Modding the Dekoni Blue ("Arch-Mod5"). Streaming audio: Apple Music to be "true hi-res lossless" stereo with multichannel (Atmos) in June.

A couple weeks back, I measured the Dekoni Blue headphone which as I noted, is itself a "modded" Fostex T50RP Mk 3 planar-magnetic (aka "Regular Phase", aka "isodynamic") headphone. It would be completely remiss of me if in a review/discussion of the Dekoni/Fostex I were not to spend some time talking about modding!

Modding has an very long tradition in the 'head-fi' world and I think it's been generally recognized that the Fostex headphones are the classic headphones to perform surgeries on. Much of what I'll be talking about here will be nothing new to those deep into this stuff already. For more on modding, check out this Headphonesty article for a background.

To start, as always, we need a plan and a "vision". As I mentioned in the last article, the Dekoni Blue sounds pretty good for the most part, so my job isn't to change the performance completely, but to rather refine what is there. If a headphone sound highly objectionable, one should really just move on and buy something else! Let's review the measurements I published last time for the Dekoni Blue with the supplied earpads and see what we can try...

I showed the measurements using the Elite Velour earpads first because those are what I preferred so any mods will be targeted to improve performance with the Velour. Looking at the measurements above, there are a few things we could try to do that could meaningfully impact the performance in a positive way.

1. I'm not a "bass head" so I think the bass is too strong. I also don't like that 100Hz bass hump; let's see if we can reduce the bass response by a few dBs and push the 100Hz mid-bass hump to a lower frequency. 

2. Notice the 10ms waterfall plot and all those lingering resonances below 1kHz. Let's see if we can improve the time domain performance a bit and reduce what looks like resonances inside that hollow ear cup.

3. The headphone is rather "hot" in the upper treble above 7kHz, especially with the Velour pads. Let's tone that down a few dBs if we can.

While I would love to show you all the measurements I went through and materials I tried, let me get to the point and talk about the simple "Arch Mod5" changes I finally settled on. It works well and sounds great to me. In the process, it taught me a few things about personal preferences when it comes to frequency response.

DISCLAIMER: Modifying your headphone will break your warranty. And doing this may damage your headphone as well. Use your discretion. I'm obviously not responsible for what you do. While I very much like the final results with mine, you are the final arbiter of sound quality and each headphone will be different! While everything can be easily reversed, proceed with appropriate caution.

-------------------

For this simple mod, all you need are a couple of items as "materials". 

1. I used inexpensive 1.5mm thick felt fabric sheets I found at the local dollar store (light blue of course for these headphones, MultiCraft Imports brand used). Make sure this is the soft, low-density fabric and not stiff felt.

2. I looked around for "meta-materials" but came up short ;-). So the other material I used was some SilverStone EPDM absorption foam which is usually applied to computer cases to keep the machine quiet. For headphone modding, we need only a little bit of the stuff.

Otherwise, you'll just need a small screwdriver (to open up the headphones), pen, ruler, and scissors for marking out and cutting out the "craft" supplies.

I. Opening the Dekoni Blue / Fostex T50RP Mk 3.

First thing you do is remove the ear pads - it should be obvious how to do this. Stick your fingers into the ear cup, push on the edge from the inside, and release the ear pad "lip" from the headphone carefully. Make sure not to rip the pads (it's stretchy so should not be an issue).

Afterwards, you'll see the driver and baffle as below. There's a loose foam baffle pad around the driver that will just fall off (not shown), and you'll see the 4 small Phillips cross-head screws you need to remove:

This is why the Fostex headphones are popular with modders - these are the only screws you need to remove to access everything inside.

From here, we can easily remove the driver and baffle unit to reveal the rear and see inside the earcup:

II. Dampen the cup, slow down the bass decay.

As you can perhaps imagine, there's quite a bit of space inside the ear cup and this lack of damping is contributing to some excess vibrations seen in the waterfall plot. So, let's use that SilverStone foam material to reduce the magnitude and persistence of vibrations against that rear "cabinet":

You can see that I haven't bothered with any special shapes other than just rectangular cut-outs of the SilverStone foam material with the dimensions noted for the left and right. You can look at this Mayflower video and see that they have traced out some shapes if you want to get fancy. The difference between the left and right ear cup is the presence of the headphone cord jack in the left cup, so one of the long pieces has been reduced from 2.5" to 2" to fit.

Now, let's use the 1.5mm low-density felt fabric to dampen the space; reducing the air compliance within the cup. Cut out two 3" x 3" felt fabric squares, one for each ear cup. Take one of these 3" x 3" sheets and lay it across the inner cup like so:

Notice for the left earcup, I cut off the edges of the felt a bit to fit better (optional, did not make a sound difference) and also a slit where the connector is.

At first when I was experimenting, I used some cotton batting instead of the sheet of felt fabric but didn't quite like the inconsistent density. I also tried 2 layers of this felt material but it dampened bass amount and subjective "kick" to the sound too much to my liking. One could experiment with thinner 1mm fabric sheets as well.

Make sure the fabric doesn't cover the screw holes. There are 4 dense foam "bumpers" (you can see these sticking out in the pictures above showing the SilverStone foam in the ear cup) that will pinch the fabric in place when you close the cup so don't worry that over time there will be any significant shifting of the fabric inside the cup.

Here's a look at the back of the headphone driver before I closed it up - notice there's white paper-like damping material behind the driver with a selected open square portion. I believe the collective wisdom is to leave that paper alone.

We can see the plastic ridges around the driver on the baffle plate. I know that some modders will "mass load" this section with plasticine, Newplast, modeling clay or just BluTack (again, as shown in this Mayflower video). I tried this, it increased the weight of the headphones a little bit, but ultimately I didn't really hear much difference nor did the measurements improve significantly, so I'll leave it to you if you want to do this. I also tried putting some SilverStone foam in these areas; again, not much difference.

So what happened to the frequency response by just damping the headphone enclosure and reducing air compliance?

The quick measurement above shows me that the bass level is slightly attenuated by the mod (red) and the treble was mildly accentuated. I like that the mid-bass hump is no longer centered at 100Hz, but has a smoother, more extended profile down into the sub-bass with a peak at around 75Hz now. We can imagine that the headphone is now even less "semi-open" than before.

The other improvement we want to see is a smoother/quicker waterfall plot. Let's compare stock (top) vs. earcup mod (bottom):

Yup, damping has an effect particularly in that stretch from 200Hz to 2kHz. No, it's not like the smooth and clean decay we see with the open-back Sennheiser HD650/HD800, or the AKG Q701. An improvement nonetheless.

III. Time to tame that treble!

The headphone actually sounds pretty good at this point. In fact, if you mod one of the sides only and put the earpad back on and listen, you'll notice that the modded side does sound a bit "clearer" in the bass and mid frequencies, presumably on account of the improved damping. You'll also notice that it's quite "airy" due to the extension of the "brilliance" frequencies. To me, this sounds somewhat like a bass-accentuated Sennheiser HD800 but without as expansive a soundstage.

So to tame the treble, I again used the SilverStone foam damping material and cut out a couple of small rectangles 3cm x 1.5cm (~1.2" x 0.6").

Screw the headphone driver back to the earcup, and then stick the foam damping over the driver like so:

There's a fine mesh in front of the driver so the foam is only adhering to that mesh, not touching the driver itself. Notice in this picture that I have put the foam driver surround back on. It's interesting to see that the stock foam material is almost exactly the same density and thickness as the SilverStone material. I know this looks somewhat scary covering up 1/3 of the driver surface area! Due to audio wavelengths, only high frequencies are affected.

I know a few years ago, MrSpeaker sold stickies called "Doggie Treats" that seemed to do the same thing.

Okay, finish up and put the earpad back on and have a listen!

What does it sound like? Well, the high frequencies are now tamed and there should be an obvious reduction in treble harshness. On tracks like Quincy Jones' "The Places You Find Love" (on Back On The Block, 1989), Chaka Khan's vocals which can be quite harsh when a headphone leans too bright should no longer strip paint off the walls ;-). The dynamics on that track, bass and treble detail should be able to "tickle" the ears. Don't forget that these are low-efficiency headphones, I listened with my Drop + THX AAA 789 amplifier for excellent results.

And as for the frequency response after doing this including measurements for both the Elite Velour (dark green) and Elite Hybrid (magenta) pads...

Notice the stock frequency response is plotted as dotted lines and the modded response as solid lines. I have pretty well achieved my goal I think. If we specifically look at the Elite Velour magenta lines, we can see that:

1. Peak bass is toned down 2-3dB. Rate of sub-bass roll-off less steep.

2. Bass bump now in the low-bass, around 70-80Hz instead of mid-bass 100Hz. 

3. Treble tamed a bit to be consistent with the lowering of bass. There's a slight "harshness-dip" into 2-4kHz, and then the reconstitution of the "presence" and "brilliance" frequency ranges to give us "sparkle", "air" for higher order harmonics are reproduced but without too much energy into 10kHz and above.

While I focused on the Velour pads mainly, notice the effect of the mod with the Elite Hybrid pads in green. It's a significantly more "bassy" and "darker" sound. Not my favourite but some bass-heads might enjoy that kind of tonality or if you have a "bright" amp for some reason, this sound profile could be preferred. Instead of a 3cm x 1.5cm Silverstone piece over the front of the driver, something slightly smaller like 2.75cm x 1.25cm would brighten up the treble "sparkle" with the Hybrid pads. Experiment!

The treble mod does not change the waterfall plot and doing all of this did not change the headphone distortion characteristics to any significant degree.

For completeness, we can look at the change in impulse response:

I did not get a reading before modding the headphone unfortunately, but here is the "non-aliasing" 100Hz square wave (band limited to 20kHz) through the left channel at a playback volume of 95dB SPL with source square wave overlaid.

Interesting. Certainly not the "fastest" headphone in my collection as we see the reproduced waveform reaching its peak around 1.75ms. Can't say I've noticed any audible issues with transients and the dynamics hold up well with these headphones. I'll make sure to look at how other planar magnetic-based headphones perform in the days ahead.

One last check! I wanted to make sure that the mod did not cause any linearity issues. Let's play some sweeps at different levels from 65-100dB SPL:

Good linearity between sweeps spaced 5 and 10dB apart.

IV. Summary

Here then are the "core" headphone measurement graphs of the "Arch-Mod5" Dekoni Blue, averaged with 3 or 4 measurements each side. My preference is the modded Elite Velour version:

My preferred earpad.

With the mod as described, the Elite Hybrid pads were a bit bass heavy, and the treble lost too much "zing" for my taste. As suggested above, a smaller foam piece over the front of the driver likely would restore the treble amount to one's taste.

You can compare these with the stock Dekoni Blue as shown at the top of this article. As you can imagine, the reason I call this "Mod5" is because there were 4 other incarnations before I settled on this version. Five versions isn't bad, and that's the thing with modding, it can get quite addictive if one is obsessive and perfectionistic (check out this amazing thread on the Fostex T50RP - too bad most of the images are missing unfortunately). Sometimes, I've had to roll changes back in the process which is why doing measurements each step of the way IMO is essential to making sure I'm proceeding as envisioned!

By the way, the impedance curve has changed a little with the modding. While planar-magnetic drivers maintain very flat impedance, at the resolution of the measurements, we can make out a slight shift lower in the resonance peak which correlates with the shifting of the bass "hump" in the headphone's frequency response:

With the tweaked bass and treble amounts, sensitivity at 1kHz still remained essentially unchanged just shy of 89dB SPL at 0.25V or 101dB/V. This is quite low voltage sensitivity, and with ~50Ω impedance means one needs to supply both adequate voltage and current to have these headphones sound their best. A reasonably powerful headphone amp is important.

I've heard that some Fostex headphones including the latest T60RP can have audible "rattling" when playing loud bass frequencies. After modding, I think it's worth taking a few minutes to have a listen with a slow sweep like this one on YouTube to make sure no rattling, weird resonances or "lumpy" irregularities are heard in the higher frequencies. I know, it's not fun listening to test tones but it's something that needs to be done for "quality control". I had no issues with my "Arch-Mod5" Dekoni Blue.

Although not the most "airy" (I would place these between the Sennheiser HD650 and HD800 in terms of perceived size of the "soundstage"), are not the most precise headphones, nor have particularly low distortion level, on balance, I really like the result of the modified sound! Of course, I'm biased. ;-)

I mentioned Yosi Horikawa's track "Bubbles" on Wandering a couple weeks ago as a great track to listen to on the Sennheiser HD800(S). Well, it's also a great track to have a listen with these modified Dekoni Blues! While not as transparent as the HD800, the added bass weight on these add another perspective to the sound which actually makes the music more tonally balanced IMO.

As I was researching the Fostex T50RP Mk 3 during the modding process, something Tyll Hertsens said back in 2015 resonated with my impression while experimenting:

"The gentle rise from 400Hz to 2kHz of the Mk3 is highly desirable, in my opinion, and is quite similar to the Harman Target Response. My guess is achieving this rise is a tricky combination of a number of things, and may suffer during modifications. If you're a DIYer with a measurement rig, take care to note exactly what this area looks like on your measurements (they may look quite different than mine on a non-IEC-spec measurement system), and try to retain the profile between 400Hz and 2kHz during your mods."

Notice that the Dekoni Blue frequency response has a subtle rise from ~500Hz to 1.5kHz with a relative dip from 2 to 4kHz. While modding, I kept an eye on this mid-range region and made sure to leave these frequencies alone. This relatively gentle rising mid-range accentuation from 400Hz to 2kHz sounds good to me for vocals, adding a bit of "presence" and harmonic richness.

People talk about a "V-shaped" frequency response. As you can see, I like mine more of a "w" with the gentle mid-range accentuation. It seems to add sweetness and presence into female voices (like with Anette Askvik's album) but pulled down at bit by 3-4kHz to reduce harshness. Similarly, male vocals like Gregory Porter's "Don't Misunderstand" on the late Jimmy Heath album Love Letter sounded great - more full, resonant, and natural with that little bit of extra mid-range. 

I've mentioned it elsewhere in my posts. I think that dynamically compressed rock remasters these days can benefit from a little bit of 3-4kHz dip to reduce the sense of "congestion". I recently listened to a 2008 remastered AC/DC's Razors Edge CD which benefitted from a little bit of an upper-mid dip to improve clarity which one could also tweak to satisfaction with extra EQ.

Alright then. Certainly "good enough" sound quality from my modded Dekoni Blue headphones for hours of enjoyment which is ultimately what it's about! These are currently my favourite "go to" headphones for general listening with a nice balance of comfort, sound quality, some semi-open-back noise isolation with reduced leakage, and decent price. Again, make sure to have enough power on the headphone amp for these.

--------------------------

The future of music streaming has changed substantially with recent announcements by Apple Music of lossless streaming starting June and a concomitant price drop in Amazon Music to match. The "800-pound gorillas" have made a move and the little guys will be taking notice. As of the publication date for this post, let's look at the monthly cost (in US$) for the most popular audio streaming services and technical specs of each:

Amazon Music - lossless FLAC up to 24/192: $9.99 ($7.99 if Prime member) 

Apple Music - lossless ALAC up to 24/192 for "Hi-Resolution Lossless" if one opts in, some Atmos content: $9.99

Spotify HiFi - likely 16/44.1 lossless (unclear if FLAC) to launch this year, possibly soon and stay at $9.99 

Qobuz Studio - lossless FLAC up to 24/192: $12.49

Deezer HiFi - lossless FLAC to 16/44.1: $14.99 

Tidal HiFi - lossless FLAC to 16/44.1, partial-lossy pseudo-hi-res MQA (bastardized PCM) up to 24/48 data rate, some Atmos content: $19.99

In my household, the girls (wife and daughter) use MacBooks and iPhones. The guys (me and my son) prefer our Windows PCs and laptops. While I'm no fan of Apple's closed ecosystem model, the way they're going with Apple Music is elegant and thoughtful for the consumer.

As you might already know, I have never really been a fan of all this "hi-res" music hype. Back in 2014, I was already critical that "hi-res" offered little benefit in terms of actual perceivable quality. Some of the earliest "hi-res" albums like Beck's Morning Phase were not hi-res and didn't even look lossless! All that hype around Pono and its graphs, and technical talk about stuff like 24/192 was typically meaningless. And last summer, we talked about being "post-hi-res audio" as audiophiles. Don't get me wrong however, there are benefits with hardware like DACs being hi-res capable for various reasons, not necessarily tied to the music itself being of hi-res though.

While we as audiophiles can argue about these technical details and minutiae as part of the perfectionistic, obsessive, "Golden Ear" hobby, don't forget that the world at large doesn't care. IMO, there was never much money audio companies were going to be able to extract from the market in music consumption just because they advertised something as "hi-res". As I discussed years ago, the value of "Hi-Res Audio" was low and as such the price has to reflect this! Now in 2021, this is exactly what we're seeing with Amazon Music and Apple Music... Lossless and high-bitrate lossy really do not sound very different, and hi-res lossless doesn't demand an extra monthly fee either - a flat rate of US$10/month is great. This has always been the honest trajectory for those who bothered to carefully listen or perform a blind test (whether something like this, this, or this).

All this IMO should also have been clear to the audiophile press including those online sites who have been touting "hi-res" as being "obviously" audible. The press has always been interested in promoting hype, not truth when it comes to a "new" money-making scheme, including ethically questionable advertising like with MQA over the years.

When it comes to truly high-resolution music, I contend that it's still inextricably linked to the production and mastering processes. As discussed in late 2015, I still think if an album deserves to be available as high-resolution audio, it should be mastered with that in mind. As with the world of high-definition video, you can have both Standard Dynamic Range (Rec.709) movies alongside High Dynamic Range (Rec.2020) content that use HDR10(+) and DolbyVision. For example, Taylor Swift's evermore is available as a 24/88.2 hi-res version but sounds no different than the usual 16/44.1 CD currently. What's the point?

If at all possible, why not release both the loud, compressed CD version and also offer an actual potentially "hi-res" version with greater dynamic range, and lower noise floor? What I suggested as resurrecting the "Advanced Resolution" idea. This will clearly sound different and music lovers can decide if a "non-Loudness War" version might actually be of value for them and to their tastes.

Good job Apple for finally jumping on the true lossless+hi-res train in style with no nonsense single-tier music delivery that's flexible, able to handle multiple bit-depths and samplerates, even multichannel Atmos all in one swoop! The competition among companies is great for consumers and I suspect will further erode Tidal's user base - not a bad thing as this will likely finally remove any hopes of the worthless, money-wasting MQA having a significant presence in the streaming space.

If Tidal goes down, this could be bad for Roon (which I use and I like, with a lifetime subscription). More than ever, it would be good to see Roon integration with other streaming services like Apple Music and Amazon Music particularly. The reality unfortunately is that Roon is a minnow in the ocean among whales, so the question is whether any of the "big boys" would be interested. I suspect some at Roon may be experiencing sleepless nights thinking about how things are going to play out.

Stay safe audiophiles. Hope you're enjoying some wonderful music!

BTW: I saw this SNL skit the other night. Couldn't help but think this kind of psychology might be at play when some "high-end audiophiles" fail a blind test and told the truth. ;-)