• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required as is 20 years of participation in forums (not all true). Come here to have fun, be ready to be teased and not take online life too seriously. We now measure and review equipment for free! Click here for details.

Revel M16 - quasi-anechoic spinorama and misc measurements

dominikz

Active Member
Forum Donor
Joined
Oct 10, 2020
Messages
261
Likes
978
This is a quasi-anechoic spinorama measurement of the Revel M16 passive bookshelf speaker.
Although this speaker was one of the relatively early ones measured in high precision using Klippel NFS in a review by @amirm (link), I figured it is always nice to have another independent data point so I measured one of mine :)

A few notes on measurement methodology:
  • The loudspeaker has been measured at ~120cm distance from any reflective surfaces, with the microphone at 75cm distance from the speaker, so the first reflection was around 5ms after the main impulse.
    • I typically try to measure at 50cm to maximize the gate length, but with this loudspeaker I had a null in the gated measurement around 5kHz that only disappeared when I moved a bit further. Moving to 1m didn't show appreciable difference vs 75cm, and a shorter length gives a bit more window flexibility - hence the choice.
  • The reflections were removed by temporal gating at ~5ms in REW to get some (low) resolution of quasi-anechoic measurement down to about 200-300Hz range (resolution is much better above ~1kHz).
  • LF response (<300Hz) is spliced from nearfield measurements (baffle step corrected) and therefore not absolutely accurate. Take the absolute shape of the response there with a few grains of salt.
  • The loudspeaker front baffle was vertically aligned with the centre of speaker stand/turntable rotation
  • Loudspeaker stand has been covered with a blanket to reduce potential reflections and care was taken to angle the microphone boom to minimize boom reflections
  • Microphone was pointed to the tweeter axis
  • REW, Cross-Spectrum labs calibrated Dayton EMM-6 microphone and RME Babyface interface was used to measure, and VituixCAD to generate the spin (and exported with 1/24 octave resolution)
  • Measured without loudspeaker grille

Basic measurements
Anyway, here's the full spinorama for one of the units I own:
Revel M16 - Spinorama at 75cm - tweeter axis.png

Preference rating as calculated by VituixCAD is:
  • Speaker alone: 5.3
  • Speaker with sub: 7.5
Driver components:
Revel M16 - Driver Components Nearfield.png

Note that the woofer nearfield measurement shows the ~450Hz resonance that also shows very clearly in Amir's measurement. In my spin this is lost due to low resolution of gated measurements at these frequencies and short window lengths.

Horizontal directivity to 90 degrees:

Revel M16 spin 2021-09-19 Directivity (hor).png

Note that the on-axis peak around 5kHz is attenuated as we move off-axis - which is also why the LW looks smoother than the on-axis curve.

Horizontal directivity (normalized to on-axis response):
Revel M16 spin 2021-09-19 Directivity polar (hor).png

Note that the apparent 'narrowing' of dispersion around 5kHz is actually caused by the on-axis 5kHz peak explained above - since the other angles don't have it as severe and the plot is normalized to on-axis, it looks like narrowing. Therefore the non-normalized plot may be more intuitive in this case:
Revel M16 spin 2021-09-19 Directivity polar (hor) - not normalized.png

Vertical directivity (normalized to on-axis response; we can see here that crossover and acoustic response is indeed optimized for the tweeter axis):
Revel M16 spin 2021-09-19 Directivity polar (ver).png


Comparisons
Here's how the spin compares to the ASR Klippel NFS spin:
Revel M16 - spinorama - Compared to ASR Klippel NFS spin.png


The overall match is pretty nice and I believe the few differences can be easily explained:
  • My measurement is a splice of a single nearfield LF measurement with gated polar HF measurement so absolute level of the <300Hz response is not 100% precise. In addition, due to temporal gating, precision between 300-1000Hz in my measurement is pretty low (but much better as we move higher in frequency). In contrast ASR measurement is very high quality and high precision at these frequencies.
  • The difference in ER and ERDI curves is a calculation error in older Klippel SW version (since corrected), identified by @napilopez in post #6 of my Neumann KH 120A measurement. If I manually calculate my ER / ERDI curves in the same (incorrect) way there is an almost perfect match:
    Revel M16 - Early Reflections - Correct vs incorrect ER ERDI calculation compared to old Klipp...png
  • I'm not sure about the differences above ~8kHz - I'm assuming maybe in the early ASR measurements a non-calibrated microphone with HF roll-off was used? Anyway, most other M16 measurements (including Revel's own) seem to show rising HF response. Also, I believe the HF wiggliness (a 100% formal technical term) in the Klippel spin was likely caused by the microphone holder and protective cage - if I'm not mistaken @amirm modified the microphone holder and removed the cage in the meantime to avoid such reflections.
    Anyway here's a comparison to measurements from a few different sources:
    Revel M16 - On-Axis Measurements Compared.png

    All of the above measurements (of different physical units) are IMO actually very close. In my old measurement of both units' on-axis frequency response we can also see that the unit-to-unit match of M16 appears to be very good - a big plus for Revel in my book :)
Overall I'd say the match to other published results is pretty good so I'm happy :)

Distortion
[EDIT 2021-10-03]
Additional distortion measurements added in post #4.
[EDIT 2021-11-19] Distortion measurements in post #4 were incorrect and are now replaced by corrected ones in post #5
Haven't really done proper distortion tests on these so I'll just post this single tests done at 76 dB SPL at 1m (from spin):
Revel M16 - 76 dB SPL at 1kHz at 1m (normalized).png

Revel M16 - 76 dB SPL at 1kHz at 1m.png

Hopefully I'll do more tests at somewhat higher levels in the following days.

In-room measurements
We use these in our pretty reflective living room at about 2,3m listening distance. MLP is about 6° below tweeter axis. At this distance I expect the in-room response to track the PIR reasonably close, here's a comparison:
Revel M16 - spin PIR vs in-room MMM - without EQ.png

There is indeed a pretty good qualitative match above ~700Hz or so.

Equalization
[EDIT 2021-11-19]
Additional EQ profiles in post #5.
Next I designed a flat-LW-target EQ in REW:
Filter Settings file

Room EQ V5.20.2
Dated: 02.10.2021. 22:49:56

Notes:Revel M16 - spinorama LW EQ by dominikz 2021-10-02

Equaliser: Generic
LW
Filter 1: ON PK Fc 100.5 Hz Gain -3.50 dB Q 2.070
Filter 2: ON PK Fc 141.0 Hz Gain -2.00 dB Q 2.000
Filter 3: ON PK Fc 1511 Hz Gain -0.70 dB Q 1.000
Filter 4: ON PK Fc 1917 Hz Gain -1.10 dB Q 5.000
Filter 5: ON PK Fc 2511 Hz Gain 1.40 dB Q 5.000
Filter 6: ON PK Fc 3225 Hz Gain 0.70 dB Q 1.001
Filter 7: ON PK Fc 5350 Hz Gain -1.70 dB Q 4.116
Filter 8: ON PK Fc 5750 Hz Gain 1.30 dB Q 5.000
Filter 9: ON PK Fc 7541 Hz Gain 0.90 dB Q 5.000
Filter 10: ON PK Fc 13821 Hz Gain 2.60 dB Q 3.397
This EQ improves the spin a bit (simulated in VituixCAD):
Revel M16 with EQ.gif

The preference score also rises to 6,1 (8,4 with sub). Note that if you're using any kind of room EQ I personally wouldn't use the first two EQ bands shown above, and would instead correct the in-room measured response (below ~300Hz). The loudspeaker's ~100Hz hump may give you a bit more bass headroom so the EQ cut filters are more effective.
So far I was only using <300Hz room EQ, but I might also try to play with the speaker correction based on LW. If I do, I'll try to report back with some measurements and impressions.

Conclusion
Using them with a single SVS SB-1000 sub and miniDSP OpenDRC-DI for room correction in a not-too-big living room, I have to say I'm so far very happy with our M16s :). They came on my radar after seeing the ASR review and ended-up being the one that impressed us the most of the loudspeakers we listened to. A year of listening (and many measurements) later it still feels like the right choice. :)

In case @MZKM and/or @pierre (or others) might be interested to do their magic, CTA-2034-A exports from VCAD are attached as well.

[EDIT 2021-11-19] Additional distortion measurements, analysis of port plugging and additional EQ profiles were added in post #5.
[EDIT 2021-11-20] Impedance magnitude, phase and EPDR plots are added in post #8.
 

Attachments

  • Revel M16 CTA-2034-A by dominikz.zip
    69.8 KB · Views: 24
Last edited:

napilopez

Major Contributor
Forum Donor
Joined
Oct 17, 2018
Messages
1,923
Likes
7,292
Location
NYC
Awesome work @dominikz! I remember a few of Amir's early measurements had an HF roll off not present in harmans; i forgot if the cause was ever determined.

I also agree I wouldn't necessarily EQ away the bass hump in unless you're planning to use a sub or keep them very close to a wall -- Harman seems to do this in almost all of their non-pro bookshelf speakers. Clearly intentional.

anyway, about as good of a match as you can hope for other than the HF roll off. Everyone, join the quasi-anechoic measurement party please =]
 
OP
dominikz

dominikz

Active Member
Forum Donor
Joined
Oct 10, 2020
Messages
261
Likes
978
Distortion
Haven't really done proper distortion tests on these so I'll just post this single tests done at 76 dB SPL at 1m (from spin):
Revel M16 - 76 dB SPL at 1kHz at 1m (normalized).png

Revel M16 - 76 dB SPL at 1kHz at 1m.png

Hopefully I'll do more tests at somewhat higher levels in the following days.

As promised, here are a few additional distortion graphs.
Measurements were done in-room (around 1,2m away from hard surfaces) so SBIR cancellations and room mode effects are visible - note that these heavily impact the normalized graphs so I will show both versions.

[EDIT 2021-10-04] In the meantime I realized that I forgot to turn-off the subwoofer when doing the below distortion plots (hint: 35Hz peak and the dip after it :)). Since I'm not high-passing the M16 and the sub crossover is set around 55Hz the plots should not be affected much at frequencies above 80Hz or so - please disregard the LF response.
[EDIT 2021-11-19] New and correct distortion graphs are now posted in post #5

70 dB SPL
Revel M16 - 70 dB SPL at 1kHz at 1m.png
Revel M16 - 70 dB SPL at 1kHz at 1m (normalized).png


80 dB SPL
Revel M16 - 80 dB SPL at 1kHz at 1m.png

Revel M16 - 80 dB SPL at 1kHz at 1m (normalized).png


90 dB SPL
Revel M16 - 90 dB SPL at 1kHz at 1m.png

Revel M16 - 90 dB SPL at 1kHz at 1m (normalized).png
 
Last edited:
OP
dominikz

dominikz

Active Member
Forum Donor
Joined
Oct 10, 2020
Messages
261
Likes
978
As promised, here are a few additional distortion graphs.
Measurements were done in-room (around 1,2m away from hard surfaces) so SBIR cancellations and room mode effects are visible - note that these heavily impact the normalized graphs so I will show both versions.

[EDIT 2021-10-04] In the meantime I realized that I forgot to turn-off the subwoofer when doing the below distortion plots (hint: 35Hz peak and the dip after it :)). Since I'm not high-passing the M16 and the sub crossover is set around 55Hz the plots should not be affected much at frequencies above 80Hz or so - please disregard the LF response.

70 dB SPL
View attachment 156977
View attachment 156978

80 dB SPL
View attachment 156979

View attachment 156980

90 dB SPL
View attachment 156981
View attachment 156982
Today I repeated these distortion measurements, this time making sure that subwoofer was completely turned off :)

As expected the distortion responses for the most part look pretty similar though:

Distortion

70 dB SPL

Revel M16 - 70 dB SPL at 1kHz at 1m.jpg

Revel M16 - 70 dB SPL at 1kHz at 1m (normalized).jpg


80 dB SPL
Revel M16 - 80 dB SPL at 1kHz at 1m.jpg

Revel M16 - 80 dB SPL at 1kHz at 1m (normalized).jpg


90 dB SPL
Revel M16 - 90 dB SPL at 1kHz at 1m.jpg

Revel M16 - 90 dB SPL at 1kHz at 1m (normalized).jpg


Plugging the port

Next I wanted to see how the speaker behaves with the port plugged - first let me show how the LF response changes with the port fully plugged (splice of nearfield woofer measurement with gated HF measurement):
Revel M16 - port open vs plugged.jpg

We can see the bass bump is reduced a little, some LF extension is lost, but the bass roll-off now has a smooth slope. Since my stereo received doesn't have a high-pass filter, my hope is that the change in slope might help me get an even better subwoofer integration - this is something I plan on testing going forward.

Looking at THD there is just a slight change when plugging the port:
Revel M16 - 90 dB SPL at 1kHz at 1m (port plugged).jpg

Here's a comparison with and without the 'plug' (a rolled-up sock, naturally :)) that makes the difference in distortion profile easier to see:
Revel M16 - THD at 90 dB SPL at 1kHz at 1m (port plugged vs open).jpg


It was also interesting for me to see how this affects the spinorama - especially since the bass bump seems to significantly decrease the preference score of M16 (if you just flatten-out the bass bump with a 2-band PEQ the M16 preference score jumps from 5.3 to ~6). So here's a comparison of how the full spin should look with and without the port plug:
port plug.gif

So by just plugging the port preference rating as calculated by VituixCAD jumps to:
  • Speaker alone: 5.6 (was 5.3)
  • Speaker with sub: 8.0 (was 7.5)
Note that here there is also some penalty due to less LF extension with the port plugged - which is why the score with sub increases more than the score of the speaker alone. IMHO the extra LF headroom from the M16 bass bump is beneficial if one uses EQ to battle room modal effects and I personally don't see it as a negative at all.

Loudspeaker response optimization with EQ
Lastly, I've played a bit with the amazing 'optimizer' function of VituixCAD to generate a few PEQ loudspeaker correction profiles for the M16. Here's what I managed to generate.

Note that if anyone were to try them, in both of these EQ presets I'd suggest to disable the first 2-bands of PEQ completely (these just flatten the bass-bump) and instead use in-room MMM measured response to correct below ~300Hz.

1) EQ optimized for highest preference score:
This EQ profile is optimized to achieve the highest preference score; 10-band PEQ, keeping boosts below 3dB, and Q below 5.
Code:
Filter Settings file

Room EQ V5.20.4
Dated: 19.11.2021. 22:52:53

Notes:Revel M16 - EQ optimized for highest preference score

Equaliser: Generic
Filter  1: ON  PK       Fc   96.80 Hz  Gain  -3.30 dB  Q  2.120
Filter  2: ON  PK       Fc   130.0 Hz  Gain  -2.10 dB  Q  2.000
Filter  3: ON  PK       Fc    1510 Hz  Gain  -0.50 dB  Q  0.997
Filter  4: ON  PK       Fc    1890 Hz  Gain  -0.90 dB  Q  5.000
Filter  5: ON  PK       Fc    2068 Hz  Gain   0.90 dB  Q  0.991
Filter  6: ON  PK       Fc    2524 Hz  Gain   1.50 dB  Q  5.000
Filter  7: ON  PK       Fc    5562 Hz  Gain  -1.70 dB  Q  3.720
Filter  8: ON  PK       Fc    5726 Hz  Gain   1.30 dB  Q  5.000
Filter  9: ON  PK       Fc    7619 Hz  Gain   0.60 dB  Q  5.000
Filter 10: ON  PK       Fc   12775 Hz  Gain   1.00 dB  Q  5.000
pref score EQ.gif

Preference rating as calculated by VituixCAD jumps to:
  • Speaker alone: 6.4 (original is 5.3)
  • Speaker with sub: 8.7 (original is 7.5)
2) EQ optimized for LW (weight=70%) and PIR (weight=30%) linearity:
This EQ profile is mainly optimized for LW linearity (70% weight) and with secondary focus on PIR linearity (30% weight). The ratios are not based on any research per-se, but my intention was to get a 'flatter' LW than what the preference score optimization gave. Again 10-band PEQ, keeping boosts below 3dB, and Q below 5.
Code:
Filter Settings file

Room EQ V5.20.4
Dated: 19.11.2021. 23:10:13

Notes:Revel M16 - EQ optimized for LW (70%) and PIR (30%) linearity 2021-11-19

Equaliser: Generic
Filter  1: ON  PK       Fc   100.0 Hz  Gain  -3.10 dB  Q  2.380
Filter  2: ON  PK       Fc   136.0 Hz  Gain  -1.90 dB  Q  2.040
Filter  3: ON  PK       Fc    1570 Hz  Gain  -0.50 dB  Q  1.040
Filter  4: ON  PK       Fc    1949 Hz  Gain  -1.10 dB  Q  5.000
Filter  5: ON  PK       Fc    2478 Hz  Gain   1.40 dB  Q  5.000
Filter  6: ON  PK       Fc    2804 Hz  Gain   0.80 dB  Q  1.080
Filter  7: ON  PK       Fc    5706 Hz  Gain  -1.80 dB  Q  2.660
Filter  8: ON  PK       Fc    5888 Hz  Gain   1.40 dB  Q  5.000
Filter  9: ON  PK       Fc    7400 Hz  Gain   0.60 dB  Q  5.000
Filter 10: ON  PK       Fc   13610 Hz  Gain   1.30 dB  Q  5.000
LW PIR EQ.gif

Preference rating as calculated by VituixCAD jumps to:
  • Speaker alone: 6.3 (original is 5.3)
  • Speaker with sub: 8.6 (original is 7.5)
A few comments on EQ
I should quality the above EQ profiles by saying that, even if the score jumps seemingly by a lot from these EQs, most of the gain comes from removing the bass bump (~0,7 of preference score increase).
As I stated before, since the bass bump is in the modal frequency range of most rooms it can be seen as extra headroom for subtractive room EQ and IMHO should NOT be removed based on anechoic measurements alone. The rest of the response correction accounts for only a small remaining gain in preference score (0.3-0.4) and most of it may be negligible in real use.

Next steps
In the coming weeks I plan to do some listening and in-room measurements to see how all this comes together :)
 
Last edited:

ernestcarl

Major Contributor
Joined
Sep 4, 2019
Messages
1,636
Likes
1,143
Location
Canada
In the coming weeks I plan to do some listening and in-room measurements to see how all this comes together :)

Is this a pair of speakers? I'm wondering if the speakers are truly, identical for you to apply the exact same EQ. Or are you also planning on measuring (quasi-anechoic) the other one as well -- even just on-axis -- to make sure they match close enough then apply the "universal" LW EQ?
 
OP
dominikz

dominikz

Active Member
Forum Donor
Joined
Oct 10, 2020
Messages
261
Likes
978
Is this a pair of speakers? I'm wondering if the speakers are truly, identical for you to apply the exact same EQ. Or are you also planning on measuring (quasi-anechoic) the other one as well -- even just on-axis -- to make sure they match close enough then apply the "universal" LW EQ?
That is of course a very valid question in general. However from my previous measurements it appears that unit-to-unit consistency of M16 is very good. E.g. see my older measurements of my two units linked in OP here:
All of the above measurements (of different physical units) are IMO actually very close. In my old measurement of both units' on-axis frequency response we can also see that the unit-to-unit match of M16 appears to be very good - a big plus for Revel in my book :)
Graph from link:
index.php

And here's another example for the same pair - here each unit was measured on a separate occasion and in a slightly different location - however still the match is very close up to almost 20kHz:
1637369726176.jpeg

Slightly different wiggles below 2kHz here could be due to small differences in mic position and measurement setup between the two takes - but even these are less than 1dB in magnitude.

Note that there's also pretty good agreement with the measurement of a completely different unit from @amirm (apart from >10kHz). All that said, IMHO the above EQs should be pretty universal.
Anyway I plan to validate these with in-room measurements when I get the chance, to see how well the measured responses track the calculated PIR after EQ.
 
OP
dominikz

dominikz

Active Member
Forum Donor
Joined
Oct 10, 2020
Messages
261
Likes
978
Forgot to post these yesterday - I also measured the impedance of my pair.

Impedance and EPDR
Revel M16 - Impedance and EPDR (unit 1).jpg

Revel M16 - Impedance and EPDR (unit 2).jpg


Here's how the impedance magnitude of my pair compares to the (digitized) ASR plot:
Revel M16 - dominikzs pair vs ASR unit.jpg

Very good match above ~40Hz, and some variability of the peak below that - which I assume is due to slight unit-to-unit variations of the box and woofer resonant frequencies.
 
OP
dominikz

dominikz

Active Member
Forum Donor
Joined
Oct 10, 2020
Messages
261
Likes
978
Next steps
In the coming weeks I plan to do some listening and in-room measurements to see how all this comes together :)
Last weekend I managed to experiment a bit, so let me report my findings.

Plugging the ports
First I plugged the ports of the M16 and tried to achieve better subwoofer integration. After a few hours of trial and error I gave up, as I simply couldn't get a smoother LF response than what I already was getting with ports unplugged.

Room and loudspeaker EQ measurements
Next I tested the two EQ profiles described in in post #5.

First here's a reference of how the loudspeaker + subwoofer in-room response looks without any EQ applied, and compared to spinorama calculated PIR; measured with MMM:
Revel M16 pair and a single SVS SB-1000 - without EQ.jpg


Next here's how the response looks when I apply the following EQ profiles:

EQ #2: Room EQ variant #1 (simpler) + EQ optimized for preference score
  • Room correction EQ:
    • 10-band per channel IIR PEQ below 200Hz
    • Manual high-shelf filter at 40Hz with -4dB gain and Q=0,8 used to generate some low bass headroom before EQ
    • Only subtractive ('cut') filters used, sharp filters NOT allowed (i.e. Q<=5).
    • Filters manually tuned after generation for better adherence of psychoacoustic smoothed response to target slope.
  • Loudspeaker correction EQ:
    • "Revel M16 - EQ optimized for LW (70%) and PIR (30%) linearity 2021-11-19" preset from post #5.
Revel M16 pair and a single SVS SB-1000 - Room EQ1 and loudspeaker EQ optimized for maximum pr...jpg


EQ #2: Room EQ variant #2 (complex) + EQ optimized for LW (70%) and PIR (30%) linearity
  • Room correction EQ:
    • 14-band per channel IIR PEQ below 400 Hz
    • Manual high-shelf filter at 250Hz with -4dB gain and Q=0,8, used to combat the SBIR dip
    • Only subtractive ('cut') filters used but sharp filters allowed (i.e. any Q, including >5).
  • Loudspeaker correction EQ:
    • "Revel M16 - EQ optimized for LW (70%) and PIR (30%) linearity 2021-11-19" preset from post #5.
Revel M16 pair and a single SVS SB-1000 - Room EQ2 and loudspeaker EQ optimized for EQ LW (70)...jpg


Summary
As you can see in all three cases (no EQ, EQ#1 and EQ#2) the measured responses track pretty closely the respective spinorama predicted in-room responses. As expected from spinorama prediction, main difference vs uncorrected loudspeaker response is in the 2-3kHz range.

As far as my subjective impression of these EQs is concerned, here's my thoughts:
  • Room EQ: Personally I find room EQ mandatory in most cases. Both my systems have audible (and distracting) low-frequency resonances without it. Both room EQ strategies shown above sound equally good to me (and much better than uncorrected), and I'd be hard-pressed to select a favourite based on sound quality alone.
  • Loudspeaker EQ: While I'm of the opinion that room correction EQ is mandatory in most cases, I'd say loudspeaker EQ is probably not really needed with loudspeakers as good as these. I can't say that I find the sound particularly better nor worse with either EQ preset vs uncorrected - I'd actually have to do a blind ABX test of the EQ profiles first to make sure if I can even consistently detect the difference. This I haven't done, at least not yet. :) That's kind of expected IMHO, as the EQ filters themselves are not really dramatic.
 
Top Bottom