What causes frequency response variations above transition?

[kifeep]

There are thousands of posts on ASR about the effect of rooms modes below Schroeder, what can and can't be done to address them, and so on. There's less discussion about the transition region, but still enough to get the general sense that most issues are caused by things like SBIR/LBIR, and other sorts of non-standing-wave effects.

Above transition, there seems to be a generally (but not universally) accepted consensus that one should not use EQ to correct frequency response issues.

Ok, fine, but what causes those issues in the first place? Is there's anything one might do to address them with room changes, etc.?

For example, here are my main speakers with no processing/EQ applied (smoothed MMM measurements to focus on the large-scale variances).

And here is the manufacturer's measurements for the speaker.

Even if there's nothing to be done, I'd like to at least understand what's happening up there.

Thanks.

 

[bmc0]

A common cause is uneven off-axis response, but I don't think that applies here. The Philharmonic BMRs seem pretty well behaved in that regard from the measurements I've seen.

Is there perhaps a couch or chair (especially one with a headrest) near the microphone?

 

[DVDdoug]

above transition, there seems to be a generally (but not universally) accepted consensus that one should not use EQ to correct frequency response issues.

I believe the idea is not to correct all of the little wiggles... It's OK to boost or cut the highs if it's "too dull" or "too bright", etc. It's probably best to use your ear for any "final adjustments.) It doesn't help if it measures good but isn't as enjoyable to you as it could be.

Speakers are not perfect on-axis or off-axis and the high frequencies DO reflect and combine in-and-out of phase. But it's harder to make a useful-meaningful corrections.

Try making a high frequency sine wave in Audacity (maybe 10kHz) and walk around, or just move your head while you're listening. You'll hear huge variations with slight movements as the waves combine in or out of phase. You'll get similar wildly varying measurements with slightly different mic positions. I was doing some high-frequency experiments once with an SPL meter on a microphone stand and when I moved around behind the meter the readings changed by several dB.

I ASSUME it's worse with stereo because if the distance is different with each speaker those distances (speed of sound) create phase differences. And even if your head is in the middle between the speakers, your ears are not!

I think you'll hear similar high-frequency variations with white noise but it's been awhile since I've listened to white noise.

BE CAREFUL with high-power high-frequency test tones. Tweeters can't handle as much power as woofers and you can burn-out tweeters that don't sound that loud, or with tones you can't hear or that the tweeter can't reproduce.

 

[boxerfan88]

IME when speakers are placed close to the front wall will result in a bump around 250-400. one can tame it via EQ

 

[kifeep]

Is there perhaps a couch or chair (especially one with a headrest) near the microphone?

These are MMM measurements, but yes the area covered was near the couch. If that's the cause, I guess I'm a little surprised that would be so frequency-dependent. I would have imagined the absorption would be more broadband

Try making a high frequency sine wave in Audacity (maybe 10kHz) and walk around, or just move your head while you're listening. You'll hear huge variations with slight movements as the waves combine in or out of phase

I ASSUME it's worse with stereo because if the distance is different with each speaker

These are MMM measurements of each speaker separately.

IME when speakers are placed close to the front wall will result in a bump around 250-400. one can tame it via EQ

Thanks, but I was mostly asking about the region above 600.
.

 

[kifeep]

Is there perhaps a couch or chair (especially one with a headrest) near the microphone?

Just thinking aloud here, if the issue is reflection from the couch, wouldn't the MMM average those out? And if the issue is absorption by the couch, then I'm still surprised it's not more of an even/wide effect.

 

[bmc0]

Just thinking aloud here, if the issue is reflection from the couch, wouldn't the MMM average those out?

To an extent. You might try doing another measurement in a slightly different (average) location to see if the dip moves around. Alternatively, doing a number of sweep measurements at different points around the area may be informative.

 

[kyuu]

You didn't say what your speakers are. bmc0 mentioned the Philharmonic BMRs, are those in fact your speakers? If so, that dip from around 2.5kHz to 4kHz seems to line up exactly with the measurements:

Those other measurements you posted, regardless of what speaker they are for, are way too smoothed to be useful honestly.

 

[kifeep]

You didn't say what your speakers are. bmc0 mentioned the Philharmonic BMRs, are those in fact your speakers? If so, that dip from around 2.5kHz to 4kHz seems to line up exactly with the measurements:

I have the BMR towers, not the BMR monitors (the hint bmc0 picked up on was the filename of the graph). I don't think there are any spins (because the speakers are quite large), but the manufacturer's measurements are here: https://philharmonicaudio.com/pages/measurements/#bmr-tower

 

[kifeep]

Those other measurements you posted, regardless of what speaker they are for, are way too smoothed to be useful honestly.

Yeah, that was deliberate. I was really just trying to understand the stuff above transition and didn't want to bogged down in general advice about the bass regions.

 

[kyuu]

I have the BMR towers, not the BMR monitors (the hint bmc0 picked up on was the filename of the graph). I don't think there are any spins (because the speakers are quite large), but the manufacturer's measurements are here: https://philharmonicaudio.com/pages/measurements/#bmr-tower

Ah gotcha. Well I still maintain that the measurements on the Philharmonic site aren't useful for determining whether or not that dip in your in-room response is just the speaker's natural response. Look at their posted measurements for the BMR Monitor and compare it to spin I posted up above:

3

Can't see that dip that's clearly visible in the spin at all, at least to my eyes. Given that I believe the tower uses the same drivers as the monitor, just with an additional midrange and of course a much larger cabinet, it's not inconceivable that in reality it has a similar dip in its response. Of course it's difficult to know for sure without an actual spin. If other owners of those towers posted their in-room measurements and they all showed a similar response above the transition region that'd be a pretty good sign.

 

[dweeeeb2]

If you have the time put a hard surface on the couch like a large book or chopping board, small coffee table, angled to maximise reflection then remeasure.

 

[kifeep]

Ok, that's interesting. If the dip is, in fact, part of the speaker response, that suggests it might be a good idea to provide correcting EQ after all. I can't do a spin or anechoic, obviously, but I might try a 1m gated measurement just to see if it's closer to my in-room or closer to the Philharmonic graph. If it's not the speaker, I guess I'll need to experiment with different approaches to suss out if it's a room effect or measurement effect.

FWIW, I can kind of see the dip the monitor has at 3k - the scales are a little different and the spin graph is printed taller, so if I imagine the on-axis graph stretched out, it sorta matches...

 

[dlaloum]

Your response is the sum of reflections, refractions and resonances...(in addition to the desired signal)

the vast majority of speakers have resonances associated with their cabinet structure - and if you pull down the peaks caused by such resonances, you will also reduce the SPL's of the core signal at that frequency... so the resonance is reduced, but now you have distorted the main signal.

The same thing happens with reflection based boosts...

The thing is, with these higher frequencies, they can relatively easily be controlled with carpet, curtains, bookshelves full of books etc... (or specialised audio treatments).

So you can control the reflections within the room via furnishing relatively easily, but you cannot do much about resonances without substantively altering the voicing of the speaker.

EQ is simply the wrong tool to use with regards to resonances, and it is a very crude tool to use when trying to tame reflections.

The right place for EQ is to adjust the core signal - but to achieve that you need to first seperate it from the resonances, reflections, refractions.... and that is frequently impossible - hence the advice to leave it alone!

 

[boxerfan88]

Try measuring using MMM (configure for >50 averages); the mic tracing a reasonably large space around listening area. Then recheck if the 3kHz dip is still there or not.

If it’s still there, then likely the dip is caused by the speakers.

 

[kifeep]

Try measuring using MMM (configure for >50 averages); the mic tracing a reasonably large space around listening area. Then recheck if the 3kHz dip is still there or not.

If it’s still there, then likely the dip is caused by the speakers.

My measurements were MMM at exactly 50 averages, but maybe not over a large enough area. I'm thinking of taking gated measurements and/or MMM measurements between the MLP and the speaker (and thereby not close to any furniture). Neither would be generally useful, but they might help confirm whether or not the dip is a measurement artifact.

 

[bmc0]

I don't think there are any spins

No full spins that I'm aware of, but there are third-party measurements from Audioholics. I don't see anything that would explain the dip around 3kHz.

 

[kyuu]

No full spins that I'm aware of, but there are third-party measurements from Audioholics. I don't see anything that would explain the dip around 3kHz.

Good lord. Is making the measurements as difficult as possible to interpret one of Stereophile's objectives?

In any case, I've marked it up as best I could manage on my phone:

Does it look like maybe there's a dip in the amplitude there where I've circled in green in about the right frequency range?

 

[bmc0]

Does it look like maybe there's a dip in the amplitude there where I've circled in green in about the right frequency range?

It's a bit easier to see with this view (in my opinion):

There's a dip off axis relative to the axial response, but it's quite small (~1.5dB) and isn't quite in the right frequency range.