Headphone Reviews

Philips Fidelio X2 (~$299, expected availability November 2014 on Amazon)
About a year and a half ago I reviewed the Philips Fidelio X1, the precursor to the X2. I felt at the time (and still do) that it was one of the very few open headphones that might satisfy bassheads, but I didn't consider it an audiophile headphone like the Sennheiser HD600 as the bass was too warm to be considered neutral. I also had some issues with some aspects of the build: the headband was too small; the cable resistance was too high; and the earpads were not replaceable. These issues were noted widely amongst headphone enthusiasts on-line as well.

I'm very happy to report that the folks at Philips were listening, and, in my view, went above and beyond simply fixing the issues—while the X2 shares a lot in common with the X1 at first glance, deeper investigation reveals significant changes and real improvements with this new release. The X2 is an excellent headphone...yes, even for audiophiles...maybe especially for audiophiles. For the first time, in a long time, I think we've got a headphone to rival the Sennheiser HD 600 in the mid-priced open headphone category. Yes, I think it's that good.

Physical Description
The Philips Fidelio X2 is a full-sized, circumaural, open headphone. The intended use for this headphone is as a mid-level audiophile headphone for use in quiet places like home or office.

Cable entry is on the left earpiece and uses a flush mounted 3.5mm TRS jack—aftermarket cables will be easily accommodated. The previous X1 cable had a somewhat higher than normal resistance of about 1.8 Ohms, which can be problematic for a headphone with a low impedance (35 Ohms in this case). The 3 meter, Kevlar reinforced woven cover OFC cable of the X2 has about 0.7 Ohm impedance. The cable is terminated on both ends with a 3.5mm TRS plug; there is no remote control cable. A 3.5mm to 1/4" adapter and a cable management clip are the only accessories included. This is a headphone for home use, so I've got no problem with the money going into the headphones and not accessories.

Another gripe I had with the X1 was that its headband was not spacious enough for folks with larger heads. The outer calf-skin leather covered headband loop of the X2 is significantly increased in size and should now fit large heads much better.

The headband pad of the X2 is a large, cushiony, mesh fabric suspended by an elastic strap. I'm really not a fan of this design in general—one-size-fits-all usually doesn't. I found the X2 to be a pretty good example of this type of suspension done fairly well, though a tad on the tight side for me and my slightly larger than normal head.

Continuing with the comfort theme, the large circular velour earpads are a treat for the ears. Though the outside of the pad is circular, the inside opening is oval shaped measuring 50mm X 60mm and about 15mm deep. The inside wall of the cushion opening is extremely soft and compliant so when your ears do touch it's no bother at all. Unless you have enormous ears these pads should be very comfortable.

Another very important improvement over the X1 is the pads are replaceable on the X2. The pads have four holes that receive the four posts on the earpiece. There's also an "align pin" on the rear of the pad and a small hole in the baffle plate to receive it. At first I was convinced the "align pin" was some sort of acoustic vent, but the hole into which it sits is closed. There also appears to be vent holes built into the base of the pad attachement posts. I did ask a Philips engineer about acoustic trickery that might have been built into the pads, he assured me repeatedly that though there is some mechanical damping built into the pad to ensure a tight, secure fit, the pad attachement features in this paragraph do not have any acoustic tuning characteristics. He's going to laugh when he reads this, but I still don't buy it. My skepticism arises from nothing more than the impression that the Philips audio engineering group seems wickedly smart to me, and all those holes must be doing something. Ah well, there's bound to be many secrets in this type of work, guess I'll just relax and enjoy life's little mysteries.

Like the X1 before it, the X2 has a 15 degree angled 50mm driver. While there are certainly many similarities when you inspect these two headphones, a deeper look—and of course chatting with one of the engineers who designed it—reveals a significantly changed headphone. Yes, I would say the X2 evolved from the X1, but I'd also say there's a lot of evolution going on there. Probably best just to quote a few lines from his email:

"The major acoustic changes between the X1 and X2 are a reconfiguration of the damping components in front of the driver, and also the driver diaphragm itself, which is now a composite design. The new diaphragm materials enabled the driver's raw high frequency response to be very smooth and so allowed us to finely tune the upper-mid and high frequency response of the headphone using the front damping components.

The LMC (Layered Motion Control) diaphragm is our special recipe for a polymer-gel-polymer diaphragm. The main reason we use it is to dampen the bending modes of the diaphragm. Bending modes abruptly affect the equivalent surface area of the diaphragm as well as possibly contributing to distortion. In the case of the X2 driver we have a composite type diaphragm, meaning that the diaphragm has different components. We use the LMC material for the dome of the driver, and single layer polymer material (same as X1 driver) for the torus. The bending modes of the dome will affect the frequency range approximately above 8kHz, so the LMC is benefiting that range and helped make it smoother as was our intention. While we found that for this size driver we preferred not so high mechanical damping for the torus (suspension) and kept it the same as the X1.

When I speak about the damping components in front of the driver I'm referring to the configuration of holes in the speaker plate that sits in front of the driver including the 'felt' that covers two of the holes. This has changed slightly since the X1 i.e. thinner 'felt' and less of it, and compared to the X1 we also have a smaller hole in front of the driver's dome.

These changes to the damping components took place during the many rounds of tuning, as we tried many different options and unanimously preferred this configuration during our blind Sound Quality Process tests (we talked about this process when you visited last year).

This design change, and the changes to the driver are related to the different responses that you observed between the X1 and X2 including the dip at 6-8kHz and the increased energy above. These changes deviate slightly from our previous sound target curve for a large circumaural headphones. Which goes to show that despite all of our efforts on highly accurate simulations with the aim to design our headphones to meet our current target curves, the listening part of the development cycle is a stage that still throws up some small surprises, and is essential for refining future target responses.

So there you have it, numerous very pointed and careful adjustments have been made evolving the X2. But there's an undercurrent of meaning that I'd like to draw your attention to with that last bolded bit.

During my visit to Philips Research Labs quite a bit of time was spent pointing out their efforts to include blind testing of a wide variety of listeners. They spent a great deal of time developing software tools to survey unsophisticated listeners about their listening preferences, which included methods meant to ensure listeners were accurately reporting their experience. They've developed software tools to train and grade audio acuity for both the public and internal listeners (who have access to a much more comprehensive training and evaluation program than the Golden Ears training tool available to the public on-line). And while measurements and computer modeling are used heavily in the development process, it's the blind listening evaluations that steer the end product.

I have to add here that I've seen similarly intense and careful efforts at Harmon and Sennheiser. The approach is methodical and scientific, but the focus of attention is on the subjective listening experience and not on measured performance. Measurements certainly play an important roll in the process—having objective target responses is considered an important engineering goal—but in the end, it's what the ear hears that counts.

If I might interject a personal note here, it occurs to me that while in the process of evaluating an unnatural product that's developed empirically by subjective impressions (headphone audio coupling is very artificial compared to listening to sound propagated through air), we're likely to find that headphones that sound good have measured response plots that are far from elegant. What I mean by that is we're very unlikely to see a generic target response curve that looks simple at all. And that makes my attempts at performing objective measurement analysis very difficult. I'm going to have to do a lot of learning.

I suggest also that other advanced enthusiasts doing measurements needs to be particularly careful not to let your own developing understanding of the measurements become a confirmation bias of sorts tainting what you think you hear. This experience has served as a wake-up call for me; I get a bit lost as I try to correlate the things I hear when switch between the X1 and X2 with the things I see in the measurements other than in a few very basic ways. (Course EQ changes, bass problems below the primary driver resonance, for example.)

Worse yet, when I mention my perception of this difficulty with the Philips engineer, he first agreed that my perception of the problem was probably correct, and then went on to say:

There is some logic that there will be slight changes to the target curve for a measurement on a dummy head for different headphone architectures (e.g. big vs. small, open vs. closed). This is easier to think about when you consider the difference between an insert earphone and a headphone. You'll probably recognize from your own measurements and listening that an accurate sounding insert earphone vs. a similar sounding headphone will have a significantly dis-similar dummy head response. In a related way there will be differences, though much more subtle, between different headband type headphones depending on the size of their front cavity and cushions, the position of the driver and the associated acoustic impedances (open/closed headphones may have significance).

On top of that there is inter-person ear drum response variation due to the physiological differences in their ears (which is fine when designing speakers based on measurements in the freefield because our brains are calibrated to our own ears, but cannot be ignored when considering a headphone design based on measurements at the microphone position of a dummy head). These human ear variations are most significant from 5kHz and upwards. It should also be noted that most dummy heads have a limited accuracy at high frequencies with respect to how they represent the acoustic impedance of a human ear.

We are trying to justify these points more quantifiably in our current research. Harman has done some commendable published work in this area, but there are plenty of angles still to explore, including the effect of different sized architectures and acoustic impedances of the components and non-linearities of the systems involved. From our experience I suggest that there is not a pressure magnitude/frequency dummy head measurement curve that fits all headphone architectures that is a perfect target to satisfy all. Hopefully we'll be able to share some of our research on this with you soon.

And sorry to say, I do believe this makes your job (and ours) of interpreting headphone responses a little more difficult.

*sigh*

So...I guess we better go to the next page and get on with the listening tests.