Katz's Corner Episode 16: The Smoking Gun

This story originally appeared at InnerFidelity.com

[Editor's Note: Through a long and tedious process it was found that the Focal Utopia and Audeze LCD-4 reviewed here at InnerFidelity in numerous articles were found to be not representative of currently manufactured product. This article is therefore not entirely indicative of our current impressions of this product. To get a complete understanding of our evaluation the reader should start with this summary article and work back through the articles leading up to our current understanding.]

Looking Back
Episodes 13 and 14, The Big Shootout, generated a lot of controversy in the headphone community. As you may recall, in Episode 13, I picked the Audeze LCD-4 over the Focal Utopia, and it was a very clear victory. The Utopia sounded bright, thin, and edgy, especially compared to the smooth and natural-sounding LCD-4. I knew that Focal fans would not take this result sitting down and that was certainly true. So I decided to bring in a panel of listening experts in Episode 14, to express their reactions to the same phones. The amplifiers we used remained the same as I used for my shootout, but I did have to replace the DAC since I learned the Oppo portable DAC could not be run and charged at the same time. Bummer. Regardless, the consensus of the expert listening panel was very much in line with my own opinions.

After those episodes, conspiracy theories have abounded: Was it valid for Bob to use different amplifiers for the two headphones? First of all, it's necessary in this kind of comparison to match the loudness of each can for an A/B comparison. Not having two matched amplifiers, I used a custom-built AMB to feed the Focals and the Audeze Deckard to feed the Audeze. You may recall that my assistant and I did some careful comparisons of the two amps using the Focals and concluded that the amps seem to sound identical. But since I was not able to get enough driving voltage to feed the LCD-4s with the AMB, the AMB fed the Focals, and of course Focal fans on the net have complained that the AMB I used must be a bright amp.

The next controversy centered on my Svengalian "influence" on my assistant and friends as I administered the test. Somehow my brainwaves have managed to influence five strong-minded individuals in my direction. But don't forget that Paige Coley actually preferred the Focal for some purposes although she agreed with the rest of us on the nature of the sonic differences. Other than the few complainers around the net, most of you have accepted the results of this shootout. Some of you cited your experiences with different samples of the Utopias, one sample a reader auditioned mirroring our judgments. Keep your cards and letters coming, guys, we do appreciate them!

My shootout has awakened the attention of a couple of behemoths: Focal (the company) and Audeze (the company). Obviously, Focal is very concerned, and Audeze is very pleased. But we have to dig further, to be fair to both companies. For example: What's going on with Tyll's Utopia on the Wall of Fame? What's going on with his not-so-favorable review of the LCD-4s? What's going on with the estimable John Atkinson's rave LCD-4 review?

In this episode I intend to get to the bottom of this controversy! Here's how: Focal has asked Tyll if he would be willing to measure the Utopia that was used in the shootout. Ostensibly, if the shootout sample measures like Tyll's initially reviewed pair, then we would just have to accept the opinions of my six expert listeners as another data point in a long line of data points. Likewise, I think it important for Tyll to measure my LCD-4s which I sent along for testing.

The Test Results
Frequency response is not the only objective criterion we should pay attention to. And it's very subject to interpretation...we're very early in the science of headphones. Besides, the ears smooth out a lot of the ratty anomalies we can measure in headphones. So what parts of the measurements do we pay attention to? What parts do we ignore? Tyll and several other researchers are currently attempting to advance that part of the science. But I think that if we find some correlation between a measurement and an audition, that can buttress both the objective and the subjective evaluations. To be fair, we are biased ourselves, because we have to interpret the complex measurements. As you and I and Tyll examine the following measurements of the controversial headphones, keep in mind that I'm interpreting them through my own rose-colored glasses. Therefore I invite you, my readers, to present your view of these same measurements. I'll also be happy to analyze the data in a different way if you request it. All in the interests of science and progress.

Tyll has kindly provided me the raw data from three Utopia sample measurements: his initially reviewed sample (probably from the very first production run), a second Utopia sample he recently received from Focal, and the Utopia used in the shootout (both of which are likely from subsequent production runs. He's also given me the measurement spreadsheets of the LCD-4 sample that he initially reviewed and was not too thrilled with, and of course that of my own LCD-4. Let's look at some very revealing frequency response measurements. I have taken Tyll's average of 5 different headphone positions, then smoothed these with 1/6 octave smoothing.For example, here is a comparison of the left and right ear of Katz's LCD-4 averaged and smoothed:


Fig 1: Katz LCD-4 left channel (red) versus right channel (green). Average of 5 different headphone positions. 1/6 octave smoothing

As you can see, the measurement shows that the left ear is brighter, sometimes by more than 2dB, from 2 kHz through about 8 kHz. To be honest, my listeners and I did not notice that in any listening test, though I'll be on the lookout for it when I get the cans back from Tyll. It appears that Audeze's manufacturer's tolerances ear to ear are not that great. For a $4000 headphone I would expect better L-R matching, at least up to 10 kHz.

Here's a similar measurement of the shoot Utopia:


Fig 2: Shoot Utopia left channel (red) versus right channel (green). 1/6 octave smoothing, average of 5 headphone positions.

These two phones measure distinctly differently! We should not expect them to sound alike. In addition, the right channel of the shootout Utopia is consistently louder than the left by as much as adB over a wide range of frequencies, which I noticed in listening and which I compensated for during the shootout. So this measurement confirms one subjective judgment. Fortunately, I believe that the brain integrates the response of both ears (somewhat). Well, at least it's easier to interpret a single reading than a pair! Here (figure 3) is a comparison of the combined (average) response of both ears of the Katz LCD-4 versus the shootout Utopia:


Fig 3: Shoot Utopia (blue) vs. Katz LCD-4 (green). Average of both channels, 1/6 octave smoothing

If this is confirmation bias, then I'll eat my hat! I think this comparison speaks for itself. It clearly confirms the sonic differences that we noted between the Katz LCD-4 and the shootout Utopia. I urge you to re-read Episodes 13 and 14 to see how eerily Figure 3 mirrors our subjective judgments.

Since "measures flat" is not the goal, I'll redraw this comparison between the two headphone models with 0dB representing "perceived flat." My compensation curve is definitely in progress. It's based on the Harman curve plus some fudging I did by comparing a set of LCD-X with my reference speakers until they matched as close as humanly possible. It's far from an accurate curve, but it's the closest I can come at this date. I took both of the headphone measurements, normalized each one's gain to match the Katz-Harman curve in the midrange, and then determined the difference between the measurement and the Katz-Harman curve. In theory, the closer each trace comes to the 0dB line the flatter the perceived response of the phone:


Fig. 4: LCD-4 Katz (brown) vs. Focal shoot (green). Katz-Harman curve represents 0dB. 1/6 octave smoothing, both ears combined.

This compensated measurement also seems to correlate with the perceived responses. 0dB would be the "optimum" perceived response, provided that the compensation curve is accurate. The first thing we notice is that the Focal has a distinct steep rolloff below 200 Hz, while the LCD-4 stays a lot closer to flat. This correlates very well with the shootout descriptions of the Focal's weak bass response.

In the lower midband the Focal is boosted compared to the Audeze, which further increases the audible difference between the midband and the bass. From 1.2 to 2.5kHz both cans dip, but the Focal begins with a steep peak circa 1.3 k and a steeper dip while the Audeze's is a lot smoother and gentler. In general, sharp peaks and dips translate to a less smooth and ragged sound quality.

The real tattletale is in the 2kHz to 10kHz range, where the Focal trace is distinctively brighter and more ragged (bigger ups and downs) than the Audeze and in some ranges, significantly brighter than the 0dB line, brighter even than the compensation curve. Tyll is more of an authority on this but I suspect these ups and downs represent comb filtering inside the headphone cavity. These measurements correlate extremely well with our subjective judgments in the shootout.

The Smoking Gun
Now we compare various samples of the same model headphone. Below we compare the Katz LCD-4 sample against Tyll's review sample. Here's a left-right combined, 1/6 octave smoothed amplitude difference display:


Fig 5. Amplitude difference between Katz LCD-4 and Tyll LCD-4 sample #1. Both ears averaged, smoothed 1/6 octave.

Below 1kHz the two samples are extremely close, far less than adB for the most part. The slight bass difference means that the Katz sample is about 1dB hotter than Tyll's sample below 60 Hz. Above 1kHz there is a significant difference between the two samples of the same model headphone. The Katz sample is depressed from 1kHz to about 4kHz by as much as 2dB compared to Tyll's sample. But what I think is significant is that the Katz sample is distinctly brighter than Tyll's sample from 4k to 10k. This could easily explain why I and my listening panel really liked the LCD-4 while Tyll felt it sounded closed in and not clear enough. The narrow difference circa 10k could be Fazor design? Fortunately, it's a narrow dip and above that we hope the differences are not that significant to the ear.

Editor's note: I hope Bob doesn't mind me jumping in here for a moment. While his plot above does a good job of showing the the differences between the two headphones, I also think it hides the balance of the headphone in and of itself. For clarity I'm going to add the raw frequency response plots of the two LCD-4 headphones reviewed.


In my review I wrote:

"Well, the problem is the LCD-4 to my ears does so well in the bass and mid-range run-up to this frequency that when it all of a sudden goes missing (it's about 8dB down from where it ought be) it begins to stick out like a sore thumb. I have to say that all my commentary on the bass and mid-range performance was burdened by having to evaluate while being very conscious of this missing octave. When I switched form the Sennheiser HD 800 S to the LCD-4 it was like someone draped cloth over my ears...sort of."

I followed that by saying I also thought the top octave between 10kHz and 20kHz was too bright in comparison. Looking at the plots above, you'll see the very steep descent on my pair at 4kHz, followed by a rough descent to 8kHz that is overall steeper than Bob's unit. You can also see that the top octave is higher in level relative to the midrange on my review unit. The response between 3kHz and 20kHz is more "U" shaped in my review unit, and falls more evenly and closer to the shape of the Harman target response with Bob's LCD-4.

After this I compared the shoot Utopia against Tyll's early Utopia...the one he has on his wall of fame:


Fig. 6: Tyll's Utopia sample, green. Utopia shootout unit, blue. 1/6 octave smoothed, both channels combined.

Figure 6 tells us that the shootout Utopia may be literally - defective, except that it closely resembles Tyll's sample #1 from 3 kHz on up. But the depression below 2 kHz would clearly make the shootout Utopia sound thin and bright compared to Tyll's early sample. This is a giant manufacturing discrepancy. Tyll has measured a third sample of the Utopia sent to him recently by Focal. Let's compare the shootout Utopia to that third sample:


Fig 7: Utopia sample #2 (brown). Utopia shootout (blue). 1/6 octave smoothed, both channels combined.
To quote Groucho Marx, "Who are you going to believe? Me or your own eyes?" Utopia sample #3 and the shootout Utopia are for all intents and purposes, identical. Bravo for manufacturer's consistency, except that we now know these sound different than Tyll's Wall of Fame sample. Fig. 16-6 shows a big difference between Tyll's early unit and the latter samples.

Furthermore, the shootout Utopia has a weaker bass below 80 Hz, which would exacerbate our subjective judgment of the weak bass range of this can. Will Tyll audition Utopia sample #3? Does the Utopia still merit the Wall of Fame? If possible, Tyll needs to audition my LCD-4 sample, and if possible I need to audition his early Focal. (Sorry, Bob, Long gone.) But I'd rather wait till Focal send him a few more samples to determine consistency and which direction they may be trending. If they trend more towards the early version, so much the better. Likewise with the Audeze: Tyll reports that Audeze are sending him a few more LCD-4s for measurement and review.

We are living in interesting times. This comparison between headphone samples is the real smoking gun: evidence of manufacturer's inconsistencies. My friend has already contacted Focal for a replacement headphone and they are being very cooperative. For his sake, I hope he receives something more like Tyll's initial sample. For those of you considering the lovely LCD-4s, I hope you receive one like my own golden reference! Caveat emptor.

Editor's Note: Well then...wowie zowie! Past little while as Bob has been publishing his findings, I've been shaking my head wondering if we were even listening to the same headphones...turns out we weren't. Glad I'm not going deaf.

Thought I'd post this graph to put it all in one piece.


The thing that strikes me most strongly is how much Bob's Utopias deviate from the rest of the pack. It's got a 4dB hotter peak at 3-4kHz; it's got the most uneven run-up to the peak; and the spike at 6kHz and 10kHz are about 8dB hotter.

I've got three LCD-4 coming in from Audeze. I'm going to do some more listening and measuring, and then I'll probably post up some adjustments to the Wall of Fame and notes in past articles.

Thanks for digging into this one, Bob.

[Editors Note: See this post for Focal's comments.]