonances in any competent subwoofer, since the bandwidth is so narrow and so low in frequency that the cone nor any part of the assembly should resonate in that narrow bandwidth. It's an entirely different story with mid-range drivers, however.

-Chris

I agree that Impulse response is more useful than using music, but it was interesting to see the differences between the two subs tested. I just read Ilkka's explanation of the use of Impulse response

It is very important that the subwoofer follows the input signal as accurately as possible. This means that the output signal has to start and stop as quickly as possible without any overhang or ringing at any frequency. Impulse response measures the accuracy of the subwoofer. Good impulse response often correlates with accurate and snappy (some say 'fast' ) sound quality.

Subwoofer's (system) impulse response should follow sound card's impulse response as accurately as possibly - both pre and post impulse.

I've just been looking at all the impulse response tests for all the subs tested and it looks like all of them show that the peak of the sub output reaches the peak of the soundcard baseline. Correct me if I'm wrong, but surely they can't all be identically dynamic, it's as if they are all modified(stretched or shrunk) so that the peak of the particular sub tested reaches the peak of the soundcard. Wouldn't that be defeating the purpose of the test and not show which subs overshoot the original signal and which ones don't reach it. In the case of stretching the subs that don't reach the full signal this would exaggerate the overhang. I realize that the graph is shown as a percentage, but maybe it should be shown differently so we can see whats actually happening. I think I just rambling now, but I couldn't find answers in Ilkka's explanation.

cheers

 

[BMXTRIX]

You know, one of the things I've always wanted to do is build my own subwoofer and speakers. Maybe someday I'll actually give it a try.

But, I'm having a tough time justifying it. You see, I have a Velodyne HGS-18 and I am wondering exactly what I might possibly gain, other than experience, if I go from that subwoofer to something I build myself. Or, more accurately, if I build a sub, and try to add it to the HGS-18, will it just sound like garbage in comparison?

In my experience, I simply haven't heard enough subs to make an honest statement about what a lot of other brands have to offer or how good DIY is compared to store bought. But, I think if you are going from cheap to DIY, you can see a big improvement with DIY for a fair investment. But, certain brands/products are excellent and not only are tough to beat with DIY methods, but include new technologies that are damn near impossible to recreate for the average DIYer.

 

[furrycute]

People always say that DIY is cheaper than buying brand names. Well, that is only true, to a certain extent.

In DIY projects, you only factor in the cost of materials, but you are not factoring in the cost of needed equipment, potential cost of the amount of time you spend working on the project, etc.

For myself, I have never been very good with tools. So for me to build a DIY subwoofer, it'll probably take me three or more tries before I can build a decent DIY sub. So I will be left with three deformed cabinets, tons of time wasted on making those cabinets, and hours upon hours of endless frustration working with tools I have no idea how to use.

If that were to be the case, I would rather spend the money and buy a good brand name subwoofer, have it delivered, unpack it, hook it up, and start enjoying the sub right away.

 

[BMXTRIX]

People always say that DIY is cheaper than buying brand names. Well, that is only true, to a certain extent.

In DIY projects, you only factor in the cost of materials, but you are not factoring in the cost of needed equipment, potential cost of the amount of time you spend working on the project, etc.

For myself, I have never been very good with tools. So for me to build a DIY subwoofer, it'll probably take me three or more tries before I can build a decent DIY sub. So I will be left with three deformed cabinets, tons of time wasted on making those cabinets, and hours upon hours of endless frustration working with tools I have no idea how to use.

If that were to be the case, I would rather spend the money and buy a good brand name subwoofer, have it delivered, unpack it, hook it up, and start enjoying the sub right away.

Ha-ha - That's very true! Working with the wood may be very difficult for some people to do well at all. Cutting the circle for the speakers is not straightforward and since I haven't done it, I'm guessing I would need a circle cutout tool for my (non existent) router.

I think though, that there are two (main) types of people who make speakers:
1. Those who know a fair bit about tools and woodworking and feel 90%+ up to the task, already own a bunch of the tools necessary, and will likely save a few bucks to make the speaker themself.

2. Those who simply are looking to learn something new about speakers and the entire process as an extension of their hobby. This second group is often the ones that go on to make 3, 4, or more speakers just to experience it all and often end up with a final product that is incredible.

But, I would think that speaker building is as much an art as many other things are and for most people, it can make far more sense to simply buy the speaker complete.

The Parts Express subwoofer kits seem like a lot of fun though...

 

[WmAx]

I've just been looking at all the impulse response tests for all the subs tested and it looks like all of them show that the peak of the sub output reaches the peak of the soundcard baseline. Correct me if I'm wrong, but surely they can't all be identically dynamic, it's as if they are all modified(stretched or shrunk) so that the peak of the particular sub tested reaches the peak of the soundcard. Wouldn't that be defeating the purpose of the test and not show which subs overshoot the original signal and which ones don't reach it.

It is proper to scale the signals to be equal. If you did not, then it would be difficult to compare. The soundcard signal is a reference signal; the point is to see how close the recorded signal tracks this reference. Two things that must be done are to scale the amplitude to be equal and align in time (to compensate for the recording/measurement delay, etc.) .

-Chris

 

[MACCA350]

It is proper to scale the signals to be equal. If you did not, then it would be difficult to compare. The soundcard signal is a reference signal; the point is to see how close the recorded signal tracks this reference. Two things that must be done are to scale the amplitude to be equal and align in time (to compensate for the recording/measurement delay, etc.) .

-Chris

But what if the amplitude is not equal in reality, why scale it to show something that is inaccurate. Why not use DB as the scale and not a Percentage of (possible)different peaks. That would make more sense to me, and also show whats really happening. If the input signal has a peak value of 85db then the sub should reach 85db, if it doesn't then that should be shown in the graph, after all this test is meant to show the accuracy of a sub.

cheers

 

[WmAx]

But what if the amplitude is not equal in reality, why scale it to show something that is inaccurate. Why not use DB as the scale and not a Percentage of (possible)different peaks. That would make more sense to me, and also show whats really happening. If the input signal has a peak value of 85db then the sub should reach 85db, if it doesn't then that should be shown in the graph, after all this test is meant to show the accuracy of a sub.

cheers

There is no absolute amplitude value in which to refer. The sub will go as loud as you set the volume control, limited by it's physical capabilities, of course. The impulse response is not intended to be a 'dynamic range' test.

-Chris

 

[MACCA350]

There is no absolute amplitude value in which to refer. The sub will go as loud as you set the volume control, limited by it's physical capabilities, of course. The impulse response is not intended to be a 'dynamic range' test.

-Chris

It may not be intended, but surely if scaled differently it could show both, I thought that would be part of showing accuracy. Surely the test can be calibrated to a certain level. It just seems that different subs will have different peak values and if they're all modified to reach the same peak value then any overhang can't be compared because they've been modified to different degrees.

cheers

 

[WmAx]

It may not be intended, but surely if scaled differently it could show both, I thought that would be part of showing accuracy. Surely the test can be calibrated to a certain level. It just seems that different subs will have different peak values and if they're all modified to reach the same peak value then any overhang can't be compared because they've been modified to different degrees.

cheers

No other option. Amplitude for reference signal to subwoofer = amplitude for the subwoofer recording trace. This is the reference point. The only way the subwoofer becomes non-linear to the point where this reference point is not correct, is if you drive at high enough SPL to drive the system into power compression. However, an actual impulse should not be used as the stimulus, as it's duration is so short that the dynamic compression might be under-stated, since the voice coil may not have had time to heat sufficiently to demonstrate the full effects of thermal resistance increase. But you can use MLS signals(which are long duration pink noise signals) and calculate an impulse from this signal, and in addition, it will be highly immune to environmental noise, unlike a raw impulse as the stimulus signal. Or, just run sine sweeps. The real dynamic range ability of the subwoofer is shown in the power compression graphs for all practical intents and purposes.

-Chris

 

[MACCA350]

Amplitude for reference signal to subwoofer = amplitude for the subwoofer recording trace.

Thats what we all hope for, but with the current tests there is no way of knowing if all subs correctly reach the amplitude of the reference signal. If the test system is calibrated(which has already been done for the previous tests)and maybe using multiple reference traces with various peak values(similar to the SPL test) and maybe try shortening the time of the impulse(just to see what happens, because if the spike is too quick we'll get to a point where maybe some subs wont even react). It just seems so simple in my mind, this is what the sub should do and this is what it did do.

However, an actual impulse should not be used as the stimulus, as it's duration is so short that the dynamic compression might be under-stated, since the voice coil may not have had time to heat sufficiently to demonstrate the full effects of thermal resistance increase. But you can use MLS signals(which are long duration pink noise signals) and calculate an impulse from this signal, and in addition, it will be highly immune to environmental noise, unlike a raw impulse as the stimulus signal. Or, just run sine sweeps. The real dynamic range ability of the subwoofer is shown in the power compression graphs for all practical intents and purposes.

-Chris

I'm not talking about long term peak output, I'm talking about the ability of a sub to follow the input signal accurately including reaching signals peak, which the current impulse response tests don't show.

cheers

 

[WmAx]

Thats what we all hope for, but with the current tests there is no way of knowing if all subs correctly reach the amplitude of the reference signal. .

There is no actual reference peak in SPL value of importance here. The incremental power compression tests demonstrate the linearity of the system in response to amplitude. If you want, you can use an impulse in SPL increments, and also compare the calculated frequency response values in this manner, but refer to my 3rd answer down.

If the test system is calibrated(which has already been done for the previous tests)and maybe using multiple reference traces with various peak values(similar to the SPL test) and maybe try shortening the time of the impulse(just to see what happens, because if the spike is too quick we'll get to a point where maybe some subs wont even react). It just seems so simple in my mind, this is what the sub should do and this is what it did do

If an impulse response is not followed, it results in distortion of the impulse response. It's this distortion of the impulse response that allows you to calculate what the speaker is distorting. Of course, any speaker will distort the signal by some degree. By performing a Fast Fourier Transform on the impulse, you can derive the amplitude response and phase response, which will specifically show such factors. In addition, you can also calculate various energy-time plots, such as energy time curves, or cumulative spectral decay graphs, to illustrate how the energy decays.

I'm not talking about long term peak output, I'm talking about the ability of a sub to follow the input signal accurately including reaching signals peak, which the current impulse response tests don't show.

Refer to the power compression graphs. These are more useful. You can repeat the same test using impulses as stimulus signals, and increasing the SPL incrementally, but the result will simply show a little less power compression than the normal plot methods, since the impulse will not heat the voice coil to the same degree. Or you can use MLS as the stimulus, and derive an impulse response from this. This will be a long term signal, and show about the same power compression (or more) than the sine wave sweep method. You will derive the frequency response from the impulse calculation, and compare in the same manner as the power compression graphs at the linked sites.

The incremental SPL frequency responses demonstrate the change in linear distortion(time and frequency response based) as the SPL changes. The THD vs. SPL tests demonstrate the non-linear distortion(harmonic distortion) as the SPL changes.

-Chris

 

[MACCA350]

Maybe I'm not explaining myself well enough. I'm not looking for THD specs or outright SPL specs, as you said they've been covered in the other tests. I'm wanting to directly compare input to output with db as part of the reference, I thought the Impulse Response test covered that but now I know it doesn't, so lets move on shall we.

Just to try and illustrate what I'm getting at. Below are the graphs from the experiment I mentioned earlier(I still cant connect to the AVTalk site so I cant compare specs or get the specific model number of one of the subs, what I do know is they were done by one of the sub testers over there). IIRC both subs were calibrated to the source within their capabilities so power compression was not an issue. So all things being equal(and I realize they may not be) they should play identically and their peaks should be the same.

Input(blue) Vs Sub1(yellow)

Input(blue) Vs Sub2(yellow)

Sub1(blue) Vs Sub2(yellow)

cheers

 

[WmAx]

IIRC both subs were calibrated to the source within their capabilities so power compression was not an issue. So all things being equal(and I realize they may not be) they should play identically and their peaks should be the same.

They would not reproduce an input waveform identically, unless they measured perfectly[no linear distortion]. They can not follow the input signal exactly, as there will be some degree of modification of frequency response and phase when a subwoofer [regardless of manufacturer, design, or cost] reproduces the input signal. One could get very close[on the plots, visibly], if they were to use an exceptionally flat and deep response subwoofer, and then implement some DSP processes for phase correction, frequency response correction and crossover(for both the main speaker high pass and subwoofer low pass).

The following graphs show a series of 3 identical bursts, each burst contains a 20Hz-100Hz range of cosines. The top graph is the original. The bottom graph is a band-pass filtered version of the top graph, using a 4th Order Butter-worth filter at 25 Hz to high pass the signal, and a 4th Order Butter-worth filter at 100Hz to low pass the signal, thus simulating the linear distortion of a typical subwoofer. The act of filtering distorts the signal in amplitude and time; this is linear distortion.

-Chris