Topic: Back to Speakers! And so it begins... again! Madisound BK-16 Fostex fe166 kit. (Read 37270 times)

Marvey · « **Reply #190 on:** July 30, 2013, 12:39:48 AM »

Usually, you don't want to apply PEQ based on the nearfield response taken outdoors, unless of course you actually hear the speakers under those circumstances. The nearfield response is good for identifying issues with the driver or crossover. It doesn't look like there is anything serious. (I do note among the various measurements a narrow depression at 3.5k.) The nearfield FR with its broad peak around 2k, looks like it would be perfect sitting on a table with some reinforcement from a back wall. I would expect everything 1k and below to be bumped up a bit.

As for the mid-bass hump, the speakers in that photo could be too close to the wall. Trying moving them out so there's about 12". I suspect the mid-bass boost is intentional. Many bookshelf speakers are designed with a mid-bass boost to make up for lack of low and sub bass.

Take a listening position measurement (sweep or RTA) with both speakers playing. This is going to be the one that counts the most.

OJneg · « **Reply #191 on:** July 30, 2013, 02:38:02 AM »

I agree with you purrin. But the other guy I'm talking with is kind of a nut for near-field measurements for some reason. The PEQ he suggested clearly isn't working for me at my listening position. Maybe if I listened on-axis it would. My next step will be seeing if I can improve the response at my desk since that's where I usually listen to these. That includes trying to play around with some foam in the port to cure the boominess. But until then, here's what I did today.

Surmise to say distortion was very high in the bass. No surprise to me. THD is 11% at -20dB and 23% at -10dB

3% -> 7%

.1% -> 1%

.2% -> .5%

.5% -> .6%

I read up on the RTA FFT parameters and I feel that these settings got me the most resolution and usefulness, but I'm not sure how to interpret loudspeaker distortion in a practical sense. I have questions.

Where does distortion in loudspeakers stem from in the first place? I understand how certain acoustic devices like horns and diaphragms can have natural harmonics related to their size and specs, but why do we see distortion over a broader band of frequencies?
So how accurate is this type of test anyway? What's all the stuff that looks to not be harmonically related to the signal but is clearly not the noise floor? Would it be fair to say these loudspeakers are dominated by 3rd order distortion more than any others?
How useful are these for figuring out how the speakers sound? I've heard some guys like Zaph say that he voices his speakers with a downward tilt because distortion components make them sound brighter. Similarly, if I'm playing pink noise and set my eyes on flat, won't I actually have a warmer tilt if the distortion components are summing with my input signal an octave above?
What does rising distortion tell me, as opposed to when distortion remains relatively constant with SPL?
Does anyone ever try to EQ to correct for distortion or is that just total stupidity?

Marvey · « **Reply #192 on:** July 30, 2013, 03:56:40 AM »

Quote from: OJneg on July 30, 2013, 02:38:02 AM

I agree with you purrin. But the other guy I'm talking with is kind of a nut for near-field measurements for some reason. The PEQ he suggested clearly isn't working for me at my listening position.

A lot of speaker builder guys like to use the nearfield measurement because it's the most easily replicated "standard" from one person to the next. (Its' not reasonable to replicate individual listening environments.) It's also somewhat of a "standard" used by JA @ Stereophile. This measurement actually works well if everyone puts their speakers at least several feet from the back and side walls, and has a good deal of room between the listening position and the wall behind them.

Listening position measurement is best, but for obvious reasons not always possible. Studios (and custom installations) will try to aim for a slight downward slope from 20Hz to 20kHz. 10db is often quoted. I've used anywhere from a 5 to 10db slope depending upon intended purpose and room parameters.

The good speaker designer designs for an expected case use (against walls, away from walls, etc.) and documents this. Personally, I feel all top-end speakers should have baffle-step compensation adjustments like the fancier pro-monitors.
Where does distortion in loudspeakers stem from in the first place? I understand how certain acoustic devices like horns and diaphragms can have natural harmonics related to their size and specs, but why do we see distortion over a broader band of frequencies?
So how accurate is this type of test anyway? What's all the stuff that looks to not be harmonically related to the signal but is clearly not the noise floor? Would it be fair to say these loudspeakers are dominated by 3rd order distortion more than any others?
Most speakers are dominated by 2nd order distortion. The HiVi woofer you have could a bit funky with its design.
How useful are these for figuring out how the speakers sound? I've heard some guys like Zaph say that he voices his speakers with a downward tilt because distortion components make them sound brighter. Similarly, if I'm playing pink noise and set my eyes on flat, won't I actually have a warmer tilt if the distortion components are summing with my input signal an octave above?
One can always say that distortion components can make something sound warm, especially with drivers which produce higher bass distortion in proportion to mid/treble distortion. At reasonable volumes, the distortion products will be too low to influence an RTA measurement. Besides, the distortion products will likely be be wide band. I don't think about stuff like this too much. You start to miss the forest for the trees.
What does rising distortion tell me, as opposed to when distortion remains relatively constant with SPL?
Rising distortion with increasing SPL is normal. Distortion that tends to not change with SPL probably means the driver is performing better at louder volumes where the distortion as a percentage of the fundamental is lower.
Does anyone ever try to EQ to correct for distortion or is that just total stupidity?
You can use EQ to correct for linear distortion (FR), but you cannot use EQ to correct for non-linear distortion.

I'm not big into non-linear distortion so long as it's sufficiently low. (Although small driver attempting to play bass = bad / distortion.) The masking effect will hide the audibility of most distortion.

I'm more of a stored energy / CSD guy for regular listening. The distortion stuff gets critical for high SPL pro applications though.

OJneg · « **Reply #193 on:** August 03, 2013, 05:01:41 AM »

I was a sad panda for a while there. I was trying to unscrew my port and I dislocated the two pieces of it. I was getting a real nasty vibration whenever I played LF tones (or music for that matter). Had to cut open the baffle and glue the port back together. All good now.

Here's off axis measurements. I kept the mic in the same position and used a right angle to orient the speaker at each angle:

So yeah, off-axis response is pretty wacky for the BK12. OSMT is very nice in comparison.

Instead of putting the BK12's THD by itself, I'm going to compare it to the OSMT. I think these plots are fairly useful in figuring out how each of these speakers sound actually. The OSMT's own inefficiency is working against it in this case, as it needs a lot more power to reach the same SPL as the BK12 for these measurements (Even though the fundamentals are not perfectly matched in any of these). Let's see. Note that the colors switch, sorry.

So even though the BK12 fails in treble smoothness and off-axis, it absolutely kicks the OSMT's ass in distortion. Especially where it counts, the midrange. As the fundamental goes up, the OSMT looks slightly better in comparison. It's interesting to note:

The BK12 switches from a dominant 3rd harmonic to a dominant 2nd and non-existent 3rd from 100Hz -> 440Hz. ?
The OSMT is absolutely littered with higher-order components and hash. And I think the metal cone is the culprit here. Are metal cones the balanced armatures of the loudspeaker world?

Also here's the BK12 near-field from different positions, as well as the horn output.

Notice how smooth the 1in and 2in measurements are. And note how this is only 1/24 smoothing. Even without smoothing it looks good. From 3ft away it's not nearly as clean. Why is this so? Is this just some sort of measurement artifact, or is there something that messes up the curves the farther you go away?
I'm not sure what to really glean from the the horn output plots. Might be seeing the horn peaks that are evenly spaced here? Not sure.

Marvey · « **Reply #194 on:** August 13, 2013, 04:06:59 AM »

Posting this fe166 / BK-16 graph for reference. 24" from driver. 1/24 octave (basically no smoothing.)

Marvey · « **Reply #195 on:** August 19, 2013, 09:21:34 PM »

Gated CSD 3.02ms at 24" from driver.
Fe166 Stock.
No Smoothing.

Marvey · « **Reply #196 on:** August 19, 2013, 09:30:12 PM »

Here's the group delay of the driver in the bk-16 enclosure with the mic level to the cone at 24" away.

As we can see, the bass below 300 Hz is delayed a bit by up to 10ms.

OJneg · « **Reply #197 on:** August 19, 2013, 09:39:09 PM »

I'm going to try both of these with the BK12.

AstralStorm · « **Reply #198 on:** August 20, 2013, 10:16:16 AM »

Are you sure you're not measuring your room's CSD there? Looks way too terrible for a driver.

OJneg · « **Reply #199 on:** August 20, 2013, 07:45:49 PM »

AS: A room CSD would have much more energy time wise, no? More like 300ms rather than 3ms.

I'm not getting the same "ringing" at my breakup nodes like you are. My 7k peak seems to decay equally with the rest of the treble. But REW generates waterfall plots from swept tones. Are you using an actual impulse (time-domain) to get your waterfalls?

I am finding some interesting things with regards to group delay though. The full-range drivers seem to be very much minimum phase from 1k and up. But by generating a minimum phase (with respect to amplitude response) and finding the "excess" phase, I get to see where my group delay is. Sure enough, it seems to be caused by the horn dicking around in the midrange.

OSMT

BK12

EDIT: How much space does the out of phase signal from the horn need to travel in order to combine in phase? If I try to take the same plots with the mic farther away, would I see less group delay? Or just better frequency response in the upper bass?

Author Topic: Back to Speakers! And so it begins... again! Madisound BK-16 Fostex fe166 kit. (Read 37270 times)

Marvey

Re: Back to Speakers! And so it begins... again! Madisound BK-16 Fostex fe166 kit.

OJneg

Re: Back to Speakers! And so it begins... again! Madisound BK-16 Fostex fe166 kit.

Marvey

Re: Back to Speakers! And so it begins... again! Madisound BK-16 Fostex fe166 kit.

OJneg

Re: Back to Speakers! And so it begins... again! Madisound BK-16 Fostex fe166 kit.

Marvey

Re: Back to Speakers! And so it begins... again! Madisound BK-16 Fostex fe166 kit.

Marvey

Re: Back to Speakers! And so it begins... again! Madisound BK-16 Fostex fe166 kit.

Marvey

Re: Back to Speakers! And so it begins... again! Madisound BK-16 Fostex fe166 kit.

OJneg

Re: Back to Speakers! And so it begins... again! Madisound BK-16 Fostex fe166 kit.

AstralStorm

Re: Back to Speakers! And so it begins... again! Madisound BK-16 Fostex fe166 kit.

OJneg

Re: Back to Speakers! And so it begins... again! Madisound BK-16 Fostex fe166 kit.

(adsbygoogle = window.adsbygoogle || []).push({});

Author Topic: Back to Speakers! And so it begins... again! Madisound BK-16 Fostex fe166 kit. (Read 37270 times)