Magico S1 Mk II - The measurements

Sound Stage Network released their own measurements and had the chance to evaluate the S1 Mk II for a longer period of time than I have.

It's time for a closer look and to see where I was right, and where I was wrong. Hans Wetzel's review was overwhelmingly positive, while mine was positive with nuances.

Tonality

Honestly, I hate the word "tonality." Is it even a word? So many have heard the S1's and proclaimed them to have "spot-on tonality" to my surprise. I didn't think so, but hey, let's see what the measurements say.

First, remember I said it was bright, with excess energy around the bottom of male voices?  Well, based on my reference speaker curve from Bruel and Kjaer, I was right.

Take a look here, we compare the two.

At least from my own reference perspective, I was spot on. The treble runs a little high for me, and the peak energy is around 200 Hz, a little over an octave higher than I would like.  I was also right about the treble being very smooth and unblemished (more on that in the next chart). I was wrong about the overall balance being smile-shaped though. This is in fact pretty flat, just not necessarily going exactly where I'd like, but that can be adjusted for with reduced toe-in or DSP. When I listened to these speakers it was practically on axis.

In the end there's really no cheating physics and there is only so much you can do with a small driver but I am really surprised the energy falls off that soon, it does sound better than that, but this early bass peak does explain why I though the mid bass was a little too much.  At the same time, that early drop off is going to give a sense of exaggerated bass clarity, since you won't activate troublesome room modes below 100Hz as much as with a true full-range speaker. 

Compression

One measurement I was surprised by was the amount of compression between 70 dB and 90 dB.  That is, at higher volume levels the top treble will stand out more due to compression in the mid-treble.  It is going to be really important to evaluate these speakers at the right volumes for you.  If you listen at low volumes, listen to them that way, and vice-versa.  This undoubtedly contributed to my impression the speaker was smile-shaped.  At 90 dB, it starts to go that way.

Based on the early bass drop off I would try to keep the speakers within 2 feet of the rear wall, but far far from the sides to get a little bit more bass without compromising the imaging and clarity of those wide dispersion tweeters.

Comparing the S1 to the B&W 802 D3

Let's do another graphing trick.  Let's compare Stereophile's measurements of the B&W 802 D3 to the S1 measurements from the NRC of Canada. There are many reasons to be critical of this chart, as the measurement methods are undoubtedly different and I never had access to the raw data.  Still, I think it's worth using as a rough guide to what we could expect, and at least one dimension of how these two speakers could be alike, or different:

Notice the Stereophile Curve is in play with the B&W. See the dip at 2.4 kHz and the peaks at 4kHz and 10 kHz? Not a neutral speaker at all.  The Magico S1 is in fact much closer to neutral, without any gimmicks applied. Of course, the B&W has more bass, and about 7 dB more sensitivity, among many other differences. This technique is fun to play with and as long as lives are not at stake, worth showing.

The Impedance Curve

What about the minimum impedance? I was wrong to worry too much about it going too low. In fact, the impedance measured by Sound Stage is very good, reaching a minimum of around 3.6 to 3.8 Ohms at 150 Hz with a near 0 phase angle.  Quite benign for most modern amplifiers.  Further, that same impedance chart is evidence of a simple, no-nonsense crossover based on well meshing drivers. Fabulous!

The Amplifier

That being said, I was not wrong about these speakers needing a beefy amp. Take a look at the 1 kHz sensitivity. 82 dB?? Wow, that really is pretty low.  In fact I can't remember the last time I saw a 2-way design with such a low sensitivity rating.  The S1 Mk II's will in fact need a big amp to sound loud.

So is this worth $16,000? I think it's a pretty nice speaker, with my objections noted. Listen for yourself. With proper toe-in (as in none) and a subwoofer to support them, and space, lots of space, the S1's are pretty nice.  Whether you find the compression or limited bass objectionable or delightful however is going to be entirely up to you.

Last, unlike some other mid-price speakers introduced this year, they are unlikely to melt in the rain.  Bam! Hahahaha.

Poor Music Lover's Alternative

If $16,000 for a two-way, limited bass speaker pair is never going to be realistic let me suggest three very good sounding kits you can build from $500 to $3,000 on my other blog, here.

The LM-1 vs. My Laundry

As a side effect of the movers taking most of my furniture, I had to move the LM-1s off their stands 2 feet from the rear wall to half that distance, sitting on books on the TV stand.

In this location I find them too boomy. What's the solution? Socks. I put a big clean fluffy white sock to plug the port in the back of each. Problem solved!  They don't sound as good as they did further away, and further apart, but they have regained their tonal composure.

The LM-1 vs. Your Chair

Quick setup instructions.

A number of readers have asked me about the horizontal layout of the drivers. I was just being creative with the pictures, but I personally listen to the LM-1's with the traditional tweeter-over-woofer, "portrait mode" arrangement.  Of course, you can lie them down to fit tight spaces, or to keep a lower profile over a mixing console, but I am not intentionally designing "landscape mode" speakers.

The LM-1 has a wide and natural dispersion but is best with the tweeters on the outside and either parallel or toed in. Vertically however the LM-1 sounds best with your ears at tweeter level or lower. Any higher and it sounds dull and lifeless.

The LM-1C is similar. Used vertically keep the tweeters to the outside of center, toed in slightly. Around 15 degrees. This will give you a big sweet spot in the center, and limit reflections to the side.  Used horizontally keep the tweeter pointed at or slightly above the tweeter towards the listener location,  You may wish to experiment with the tweeter on the "outside" of the TV.  In other words, keep the woofers closer to the TV than the tweeter but tilt the speaker vertically. Play around with it  a little and the best tuning will become pretty obvious.

Experiment!

The LM-1 vs. Bruel and Kjaer

The LM-1 speaker kit has only been promoted for a little while and already it's evident there's a lot of misinformation and attempts to discredit the design. I hope it doesn't become the Kardashian (any of them) of speaker kits.  In any event, one completely inaccurate criticism is that the frequency response is below-average. Nonsense. Could you use different parts and asymptotically approach perfection? Probably. You could get different, but getting noticeably better will be very difficult.

Let's go over some background.

Who is Bruel and Kjaer?

When scientists and engineers think of high-end acoustical instruments Bruel and Kjaer is among the most respected names there is. One famous bit of knowledge that has come out of that organization is the ideal B&K speaker curve, which I happen to like a lot (that's a personal preference). It is meant to be as the ideal response at the listening position. I copy it here, below:

Note that it is not flat! A speaker that measured flat at the listening location would be an ear drill. The B&K curve is about +3dB at 70 Hz or so and -3dB at 20 kHz. Quasi-anechoic measurements measure the speaker driver in near-field, with the expectation that at a distance the measurement would become like you see here. The full discussion of why this is beyond the scope of this posting, but do your research and you'll find much more written about it. You may also find that the Dirac Live target curves, also follow the B&K model of a gently descending response. The point is, I don't make this stuff up. Flat at the listening location is not actually ideal. What is ideal is open to some interpretation.  I'm choosing the B&K curves as a matter of taste.

So, how does the LM-1 do by comparison to this fabled B&K curve in real life? Let's comapre (LM-1 level is offset and measured in a bookshelf):

Damn well! This is the plot with R1 = 4.2 Ohms (2.7 + 1.5 actually). I'm packing and that's all the resistors I have on hand.

So, what does this say? This is an objectively neutral speaker kit. If you are looking for a true bookshelf speaker system with reference grade frequency response the LM-1 will do it for you, within the limits of what a small monitor can do.  Also, as discussed, the treble balance is up to the builder. If you don't like it, change it. :) I have a feeling that most listeners will like a value closer to 3.9 Ohms or so.

The LM-1 and Clarity ESA Notes

June 4, 2016

My usual go-to medium grade capacitor is Mundorf MKP, but since I had such good results with the Clarity MR caps in my reference speakers I decided to try the Clarity ESA's. 24 hours into the experience and it's still too hard.

I've increased R1 from 2.7 to 4.2 Ohms and it's much better. The LM-1 curves match the center channel curves more closely. However I'm still not quite satisfied with the treble.  Of course, this is a $35 ring radiator tweeter, vs. the $500 Mundorf AMT I've gotten used to. The two may never closely align, but I have hopes. I've heard lesser tweeters smooth out with better caps, and I've heard caps mellow out after a few days, so this may take a while before I'm ready to say I'm done.

June 5, 2016
At about 48 hours the caps got noticeably less harsh. It's also interesting to note that changing R1 from 2.7 to 4.2 made only about 1.6 dB of difference at 13 kHz, but it's all the difference in the world to my ears.

June 6, 2016
In spite of the packing going on around here, I closed some drapes and listened more last night. I'm ambivalent over the total treble balance, but the tweeter and C1 seem to have settled in quite nicely. I sometimes feel it may still be a little bright, but the measurements don't lie, I'm actually just under the B&K curve, so I'm going to leave it alone. Honestly I think my reference speakers are a little dark, so this may be a good chance for my own ears to adjust. My PC is getting packed up today, so I'll be unable to do more measurements or design as I won't have access to my only Windows 7 PC. This will be a good chance for me to explore what else is available on Ubuntu/Linux! :)