Auditioning the KEF LS50 Black Edition, powered by SMSL SA300 Class D amp on computer workstation. |
With autumn here, sunlight hours reduced and looking towards the rainy months ahead, I thought it would be good to start expanding measurements and be a bit more "serious" with evaluating loudspeakers.
As you know, speaker measurements have been done since the start of the audiophile hobby and there are all kinds of ways to get the job done to varying levels of precision and reliability. Stereophile has been doing this for years with John Atkinson's use of the MLSSA (DRA Labs) system as he summarized in this review which formed his 1997 AES paper. More recently, it's great to see Audio Science Review's use of the Klippel Near Field Scanner (NFS) for many speaker measurements already!
For many of us interested in speaker performance, probably introduced to the ideas with Floyd Toole's book Sound Reproduction - Loudspeakers and Rooms (first edition 2008, now in 3rd edition 2017), there has been a rise in loudspeaker measurements using the "Spinorama" technique. This measurement method plus its standard representation of data has since becoming codified in the ANSI/CTA-2034-A "Standard Method of Measurement for In-Home Loudspeakers" (you can get a copy of the current CTA-2034-A R-2020 here), first published in 2015.
I have a copy of the 2nd edition of Sound Reproduction which I believe is pretty much the same in content to the 1st edition, and Chapters 17 on "Subjective Evaluations" and 18 on "Objective Evaluations" are essential reading. In these chapters, Toole lays out both the "genesis" of his work from the mid-1960s to work in the 2000s across anechoic facilities at Canada's National Research Council and at Harman. For further history, check out this well-done timeline, this page for more on the "Spinorama", and here's a very nice collected "gallery" of speakers with respective data.
As noted in the graph, the rated anechoic sensitivity for this speaker is 85dB/2.83V/m. Stereophile estimates that the LS50 is around 84.5dB/2.83V/m. As you can see, I've put the cursor at 85dB and indeed it looks like the sensitivity on average would be slightly lower than 85. One technique I've seen to arrive at a specific number is to average the results from 300Hz to 3kHz. Thanks to recent REW updates as commented by JohnM below in the comments, we can easily calculate the average sensitivity from within the software using its "Graph Metrics" feature! With that, measuring from 500Hz to 7kHz which will cover the most sensitive portions of human hearing, we arrive at a number of 84.7dB/2.83V/m. Nice!
There are a few steps involved in compensating for port amplitude which needs to be attenuated before "adding" the Woofer + Port on the LS50. I refer you to this page for nearfield measurements and calculations.
While frequency anomalies are more audible, we can look at the time-domain characteristics for the speaker as well. Here's the LS50 step response:
A couple of notes about this. First, the miniDSP UMIK-1 USB microphone I'm using is limited to 48kHz sampling rate and appears to implement a sharp linear phase filter for the ADC resulting in visible pre-ringing. Also, the microphone polarity is inverted so I had to flip the graph to what you see above. Other than the pre-ringing, this is basically what Atkinson found in Stereophile. As we can see, the tweeter (initial negative deviation) is hooked up with opposite polarity to the woofer (larger positive low-frequency step and roll-off). This is a common arrangement with passive 2-way speakers. While the crossover between the drivers are not "time/phase coherent", the step response is nonetheless temporally nice and "tight" as a coaxially arranged driver.
For those who like videos, here's Dr. Toole from 2015 summarizing much of the ideas:
For me, the core points which audiophiles need to heed are these:
1. Among listeners, there tends to be repeatable preferences between different loudspeakers even when blind tested. This is an important point for audiophiles, many of whom for some reason have issues with blind test methodology thinking that the method results in "it's all the same" data. Logically, true scientific advancement of sound quality requires that we embrace every beneficial method of testing, and the irony is that many audiophiles dismiss controlled blind testing when in fact this is in all likelihood the most powerful tool available for honest research - just as it is for every scientific discipline involving human subjects!
Since preferences can be repeatably observed, objective means can be used to quantify what "sounds better". This is the empirical foundation from which science can provide guidance. While subjective preferences affect every aspect of human life, factual knowledge helps us to hone in on more interesting products and better decision-making.
2. There needs not be (nor is it realistic!) to have a single "curve" from which to draw conclusions for loudspeaker performance. Toole talks about this in Chapter 17.1 in the 2nd edition of the book when discussing the "genesis" of his thoughts and the ensuing work. This is mainly referring to the frequency response curves of loudspeakers, but we can see this general principle for the various dimensions of performance. As usual, not all dimensions are of equal importance. Frequency response for example is essential and typically more important than time performance if we are to prioritize.
3. Nobody listens in anechoic chambers or have their hears in a vise such that the sound coming out of speakers can be precisely quantified. Obviously it would be no fun listening in such rooms or be physically constrained (restrained!). As such, the room-loudspeaker forms an inextricable dyad. The sounds we hear are very likely, if not always, a combination of on-axis (straight ahead from speaker to ears), off-axis (angles off center) and sounds bounced off the room depending on placement geometry, distance, furniture, etc. (some of which discussed in a previous post). Headphone listening of course has the benefit of breaking this transducer-room connection.
Related to this, Toole talks about the "Central Paradox" (see Chapter 5). While music source hardware can be well defined (we've measured and tested cables, DACs, amps, streamers, etc...), and we know much about the limits of the human perceptual/cognitive systems, the middle part involving "transmission" of sound between the hardware/speakers and our ears across rooms remains complex, with idiosyncratic characteristics. I'm not sure if the word "paradox" is the best one, but it does highlight the fact that while knowledge of hardware and human subjects can be better defined, the moment we talk of sound quality in a room, uncertainties will rise substantially. Maybe saying "Central Ambiguity" might be better than "Central Paradox"?
Taking those 3 items above, the job of speaker measurements is an attempt at presenting data so that we can drive realistic expectations. The hope is that those characteristics captured represent a significant portion (never all) of what is most important to account for the preferences as per item (1) above.
While it's always tempting to think simply in "black & white" about objective results as if they can "definitively" tell us what speakers to buy and which never to consider, having said the above, I trust that rational audiophile will realize that in reality this is not the exercise. The hope is not to just have data in themselves tell us what to think or what to do. Rather, I hope the measurements spur us to think for ourselves and "do the work" when we consider our rooms, our needs, and of course our practical limits to space and price. If the data provides meaningful guidance, it will have served its purpose. I think measurements can also over time provide a beneficial feedback mechanism. There is such a thing as "ear training" such that knowing about the speakers' measured characteristics might train us to become better listeners with experience and within a competitive marketplace, also encourage manufacturers to test and publish the objective performance of their loudspeakers in recognition of the educated audiophile consumer.
Finally, unlike a piece of copper wire that conducts electricity as a cable, or even what clearly constitutes as accuracy for a high-resolution DAC, the "Central Ambiguity" of loudspeaker-room evaluation demands that we appreciate the subjective as well. Unless truly egregious or clearly the price-to-performance ratio appears to be way off any semblance of "value", there will need to be many shades of gray between "good" vs. "bad" performance when we humbly present opinions.
I. Let's get practical...
Over the years, I think I've basically measured and tried to express what I've learned for essentially all the hardware, software, and even accessories I use daily here at home. I've regarded this blog as a type of "personal learning project" as an audiophile so it's only natural to expand and try to do something more, the most obvious being loudspeaker measurements as a hobbyist.
Remember that the community of DIY hardware audio guys and gals have been doing various experiments and measurements for a long time and in doing so, have laid a path for typical audiophiles to explore should we desire. As you'll see, the software I'll be using will draw from the world of "donation ware" and the work of those who have freely shared over the years. The job here then is to cobble together some ideas and tools to help measure speakers in a way that hopefully will help us distinguish speaker contributions from the room. Ultimately there is no absolute substitute for a proper-sized anechoic chamber if we are to isolate the sounds coming out of our speakers and measured at an appropriate distance like 2m especially for large speakers requiring proper driver integration. For that, I think we should thank SoundStage! for publishing the NRC anechoic chamber results over the years. At best, what I'm doing here will hopefully be a close enough approximation.
As usual, I like keeping things practical and transparent so those of you interested in following along with this stuff can also give it a try at home. I'll be doing these measurements with these basic pieces:
In the picture, you see that I'll be doing all of this in the basement of my home in the kitchen area which is currently not rented out and has become an audiophile "work area" of sorts for me. The room itself is not large - about 13' wide x 20' wide with 8' ceiling with a more spacious area in front of where this picture was taken without the fridge and stove to the sides. Laid out are the tools I'll be using to create a simple "quasi-Spinorama" or I'll lovingly call "qSpin"; distinguishing what I'm doing with the real "Spinorama" as laid out in the CTA-2034A guidelines. ;-)
In total, the test "rig" didn't cost much - a pair (of which I only needed 1 for this) of Sanus BF-31B 31" speaker stands (<US$50/pair on sale), microphone stand (~$20), miniDSP UMIK-1 USB microphone (~US$100, all measurements with calibration applied), a metre stick, inexpensive 12" rotating stand (~$11), large plastic 18"x24" cutting board for base (~$15), optionally for extra precision a laser distance measure (~$30), and I forgot to show my multimeter that can measure AC volts. Most of this stuff I have already around the home and for other projects so I didn't spend any money on this other than a few bucks on the 12" rotating turntable.
I printed out some protractors to scale and taped them to the cutting board base to measure relative angles (nothing fancy, one like this is great). That's how I derived those markings on the green tape for the rotation angles marked on the white cutting board base - 0° up front all the way to +/-180° behind. CTA-2034 states to measure every 10° which I'll do for the front hemi-circle, but since I'm kinda lazy and don't really believe I need this kind of "resolution" for the rear hemi-circle, I marked off every 15° instead.
I have an old "Compendium of Pharmaceuticals and Specialties" (Canadian medical people will know what this is) to lift the stand up if needed, plus Alan Greenspan's memoires can serve as some extra lift if needed. ;-|
In the image above, you also see some 14AWG speaker wires to the right that connects to one of my 250W/8Ω Emotiva XPA-1L monoblock amplifiers more than 1m away for passive loudspeakers. Remember that I measured the XPA-1L previously and it scored well with good damping factor of >80, better than -80dB on my "triple tone TD+N" distortion test at 2V, and enough power to provide >32V <0.1% distortion, all into 4Ω. There should be no concerns even for hard-to-drive loudspeakers. One caveat to keep in mind is that the XPA-1L does dip a bit into 20kHz by -1.75dB into 4Ω; not a problem for real life playback, but good to keep that in mind when reviewing graphs.
As for the rest of the signal chain, I'll use the inexpensive but accurate Topping D10 DAC --> passive Nobsound preamp --> Emotiva XPA-1L --> speaker. From Topping D10 to Emotiva amplifiers, distortion will be <0.01% at the kind of power levels I'll be feeding the speakers - my job is to test the devices within reasonable listening levels, not blow up the speakers!
Before moving on, one last important reminder is that when we place the speaker on top of the rotating stand, we need to be mindful of the Point Of Rotation (POR) such that the front of the drivers are rotating around that axis rather than tracing out an arc. Also, I'll make sure that I identify the Reference Axis which typically will be pointed straight at the tweeter unless specified otherwise:
As per CTA-2034. There's also discussion of height variability and front baffle to POR maximum distance of up to 5cm/2". |
II. The subject of investigations today - the KEF LS50
Thanks to linnrd for letting me borrow his KEF LS50 "Black Edition" speakers for this inaugural measurement "spin"!
As you can imagine, there is a good reason to start my loudspeaker measurement endeavor with this. Though not by any means historically "classic" since they were only released in 2012, this speaker has made a splash in the audiophile community with their marketing and almost universal coverage. The speaker was first release in celebration of 50 years for KEF (hence the name), but since has been made available in all kinds of other "editions" of various colors and contrasts between driver and enclosure. They're easy to find at a good price - I've seen recent sales on new units since late 2019 down below $900 Canadian dollars. This often happens as companies prepare to release new models to sell off stock. KEF has recently released the passive LS50 Meta and LS50 Wireless II to much fanfare of course.
Numerous measurements on these speakers have been released as reported in Stereophile, SoundStage!, ASR, among others. Having a variety of results like this will provide the opportunity to compare and contrast what I find.
Externally, the KEF LS50 I think is rather handsome. It's made of some kind of dense plastic polymer and as you can see, the Black Edition is indeed black throughout with tasteful colored lettering. This would look great in a dark soundroom or home theater system where you don't want distracting color contrasts. There appears to be some reflective "specks" in the black color which adds highlights at certain angles. Because the texture is matte black, it will pick up light-colored dust and indeed you can see some of this on the side and finger marks as seen in the front speaker image on the left.
The speakers came as a pair and weigh a healthy ~16lbs each. For consistency, I arbitrarily picked one of them and called it the "LEFT" speaker, this is the one most of the measurements will be performed on. Dimensions for each speaker are 11.9"x7.9"x10.9". See the original LS50 brochure for some slick marketing on the technology behind the design. While the "Uni-Q" driver array is mounted centrally, note that the rear port is situated to the top left which might result in some mild asymmetries - we'll see. An interesting property of the port is that it's soft inside which is said to help with dampening and the shape reduces turbulence. The curved front baffle is designed to reduce diffraction and the LS50 whitepaper (see page 15) implies a smoother response with less "ripple" although the graph is rather low resolution.
III. Let's Test!
With that, let's give it a go... Here's the "LEFT" loudspeaker subject used in the majority of these test. I'll have a look at the "RIGHT" speaker a little later for comparison.
"LEFT" KEF LS50. For a bit more stability, I put a small weight in front of the speaker stand. |
1. Speaker sensitivity. Let's have a look at how loud this speaker gets in this untreated 13' x 20' x 8' room. We'll run a typical 2.83V sine sweep through the speakers, measured at 1m.
[Updated: Previously I was showing the full room response; let's gate the measurements to exclude all the reflections which would add to room gain and see what I come up with.]
2. Speaker Impedance & Phase. Another important parameter we've discussed in the past is speaker impedance. This tells us how difficult the speaker load is to drive.
Looks like the box resonance frequency is around 56Hz and the minimum impedance is almost exactly 4Ω with the lows at around 150-400Hz. These results are almost identical to Stereophile's. Clearly it would be best to have an amplifier capable of driving 4Ω loads paired with these speakers even though "rated" at 8Ω nominal in the specs sheet. Since KEF specifically points to LS3/5A speakers, compared to something like the Falcon LS3/5A, the LS50 is a significantly lower impedance design (ie. harder to drive than true LS3/5A).
[Addendum Dec 6, 2020] I've updated the graph above with the "Equivalent Peak Dissipation Resistance" (EPDR) curve which incorporates both the impedance and phase to more accurately reflect the "resistive load that would give rise to the same peak device dissipation as the speaker". I'll try to include this in upcoming speaker measurements. Notice that the EPDR dips down to 2Ω with the LS50 in the low frequencies and around 150Hz.
Together, the sensitivity and impedance/phase curves can give us an idea of the type of amplifier we might need to get the speakers playing at the appropriate amplitude with good control. Thankfully these days, amplifiers don't have to be expensive to be excellent with plentiful power and able to drive 4Ω loads.
3. Gated Frequency Response, "qSpin" CTA-2034-inspired Chart, and Directivity Polar Maps.
As you can see in the "Speaker Sensitivity" graph above, it doesn't really help us to understand how well the speaker reproduces frequencies if it's distorted by room effects. In a small space like this, we can use the "gating" technique to examine the brief initial direct sound "window" from the speaker to the microphone before reflections from the nearest surface. Since I know that the microphone is 1m away and all other surfaces in the room are >1m away from the speaker, the first reflection must be from the speaker driver to the floor which is shy of 1m with the LS50 sitting on top of the platform and 31" stands.
For more information on using the gating window, have a look here.
Next, with the appropriate placement aligned for POR and intended Reference Axis, I captured all the relevant angles I want to examine with the speaker around that "turntable". From 0° to +/-90°, I'll measure every 10°. From +/-90° to 180°, I'll just do every 15°. Then I lay the speaker on its side like this:
And capture the off-axis sweeps above the speaker (+) or below (-). BTW, you can seen in the image above, the Emotiva XPA-1L amplifier with its blue LED under the table >1m away. I've tried to target between 80-85dB SPL which I think is a realistic upper-end "reference" listening level. Personally, I'm more comfortable listening around 70-75dB average...
So, with the angles captured and gating applied, the REW screen looks like this for the vertical measurements (REW measurement length 512k, acoustic timing reference to help with time alignment). Notice the tracings are closely paired suggesting symmetrical performance +/- vertically:
These are only half of the measurements; there's another "family" of curves for the horizontal tracings. For this first post, I've decided to take the full measurement set. In the future, vertical tracings do not really need to be this detailed and I suspect just doing +/-90° and not worry about >90° vertical angles would suffice.
Notice that with gating, because we only have a limited time window to work with, we will not be able to accurately capture all the low frequencies, and resolution will diminish as we approach the limit. Hence in the diagram above, we see that the frequency response ends around 350Hz for just 3.2ms of echo-free window. We'll deal with filling in a measurement of the rest of the low-mids and bass response later.
After all that, I'll then export the data to .txt format and use the amazing VituixCAD software to calculate CTA-2034 curves using all the data - vertical and horizontal. Voila:
I highly recommend reading the excellent Sausalito Audio "Interpreting Spinorama Charts" primer to figure out what these curves mean.
Also within VituixCAD, we can easily plot out the "polar maps" for the LS50 to help us visualize the horizontal and vertical directivity of the speaker:
3dB contour lines turned on. +/-180° range. |
We can see that as per the speaker design with concentric drivers stuck in the middle of the box, both vertical and horizontal maps look highly symmetrical. Remember that the rear port isn't symmetrically placed so that might account for some minor difference between + and - angles. Based on the "map", we see that it's best to listen to these speakers within +/-60° both horizontally and vertically in relation to the tweeter axis (0°).
Directivity indices reflect the consistency of the sound as one moves off-axis. As you can imagine, for most speaker, as one moves laterally or vertically from the on-axis sound, amplitude will start dropping off. For a narrow dispersion speaker where there is a steeper off-axis drop-off in amplitude, the value of the DI will be higher or more "forward biased". In bass frequencies, DI is close to 0 because large wavelengths / low frequencies will diffract around the speaker for an omnidirectional radiation pattern (here's an interesting technical PDF with simulation code). As frequencies increase, the radiation pattern becomes more unidirectional and hence the DI number increases. If the tweeter "beams" and results in a progressively "pencil thin" pattern by 20kHz, the slope of the DI will end up being more acute compared to a speaker with more constant directivity and flattish slope to the curve. I'm not sure about the significance of differences between the DI compared to Early Reflections DI and how this translates to real-life listening.
While we could debate the preference for narrow vs. wide dispersion speakers, what is important in these DI graphs is that they appear smooth. Irregularities suggest that the off-axis sound changes timbre and in a room where the sound heard is a composite of both on and off-axis spectra, a "lumpy" DI implies that various reflections have different timbre and when combined, could color the eventual sound. It's useful to look for these irregularities as they also tell us in which frequencies the speaker might be resonant. Another reason for DI irregularity might be due to crossover issues.
In practice, a smooth DI speaker will naturally take well to equalization - EQ changes to the on-axis frequency response will be proportionally reflected in the off-axis sound, maintaining tibral coherence. Remember that human ears quickly lose acuity below about 400Hz and above 7kHz so be mindful of not getting too obsessed about minutiae in the upper frequencies especially.
For the LS50, we can correlate the color polar maps with the DI and see that directivity has an irregularity mainly centered from 1.5-2kHz but beyond that, certainly by 5kHz, the tweeter maintains quite constant directivity essentially into the rest of the audible spectrum.
4. Blending Nearfield Lows and Gated Highs to create the quasi-anechoic frequency response
Clearly, we now need to account for the lower frequencies. In a home situation with limited space, there's really only one practical way to get the job done - nearfield measurements to capture the output from the low frequency woofer(s) and adding the amplitudes with the output(s) from port(s). This technique has been described since Keele in 1974 with a long track record.
This nearfield technique requires that we place the microphone as close to the center of the driver as possible (within 0.055 * D, where D is the cone diameter in cm to achieve ~1dB resolution). Therefore with the LS50, the cone diameter (not including surround) is 9.5cm which means we want to keep the microphone within about 0.5cm distance but of course not touching. As for the port, place it centered, flush but not inside. Notice that the concentric driver arrangement with the LS50 means there's only one driver to measure. Typically we can also do nearfield captures of the tweeter and an overlay will give us an idea of the crossover frequency.
Whereas the gated windowing limits low-frequency resolution, nearfield technique typically is only accurate for low frequencies up to the "Fmax" (calculated as 10950/D in cm for infinite baffle, or 10950/baffle width as an estimate in real life speakers). For the LS50, the nearfield plot should be conservatively accurate to 550Hz.
Now, let's then do a "blend" between the near-field and far-field around the overlap between 350Hz where the gated measurements end and 550Hz where the nearfield results remain valid. Thanks to our DIY friends, we can use Charlie Laub and Jeff Bagby's excellent FRD Response Blender and Minimum Phase Extractor v. 2 to get the job done.
FRD Response Blender is an Excel spreadsheet where not only can we blend the two responses but also apply a baffle diffraction step correction, typically a -6dB attenuation approaching 0Hz. Since the LS50 has a specially designed front baffle to reduce cabinet edge diffraction, I'll be reasonably generous and assume that the diffraction "rippling" is minimal. (For even more advanced baffle diffraction calculation, check out the Baffle Diffraction and Boundary Simulator.)
I can then export that blended frequency and phase response out and use REW to create a "final" graphic of the CTA-2034 "inspired" tracings for the KEF LS50 that looks like this:
It's important to standardize the aspect ratio for curves due to the potential to change perception if the image is spread out too wide or the Y-axis too compressed. I'll use 25dB/dec for my graphs. |
Overall we can see that the LS50 achieves a respectably flat quasi-anechoic frequency response across the CTA-2034 "Listening Window" (defined as on-axis plus +/-10/20/30° horizontal plus +/-10° vertical off-axis), with a gradual loss in bass power below 150Hz and steepens below 50Hz. As a small speaker, it is obviously missing the lowest bass.
If we take some time to look at that frequency response, we see what appears to be a frequency "step" from 1.5-2kHz that correlates with irregularity in the directivity indices. The crossover point is just above this at 2.2kHz. The frequency response also shows a relative accentuation of the treble from 2-7kHz evident particularly in the "On Axis" and "Listening Window" curves.
I'll need a bit more experience with small speakers like these in my rooms before I can get a practical "read" on how much the Directivity Indices affect the sound quality. With near-field listening like on the desktop, look to the "Listening Window" for an approximation of the frequency response. In larger rooms where you're sitting in the far-field, what is heard will be more affected by room reflections so look at that "Early Reflection" curve for guidance. On a side note, sometimes you'll see on these Spinorama diagrams a "Predicted In-Room" (PIR) frequency response which is defined as a mix of 12% "Listening Window", 44% "Early Reflections", and 44% "Sound Power" curves. For the most part, PIR curves looks like "Early Reflections" but typically about 1-2dB lower in amplitude. Unless I see something unusual, I think it's good not to clutter the graph with too many lines.
[Those looking for a more technical description of what is "near-field" vs. "far-field" as well as terms like "free-field" and "diffuse field", check out this Siemens page. As you can see, that definition of near-field as <2 wavelengths is a scientific one describing behaviour of the acoustic wave. In practice, typically we speak of "near-field listening" as somewhere between about 2-5'; beyond that, we're looking at "far-field listening". Furthermore, there's the matter of speaker size such that "far-field" measurements typically would be ≥3x the length of the longest speaker dimension for proper summing of the drivers.]
5. Step Response and Cumulative Spectral Decay
A couple of notes about this. First, the miniDSP UMIK-1 USB microphone I'm using is limited to 48kHz sampling rate and appears to implement a sharp linear phase filter for the ADC resulting in visible pre-ringing. Also, the microphone polarity is inverted so I had to flip the graph to what you see above. Other than the pre-ringing, this is basically what Atkinson found in Stereophile. As we can see, the tweeter (initial negative deviation) is hooked up with opposite polarity to the woofer (larger positive low-frequency step and roll-off). This is a common arrangement with passive 2-way speakers. While the crossover between the drivers are not "time/phase coherent", the step response is nonetheless temporally nice and "tight" as a coaxially arranged driver.
BTW, here's an interesting discussion on CSD audibility (or lack thereof).
6. Linearity from 70dB - 100dB SPL
Since I'm borrowing this speaker from a friend, let's not push things too far but still check out how linear the output is playing a 200ms multitone starting at 70dB, gradually increasing by 3dB up to 100dB SPL. Here is the multitone I'll be using:
As you can see, we have 5 tones here ranging from 300Hz to 2.5kHz for the speaker to reproduce. I will record the tone through the UMIK-1 calibrating the 85dB SPL tone as the 0 point. In post-processing, I will calculate relative peak amplitude deviation from linear across 3dB steps up to 100dB and down to 70dB SPL.
Here's the LS50:
The assumption here is that the whole system from DAC to preamp to amp to measurement microphone is more linear than the speaker itself when reproducing tones of various amplitudes (probably a safe bet!). While I cannot know the accuracy to an absolute degree, I can still make relative comparisons between devices. I've measured other speakers with this technique and what I can say is that the result here with the LS50 is very impressive. Within the 30dB amplitude range of 70-100dB SPL; there <0.7dB deviation from ideal. We'll talk about the results from other speakers in the days ahead for comparison.
7. Total Harmonic Distortion (THD) & Intermodulation Distortion (IMD)
Using a frequency sweep in REW, we can get a sense of THD with the speaker playing at ~85dB SPL level:
So as not to clutter the graph, plot down to 6th harmonic. |
You can see that I put the cursor at 1kHz with a horizontal level at -40dB (1%). Below around 500Hz we're likely seeing a bunch of room effects. Above that, it's nice to see consistently clean, <0.3% THD level. This gets back to the question posed by our blind test a few months back; is THD actually all that audible? Though perhaps not absolutely necessarily, I think it would be good to see <1% THD as a sign of an "accurate" speaker. Note that the the UMIK-1 microphone has limitations so I'd be using this as a quick screen rather than an absolute measure.
To get a sense of whether intermodulation changes significantly across amplitude levels, here's a panel of four IMD readings from 60 to 90dB SPL:
Looks quite reasonable. Notice the presence of a 700Hz tone showing up which is the highest distortion peak from 70dB SPL up. Not sure what to make of that and will keep an eye on it with other speakers. In theory at least, these non-harmonically related intermodulation products should be more audible than THD.
8. Stereo Pair
Most speaker tests I've seen published are done as a mono evaluation only. I wanted to make sure to check that both speakers in the pair measured closely as this will have significance for channel balance and for precise imaging. The LS50s came in a box of 2 presumably matching speakers - here are the overlaid/side-by-side comparison results from my friend's "LEFT" and "RIGHT" KEF LS50s.
Impedance match:
Frequency response - in room and gated window (gated response attenuated by -6dB to separate):
Cumulative Spectral Decay:
Looks great! The two speakers are almost identical. Impedance is precise suggesting well-matched crossovers and drivers. Frequency response is basically within 1dB of each other. Time domain CSD waterfall over 3ms looks clean for both with similar pattern. This speaks well for KEF's manufacturing standards (made in China) and suggests that stereo performance should be excellent with precise soundstage and imaging.
IV. "So how does the KEF LS50 sound?"
As usual, although I show the objective results first because I believe these are much more useful/interesting than my subjective opinions, I listened to these speakers first before measuring them. In fact, I had them up on my computer desktop within an hour of getting them home from my friend's place. ;-)
As I write this having both listened to and measured the speakers near the end of my evaluation, I'm sitting on my computer workstation (as per the picture at the top of this post) listening to Louis Armstrong singing "St. James Infirmary (Gambler's Blues)" from Satchmo Plays King Oliver on these LS50s. I'm using a class D SMSL SA300 on the table which sounds very good by the way with a variety of speakers I've paired them with. DAC duties provided by the Topping DX3 Pro. Yup, that's ol' Satchmo singing alright.
As a nearfield monitor these speaker are precise, and render a wide soundstage. Those QSound enhanced CDs like Sting's The Soul Cages really do come alive with the sense of envelopment. A track like "One Day More" from the Les Misérables soundtrack (original London cast recording) where there are multiple parts throughout the song with each singer occupying a specific place in the soundstage was rendered beautifully. You can close your eyes and imagine the placement on stage as the singers take turns in and out of the track maintaining their positions without getting congested at the overlaps when multiple voices gather yet with independent parts.
As a critical review, there are obviously 2 limitations I hear with these speakers:
First, as expected for small speakers, they don't have much low bass power. On a desktop, using some EQ to boost the bass could help but it's too thin for my taste below 80Hz. Definitely consider options like placements near walls for bass reinforcement (listen for bass bloat) and consider sub(s) to fill in from 20-100Hz.
Second, beyond the bass limitation, I agree with some subjective reviewers that the tonality lends more towards neutral-cool than neutral, and certainly not "warmth". I had this thought when I first heard them prior to the measurements but after performing the tests, I can't help but think that the "step" up around 1.5-2kHz with elevated upper-mid frequencies to 7kHz in the "Listening Window" could be tamed a bit. Human hearing is highly sensitive around these mid/upper-mid frequencies so it's important to get these frequencies "right". Remember the old "BBC Dip" where a drop ~3kHz may be preferred by some or may correct for differences between microphone recordings and hearing a live performance as per Linkwitz's discussion here (see "Psycho-acoustic 3 kHz dip"). Interestingly, the measurements at ASR as well as SoundStage!'s NRC results both show this upper-mid boost, but it's not as well demonstrated nor discussed in John Atkinson's Stereophile measurements.
Believe it or not, I agree with Mr. "I hear cable differences" Audio Bacon's review more than Stereophile's almost universally positive comments!
I hooked these guys up in my main soundroom for a listen as well (no subs):
As you can see, I'm running these with the Hypex nCore NC252MP amp, some homemade speaker cables (bi-wire plugs disconnected), Raspberry Pi "Touch" streamer or Surface laptop, Emotiva XSP-1 balanced preamp, and the RME ADI-2 Pro FS R Black Edition DAC. In this configuration, I'm not running the speakers close to any walls for bass reinforcement.
In general, the LS50s sound pretty good >6' away at the "sweetspot". Again, like with nearfield listening, I remain very impressed by the soundstage these speakers project. They remain agile, conveying good detail - Joe Beck & Ali Ryerson's "Scarborough Fair / Norwegian Wood" is an example of such a sprightly track on the TAG McLaren Audio Test disk.
The weakness clearly remains the relative lack of bass if one listens to any pop or rock. I was in an old-skool pop mood when I had these in the sound room and thought Debbie Gibson's "Foolish Beat (Extended Mix)" (off an old CD single) sounded amazing! Good female vocals, plus the LS50 rendered the "classic" '80s pop sax nicely. There's a percussion solo part in the middle which came through with good punch, demonstrating that bass is there; just not very strong and easily overshadowed. (I know... Debbie Gibson isn't exactly typical audiophile music, let's listen to some Rick Astley next time. ;-)
Remember that the LS50s are not particularly "forgiving" speakers and poorly mastered stuff like Bryan Adam's Anthology (1980-2005) sounded nasty, overly bright, and "strident" to use a common audiophile adjective. Granted, I had never thought of this album as a model of sound quality, but the LS50 simply laid bare and seemed to accentuate the deficiencies. While the production is much better with Donald Fagen's The Nightfly, tracks like "I.G.Y." are already on the "bright" side which the LS50s further pushed in an unpleasant way for me. I also wondered with some of these pristine-sounding tracks if the upper-mid/lower treble might be a tad congested during dense passages.
I definitely would try some EQ or run DSP correction for regular listening if I owned these (not to forget to use a sub or two!).
V. Summary...
In summary, the KEF LS50s are great looking speakers wrapped in a modern, stylish design that feel solid and I can certainly see their appeal! The price hovering around $1000/pair is also quite good for the quality you get. The small form factor allows it to be placed conveniently even on a computer desktop and certainly if you need surround speakers these would be great options since you'll likely be supplementing them with full-range speakers and/or subwoofer(s).
Having said this, and with the expectations I had when borrowing this pair of speakers, I ended up not enjoying them as much as I had hoped. The universal praise they got when first released seems a bit much, I think. I don't know about the competition in 2012/2013, but Stereophile declared them the "Overall Component of 2013", and Guttenberg likewise called these "The Audiophiliac's speaker of the year" in December 2012.
Based on the sound I'm hearing, they're too bright for my taste. As I mentioned above, I honestly would have to apply EQ/DSP to these for longer term listening. I came across this page with more measurements of the LS50 and I love the work the poster did around crossover modifications (gated window measurements in a gym with speakers well elevated - nice!).
KEF states that the LS50s are inspired by the classic BBC LS3/5A speakers. In fact, their whitepaper puts up this comparison graph:
While the LS50 looks smoother overall and has better sensitivity, notice that for this specific LS3/5A comparison (remember, LS3/5A-type speakers were produced by a number of companies), other than the 4-5kHz dip, much of the roughness in the LS3/5A's frequency response is above 7kHz where for men >40, our hearing acuity will have already dipped in sensitivity and we might not overly object to the irregularity (see here for some data from the Normative Aging Study, 2010). While on this company whitepaper graph, we don't see as much of a "step up" in the frequency response around 1.5-2kHz as in my LS50 "Listening Window" measurements, it does still show a relative rise in frequency response from 1.5 to 6kHz.
Both the LS50 and LS3/5A are rather bass-shy. Remember that the LS3/5A is a closed speaker so the ported LS50 could provide a little more bass extension and that bump around 50Hz; clearly not much of a difference though even with the company's graph.
It has been fun putting together my inexpensive "ghetto" turntable measurement rig and grabbing some data with these "modern classic" KEF LS50 speakers. The objective performance of these speakers have been well characterized by others and the results here look comparable. Over time, I'm sure there will be ways to optimize how I measure and maybe dream up new tests similar to how my DAC measurements evolved over the years. Obviously, I have space constraints which will limit resolution and the speaker size I can work with in the basement kitchen. :-)
Perhaps more importantly, I hope that some of the thoughts here can be helpful for the audiophile community - especially those who are "subjective-only" at this point or have not heard of or reviewed the value of those CTA-2034-A graphs. Consider the importance of speaker testing and correlating the graphs with one's own perceptions as an educational exercise. Just looking at the graphs above, I hope we can all appreciate just how "imperfect" speakers are compared to the rest of our electronics! Even with a good pair of modern speakers like these, frequency response is clearly far from flat (unlike DACs and decent amps), a 1dB difference between stereo pairs here and there would still be considered fine, and >1% distortion is rather routine depending on what frequencies we're looking at. When we consider the cost of a good sound system, it is in this context that I've advocated that audiophiles put more money into speakers (and the room) above any other audio component simply because that's where you're most likely to find "benefit per buck".
As usual, I believe there is much to gain if we embrace both subjective preferences and objective performance as "music lovers" and "hardware audiophiles".
--------------------
As we close off today's post, recently, as you know, the KEF LS50 Meta passive loudspeaker has been released. It's interesting to see that the measurements posted on ASR are showing that the new model indeed has a different frequency response profile with a dip into the 2.5-3.5kHz range, along with less accentuated upper-mid and smoother Listening Window and Directivity Index curves. The measurements for the LS50 Wireless II are even more impressive with what looks like essentially ideal Listening Window frequency response!
This is encouraging but for now I'll take Steve Guttenberg's exuberant video "review" of the LS50 Meta with multiple grains of salt. "So much more transparent..."? "Mega improvement..."? "Sounds like an open baffle speaker..." - is that supposed to be a good thing? I'll wait for the opinions of cooler heads (as I finish this article, it's encouraging to hear that Agitater likes 'em).
Kudos to KEF for using the "metamaterials" idea as a powerful USP (unique selling proposition) to build the product marketing upon even though I would not be surprised if other design changes have had greater beneficial effects. The crossover I suspect is a big one with optimizations and I see they slightly lowered the crossover frequency from 2.2 to 2.1kHz.
Bass response unfortunately doesn't appear to have changed much with this new passive Meta model however. KEF's specs lists +/-3dB at 79Hz-28kHz for the Meta and better for the Wireless II active (+/-3dB 53Hz-28kHz). Check out the fancy whitepaper for both these models.
Thanks to my friend linnrd for letting me borrow these along with a haul of other speakers which I'll run some measurements on in the weeks ahead. ;-)
And with that, I wish you health and may your sound room be blessed with beautiful music...
Addendum December 20, 2020:
The KEF LS50 doesn't much of anything due to its rather flat frequency response already! However, here's a FIR filter to try out that should give you a little more on the low end and smooth out the mid further:
Addendum March 13, 2021:
Variations of this technique with complex summing of near-field and splicing with far-field data is used commercially as well. Here's a YouTube video from Listen, Inc. for example.
"This is an important point for audiophiles, many of whom for some reason have issues with blind test methodology thinking that the method results in "it's all the same" data. Logically, true scientific advancement of sound quality requires that we embrace every beneficial method of testing, and the irony is that many audiophiles dismiss controlled blind testing when in fact this is in all likelihood the most powerful tool available for honest research - just as it is for every scientific discipline involving human subjects!"
ReplyDeleteAn audiophile *cannot* conduct a blinded scientific study - that requires far more than just a single person to do with any scientific soundness. Audiophiles correctly dismiss a specific kind of blinded testing - any variant of ABX testing that requires memorisation and comparison of audio - because these tests are garbage when misused for the purpose of trying to ascertain ultimate audibility, which is how they are typically used by this ideological tribe.
Pseudoscience-peddling "objectivists" are in reality not objectivists at all: they rely principally on a subjective method of testing (ABX testing) as their means of ascertaining audibility. This method has no resistance to Type 2 error, which allows them to continue revelling in their head-in-the-sand delusions. It's the modern pseudoscientific equivalent of the medieval Trial by Ordeal. I strongly suggest that you remain on guard for this mob and its zealots, as they are common the internet.
I hate the word "Audiophile" as over the last few decades it tends to describe one that believes in any old folly, semi-magical cures, and has very little logical thinking or understanding of electronics or physics. They tend toward "beliefs" based on what peers tell them or they have read about online. They can never demonstrate these magical abilities to hear difference in wires, cords and other non sensical products that are severely overpriced.
DeleteLet alone passing a blind or double blind test, they rely completely on what they "think" they hear, as opposed to what they can demonstrate to themselves they know they hear. If they know for sure, it could be extrapolated out to show, convince, or demonstrate to others, just how valuable their speaker cables and power cords are.
Thanks for the note Unknown and Kevin,
DeleteYes there are important points which we can parse out...
- I agree, true blind tests with reliable results are highly unlikely to be doable at home and definitely will require more than 1 person to do well. My issue is with some audiophiles who apparently automatically turn off the moment someone brings up "blind testing" even though these have been conducted in the literature for years. I hear you about the potential for Type II ("false negative") outcomes being high especially when we're dealing with comparisons of devices which are of good quality.
- Apart from instantaneous comparison trials with an ABX tester using 2 samples with freedom to select out isolated parts to listen to (like using foobar ABX Comparator), I agree that it would be unlikely beneficial trying to detect more subtle differences if it takes time and effort to change settings and cables in between. I have attended a few cable demos at audio shows for example, and clearly the "results" of those "AB comparisons" are rather silly as a common example that is sometimes touted in the magazines!
- As in most things we have to judge for ourselves the value of something using "shades of grey". All kinds of methodology out there for "blind testing", and it all depends on how systematic the testing was done, the number of people being tested, etc... It's just as silly to accept "blindly" that someone claiming to not being able to ABX a difference casually means there is definitely no difference as it is some forums where even the mere mention of a blind test results in a discussion thread being closed down!
- I do hope that we can somehow "sanitize" the word "Audiophile" Kevin. If the "Mainstream 21st Century Audiophile" in time can be seen as a new generation of audio hardware enthusiasts open to truth, resilient to hype, grounded in scientific principles, yet sensitive to subjective preferences and opinions around the art of music, this would be a nice vision of the future to head down as rational hobbyists IMO. I personally think this can be done and IMO looking around, it has started...
Excellent article on blind testing by Brent Butterworth.
Deletehttps://www.soundstagexperience.com/index.php/features-menu/pulse-menu/834-the-problem-with-blind-testing
You can calculate an average over a frequency range in REW using the Graph Metrics feature: http://www.roomeqwizard.com/betahelp/help_en-GB/html/graph_splphase.html#metrics
ReplyDeleteNice John!
DeleteDidn't notice this feature as I was using one of the older beta 5.20 builds on my "analysis" machine. Again, simply exemplary work...
Updated the sensitivity calculations using "Graph Metrics" nice! Sure beats exporting the numbers in to Excel to calculate ;-).
DeleteHi Archimago.
ReplyDeleteA great write up with some very nice measurement calculations. I enjoyed reading this, as this is a very good summary how to do home measurements and calculations to create an accurate presumption of the performance. Great. Thank you.
Juergen
Thanks Juergen,
DeleteObviously not a professional process here but for me it's a fun start and hopefully can yield some insights over time. Already I have a few other speakers I've looked at and the experience has been enlightening!
More great work as you have used your pandemic time wisely...I think. the LS 50 is a very good bookshelf speaker for sure.
ReplyDeleteAs one who records, just not much in the last nearly a year, I am most concerned about realism as it seems you are. before I have even begun a session I get to the venue way early get set up some, but then when the artists show up I spend much time listening to them warm up in various places in the room or concert hall, church, where ever we are to get a sense of the "real live sound of the musicians and instruments in the space". That is what I what to recreate in my recordings.
I love omni mics if the room is large, tall ceilings, and the ambient noise is the room is mostly absent, meaning the HVAC blower is off and the outside noise is mostly blocked out. Omni's pick up everything, but I love their sounds for small to medium ensemble work.
If we are using a piano and play it for about 15-20 minutes to get a sense of the overall sound and tune of the piano. Recording wise there are some decisions about recording perspective. Do I want what the pianist hears or do I want what the audience hears, or something in between? I generally like uni mics, condensers at C3 and C5 about a foot about the dampers, panned at 10 and 2 o'clock with C3 on the left and c5 on the right...the pianist's perspective and then I might add a distant mic to capture some room on a separate track to add later if the artist wants it. The room is such a huge part of capturing sound and it is no different with listening to speakers...lively, damped, big, small, tall ceilings, or not so tall; and then were do you get to sit?
I personally like the use of subwoofers and bookshelf speakers and getting as much of that low bass out of the cabinet to not muddy up the sound of the bookshelf speakers. It also allows you to add as much bass to the presentation as you like without adding effects to the smaller speakers. I prefer a 70hz crossover to the sub.
By your great work here I can see this would be a very good speaker for my 73 year old ears as a rise in HF for me is NOT a bad thing. I can see how in a small to medium size room how this LS50 with a sub would give some full range floor standers in the $10K range a run for their money. Great work as usual.
Hi Jim,
DeleteThanks for sharing your recording experience. Excellent stuff and it's great that you're thinking through the final mix with how it's eventually going to be presented... Whether sounding like a more close-up event (on-stage with the artists perhaps), or more an "in the audience" perspective.
Yup, there is certainly something to be said about the beauty of revealing bookshelves like these + sub sound. For years I was doing this as well and always was impressed once everything was "tuned in" how spacious and "large" one can make the bookshelf speakers sound. Alas, back in the day with bookshelves + sub for me, I was living in an apartment so the real trouble was not having pesky neighbors complain about "noise" :-).
Interesting you brought up the tonality and rise in HF possibly being preferred. You might be right. From a production monitoring perspective, we might want absolute neutrality so the sound isn't being mixed/mastered unnaturally, but certainly for playback and enjoyment, one might prefer that elevated upper-mid especially listening nearfield or on-axis.
This is also why I think it's important for "purely subjective" reviewers to be insightful of their own auditory capabilities. BTW, Toole's book in Chapter 17 I believe has a section on age and reliability of testing results. Obviously we can be trained to listen better. But males >60 years old typically as I recall will show significant loss in reliability as blind test subjects for his loudspeaker trials.
I think that one of the things that is often missing is the reality or the performance of the actual event captured properly. So much today is so slick, smooth, a product of all the mastering tools that can be bought as plug-ins from so many sources that "promise to fix your mix" so to speak. Compress this, expand that, sweeten this and make this instrument darker. There are often so many tracks that can be fooled with and the temptation is often too great to mess with it. I really hate pitch correction...if you can't sing, don't. Often it seems there are too many hands in the soup that if you were there at the recording and heard the final product you might wonder if it was the same event or session.
DeleteIt is true that many have a favorite mastering engineer whose work they admire, but all of them would probably master any work with their own sonic signature and many of them may be liked by many listeners. If you are doing an LP there is another hand in the soup as well and they will arrange the tracks so that the ones with greater bass are on the outside tracks of the lp.
Of the magazines I receive I am amazed at how many companies offer plug-ins at great prices and the possibilities are endless of how or what you can tweak. I have one of my first recordings of a solo pianist who did a gospel album he had wanted to do for years and we recorded it on a Steinway B 7ft grand. I had to add just a touch of EQ in the bass to make it sound like what I heard playing the piano. We both liked the recording and off it when to be mastered and when it came back the bass was so thin and the overall sound so bright neither of us really like it at all. Luckily I still had the original files and remade the CD with nothing added mastering wise. It could have been that the mastering engineer had speakers that had too much bass or his headphones did. I never track with the Audio Technica ATH 50X as they are a little warm, but I do prefer them when listening to classical music that is recorded from a distance.
My son Nick just bought a pair of Beyer DT 770's that I listened to briefly and I really like the sound of those. I had given him a pair of AT ATH 40's for his gaming with a Focusrite Scarlett 2i2 and a SE Electronics SE7 mic.
Then after all of this we have our own gear that adds something to every recording we listen to. I don't use tone controls as much these days as I did 20 years ago as I am now often listening to what the engineers decided was the "right presentation". I would say that most of the time they get it right.
I had these for a few years and loved them for the most part but I always had trouble getting the bass right, they really want to be crossed over at 120-150Hz but then you start localizing the subwoofers in a 5.1 setup, for stereo it wouldn't be a problem. Hopefully you get the chance to review the KEF R3 at some point, they still have the nice highs but the bass problems are fixed of course.
ReplyDeleteGreat, thanks for the tip!
DeleteWill keep an eye out for the KEF R3's. Not much more expensive than a current LS50 Meta and I see it has a larger 6.5" bass driver as a 3-way.
Definitely looks like it'll be an easier integration with a sub lower down... Agree, 120-150Hz crossover far from ideal.
Hi Arch,
ReplyDeleteNice work on your qSpin setup.
Initially, I was a bit skeptical about Spinorama because of its close association with Harman. (I'd feel the same if it were any other corporation.)
Although it's the closest thing we have to an objective evaluation practice for loudspeakers, I have a few remaining issues with it.
I think what it tells us is exactly what it claims: it will be a good predictor of personal preferences for mono speakers in a fairly large room playing specific standard recordings.
My main issue is that it ignores the rest of the chain. The preferred curve is, from what I've seen, pretty close to many monitors used by many producers using 'standard' recording techniques.
The thing is that recording techniques and preferences vary widely. I'm reminded of how dull and lifeless many EMI classical recordings sounded back in the days of Suvi Raj Grubb. Then I read somewhere that he used Quad ESLs at moderate volumes to monitor recordings. So... they don't sound dull and lifeless on Martin Logans. However, if everybody were to start using the same house curve that would be a really interesting development.
As for predicting room response, there again the variations are wide. For example, in my room with its enormous 60 Hz peak, the LS50s produce subjectively useful output to about 50 Hz. Rock music might sound a bit thin, but Charlie Haden's bass sounds pretty full even without subs/woofers.
And, there's user preference, the last link in the chain. E.g. a full mid-bass reduces clarity, at least in my experience, so many listeners (like me) might not find this such a big defect. As another poster noted, elevated highs might sound better to us older males. My first impressions of the KEFs were smoothness: you can get away with a peak in the upper mids if the response isn't rough (and once again, clarity is improved.)
I bought the LS50s to integrate video and music: this hasn't happened yet, largely because of room layout, but I'm quite happy with them, and haven't seen or heard anything as good (to me) near the price. I do use EQ in my surround setup and convolution or EQ (or sometimes nothing) in stereo. Sometimes switching off the EQ will improve dialogue clarity in movies with high surround channel levels. And I'm continually turning down the sub to avoid disturbing neighbours.
I'm trying to avoid the temptation of listening to the Metas: why try expensive wine when you enjoy a moderately priced grenache?
I would hate to see one practice become our sole method of evaluating speakers, and listeners missing out on the opportunity to try different flavours, one of which might be more to their taste than a Revel or JBL. And in this case, I would hate to see subjective preferences in speakers being labeled as 'wrong' because they differ from the 'standard'.
Whatever objective criteria you might use to select speakers, you're not going to know if they're right for you till you get them home and listen to them with your music (or films.)
Cheers
Phil
Edit: so many listeners (like me) might not find the absence of a full mid-bass such a big defect.
ReplyDeleteThanks for the note Phil,
DeleteWell expressed as usual... Yeah, the Spinorama technique is interesting I think for integrating many measurements and helping provide a way to appreciate the complexity of speakers in rooms. How exactly the speaker presents itself in any particular room and "sounds" will definitely be individual. And then there's of course the questions of how we'll perceive it as music lovers listening typically these days in some form of "stereo" (whether 2-channel or multi).
Agree as well that we should not end up being dogmatic with the measurement/objective technique. Should always be on the lookout for potential new and better ways as might be revealed over time!
The original LS-50 had that mild Rise, in the 2-3khz range or so, that gave it those forward and clean and clear vocals and sounds, that appeared to shimmer and be so present. I have a speaker at home, that has a boost in the same exact range, and it has that almost "magical" sound, as if singers are right there! Thing is though, it gets old fast, and gets tiring. Same with the old LS-50. It did some things fantastic, but was too forward. I do not think so much bright, but more that 2-3 khz range was irritating.
ReplyDeleteHi Kevin,
DeleteThanks for the info from personal experience. I think you've nailed it with that tendency for the LS50s to sound "forward". This was my concern as well, hence my comment about applying a bit of DSP for longterm listening (of course I do that already in my soundroom for years now!).
Not as bad by any means but it reminds me of when TVs are purposely tuned a bit too sharp or high contrast. Initially it can be alluring but as time goes on, we might want to tone things down just a tad to achieve a more comfortable experience!
Thank you for extensive info about measuring loudspeakers at home! Just some notions:
ReplyDelete- Your room (kitchen) has nothing but very reflective large panel-like surfaces. It is like worst scenario really. Is it really the only room possible? Can you take off all cabinet doors?
- Please set the measurement axis diagonally in the room, to minimize reflected energy and to get some more gating time.
- typical two-way speakers can reliably be measured at 1m distance, max 1,5m. This greatly improves gating reserve and reduces relative level of reflections compared to 2m. You can test this by changing distance of eg. 60-200cm by 10cm steps. You'll be amazed of differences!
Greetings,
Juhazi
Thanks for the note Juhazi,
DeleteYes, you're right, actually in some subsequent measurements, I have been able to rearrange the directions a bit and provide more distance to increase the gating time. As an empty space in the house, this basement kitchen is the most convenient simply because nobody in the family uses it and I can leave my stuff all over the place without anyone complaining ;-).
I had to Google "Dyad". Otherwise, very nicely done Arch.
ReplyDeleteYou've shamed me by reminding my turntable is STILL not 100% done, but I'm working on a dedicated room to put it in so, there's that excuse.
Btw, being inherently lazy, or in engineering terms, preferring a more efficient solution, I opted for one of these https://www.amazon.com/gp/product/B07WZL4W39/ref=ppx_yo_dt_b_asin_title_o07_s01?ie=UTF8&th=1
It's not made clear on webpage, but the unit can do a single 7.5 degree rotation "step". Via wireless remote. Yaay. Load/speaker depth can be increased by ringing it with https://www.amazon.com/gp/product/B07KDR49KW/ref=ppx_yo_dt_b_asin_title_o07_s00?ie=UTF8&psc=1
and creating a bridge between centerpoint and ring.
cheers
AJ Soundfield
Hi AJ,
DeleteThanks for the tip! That looks like a very useful product and the 40kg (!) weight limit would be just great for most small speakers.
Looking forward to your work, AJ!
There seems to be a trend toward eliminating criticism on the ad supported sites and John Darko is no exception. My interaction with him in person and comments when he allowed them have generally been negative. He is guy who writes well, found an audience and believed that his ideas were as good as his writing. He really can't defend himself from any criticism so he doesn't allow it.
ReplyDeleteGreat review! I agree your a point of view. root for your post from South Korea
ReplyDelete