[Stephen Scharf]

Background: while these measurements are not of Danny's wonderful GR-Research series of loudspeakers (I also own a pair), my thought was that with all the wisdom and knowledge that Danny has taught us about the importance of measurements over the years, that the folks in this forum, in particular, would appreciate seeing some real-word, in-room measurements of another monitor-type speaker, in this case my Harbeth 30.2 40th Anniversary standmount monitors.  The Harbeths are designed by Alan Shaw in the classic BBC monitor tradition.

How's this for "flat"? This is the actual in-room measured response of the Harbeths measured from the seated listening position using a NIST-traceable mic and...without the sub in the system.



The R Channel in the green trace is down a bit in the lower frequencies (below 70 Hz) compared to the L channel (shown in the red trace) because the R channel speaker fires towards a "halfwall" that separates my kitchen area from the living room (listening room) area, and causes a bit of a back-reflection, whereas the L channel, shown in the red trace, fires down a long hallway and doesn't incur a back-reflection.

My thought in posting this is that Danny and the gang would enjoy seeing this data.

Cheers, gang, and please share your thoughts.

[settador]

While these measurements and this graph are over my head, the sound of my Harbeth 30.2s saturate my brain!

[newzooreview]

That is consistent with my experience with the Harbeth 30.2 in my room in comparison to other speakers that measure flat in room (e.g., Ascend Acoustics speakers designed with the Klippel Near Field System).

It's been decades since Harbeth retained any of the "BBC dip" in 2-5k frequency range, and measurements from Stereophile, for example, bear this out.

[S Clark]

Not quite as smooth as you might think due to the scale of the graph.  There's a 10 dB variance between 60 and 100 hz.  You're going to hear that as a jazz bass makes a run down a scale.    But it is certainly flat above 250Hz. 
I haven't heard a pair of Harbeth's in a decade.  I remembered it as nothing wrong, but nothing special.  I could see how it would have an appeal for certain customers. 

[Stephen Scharf]

Not quite as smooth as you might think due to the scale of the graph.  There's a 10 dB variance between 60 and 100 hz.  You're going to hear that as a jazz bass makes a run down a scale.    But it is certainly flat above 250Hz. 
I haven't heard a pair of Harbeth's in a decade.  I remembered it as nothing wrong, but nothing special.  I could see how it would have an appeal for certain customers.

While your point above about their frequency response is accurate, it’s not a factor in my actual application because what I didn’t show above was the frequency response using them with my REL subwoofer integrated into the system, which is how I actually use them in my listening set-up. So, in my actual application, that dip between 60 & 100 Hz is not there, and I don’t hear that as a jazz bass makes a run down the scale.

Before I added the REL, I ran a DOE (Design of Experiments) in the statistical program, JMP, to determine which subwoofer settings would give me an actual, measured in-room frequency response closest to the Harman Curve. The results of the DOE are shown below.

[Stephen Scharf]

Here's the DOE analysis I did in the statistical analysis application, JMP, to determine the REL subwoofer's control settings to integrate the REL into my system with the Harbeths.



This DOE model is way statistically significant with an R-Square/R-Square Adjusted of 0.9704/0.9486 and a p-value of 0.0016 (where Alpha is 0.05)

The DOE model shows that in my application, subwoofer Gain has the most influence on the actual measured in-room low-frequency response, but there is an interaction, between the Gain & Crossover setting that, with a p-value of 0.0512* (XO, depicted in the interaction plots below the ANOVA table) that is borderline statistically significant as well.

So, in my actual listening application, the FR is not what what was depicted above using the Harbeths alone, its better than that as the result of integrating the REL with the Harbeths based on this DOE.

I'll just that, in his videos, Danny does not depict FR curves below 200 Hz when showing the measurements of a speaker.

*-depends on where you want to set alpha for the DOE.

[Danny Richie]

To get more usable data, reset the scale of the graph to a higher resolution setting. Try moving the whole range down to a 30db range from the bottom of the graph to the top of the graph. It will be much easier to read that way.

[Stephen Scharf]

Thank you, Danny, for the suggestion. 

Next time I use REW, I'll implement your recommendation.  I may want also want do some measurements of my new Buchardt S400 MkII Color Signature standmount speakers that just arrived on Wednesday!



Still burning them in, so I'll wait for that to complete before getting out my NIST-traceable mic and REW.

BTW, you'd approve of the parts Mads Buchardt used for the crossover for the S400 MkII: all air-core inductors, high-quality resistors, Cross caps, and Miflex caps.

Here's a pic of the crossover board:


No cheap sandcast resistors and iron-core inductors here. 

Also, my Color Signature Editions also use silver wire internally, as well as Argento binding posts (same as my Constellation Inspiration integrated amp uses). As we both know, the quality of the crossover parts, speaker parts, and wire used internally have a real impact on audio quality.

The Buchardts will be replacing my Harbeths in the system, which are going up for sale; apologies for the temporary digression.

Cheers and have a great rest of the weekend.
-Stephen

[Danny Richie]

Thank you, Danny, for the suggestion. 

Next time I use REW, I'll implement your recommendation.  I may want also want do some measurements of my new Buchardt S400 MkII Color Signature standmount speakers that just arrived on Wednesday!



Still burning them in, so I'll wait for that to complete before getting out my NIST-traceable mic and REW.

BTW, you'd approve of the parts Mads Buchardt used for the crossover for the S400 MkII: all air-core inductors, high-quality resistors, Cross caps, and Miflex caps.

Here's a pic of the crossover board:


No cheap sandcast resistors and iron-core inductors here. 

Also, my Color Signature Editions also use silver wire internally, as well as Argento binding posts (same as my Constellation Inspiration integrated amp uses). As we both know, the quality of the crossover parts, speaker parts, and wire used internally have a real impact on audio quality.

The Buchardts will be replacing my Harbeths in the system, which are going up for sale; apologies for the temporary digression.

Cheers and have a great rest of the weekend.
-Stephen

I did some design work for that company on that model.

They implemented some of what they learned from my sample (like the use of Miflex Copper foil by-pass caps).

One of the things they were stuck on was the sharp cabinet edges. They liked that look. The sample that I sent them had an additional side panel mounted to them with a curved front edge that did away with the diffraction issue and smoothed out the response. It also stiffed up the side panels. My design for them also lower the crossover point which improved them in several areas, but their engineers were afraid to let the tweeters play down that low. So they went with a higher crossover point.

[rfluongo]

I did some design work for that company on that model.

Danny, that's pretty cool info. BTW, do you charge yourself by the hour when designing for GR? 

[Stephen Scharf]

I did some design work for that company on that model.

They implemented some of what they learned from my sample (like the use of Miflex Copper foil by-pass caps).

One of the things they were stuck on was the sharp cabinet edges. They liked that look. The sample that I sent them had an additional side panel mounted to them with a curved front edge that did away with the diffraction issue and smoothed out the response. It also stiffed up the side panels. My design for them also lower the crossover point which improved them in several areas, but their engineers were afraid to let the tweeters play down that low. So they went with a higher crossover point.

Very cool. I had heard something to the effect that you had collaborated with Buchardt, and provided them with some design input and expertise. I agree with you on your specific points about refinements to the cabinet design, but "it is what it is", as the saying goes. Regardless, I'm very happy with them; they have impressive lower FR, are tonally accurate, and very musical and engaging to listen to. I found John Darko's YT review of them to be very accurate.

Thanks, Danny, and also, thank you for continuing to share your knowledge and expertise with all of us. As a professional scientist, I really enjoy your content intellectually-speaking, and have learned a lot from your videos. Cheers.

[paolocaminiti]

[...] The sample that I sent them had an additional side panel mounted to them with a curved front edge that did away with the diffraction issue and smoothed out the response. [...]

Out of curiosity what generates the diffraction when the tweeter is in a waveguide and the cabinet is just as wide as the woofer? I thought diffraction would come from the front baffle reflections and sharp edges.

[Hobbsmeerkat]

Waveguides don't really change how sound from the driver wraps around the edges of a speaker.
if there are still sharp edges, there will still be ripples created from those edges which will appear in the on-axis as a peak then a dip.

The best designed horns/waveguides continue their curve away from the speaker, rather than allowing for a sudden cut off.

[paolocaminiti]

Thanks Hobbes,

I see, so it was a matter of rounding the front edges.