First measurements in REW (March Sointuva)

[concorde1]

My Cross Spectrum Labs UMIK-1 just arrived yesterday.

I used it in REW to measure my March Sointuva speakers with a 256k frequency sweep.

To calibrate in REW, I used the narrow_band_response_0_degree.txt file that was supplied on the USB stick by Cross Spectrum Labs.

The results look poor. I don't know if it's the speakers, or the way I've measured. At below 100 Hz the frequency response drops instead of continuing to rise until its claimed 60 Hz +/- 1.5dB.

1. My various measurements. L+R at same time. All at same height, just in slightly different positions. Measured around my listening position / ears.

2. The average of my measurements.

3. A single measurement directly in front of the speaker (very close proximity). L speaker only. This shows that the room is not to blame - the issue is still present.

So . . . any ideas on what's gone wrong?

 

[sam_adams]

3. A single measurement directly in front of the speaker (very close proximity). L speaker only. This shows that the room is not to blame - the issue is still present.

When one measures nearfield response after Keele, you must be very close to the drivers to get accurate results otherwise the room is still interfering with your measurements. Using smoothing on the graph plot will hide the room effects when measuring in this fashion.

 

[MarkS]

Nothing is wrong. The last graph shows that the level is the same (75 dB) at 60Hz as it is as 700Hz and 2kHz.

 

[restorer-john]

You have to take into account the 5 hours time difference between New Zealand and Western Australia. The woofers still have a bit of jet lag. It'll wear off.

 

[phoenixdogfan]

How do they sound? To me that looks like a very good out of the box, in room measurement. The response below 100 hz is more about your room and its nodes than anything else.

 

[alex-z]

Getting accurate low frequency data in a room is difficult, especially with a bass reflex speaker. You need to measure the woofer itself, and the passive radiators near-field (less than quarter wavelength distance) and merge them in software like Room EQ Wizard.

You should also measure the channels individually for a better idea of the overall sound. Reflections present in one channel can cause cancellation or summation in the other channel, which is possible with real music, but less likely, as both channels won't be playing exactly the same sounds.

 

[MarkS]

Or you could just enjoy them!

 

[ppataki]

So . . . any ideas on what's gone wrong?

I agree with the others before me, there is nothing wrong here
You can use EQ to flatten this response nicely - of course the room mode induced dips will somewhat remain there
If you share your mdat file I would be happy to do some simulations on how that flattened response could possibly look like

 

[concorde1]

Thanks all for valued input.

One crucial thing I forgot to mention is there is a desk between me and the speakers - not ideal. Speakers are raised above the desk height.

Is there anything I can try to help with the dips between 65-100Hz?

You should also measure the channels individually for a better idea of the overall sound.

I will try this.

 

[HarmonicTHD]

Thanks all for valued input.

One crucial thing I forgot to mention is there is a desk between me and the speakers - not ideal. Speakers are raised above the desk height.

Is there anything I can try to help with the dips between 65-100Hz?


I will try this.

A bit, but you won’t be able to avoid the dips completely as they are simply caused by physics. Look up room modes: simply speaking the dips are caused by standing waves of sound in your room and their base frequency depend on length, width and height of your room. Which dip you perceive (measure) depends on the listing position within that room.

You can
a) accept it as is
b) reposition your speakers and listening position (if they that is even possible as your desk stand there probably for a reason). REW Room Simulation can guide you. But you will only be able to improve the effect but never avoid dips.
c) EQ the peaks (that is what all the automated Room EQ software does eg REW, Dirac, Audyssey etc )
d) (Massively thick) Accoustic panels can improve but not avoid it (and in most domestic cases are out of the question)
e) Add one or even better subs, strategically positioned in the room to „fill“ the dips. This is most likely the most effective solution (most bang for buck, next to EQ, if you are otherwise limited by room, position and treatment).
f) …others please add, if I forgotten something.

There are also some good news: Psychoaccoustics show that there is a limit for humans to hear very narrow (high Q) frequency dips in music (look it up please as I am quoting from memory). It is frequency dependent and can be graphed by setting the smoothing in REW to Psy.

I did b), c) and a) and e) which is probably not so uncommon for home office audio use.

 

[staticV3]

@concorde1 here's the Sointuva's estimated in-room response:

Source

Doesn't look too dissimilar to your measurements.
Slight midrange dip, bass peaks at ~130Hz, slopes down gently to 40Hz, then drops off quickly below that.

 

[Sokel]

I would urgently reposition everything,that dip at 100Hz alone takes away the slam,authority and "foulness" in every music style.
Can you post a rough drawing of your room with dimensions?

 

[concorde1]

I would urgently reposition everything,that dip at 100Hz alone takes away the slam,authority and "foulness" in every music style.
Can you post a rough drawing of your room with dimensions?

I just drew this rough sketch. Note the room is hard wooden floor.

 

[HarmonicTHD]

I just drew this rough sketch. Note the room is hard wooden floor.

View attachment 256666

Yes checkout REW Room Simulation or other similar simulators on the web to determine the frequency of the expected dips for your speaker and listening position.

Having speakers at 1/2 the room length (and the listening position at 1/4) leads inevitably to dips (see above - standing waves are always multiple of length, height etc).

And remember no positioning will make all dips go away, it will always be a compromise which also needs to suit your living habits.

 

[Sokel]

I just drew this rough sketch. Note the room is hard wooden floor.

View attachment 256666

Seems like you're having a 3 meter side triangle?
I would reduce it to 2.1 meters moving the speakers near the desk for star.
This will also give some room to the right speaker.

 

[theREALdotnet]

I notice that the height of the room (3.4m) is about the wavelength of a 100Hz tone. With a second order floor-to-ceiling mode at 100Hz you will have nulls at 1/4 the room height (85cm) from floor and ceiling. That seems pretty close to a typical listing height, no surprise not much is happening at 100Hz.

 

[concorde1]

A development:

I have been noticing for a long time that when I use Pianoteq to play music, with certain notes in particular and with damper pedal, I can hear "distortion" or possibly "beats" (the physics phenomenon).

I took note of a recent post here that low power electronics (DAC/preamp) should not be stacked directly on top of high power electronics (power amp). This is exactly how I've had my electronics arranged in my audio system.

I have just shifted my DAC and preamp off the power amp and about 10cm off to the side. The awful distortion-like sound in Pianoteq seems to be gone.

So it will be interesting when I measure my speakers again tonight or tomorrow to see if this action has improved the frequency response.

 

[MAB]

Your measurements don't look that bad! I'm sure the speakers actually sound good. Here is my interpretation of your first post:

I drew a dotted line on Erin's response, overlaid it on top of your measurements (after adjusting both to approximately the same scale). The deviations seem to be room modes and boundary placement. See https://www.genelec.com/monitor-placement, I thought Klippel had an even better discussion but can't find it right now.

If I do the same for your near-field measurements, you match Erin closely:

You can roughly scale the close and far measurements, subtract them, and the difference is a rough estimate of your room.