advertisement


Speaker/Room Measurement Witchcraftery

S-Man

Kinkless Tetrode Admirer
I finally got round to borrowing a microphone and attempting some measurements. Here is my first effort at recreating Fig 6 here in my own room:
https://www.stereophile.com/content/kef-ls50-anniversary-model-loudspeaker-measurements

Microphone: B&K Type 2669 with 5935 power supply calibrated with B&K 94dB 4230 calibrator
D-A: Focusrite 2i2
Software: REW at 96KHz

System: LS50 (port plugs fitted) electronically crossed to ZRB (Zero Resonance Bass).



I only averaged 15 measurements unlike JA's 20.
 
Is there an easier way to average several measurements in REW? I had to average them in pairs (using 1 twice): 16 ->8 ->4 ->2 ->1, which took ages.
 
If that response is indeed representative you have a serious bass problem.
almost 30dB difference between 30hz and 45hz!

Fair point.
However I have tried Eq'ing and achieved a much flatter response....BUT it doesn't sound any better!
I can also achieve a flatter response by moving the ZRBs nearer the wall but leaving the LS50s where they are. This sounds worse, possibly due to poorer time alignment.

My theory on this is that there is not much content below ~40Hz on 95% of CDs so the high levels of low bass don't actually occur in practice. Nulls are known to be less audible than peaks. I do like the "psychoacoustic" averaging feature on REW. Here is the same data averaged by a psycho:

LS50_L_R_15_average_psycho.jpg

5ysnpb22v
 
Last edited:
I've just realised that I measured L+R together, whereas JA measures them separately and then averages. I don't know if this makes much difference.

Keith, here is L and R (red) separately measured "centre of head" with 12th octave averaging:

Listening_position_red_right_12th.jpg


And with psycho averaging (which I think is easier to assimilate):

Listening_position_psycho_red_right.jpg
 
sweeping the low end you should hear very clearly the frequency thats taking off.

This is a problem that needs acoustic treatment not eq.

Moving the speakers around the room should make a big difference.

Maybe post a picture?
 
sweeping the low end you should hear very clearly the frequency thats taking off.

This is a problem that needs acoustic treatment not eq.

Moving the speakers around the room should make a big difference.

Maybe post a picture?

See post #4 and...
I don't habitually listen to sweeps. However the speaker positiong was orginally determined many years ago by using a sig gen and listening to how the speakers performed on their own and together. At that time I had ported speakers with 50Hz tuning freq (almost exactly the same as the LS50s from my measurements). The positioning has proved to be rather hard to improve upon in subjective terms even though the curves look scary. I can also dial in my LF rolloff as I please and believe me lower = better, by better I mean tighter and more tuneful.
 
Thanks S what are the dimensions of your room, I suspect the 30Hz is room mode, the dip is just a cancellation , moving the microphone would probably help that.
Treble is a little rolled off, was the microphone at ear/tweeter height?
Keith
 
I wish I had been able to measure the Kii3s like for like when they were here. The bass perfomance was the biggest difference, in favour of the ZRBs. That was before I had the LS50s.
As an aside, the LS50s are a serious notch up from my CAOW1s, but the sodding things have spoilt my 100% DIY system :(:(:(:(
 
Thanks S what are the dimensions of your room, I suspect the 30Hz is room mode, the dip is just a cancellation , moving the microphone would probably help that.
Treble is a little rolled off, was the microphone at ear/tweeter height?
Keith

5.65M (incl 50cm rectangular bay window) X 4.25M

Without a doubt the 30Hz is the main room mode and I like to sit in the null about 2/3 of the way down the room. The measurements are indeed much flatter in the bass if I move the microphone forwards a couple of feet.
The microphone was at ear height, which is a few inches above the KEF's tweeters. Nearfield the KEF's measure very flat out past 20K.
 
Here's a plot (psycho'ed) of the ZRB closer to the wall (green) and then closer + lower gain on the bass (blue).

Left_with_ZRB_at_180_lower_gain_blue_psycho.jpg


Looks flatter, but doesn't sound as good :confused:
 
I finally got round to borrowing a microphone and attempting some measurements. Here is my first effort at recreating Fig 6 here in my own room:
The curve looks a bit different to expected. Is this is a gated on-axis(ish) response or averaged over a long time? Have you added the calibration curve for your microphone to REW?

Are you using a sub for the low frequency response? The LS50 is a tiny ported speaker and below the high tuning frequency the port and cone will be largely cancelling each other as can be seen in the Stereophile plot. You can expect accuracy in the measurements to be progressively lost in this region.
 
Here's a plot (psycho'ed) of the ZRB closer to the wall (green) and then closer + lower gain on the bass (blue).

Looks flatter, but doesn't sound as good :confused:

This is my experience with DSP, and my personal pet theory (no evidence just a feeling) is that multiplying the data in REW/DSP takes the 'soul/life' out of the music. And maybe 'flattening' the FR takes the 'colour' out of the speakers which is what you liked in the first place. Taking that further; if DSP gives one a flatter response then all speakers would sound similar. I'm more drawn to passive room treatment and improved speaker positioning.
 
REW is just measurement software no DSP.
5.65M (incl 50cm rectangular bay window) X 4.25M

Without a doubt the 30Hz is the main room mode and I like to sit in the null about 2/3 of the way down the room. The measurements are indeed much flatter in the bass if I move the microphone forwards a couple of feet.
The microphone was at ear height, which is a few inches above the KEF's tweeters. Nearfield the KEF's measure very flat out past 20K.

I would enter your rooms physical dimensions into an online calaculator this is a good one,
https://amcoustics.com/tools/amroc?l=700&w=450&h=285&r60=0.6
Then try and track down the causes of the various cancellations/reinforcements, see if you can tie them to length,width,height .
I would imagine the bass bins would benefit from boundary coupling, but you say it doesn’t sound so good?
Keith
 
And maybe 'flattening' the FR takes the 'colour' out of the speakers which is what you liked in the first place. Taking that further; if DSP gives one a flatter response then all speakers would sound similar.

I think there is a lot more to it that just FR. Compare my CAOW1 (DIY) speakers to LS50s (red), nearfield:
IIRC this is with ports blocked and active crossover in place. Also, until I figure out gating, there's loads of room mixed in, so please take with a pinch of salt:

LS50_red_versus_Murphy.jpg


From the above I would assume the LS50s sound a bit brighter due to the increase in the presence region. But of course this is only the on-axis response.
In reality the pivotal difference is that the LS50 sound cleaner and faster... brass instruments sound brassy instead of slightly fuzzy. Maybe this is attributable to less box colouration (the LS50s are remarkably dead on the knuckle rap test) or maybe it's lower distortion (I've always wondered if the OW1 can really handle a 2nd order acoustic crossover at 2.5KHz) or both? Whatever it is, I doubt my rudimentary measurements are going to show it.
 
I would imagine the bass bins would benefit from boundary coupling, but you say it doesn’t sound so good?
Keith

After reading FT's excellent book I tried this. I didn't like the sound and don't agree with Toole on this, I think (but I'm not sure!!) that time alignment matters more than flat response at such low frequencies. I also believe that the brain is very good at filtering the room response itself.
Whilst playing with the measuing kit I tried again. On initial listening, late at night and quiet, I thought I was on to a winner (expectation bias caused by the pretty curves???). However at normal levels I soon reverted to my original "time-aligned" positioning.
 
Looking at those two plots (I'm assuming the LS50 is the top curve in HF region) I would guess both these speakers to sound very similar. But as you say they don't and there must be more to it. Perhaps the sweep tones used to measure are gentle (controlled/slow) compared to dynamic real world instruments say a trumpet or trombone? If this is correct what is the point of these measurements?
 


advertisement


Back
Top