LS50 Spinorama and EQ

[aarons915]

I realized a few people around here have LS50s and there isn't a readily available LS50 Spin so I figured I'd post one I found in a random study in AES. Thanks to @napilopez for directing me to VituixCAD so I could trace the original graphs and export the curves into REW for a proper scale and EQ capability. Anyway here's the recreated graphs in REW, I purposely left out the direct on-axis because I think it makes it easier to read and coaxials are usually best listened to off-axis, which is better represented by the Listening window.

It's already a very good speaker but it's obvious there is some excess energy around 750Hz and between 2-5k. Using the EQ function in REW, these filters will make the listening window more neutral. I've been using similar filters for awhile and it smooths the speaker out quite a bit, it also takes care of any brightness they can sometimes exhibit. Edited to update EQ, fixed 1.5k dip and changed 3.8k to -.2db based on listening tests.

 

[Ron Texas]

There are all sorts of measurements of the LS50 out there.

 

[napilopez]

I realized a few people around here have LS50s and there isn't a readily available LS50 Spin so I figured I'd post one I found in a random study in AES. Thanks to @napilopez for directing me to VituixCAD so I could trace the original graphs and export the curves into REW for a proper scale and EQ capability. Anyway here's the recreated graphs in REW, I purposely left out the direct on-axis because I think it makes it easier to read and coaxials are usually best listened to off-axis, which is better represented by the Listening window.

View attachment 46028

It's already a very good speaker but it's obvious there is some excess energy around 750Hz and between 2-5k. Using the EQ function in REW, these filters will make the listening window more neutral. I've been using similar filters for awhile and it smooths the speaker out quite a bit, it also takes care of any brightness they can sometimes exhibit.

View attachment 46029

Thanks @aarons915 ! One quick note: It's important to have the right scaling/aspect ratio in FR graphs, spinoramas in particular, as people tend to quickly eyeball the graphs. The standard is 25db per decade. You can make sure you're always sharing the right aspect ratio by selecting the "25 db/decade option" after you press the 'capture' button in REW. This will make sure the plots themselvesare scaled correctly, regardless of what you've set the SPL boundaries too. Since we're always looking for FR smoothness, this helps make sure issues aren't exagerated, or more typically masked. As it stands, your spin graph is a little prettier than it should be =]

 

[Hugo9000]

Thanks @aarons915 ! One quick note: It's important to have the right scaling/aspect ratio in FR graphs, spinoramas in particular, as people tend to quickly eyeball the graphs. The standard is 25db per decade. You can make sure you're always sharing the right aspect ratio by selecting the "25 db/decade option" after you press the 'capture' button in REW. This will make sure the plots themselvesare scaled correctly, regardless of what you've set the SPL boundaries too. Since we're always looking for FR smoothness, this helps make sure issues aren't exagerated, or more typically masked. As it stands, your spin graph is a little prettier than it should be =]

Lately you don't miss an opportunity to criticize the LS50 and nitpick every anomaly in FR measurements, in seemingly any/every thread where that speaker is mentioned. Yet you bend over backwards to try to explain away the truly horrible measurements of other speakers you review such as the Sonus Faber. Why is that?

 

[thewas]

I have done that in the past also for my LS50 based on its listening window measurements of the NRC, works quite ok above the Schröder frequency, although the question is how well our listening distance and room reflectivity correlates to it, so I now rather do it on moving microphone measurements around my listening position which correlate also quite well to the anechoic calculated measurements, showing for example the exaggerated region between 2 and 5 kHz you mentioned:

 

[napilopez]

Lately you don't miss an opportunity to criticize the LS50 and nitpick every anomaly in FR measurements, in seemingly any/every thread where that speaker is mentioned. Yet you bend over backwards to try to explain away the truly horrible measurements of other speakers you review such as the Sonus Faber. Why is that?

Woah, you're misinterpreting me! I didn't say anything bad about the LS50, I simply pointed out an issue with scaling in the OP. Scaling is part of the spinorama standard, and defined in CTA-2034A, so it's important to get that right if we're going to be sharing spins around here that we can readily compare to one another.

Improper scaling is one of the ways manufacturers get away with hiding anomalies - any speaker looks better with flatter scaling. I would've said the same thing had the OP posted about any other speaker, and I've mentioned scaling to amir in the Kali thread twice too. It's just one of my pet peeves

That said, I've never said the LS50 measure poorly. I have said a few times that I prefer the R3 and LS50W over the LS50 because they fix some of the issues with the LS50, especially if you're not planning on using a sub. I looked through my post history for mentions of the LS50 and the worst things I said about the LS50 is that it has a midrange dip and that the LS50W measures better. These are facts. But that doesn't mean I don't still like the LS50.

As for the Sonos Faber, you're free to interpret the measurements however you want; that's why I go through the effort to make them and share them here. I don't always agree with other people's interpretations of measurements myself. That said, while I don't think the Sonus Faber is the best in its price range, I also think it's far from "horrible."

I always prioritize directivity, and the Sonetto II has good directivity. Good directivity means a speaker can be EQd to be flat.

The Sonetto is too bright on-axis, but this is largely helped by listening well off axis - the treble is almost completely flat at 20 degrees off-axis, which is basically pointing straight out in my living room setup, a very common way to position speakers. That leaves the midrange scoop as the only other real issue , but that can be EQ'd (the thin dip at 1.1K is too narrow to be a serious issue, and dips are less audible than peaks). It's also a fairly wide directivity design, which many tend to prefer to narrow DI of waveguided speakers. I don't think it's the best speaker in the price range, but it's not bad either.

Anyway, if it seems I comment on the LS50 more often than others, it's probably because I'm more familiar with KEF speakers than any other brand. They're probably the hifi company I work most often with and have spent the most time listening too, and the LS50W and R3 were my daily drivers longer than any other speakers I've heard.

 

[aarons915]

I have done that in the past also for my LS50 based on its works quite ok above the Schröder frequency, although the question is how well our listening distance and room reflectivity correlates to it, so I now rather do it on moving microphone measurements around my listening position which correlate also quite well to the anechoic calculated measurements, showing for example the exaggerated region between 2 and 5 kHz you mentioned:

View attachment 46048

I agree, the only problem with the NRC listening window is it is only 15 deg off-axis horizontally, vertically and direct on-axis, so it tends to exaggerate the boosted area between 2-5k and I don't want to over EQ the range. I was also using spatially averaged in-room measurements for my EQ prior to this method and it's very close, I haven't A/B the different filters yet but I doubt I'll hear a difference. The real benefit of EQ'ing based on a Spin style measurement is you can see how you're going to affect the early reflections and sound power curve after EQ'ing the listening window. There are a few examples in Dr. Toole's latest book where he shows examples of speakers you can't EQ because you'll create problems in other places. The most common example is a speaker with a neutral on-axis response and erratic off-axis, fixing the off-axis would create a dip on-axis, doing more harm than good. Thankfully the LS50 is consistent off-axis, so EQ works very well on them.

 

[thewas]

Exactly, with an EQ you can correct a loudspeaker with flawed frequency response and smooth directivity, but not the opposite, also the reason why I have kept my LS50 since almost 8 years now.

 

[aarons915]

Thanks @aarons915 ! One quick note: It's important to have the right scaling/aspect ratio in FR graphs, spinoramas in particular, as people tend to quickly eyeball the graphs. The standard is 25db per decade. You can make sure you're always sharing the right aspect ratio by selecting the "25 db/decade option" after you press the 'capture' button in REW. This will make sure the plots themselvesare scaled correctly, regardless of what you've set the SPL boundaries too. Since we're always looking for FR smoothness, this helps make sure issues aren't exagerated, or more typically masked. As it stands, your spin graph is a little prettier than it should be =]

Thanks I haven't heard that before, I actually went and measured a few of the Klippel Harman measurements and Amir's own Klippel measurements and none seem to follow that rule though. Here is the 25db/dec graph in case anyone prefer this aspect ratio, I think it looks a bit too bunched up whereas my original may be a bit too stretched out.

 

[napilopez]

Thanks I haven't heard that before, I actually went and measured a few of the Klippel Harman measurements and Amir's own Klippel measurements and none seem to follow that rule though. Here is the 25db/dec graph in case anyone prefer this aspect ratio, I think it looks a bit too bunched up whereas my original may be a bit too stretched out. View attachment 46089

Thanks =] Honestly, it's pretty confusing, as it seems harman doesn't even always follow its own guidelines consistently, so I hate to be a stickler about it. Amir also mentioned in one of the threads that he's not exporting to a standard size yet. But the aspect ratio you used above is correct. It is the aspect ratio that matches the one used and described in the actual CTA 2034-A document, so I think it's the one we should be using.

Here is your 25 db/decade measurement overlayed over a sample presented in CTA-2034A. It's a perfect match for scaling. The only difference is you're using 60 dB limits instead of the usual 50 but the x and y axes are scaled correctly.

It also is a perfect match for the scaling used by the NRC/soundstage network, for that matter, making it further useful for comparison purposes with existing anechoic data. I can readily overlay the above measurements over the NRC's listening window of the LS50.

You can see despite the difference in the NRCs measurements and averaging technique, and the offset in SPL levels, the shape of the listening window curve is a very close match.

 

[aarons915]

]You can see despite the difference in the NRCs measurements and averaging technique, and the offset in SPL levels, the shape of the listening window curve is a very close match.

They are very close, I made the same filters based on the NRC listening window to compare to my filters from the Spin and also the filters from my in-room measurements that I've been using up until this point. They are all somewhat similar so not that easy to tell apart but it seems like the filters based on the Spin are the best to my ears. The NRC listening window isn't the standard CTA-2034 but instead is on-axis and +/- 15 deg horizontally and vertically, it ends up showing a more boosted region from 2-5k so ends up making bigger cuts in that range. I usually find minimal EQ is better than making large cuts and that seems to be true here as well.
Either way, these speakers continue to impress me. They have beat a few highly regarded speakers in my own blind listening tests, including the Revel M105 and KEF R3. Properly integrated with a sub or 2 and using the PEQ filters in the 1st post, they seem very hard to beat anywhere near their price.

 

[thewas]

I agree, got mine when they came out 7 years ago and now with the knowledge and experience of equalising them gained over the years they are by far the loudspeaker I have kept the longest and am listening at most, as now I use them for my desktop setup,. With 2 subwoofers they also worked great for me in classic Hifi setup.

 

[phoenixdogfan]

Not only do the results look good, they really sound good. I pair them with an SB2000 sub, sit about 5 ft away at the apex of an equilateral triangle in my bedroom set up, and they are extremely good as long as you don't need a rig that can rip your ears off. I've had them for 4 years and I'll keep them a good while longer.