Harbeth 30.2 in room, real-time frequency response measurements

Puma Cat

Puma Cat

Well-known member
Joined
Apr 5, 2013
Messages
2,533
Location
East Bay, CA
Okaaay....now that I've settled in with the Harbeth 30.2s, have 'em set up on the stands the way I like, I thought it was time to break out good 'ole Room Equalization Wizard (aka REW) and start doin' some measurements with an end ultimately to doing a DOE (Design of Experiments) to optimally integrate my REL R-305 with the 30.2s

The 30.2s are on Resonant Woods "Curved Series" maple stands using Brook Jensen's (of Resonant Woods) hard phenolic ball feet screwed-down & locked into the 1/4-20 threaded inserts in the bottom of the stands. The stands are resting on a particle board platform that has been stained with black Varathane urethane stain.

The speakers, stands, ball footers and the stand platform are shown here:
Harbeth%20Setup.jpg


The phenolic ball feet, which are very hard (same material as billiard balls), function analogously to speaker stand spikes in that there is a very small contact point between the phenolic ball-footer & the stand platform; they are leveled and very stable. The function of the particle board platform is to provide a firm, stable, non-rocking & "non-resonant and non-ringing" platform for the stands to sit upon on the carpet (this is my living room, not a man-cave). I tried many different materials, from regular wood to porcelain floor tiles to travertine stone, and a common hard dense particle board platform simply sounded the best.
The phenolic ball footers on the particle board platforms are shown.
RW%20stands%20footer%202.jpg


The Harbeths are coupled to the Resonant Woods stands with four 1.5" square pieces of beeswax cloth at each corner. Additionally, there is a small 4" X 1/2" strip of beeswax cloth on the rear support rail at the back of the stands. The beeswax cloth serves to couple the speakers to the stands, as well as provide a "constrained layer"- style damping. The stands are "open top" so that the cabinets can do their Alan Shaw "tuned cabinet resonance" thang. The beeswax cloth also couples the speakers to the stands so well that you can practically lift the stands by lifting the speakers. Moreover, it sounds really good.

Beeswax%20on%20Stands.jpg


With that, I set up Room Equalization Wizard and did a very quick first-pass test sweep, just to check that things are working with the REL....
Screen%20Shot%202019-08-25%20at%204.32.06%20PM.jpg


Not bad for a very first pass just to check settings and gear.

Next, I'll see if I can smooth out some peaks and nulls and get the response above 4K up a skosh...

More later, stay tuned....
 
Congratulations Stephen! The 30.2s are very special. BTW nice setup!

Did you get rid of the Dynaudios?
 
Beautiful setup Stephen, well done!

I love the Harbeth 30 series speakers. Looking forward to more of your results.
 
joeinid said:
Beautiful setup Stephen, well done!

I love the Harbeth 30 series speakers. Looking forward to more of your results.
Click to expand...

Thanks, Joe...

The next set of results will probably get...pretty statistical.

P-values, transfer functions, F-statistics, and interaction plots. :P

Stay tuned!
 
Ritmo said:
Congratulations Stephen! The 30.2s are very special. BTW nice setup!

Did you get rid of the Dynaudios?
Click to expand...

Hi Mike,
Thanks. With the arrival of the Harbeth 30.2s, new Shunyata Denali 6000 V2, NR-Venom series of power cables, Shunyata Sigma Ethernet and Alpha USB cables, the implementation of the optical Ethernet, Sonore OpticalModule FMC for streaming to the SOtM Ultra Neo network bridge (all powered by first-rate linear power supplies), the system has been bringing some nice improvements of late.

No, still have the big Dyns; they're presently residing in the spare bedroom. Thought I'd swap out speakers from time to time. Maybe on a quarterly basis or something similar.

Cheers!

-Stephen
 
Puma Cat said:
Okaaay....now that I've settled in with the Harbeth 30.2s, have 'em set up on the stands the way I like, I thought it was time to break out good 'ole Room Equalization Wizard (aka REW) and start doin' some measurements with an end ultimately to doing a DOE (Design of Experiments) to optimally integrate my REL R-305 with the 30.2s

The 30.2s are on Resonant Woods "Curved Series" maple stands using Brook Jensen's (of Resonant Woods) hard phenolic ball feet screwed-down & locked into the 1/4-20 threaded inserts in the bottom of the stands. The stands are resting on a particle board platform that has been stained with black Varathane urethane stain.

The speakers, stands, ball footers and the stand platform are shown here:
Harbeth%20Setup.jpg


The phenolic ball feet, which are very hard (same material as billiard balls), function analogously to speaker stand spikes in that there is a very small contact point between the phenolic ball-footer & the stand platform; they are leveled and very stable. The function of the particle board platform is to provide a firm, stable, non-rocking & "non-resonant and non-ringing" platform for the stands to sit upon on the carpet (this is my living room, not a man-cave). I tried many different materials, from regular wood to porcelain floor tiles to travertine stone, and a common hard dense particle board platform simply sounded the best.
The phenolic ball footers on the particle board platforms are shown.
RW%20stands%20footer%202.jpg


The Harbeths are coupled to the Resonant Woods stands with four 1.5" square pieces of beeswax cloth at each corner. Additionally, there is a small 4" X 1/2" strip of beeswax cloth on the rear support rail at the back of the stands. The beeswax cloth serves to couple the speakers to the stands, as well as provide a "constrained layer"- style damping. The stands are "open top" so that the cabinets can do their Alan Shaw "tuned cabinet resonance" thang. The beeswax cloth also couples the speakers to the stands so well that you can practically lift the stands by lifting the speakers. Moreover, it sounds really good.

Beeswax%20on%20Stands.jpg


With that, I set up Room Equalization Wizard and did a very quick first-pass test sweep, just to check that things are working with the REL....
Screen%20Shot%202019-08-25%20at%204.32.06%20PM.jpg


Not bad for a very first pass just to check settings and gear.

Next, I'll see if I can smooth out some peaks and nulls and get the response above 4K up a skosh...

More later, stay tuned....
Click to expand...

Like your Harbeth setup, very nice.


Sent from my iPad using Tapatalk
 
So, this is where I'll be heading with my "experiments" with the Harbeths...a Design of Experiments (aka DOE) to fine-tune the in-room response.

Some of you may remember this study from back 2011 when I did it for optimizing my Dynaudio Contour S3.4s (oops, can I say that in a Harbeth forum? :P).

Below is the ANOVA table and Leverage plots in JMP 10 for the 20 Hz response based on the DOE.

20Hz%20Screen%20Grab%20cap.jpg


The prediction profiler telling me where to set the Sub's controls for maximal response at the room mode peaks and nulls.
20%20HzProfiiler%20plot.jpg


And, of course. the surface plot for correcting a 155 Hz mode...;)
155%20Reaponse%20II.jpg


For the 'Beths, I'll probably be also looking at subwoofer position relative to the corner walls its sitting in. I'll use distance from the two walls as the continuous data variable.

I'll be using JMP 14 and a Custom Design so that I can add quadratics to the model to examine curvature in the functional responses.

When I showed this DOE to Stirling Trayle, he was all over it...

"The power of DOE compels you!" :P
 
Puma Cat said:
So, this is where I'll be heading with my "experiments" with the Harbeths...a Design of Experiments (aka DOE) to fine-tune the in-room response.

Some of you may remember this study from back 2011 when I did it for optimizing my Dynaudio Contour S3.4s (oops, can I say that in a Harbeth forum? :P).

Below is the ANOVA table and Leverage plots in JMP 10 for the 20 Hz response based on the DOE.

20Hz%20Screen%20Grab%20cap.jpg


The prediction profiler telling me where to set the Sub's controls for maximal response at the room mode peaks and nulls.
20%20HzProfiiler%20plot.jpg


And, of course. the surface plot for correcting a 155 Hz mode...;)
155%20Reaponse%20II.jpg


For the 'Beths, I'll probably be also looking at subwoofer position relative to the corner walls its sitting in. I'll use distance from the two walls as the continuous data variable.

I'll be using JMP 14 and a Custom Design so that I can add quadratics to the model to examine curvature in the functional responses.

When I showed this DOE to Stirling Trayle, he was all over it...

"The power of DOE compels you!" :P
Click to expand...

There has to be an easier way... [emoji848][emoji51]
 
nicoff said:
There has to be an easier way... [emoji848][emoji51]
Click to expand...

Actually, this IS the easiest way. Ask any good engineer.

I used to do this type of stuff for living as a Design for Six Sigma Master Black Belt working with molecular biologists, engineeers and scientists developing complex DNA and sample prep chemistries, DNA sequencers and genetic analyzers, and there is no other way to simultaneously examine multiple factors driving multiple functional responses, especially when there are conflicting responses (the cause of factor interactions).

Doing the experiments are easy; and JMP makes the analysis easy, too.
 
Puma Cat said:
With that, I set up Room Equalization Wizard and did a very quick first-pass test sweep, just to check that things are working with the REL....
Screen%20Shot%202019-08-25%20at%204.32.06%20PM.jpg


Not bad for a very first pass just to check settings and gear.
Click to expand...
That's remarkably smooth. Almost too much so, in the bass. Top looks about right. What resolution are you using? Where is the mic?

Puma Cat said:
Next, I'll see if I can smooth out some peaks and nulls and get the response above 4K up a skosh...

More later, stay tuned....
Click to expand...
I wouldn't do that. That is a "summed" pressure response at a single pressure mic position. It gives info, but that isn't quite what your 2 ears binaural hearing system hears. Move the mic around a bit more, watch the bass holes fill and see other peaks. I would go very gently there, especially boosting to fill pressure nulls. That will affect the direct response of the speaker, which is mixed in to that "summed" response. Possibly for the worse.
Take one range at a time, bass/mid/treble. Make gentle adjustments with one only, use your 2 ears to test the results with a wide variety of familiar tracks. It's very easy to go astray here. Based on the direct native responses of the speaker, I'd leave the bass alone, maybe add a shallow broad Q dip centered around 4k and then a bit of boost >12k....depending on your age.

cheers,

AJ
 
Stephen, Are you planning on using any type of equalizer? I assumed you were just playing with your subs crossover controls and room positioning of mains and the sub, perhaps listening position??
 
AJ Soundfield said:
That's remarkably smooth. Almost too much so, in the bass. Top looks about right. What resolution are you using? Where is the mic?


I wouldn't do that. That is a "summed" pressure response at a single pressure mic position. It gives info, but that isn't quite what your 2 ears binaural hearing system hears. Move the mic around a bit more, watch the bass holes fill and see other peaks. I would go very gently there, especially boosting to fill pressure nulls. That will affect the direct response of the speaker, which is mixed in to that "summed" response. Possibly for the worse.
Take one range at a time, bass/mid/treble. Make gentle adjustments with one only, use your 2 ears to test the results with a wide variety of familiar tracks. It's very easy to go astray here. Based on the direct native responses of the speaker, I'd leave the bass alone, maybe add a shallow broad Q dip centered around 4k and then a bit of boost >12k....depending on your age.

cheers,

AJ
Click to expand...

Thanks, AJ!

This was very helpful info, I will try out some of your suggestions. The mic is at one of my two primary listening positions, 7 ft 6 in. from each speaker. The speakers, per Harbeth's guidance are 2 meters apart, center to center.

As for resolution, I am using 1:3 in REW. I'll move the mic around a bit and see what happens.

Craig also suggested moving the speakers around, so I'll play with that a bit, as well. That was a good suggestion that I want to investigate just to learn what happens when I do that (we learn by doing experiments, that's the fun part! :D)

I'm not an acoustical engineer, so what is a Q-dip? I'm not sure how I would add it w/o some form of equalization. As for a boost above 12K, I have no way to affect that, either. I have an external tone control that I could try if inclined (The Delicious One from Alberto).

Thanks for the suggestions.
 
crwilli said:
Stephen, Are you planning on using any type of equalizer? I assumed you were just playing with your subs crossover controls and room positioning of mains and the sub, perhaps listening position??
Click to expand...

Hi Craig,
No, no plans on using an equalizer. You're correct that I'm just playing with sub's controls: Gain and Crossover. The control that has the biggest effect is the Gain control.

And yes, I am just playing with the position of the mains and sub, to an extent. I'm going to take you up on your suggestion to move the mains around a bit. Right now, they are about 31" in front of the front wall (which is a half-wall, BTW) I don't have much wiggle room as my space is small...you guys would be "shocked", its not a dedicated listening room, but...it works and...it sounds really good. I'm trying do the best I can with what I have to work with. The good thing is, Harbeths work really well with both near-field and mid-field listening. ;)
 
So, I followed up on Craig's suggestion and have been playing around with placement of the 30.2s. I turned the sub off to fine-tune the position of the mains, and I'm using a laser "tape measure" to measure the position of the mic to each speaker.

This plot is from my "#2" listening position, which is along a diagonal; there is effectively no back wall in this config, as the speakers fire down a 30' long hallway. This is how I listened to my Dyns for many years for "critical listening". The speakers are 9 1/2 feet from the mic; the speakers are ~2 meters apart per Harbeth's guidance.

This plot is double the resolution of the previous plot, with a 1:6 smoothing and is the average of both the L and R channels. Target SPL is 75 dB, so the horizontal reference line is set to that in the plot.

Averaged%20Left_Right%20SPL%20vs%20FR.jpg


I dunno, this seems to be lookin' pretty good....as Craig mentioned, looks like I want the sub to bring up just in the range under 80 Hz.
 
Sorry for late response, it's been a madhouse here due to impending hurricane.
Puma Cat said:
The mic is at one of my two primary listening positions, 7 ft 6 in. from each speaker. The speakers, per Harbeth's guidance are 2 meters apart, center to center. As for resolution, I am using 1:3 in REW. I'll move the mic around a bit and see what happens.
Click to expand...
You should use at least 1/6th octave, or the psychoacoustic smoothing option, looks like you've done the former. That will help identify any peaks in bass that may need mitigation.

Puma Cat said:
I'm not an acoustical engineer, so what is a Q-dip? I'm not sure how I would add it w/o some form of equalization. As for a boost above 12K, I have no way to affect that, either. I have an external tone control that I could try if inclined (The Delicious One from Alberto).
Click to expand...
The Q is essentially the frequency width/range of your boost/cut. Since you are not using EQ, its moot.
I would not have moved speakers based on you first graph despite the too low resolution, unless there were problems with sound/bass/imaging. Moving the mic around would show how much the bass changes over small distances...and why boosting nulls/dips isn't a great idea. Judicious cuts to peaks are ok.

Puma Cat said:
This plot is double the resolution of the previous plot, with a 1:6 smoothing and is the average of both the L and R channels. Target SPL is 75 dB, so the horizontal reference line is set to that in the plot.

I dunno, this seems to be lookin' pretty good....as Craig mentioned, looks like I want the sub to bring up just in the range under 80 Hz.
Click to expand...
Try posting individual L R (1/6th), not averaged. Agree the sub may need to come up a bit. How does it sound?
 
You must log in or register to reply here.
Back
Top