Selah Audio RC3R 3-way Speaker Review

[bobbooo]

Yeh, intuitively I still have my doubts about the scores.

Frustratingly I think the calculation is very close to being correct, it's just the lack of any effect of the slope variable (SL) on the smoothness variable (SM) that's puzzling, despite the former being part of the definition of the latter in the paper. Any chance you could ask Sean Olive to have a quick look?

 

[amirm]

Any chance you could ask Sean Olive to have a quick look?

I asked. He did not reply.

 

[bobbooo]

I asked. He did not reply.

Ah ok. Did I see Todd Welti on here are some point? I don't think he was directly involved in this particular paper of Olive's, but he might have some useful insights considering his high level of expertise in similar areas, and the fact they've worked a lot together on other projects.

 

[amirm]

Ah ok. Did I see Todd Welti on here are some point? I don't think he was directly involved in this particular paper of Olive's, but he might have some useful insights considering his high level of expertise in similar areas, and the fact they've worked a lot together on other projects.

I don't think Todd was involved. The person who was left Harman Sean told me.

 

[MZKM]

as it considers the slope itself as a ‘deviation’.

I could apply a similar thing that I did for Smoothness to the NBD_PIR. (add the absolute difference between the actual and target slope; making a perfect -1.75 slope flat, giving a perfect 0 score). This however is not stated in either paper, but since Smoothness is just R^2 and slope doesn't matter currently (and shouldn't matter for defining a linear regression), maybe Sean accidentally put the slope calculation in the wrong section, and it should actually be applied to the NBD_PIR section.

If he could respond to the message Amir sent him, that would be great.

 

[Mashcky]

The vertical cancellations occur because the radiation of the loudspeaker chassis is not coincident. This means they do not radiate from the same point, at the same time.

This leads to the formation of radiation lobes around the crossover frequency in vertical direction.

...

The midrange dome tweeter in combination with the baffle radiates very wide in the range around 1-2,3kHz. The combination with the ribbon tweeter, which bundles strongly in vertical direction, makes things even more difficult.

So this is largely about cabinet design and could not be solved by time alignment?

 

[HammerSandwich]

The problem is that distances between drivers & mic/listening position change as you move. IOW, you cannot maintain time alignment across wide angles when the drivers are in different places.

 

[bobbooo]

I could apply a similar thing that I did for Smoothness to the NBD_PIR. (add the absolute difference between the actual and target slope; making a perfect -1.75 slope flat, giving a perfect 0 score). This however is not stated in either paper, but since Smoothness is just R^2 and slope doesn't matter currently (and shouldn't matter for defining a linear regression), maybe Sean accidentally put the slope calculation in the wrong section, and it should actually be applied to the NBD_PIR section.

If he could respond to the message Amir sent him, that would be great.

That might be an interesting experiment, but how exactly would you incorporate the SL variable into the NBD_PIR formula?

 

[MZKM]

That might be an interesting experiment, but how exactly would you incorporate the SL variable into the NBD_PIR formula?

Ideal slope would be -1.75, use that to add/subtract the actual data points based on the frequency, similar to column D in the SM sheet.

That way an ideal speaker with ideal slope would get its slope turned flat, creating an ideal NBD score.

 

[ctrl]

So this is largely about cabinet design and could not be solved by time alignment?

With non-coincident chassis/sound sources, the time alignment applies to only one point in space. If the listening position is changed vertically (with vertically offset drivers), the time alignment of the sound from the two drivers changes to the new listening position and causes interference that leads to the radiation lobes.

A detailed consideration of the formation and orientation of vertical lobes can be found here (the article is actually about interference in M-T-M arrangements, but the subject of radiation lobes is also covered in the beginning):
http://www.birotechnology.com/articles/VSTWLA.html

The formation of vertical lobes can also be deliberately used, for example, to extract "sound energy" from a loudspeaker in the frequency range where human hearing is particularly sensitive - which would cover the range 1-4kHz.

Here is a simulation of the early reflections of a 6'' woofer with a 6.5'' wave guide with actually too high crossover frequency at 2.5kHz. This gives us a "bad" vertical radiation, but in return a loudspeaker that will not sound aggressive even at high sound pressure levels when properly tuned.

 

[bobbooo]

Ideal slope would be -1.75, use that to add/subtract the actual data points based on the frequency, similar to column D in the SM sheet.

That way an ideal speaker with ideal slope would get its slope turned flat, creating an ideal NBD score.

You could try it out, and see what it does to the scores. And also check if changing the target or 'measured' slope changes the NDB_PIR value and so the final score, as it should.

 

[Selah Audio]

I do find it interesting how the horizontal response (maybe the vertical, too; I haven't looked closely) is not symmetrical across the -90 to +90 measurements. Makes me question the result, TBH. I don't see a reason why one side would measure differently unless the drivers weren't perfectly center-aligned. OTOH, the difference is only a couple dB. Just one of those things that makes you think "hmm". *shrug*

(I provided my plot of the data below)

View attachment 48310

Erin,

Good to see you on here. Thanks for pointing this out. I'll address it in a post this evening.

 

[Xyrium]

I don't think Todd was involved. The person who was left Harman Sean told me.

Where does one go after Harman, don't they own everyone at this poiint?

Looks like your in room estimate cleans up the rising frequency response and makes it nicer through the HFs, and inflated in the LFs. Might be a good sound for many folks.

 

[amirm]

Thank you again for the detailed measurements, one small correction though, a similar hump in angle measurements is usually a sign of a resonance, but not a hump in the directivity indices, there it means a discontinuity in the directivity. A similar hump in the angle measurements like a resonance almost disappears in the directivity indices as those are something like the difference of the angle measurements.

Thanks. Was in too much of a hurry posting this review. Corrected.

 

[Selah Audio]

Back on January 7th I posted on a few Facebook groups that the Klippel system was going to be used by Amir. Both groups are focused on DIY speakers and have some very talented people, including some that work for speaker manufacturers. Little did I know that one of my designs would be tested less than a month following the post.

I've followed ASR some over the last year, partially due to the Benchmark reviews. I've used their pre-amps and amplifiers at audio shows and they make great equipment (plus there are really nice people that work for them). I know the Klippel system and speaker testing is new territory for Amir and I expected that there would be a learning curve. I had a chance to meet Mr. Klippel a few years back and his driver test system is one of the best things to ever be used in the speaker industry. The more recent nearfield scanner was a brilliant idea!

There's really a need for more objective speaker testing in reviews. I like that Sound Stage online magazine has long used the NRC in Canada for measurement tests. Of course much of what the NRC is doing was due to the efforts of Floyd Toole. The CEA-2034 is an extension of Floyd's earlier work. There is one thing that isn't included in the CEA-2034 and that's the compression test done by the NRC. I think that's something that can be of use, especially if you're running a system at high volume levels.

Honestly I feel like the speaker I designed was involved more in a drive-by shooting than a thorough review. I would've happily supplied information about the design if it had been requested before the review. The woofer and midrange were correctly identified. The tweeter is actually an earlier version of the Fountek CD3.0 (the ribbon element was changed for the current model). This tweeter has a naturally rising response on-axis and the on-axis graph of the speaker indicates that. Since the aspect ratio of the ribbon is different than a dome tweeter (point source) the upper end response of the ribbon will be shelved downward at angles above or below the vertical on-axis response. The curves here show that.

This design was done over 15 years ago and at that time I focused most on the on-axis response; as a result, I'm not surprised that the off-axis isn't as linear as my current work. I do wish that the comments here had been based on some listening and not just in reference to the graphs. The graphs themselves are a mixed bag and difficult to compare to the other speakers tested here. The dB scales are not consistent and the ones marked for the standard 2.83v/1M are way too low for the actual system sensitivity.

As noted by a few posts in this thread some wrong conclusions were expressed based on Amir's interpretation of the data. I know that he's not an experienced speaker designer and I don't expect perfection. I do appreciate the effort to offer an honest opinion, something sorely lacking in much of the online reviews and videos that we see. I would suggest listening to every speaker before the measurements are done. That way there won't be as much of a bias before commenting on the performance. Make notes on what you hear and see if the tests confirm what you wrote down.

There was also an observation of some differences in the off-axis left versus right. The drivers are centered on the baffle and there should be no deviation for the left vs. right angles. This indicates that 1) the cabinet was mounted off-center or 2) the machine had some issues with the correct positioning of the scanner.

I sent in some $ today to help for future reviews and would be happy to provide some of my current work if requested.

Rick Craig
Selah Audio

 

[Selah Audio]

This is my speaker Amir measured. Amir, huge thank you for taking this one for a spin!

In real life the speakers are bright like the measurements have suggested. I use treble cut and bass boost tone control, which flattens our the in room response nicely. The spinorama is pretty smooth, and to my ear they respond well to the necessary but basic EQ.

Vertical treble response is unsurprisingly odd because it is a vertical ribbon tweeter. Horizontal directivity is solid though.

It seems like there’s not clear consensus yet on what these measurements say about the speaker, and it “scores” quite poorly. This is at odds with my subjective experience with the speaker. Also it is from a talented designer that leveraged some fairly expensive drivers (which are still available), so there may be more to unpack here.

Apologies that the review from Amir is not as detailed at others, this can be attributed to logistical matters of getting the speaker back to me.

Thanks for providing the test speaker. The tweeter level can be adjusted with a change of the resistors in the crossover. If you want to do that I would be happy to help.

 

[dshreter]

Thanks for providing the test speaker. The tweeter level can be adjusted with a change of the resistors in the crossover. If you want to do that I would be happy to help.

This is a very kind offer and I will take you up on it! Thank you for joining the discussion as well and your very thoughtful commentary above.

I do wholeheartedly agree with your perspective, the speakers sound excellent. And they were ahead of their time with performance that still holds up well today.

 

[jtwrace]

would be happy to provide some of my current work if requested.

Rick Craig
Selah Audio

That's a no brainer Rick, just send it to @amirm ! I think it would be great to show what evolution does when one learns and embraces science. Unlike many small companies that offer kits, DIY assistance and fully built speakers, you put it out there!

 

[Shazb0t]

I sent in some $ today to help for future reviews and would be happy to provide some of my current work if requested.

Rick Craig
Selah Audio

I really hope that you do send in something from your current line! I've been curious how your Ceramicas would measure. I don't feel that the spin for this speaker was a bad result considering the age of the design and the ribbon used. A newer RAAL ribbon would be interesting to see tested on the Klippel.

 

[MZKM]

The dB scales are not consistent

Yep, I try to keep my graphs the same (you can see the ranking list in the link in my signature), your speaker scored a tad better than the LS50 if ignoring bass extention, and considering the similar price and your’s being older, that’s not bad at all. And as you’ve stated you focused more on-axis, I see the measurements of some of the current speakers on your website and they indeed look better, the Epico for instance. However, for the DIY Filarmonica (like the tribute), the 5dB slope down from the bass to treble has always perplexed me.