What’s the difference? Is it placebo?
Ever since I lost the headphone jack from my beloved iPhone SE (which I kept going through 4 years and 3 battery replacements) and 'upgraded' to an iPhone 12 Mini, I felt like something was missing from the new headphone dongle.
Now that I’ve tried some (well, one) of the innovations that have come to the portable audio market since the standardized removal of headphone jacks, I’m actually kind of thankful for Apple opening up this new market. We’ve seen some incredible new devices created to fill the vacuum, and they’re better than a headphone jack ever was and small enough to stay portable and unobtrusive. It’s kind of amazing.
The biggest players right now are probably the FiiO BTR5, the EarStudio ES100, the iFi Hip-DAC, and the quizzically named Qudelix-5K. As commuting is becoming a thing again (at least in my state) I wanted to finally improve my portable audio since losing my beloved headphone jack; so I went for the last one.
Why the Qudelix-5K over the others? Well, it has a nice pedigree, great features, it’s among the smallest (slightly bigger than a 9V battery) and it looked good enough.
The biggest question I had was: is this really better than the Apple headphone dongle?
As an engineer, I had to believe it was. My imagination is stretched when thinking about the constraints and challenges that must have been involved in cramming a whole digital converter (DAC; and ADC for the mic remember) and headphone amplifier in the tiny 4mm by 15mm by 9mm lightning jack. In the headphone amplifiers I build each individual capacitor and chip tends to be larger than that. But the future is an amazing place; I have to suspend my disbelief.
And the Apple headphone dongle does work. It sounds pretty good! I’ve been listening to it for over a year with various headphones, IEMs, and earbuds (my personal favorite) and my overall evaluation is still that a lot is missing.
Let’s find out.
Phone: iPhone 12 Mini
Headphones: Sennheiser HD598SE (NewFantasia silver balanced cable), Sennheiser HD-25-II, Moondrop Shiro-Yuki earbuds, Yuin PK3 earbuds, KZ ZSN IEMs.
Qobuz FLAC only (44.1 and higher).
- Clairo - Zinnias (Indie; 96/24)
- Noah Kahan - Maine (Folk; 44/16)
- Free Throw - Cloud Sick (Emo Rock; 44/16)
- Jack Ingram et al - The Wind's Just Gonna Blow (Country; 44/16)
- Brandi Disterheft - Coup de Foudre (Jazz; 96/24)
- Qudelix-5K DAC+Amp (Balanced/Unbalanced jacks; Bluetooth or USB-C)
- Apple Lightning to 3.5mm Headphone Adapter
The Listening Test
For a fair test I’ll stick to single-ended until the end. Obviously the Apple dongle isn’t going to do balanced, which is basically a second amp creating an opposite inverse ground reference for double the power. I’m not a huge believer in balanced cabling; why not just build an amplifier that’s twice as powerful? But it is twice as powerful, so that’s neat.
I’m using the ddHiFi 3.5mm single-ended to 2.5mm balanced to convert the headphones with a balanced cable to keep the cabling the same across tests. It just ties the separate grounds together, so the end result is the same as using an unbalanced cable.
Note on my cheapskate headphone collection: I buy affordable transducers. Long ago I discovered the source and drive makes a big enough difference to the sound that I can buy some underdogs and drive them really well and get what I want. These are reasonably good headphones. I hope it matches what is in the reach of an average audio enthusiast and not someone with thousands of dollars to spend, and encourages you that you can get great sound without a second mortgage: you just need to balance your investments across the signal chain. I truly believe that.
My experience with these is already that they’re hard to drive. Low current kills them; they get muddy and confused. Once I built an amp that could output 100mW, they really started to shine.
Starting out with Clario, things are sounding pretty darn good for such a small device. Considering I’ve built headphone amps that couldn’t make these cans sound good, the fact that this little thing is doing okay is already impressive. With some critical listening, the first thing I notice is a soft somewhat boomy bass. Second, a bit of a diffuse vocal sound. We’ve got her voice coming in from both ears without much focus here; it’s kind of scattered.
Moving through the playlist, on Noah Kalan’s baritone vocals I’m noticing a little grit. And when the acoustic guitar power strums come in, things are getting muddy. I’m also hearing a little tin-canniness in the vocals, especially in the sibilant consonants. It’s not a ton of real sibilance, but it’s harshness for sure. The louder Free Throw track fares pretty well and is enjoyable overall, but suffers from a lot of the same harshness I can’t shake. And when the chorus comes in it’s again a voice diffused across the soundstage, kind of smushed together. Separation between the voice and backing instruments is decent; but drums are definitely getting lost.
On the incredibly intimately recorded “The Marfa Tapes” the Apple dongle faithfully reproduces the notes and plucks, but I’m not feeling the room very much. It kinda gives off a sense of technical correctness rather than realism, and the noticeable grit is in full force on the mens’ voices especially. Finally, the jazz track from Brandi Disterheft: it kicks off with her skilled bassline which has a distinct separation between the higher bass and the low bass; it’s like it’s two separate frequencies with nothing in-between. The cymbals are veiled as well without much sparkle to them like I know is in this recording. With the female alto vocals entering, once again they feel a little diffused and hard to pinpoint. Piano sounds very good in comparison, with great resonance and tone.
One thing I’m impressed by is the volume. I have experienced the Apple dongle not quite being enough in noisy environments (my ears are probably shot…), but sitting at my quiet desk it’s got plenty of juice at around 60% volume.
Overall, it is pretty impressive that this small device can put out this kind of sound. That’s square one. It’s certainly not bad, though it has flaws in the all-important details. Could it be better?
First I matched volume as well as I could based on the last track before beginning; the best way to trick yourself into thinking there’s an improvement in sound quality is to up the volume. No thanks!
Back to the start, on Clario’s ‘Zinnias,’ the first thing I notice is the bass. It’s very different. It’s full, with all of the range from the lowest lows to the subtle mid-bass to the higher bass, but it’s also under control and kept tight and realistic as opposed to just a generic boom. The second thing that’s immediately different is the vocals: they’re focused as a laser. Where the dongle was sort of like a shotgun, this is a sniper. You can hear how she recorded her voice layered. Finally, the soundstage; it’s… it’s there. I just realized there wasn’t really a soundstage with the dongle. With the Qudelix, I can hear cymbals center left, and vocals centered and spread out just a tad, and small cymbal brushes coming from the distance; from a point, not just a general sound.
Aside: what the heck is a soundstage? To understand what it really means, think back to the unfocused vocals: when things are difficult to place spatially in a stereo field, it creates a sort of diffused sound across the board. It can sound “good,” but there’s not clear consistency in where sounds are coming from. When there’s enough detail in the recording and the reproduction, you can hear instruments, vocals, drums, cymbals in different places across the virtual space in your ears. Some might actually be across the whole field, whereas some might be points; in the center, 30 degrees to the left, or off from the general right side. Different sounds with different spatial character gives you an incredible sense of realism. That’s soundstage. It only happens when the recording and playback is extremely accurate.
Noah Kahan’s “Maine” starts out subdued with a quiet strum and vocal line. His voice is revealed as a touch rough—not artificially gritty—and doesn’t have any harshness or sibilance to it. There are seagull cries in the background, which I didn’t really notice before. All the layering in this song comes through very well, with each layer added still standing on its own and not muddy at all. The climax brings a low bass that’s smooth as butter, and cymbal hits that reverberate into the background subtly but completely separately from the other layers, bouncing from left to right, perfectly locatable.
For the energetic new Free Throw track, all the harshness is completely gone. I’m not sure if I could listen to this album on the dongle without getting stabbed in the ears feelings, but with the more even and smooth approach of the Qudelix, it’s not a problem at all. The same qualities are repeating: clear instrument separation, fantastic bass control and clarity, smooth high end.
Now a track that made me say “Wow” the first time I heard it through a good DAC: live/ambient recordings tend to reveal equipment’s true colors and this is no exception. Jack Ingram’s “The Marfa Tapes” is such an emotionally powerful but subtle album, recorded with Miranda Lambert and Jon Randall in sweet harmonies. This DAC is treating me: the thing that’s immediately noticeable is the ambiance. You can hear the room, not just the music. The little breaths, the reverb of the guitars, where each singer is sitting. It’s all there. If I have anything to complain about, it’s that the vocals are still just a touch rough. Not gritty or distracting, but not perfectly smooth either. Could just as easily be the headphones or the recording, not sure; but it’s not the same as the grit that the dongle presented pretty consistently.
Finally, the jazz track pops in with a much more realistic fast bass walk, losing all the boom and frequency peaks and just showing my ears what an upright bass looks like. The alto sax comes in a little recessed, and the vocals dead center. The vocals still have that tad of roughness to them; not a 60 grit sandpaper, but maybe a 320 grit. ASA 400 film. Fine grained, but present. The piano reverberates realistically and has a full substance to it and a recessed presence that says “I’m behind the mic.” And the drums! The best part. Especially toward the end, when vocals, sax, piano, bass, and drums are all riffing at the same time—there’s no loss in any part. The drum kit is panned across the stage, hi hat left, snare center-right, crash on the right. I just wasn’t hearing any of this before—it was just one sound.
Overall: this little thing is absolutely doing some impressive work for its size. It matches my assumptions about what would be possible with a tiny IC versus a dedicated (albeit still very small) device.
I said I wasn’t going to compare the Apple jack to the balanced mode of the Qudelix, but I can’t help trying and seeing what the differences are. And there are surprising differences.
Side note: what is a "Balanced" amp? It's basically two amplifiers instead of one, but the 2nd amp plays an inverted copy of the audio signal. The 2nd amp is hooked to the "ground" and goes against the first, effectively creating a virtual ground and twice the voltage potential of a single amp on its own. It also reduces crosstalk, since the two channels don't share a ground wire.
Surprisingly, the volume isn’t that substantially different. What feels different immediately is the mid-bass, and the dynamics. It feels like the headphones are having an easier time of quickly going from quiet lows to louds. Instrument entrances just hit harder. Bass somehow feels even more controlled. There’s just a lot more there there.
This is not altogether surprising: I know these headphones are fairly power hungry. What was surprising was the almost complete disappearance of the rough vocals. Everything feels a little bit smoother. That might owe to the dual ES9218P DAC chips that are only used in tandem when in balanced mode, making the thing a strange inverse implementation of dual-DACs, combining the analog signal from each chip at the voice coil itself. Another potential explanation is that there’s extra distortion from the two amplifiers combining, smoothing over some of the roughness. But I’m still hearing the same high level of detail, instrument separation (perhaps even more; balanced amps do have less crosstalk due to independent grounds), and imaging, so I don’t think it’s a layer of distortion.
That matches Qudelix’s own stats: balanced has a full half the THD+N (0.002% vs. single-ended’s 0.004%; likely not audible, but not worse), and significantly better and maybe even audible channel separation improvement (117dB vs. 79dB for single-ended).
So, wildly, you get more than just power and channel separation from this DAC/Amp in balanced mode. You get a second DAC.
Senny HD-25-II (Unbalanced): I mean, the character of these is so very different from the open 598SE, but I still have a place in my heart for them forever. They’re dark and warm but still show you tons of detail when it matters. The iPhone dongle does a great job with these, but some of the same things come up: the bass isn’t controlled very well and ends up feeling like it’s through a thick blanket; and it seems to me like there are three locations for the soundstage; left, center, and right. I’m also noticing a weird treble effect where it seems like there’s an emphasized layer over vocals, but 5kHz over their range. Shouldn’t be there. Overall, still impressive for the size, but nothing special and a couple key problem areas. The Qudelix-5K on the other hand shows all the detail and has such strong control over the low end. Same comments as before; but the HD-25 II’s are less revealing of any grit, so even unbalanced I’m hearing everything nice and smooth. Excellent. I’ve got a balanced cable on the way for these and I look forward to how it changes the sound.
Moondrop Shiro-Yuki Earbuds: there’s a pretty remarkable set of earbuds on the market these days that are basically the portable equivalent of open-back headphones, in contrast to the sealed IEM. These Shiro-Yuki earbuds are one of the ones I’ve enjoyed greatly. They don’t resolve quite as well as the Sennheiser HD598SE’s so it’s a bit more difficult to discern some of the fine differences in the different amplifiers, but I’ll cut this short and say that all of the same characteristics were clear. If anything, due to the fairly open feel of these earbuds, the Qudelix was a much better pairing, and the predominant feel of the Apple dongle was “narrow.” The differences were much more difficult to discern with these buds though, so in a pinch I’d probably be happy with the Apple dongle. But as I say that, I was assaulted with the harsh vocals of Free Throw through it; through the Qudelix, the screaming chorus didn’t break up at all and wasn’t fatiguing, and the soundstage was 180 degrees broad. Nice.
Yuin PK3 Earbuds: Comparing these to the previous earbuds is fun all by itself. Significantly less harsh than the Shiro-Yukis to begin with, these handle volume and power much better and have a sound much more similar to the full-sized headphones. They’re really nice. They sounded particularly good on the Jack Ingram track, revealing the atmosphere and room feel just as well as any. They also sounded pretty great through the Apple dongle to be totally honest. The vocals were more one-dimensional for sure and did have the same grit I heard before, and the layers had trouble separating, but it still had a rich depth to it that wasn’t bad. Qudelix still wins here by a nose.
KZ ZSN: My only IEMs (they’re just not a comfortable form factor for my ears). I haven’t listened to them in a long time, but even through the dongle, I’m pretty impressed. Character on the Jack Ingram through the dongle is a little 2D; the background ambiance is pretty flat, and each voice feels like it’s coming from across the stage. On the Free Throw track I’m getting the same loose and boomy bass character, not really a specific identifiable instrument. Through the Qudelix, things come together better; not a whole lot, but the vocals are more focused, the bass line is tighter but still a little boomy (attributing to these buds). Overall these are a bit harsh in the upper mids in general, so it’s hard to enjoy them completely. They’re cheap, this isn’t unexpected. Going back to the quiet music, there does seem to be more realism through the Qudelix, almost like there’s more substance between each discrete frequency, especially audible if you ‘hear between the lines’ in vocals, tiny resonances and a general sense of fullness is what it comes down to.
Overall, I certainly enjoyed the Qudelix more on every distinct headphone. In general, it had better control of bass, more fine detail, better soundstage and instrument separation, and fewer fatiguing artifacts and harsh characteristics. The end effect is that the Qudelix-5K could reach, in certain circumstances, a very realistic feel to the music. I very rarely if ever found that feeling with the Apple dongle.
One takeaway I won't forget: I need me some better IEMs.
The form factor of the Qudelix-5K is a conundrum: it’s so small and light that you can’t help but be amazed. It feels like a futuristic device, to get this type of quality and balanced drive in a matchbox. For me, it obviates the need for a separate DAP, which I’ve always hesitated to get because of the investment I’ve made in high-quality online services like Qobuz and Roon—the modern smartphone still seems like the perfect front-end for portable listening.
The small form factor, however, brings some usability quirks. There are two buttons, one on each side, toward the jack end of the tiny device. The buttons are actually two buttons each, and one lights up red and the other blue. There are no labels; only a small vertical line at the top of one button allowing you to (I suppose) differentiate them without looking. It’s still confusing, and in the first week I’ve made mistakes a couple dozen times.
I made some labels with my label printer. Feels crude, but it works. Maybe I’ll learn the muscle memory over time.
The other half of the Qudelix story is the app. It’s full-featured and has everything from details about the playback, to different options for how the buttons work, to a full parametric equalizer. It’s very useful.
The third half of the Qudelix story is Bluetooth. I hear it’s good and stable. I don’t really plan to use it; the iPhone only supports AAC for the Bluetooth signal, which is a shame, and my little Lightning to USB-C OTG wire seems to be working great. I’ll stick with it.
This is meant to be less of a review of the Qudelix-5K and more a comparison of a fully capable DAC and Amp to the modern miracle of the DAC-Amp-in-a-wire. After some critical listening here, I’m more confident than ever that miniaturizing these devices brings with it many compromises and tradeoffs. That’s not surprising in the least, but it is the truth. It’s more amazing still that the quality we used to get from discrete desktop DACs and amps can be nearly matched in a portable one the size of a 9-volt battery.
It also wouldn't be fair to end without mentioning the Apple 3.5mm dongle's incredible achievement. In the size of a pencil eraser, they managed to pack a damn decent little headphone driver. Most people will never hear the subtle difference—and I'm not saying that those who can have special ears, just that most people are sane and care a reasonable amount about the quality of their music, and that's perfectly fine. Apple has done a great job of striking the perfect balance for most of their users, which is exactly what they're good at. And because they did that, those of us who do care an unreasonable amount about the quality of sound are lucky enough to get some new products that have pushed the boundaries of portable audio quality. Can't complain one bit.
It's also worth noting that the Apple headphone dongle is very good. It produces a very pleasing, balanced sound with quite a bit of low end extension, and can drive headphones that are pretty difficult to drive from your average headphone jack. The minor differences you might pick out in critical listening are likely not audible if you're just enjoying your jams.
Finally, while I like to believe I've desensitized myself to emotional attachment to specific devices and can have a relatively impartial evaluation of what I hear, I’ll get my partner to set me up a randomized blind test sometime soon. I would love to see if I can tell these things apart like I say I can. Until then, enjoy the music!