Volume knob is of a good size and feels smooth when turned. There is a 3-level gain switch - -10dB/0dB/+10dB. Finally, there is the push button on the left which is a momentary contact switch that toggles power. There is a small LED in the center (which you can see lit up in the top image), white when ON, and when you first power-up the amp, for a few seconds it turns red to ensure a graceful startup.
In the event of a DC offset, overcurrent, short detected, or overheating, the amplifier will go into protection mode and that LED will trigger red to let you know.
|Notice that I've turned the SMSL amplifier to Bluetooth input to mute the output to the speakers when headphones are in use. I could also just as easily use the volume control on the SMSL. Tiny SMSL M100 Mk II DAC is all I need. The headphone amp barely gets warm with devices stacked like this.|
This gives me full flexibility in controlling volume through both the SMSL Class D amp as well as the Drop headphone amp. I also like that the Drop amp has an "Auto Off / Bypass" switch that will allow the unit to turn off automatically after 2 hours with no audio input; nice since I always appreciate anything that can save energy. Even when on and playing music louder than usual (+10dB gain, 3:00 position on volume knob) to my least sensitive headphones (Dekoni Blue / Fostex T50RP Mk 3), the amp is only using 6W as measured through the Kill-A-Watt meter.
The pass-through RCA out always remains active.
The metal casing is angular and feels robust, weighs a bit over 3lbs, and measures at a moderately sized 8.3"W x 9.1"D x 2.2"H to round out the overall physical specs.
When it comes to headphone impedance supported, the manual states that it will "graciously" power headphones between 16-600Ω and the design accepts a "minimum" 12Ω. However, it still states that for even lower impedance devices, it can "tolerate" 8-12Ω "if the device is driven no higher than 4V peak across the load". Important to keep in mind if you use very low impedance IEMs like these Campfire Andromeda headphones rated at 12.8Ω but some tests showing <9Ω average impedance. Yikes. Make sure not to blow your sensitive headphones with too much power.
The single-ended output has about 1/4 the potential power of the balanced out with some current limits when paired with low impedance 'phones as you can see in the specs:
From what I have seen in threads like this, internally the 789 is not "fully" balanced. Just ensuring something is fully balanced is no guarantee that it's "good". Personally, what's more important is whether the performance lives up to expectations when we listen and put it up on the test bench.
Okay then, let's run some tests and see how this baby performs, shall we?
To start, let's have a look at the output impedance:
Nice. Very low, stable impedance across audible frequencies measuring at <0.4Ω on average which is at the limits of my testing set-up. This will have no issues with all kinds of headphone loads (it'll be highly "load-invariant").
Regarding load invariance, here's a peek at the frequency response (10Hz to 40kHz) with my 3 "standard" load values (20/75/560Ω) at 0.5Vrms and also compared to the Polk Ultrafit 2000 headphone which I measured with some other amplifiers a few weeks back. For completeness, here is the Polk Ultrafit 2000 impedance curve again:
I had to split the curves apart so you can see them individually, otherwise they would overlay on each other.
Notice the 20Ω load does dip a very tiny amount more than the others towards 40kHz, which is insignificant. The Polk Ultrafit 2000 with its varying impedance from 15-55Ω has no effect on the frequency response as expected when paired with such a low output impedance amplifier.
I included the impulse response from the amplified signal fed to the 20Ω load. The headphone output is basically perfectly reproducing the linear phase signal from the RME DAC.
While it's nice and befitting of a "higher end" headphone amplifier to have balanced output, I have little reason to believe for my uses (short cable, sensitive enough headphones) that this would make any difference. Maybe there's some benefit for a parameter like crosstalk, but I don't feel that crosstalk is in any way lacking and if anything, sometimes I want to add some crossfeed into the headphone playback!
As such, for the tests here, let's focus on the single-ended 1/4" phono output. If it performs well, I think we can be quite confident that the balanced output will be just as good or better. If there are any concerns, I suppose I can re-examine this in the future when/if I get some 4-pin XLR balanced cables.
The picture above is a snapshot of what it looked like when measuring the amplifier with the Polk headphone connected as load. This is an updated image as I reworked the wiring to tap measurements right from the output instead of a few inches out. A few of the graphs have been updated, particularly the stereo crosstalk measurement below.
To the right we see some dummy loads. The measurement tap is sent to the Linear Audio AutoRanger. Output of the AutoRanger (2V nominal setting) is sent to my RME ADI-2 Pro FS ADC. There's a little Intel NUC (not shown) running Room EQ Wizard which has evolved into a fantastic tool not just for room and speaker measurements, but also with the flexibility to apply to headphones and devices in general - fantastic work, John Mulcahy!
Amplification is about clean power, let's have a look at the THD vs. Power graphs then using my "standard" loads - 20Ω for "Low Impedance", 75Ω for "Medium Impedance", and 560Ω for "High Impedance". For the test set-up, I'll use the RME ADI-2 Pro FS as DAC connected by XLR to the Drop + THX AAA 789, then measuring the 1/4" phono output connected to dummy loads or an actual headphone as the case may be. To push the output into clipping, I set the gain to the high +10dB setting. Both channels driven in these tests.
The top bright red line is the THD+N, and the next lower black line is the THD. With a 20Ω load, we see that the left channel goes >1% THD+N at about 2.7W and the right channel doesn't clip but will go into overload protection if I increase the volume any further. The specs say 3W into 16Ω so we're definitely in line with that. Notice that distortion increases steeply above 2.2W.
As in my other amplifier reviews, I like to define "high fidelity" as better than 0.1% (-60dB) THD+N which is around 2.5W with this amp.
0.5V into 20Ω (12.5mW) (L channel shown, both channels driven):
2.5V into 560Ω (11mW):
Beautiful. Realize though that those are synthetic resistive loads with the 20Ω needing more current and 560Ω fed higher voltage.
How about we try something a little more challenging, something more real-world using an actual reactive load with fluctuating impedance? Let's look at distortion into the Polk Ultrafit 2000 headphone which as stated above has an impedance curve that swings from 15-55Ω. I'll use a 0.5Vrms signal which is already much louder than I would normally listen to music with when using these headphones.
Interesting isn't it? Notice that the distortion characteristics with actual headphones are clearly a bit higher than just a flat resistor. It's still far from problematic with THD around -90dB (0.003%) or better, but more than the -100dB (0.001%) numbers we're seeing in the 2 graphs above. While "ideal" resistive loads provide useful standard values to measure, realize that in real life, how an amp interacts with the transducer is important. As you can see, it's not a problem with a high resolution amp like this, but worth thinking about with lesser amplifiers.
Just like with typical speaker amplifier tests, we can have a look at things like Intermodulation Distortion. Let's have a look at some FFT's now including the SMPTE, CCIF/ITU-R, and Linkwitz at ~1Vrms into the moderate 75Ω load (~13.3mW). For good measure, let's also throw in the TIM signal I typically use with speaker amps:
Look at how clean and "textbook" these distortion graphs look! No issues at all with that "transient intermodulation" (TIM) signal of 96kHz bandwidth (192kHz sample rate) with tiny insignificant sidebands poking through down at -128dB.
Finally, let's use a somewhat more complex tone, here's the Triple-Tone TD+N at 0.5V into 20Ω (12.5mW) which should not be out of reach of most amplifiers and we could use this as a "resolution" benchmark with other headphone amplifiers I might test out in the days ahead. As usual, both channels simultaneously driven:
|NOTE: Updated with rewired load.|
Hmmm, I'm having fun listening to this amplifier. To be honest, there's really not much more I need to say!
The Drop + THX AAA 789 basically amplifies the audio signal cleanly, with ease. I have had no problems with driving any of my headphones to more-than-loud-enough listening levels including something like my Fostex T50RP Mk 3 (Dekoni Blue), a couple of headphones I borrowed from linnrd recently - the Beyerdynamic DT 990 (600Ω) and Sennheiser HD650 - among others.
No sense of distortion or strain is heard even at loud volumes whether I'm enjoying some NiN (I was listening to the old Quake Soundtrack from back in 1996) or some intimate erhu recording (Fu Na & Huang Jiang Qin Face To Face 2: Guzheng vs. Erhu). Plenty of detail, "microdynamic" nuances, tight bass when headphones allow, excellent sense of "speed" with transients; again it's all about headphone abilities.
I generally don't hear much in the way of "soundstage" with headphones. It's mainly various degrees of inside-the-head right-left stereo separation. The exception would be binaural recordings where I can actually imagine an illusion of space. Obviously the headphones are most important but I think because of the resolution and low noise, this amplifier does a wonderful job with these recordings. A good example would be Amber Rubarth's Scribbled Folk Symphonies (2016, DR12) - have a listen to her rendition of "Losing My Religion" on some good open or semi-open headphones.
Daniel Lanois' recent release Heavy Sun has been on rotation in my playlist lately. The uplifting, soulful, gospel tinged collaboration sounds great with catchy melodies paired with my favourite headphones. Quite relaxing on a pandemic evening.
For some (generally) fun folk/country, check out Corb Lund's Cabin Fever (2012) and the more recent Agricultural Tragic (2020). (Lund's from Taber. I grew up in Alberta, so a shout-out to all the Albertans out there.)
I subscribe to the philosophy that amps aren't supposed to have their own sound. This amp does not call attention to itself and this makes it a fine hi-fi reference amplifier. When there's really nothing more to say about the device itself, I think the subjective reviewer's role is simply to offer some potentially interesting music recommendations.
|Drop + THX AAA 789 modeled with vintage 1991 "Made In W. Germany" Beyerdynamic DT 990.|
As you can see, I've put this amplifier through a number of tests to review performance. Headphone amps can be tedious to characterize simply because of the unpredictably wide range of impedances it might encounter with your headphones, hence the necessity to try different loads. All-in-all, I love this amplifier which I've now used for a number of months on my main computer Workstation. It has been a wonderful companion over the winter months listening to tunes while doing work or relaxing in the evenings after the kids go to bed.
At around US$300, this Drop + THX AAA 789 headphone amplifier is not "cheap" - but certainly affordable for anyone who has an interest in good music reproduction. The level of power and resolution available is phenomenal and IMO this is a fantastic deal if you already have a decent DAC and just need to beef up the headphone amp section with a convenient pass-through to a desktop amp if you also have speakers.
The "AAA" (see the THX series here) in the name stands for Achromatic (ie. "colorless") Audio Amplifier which refers to the feed-forward distortion reduction circuitry. Many of us probably first heard of AAA a number of years back when Benchmark released their AHB2 amplifier. Indeed, "colorless" is a good description. It doesn't color the frequency response, is load invariant, distortion is very low and will supply more than enough power for the vast majority of headphones you might want to throw at it.
At this level of objective performance, it would be ridiculous to call this a "beginner" or "starter" amplifier. Realistic, rational audiophiles do not gauge equipment primarily by what they see (eg. aeronautical milled aluminum fascia, anyone?) or how many dollars are attached to the MSRP. An audiophile who appreciates high fidelity gear could easily see this as an "end game" headphone amp, upgrading only for more features (eg. combined DAC + headphone amp like the RME ADI-2 Pro FS R BE), higher power when needed, or if you purposely want to color the sound then go for a tube amp perhaps. I suspect many "high end" manufacturers would shudder at such a thought of historically obsessive audiophile consumers getting off the incessant upgrade merry go round.
In the last few years, there have been a number of THX AAA headphone amplifiers released - SMSL SH-9 (AAA 888, <US$300), SMSL SP200 (AAA 888, US$250), SMSL SP400 (AAA 888, US$600), Monolith 124459 DAC + AAA 788 (~US$500), and the Monolith AAA 887 (very similar to the 789 here in many ways, US$350) are a few to consider if you want balanced headphone output. As I mentioned above, I actually have not even used that 4-pin XLR.
If RCA inputs and single-ended headphone out is all you need, the recent Drop + THX AAA ONE (~US$150-200) might be a good option.
Yeah. Don't waste time. Get a good hi-fi amp like this, then take your time to find good headphones, and make sure to enjoy the music along the way.
To end, I noticed that there has been a bit of a press drive lately for the THX Onyx DAC dongle (THX these days is owned by Razer, makers of computer gaming stuff mainly). You're seeing this product splashed across sponsored sites all over the place in the last couple of weeks.
Looks like the Onyx's output impedance is nice and low at 0.2Ω, maximum 2Vrms output is good for a mobile device, and 180mW into 22Ω @ 1%THD is nice. I trust it will perform well when it comes to low distortion and if it provides decent power efficiency (battery life is essential for mobile audio), then I suspect many audiophiles will be appreciative now that LG is exiting the smartphone market. The Onyx's headphone amplification ability is what is most interesting, and it looks good for such a device.
Hmmm... But why all the hype with MQA attached to this product?
Those who have been on this blog for awhile have already foreseen the inevitability of MQA creeping into products like these because ESS embedded the poor quality filters into their DAC chips starting in 2019. Notably, the ES9068 and now the ES9281Pro that the Onyx uses to provide a turnkey MQA data lock and filtering mechanism which is what they mean by "rendering". Relatively inexpensive DAC dongles have been doing this for a little while now (like the HELM Bolt or Zorloo Ztella for US$100 with the ES9281 Pro inside also but without THX AAA amplification).
There's nothing new here not already covered by my article on MQA back in 2018. In a world where consumers are enticed by new features with promises of improvement (which in this case is clearly false), as companies create new generations of DACs, they're prone to use something like MQA to justify the value of their supposedly "upgraded" products. It's all nonsense and I do hope that as audiophiles, our "job" is to know what is of value and what is hype. Who else better than audiophiles to clarify facts for the mainstream consumer, right?
I want to call out Audioholics for their unsatisfying "review" of the THX Onyx by Wayde Robson recently.
Come on... For a site that does a great job dispelling snake oil like this fantastic video on the poor AudioQuest Thunderbird Zero speaker cables (I'm still blown away by the brittle banana plugs!), what's the deal with all that MQA stuff in the Onyx review!? Honestly, for all the misinformed nonsense that Hans Beekhuyzen has spewed for years, he deserves to be ignored - the dude will promote basically anything including audiophile ethernet switches for goodness sakes!
I would like to see Beekhuyzen justify the statement that MQA is "only lossy above 45kHz" - it's not, that's a lie, and why would anyone, least of all Audioholics repeat something like this? At best it might be CD bit-depth and 24kHz "lossless" (even this is debatable as per the video below!).
Furthermore, this "THX Onyx Audio Modes" diagram parroted by the reviewer is pure fantasy (not even science fiction):
|Notice the misleading color-coding as well which plays into the listener's psychology by what they're seeing, not actually hearing!|
Seriously? Pseudo-hi-res, partially lossy MQA is "best quality" compared to even true >48kHz (24-bit) PCM or DSD? Talk about pseudo-science audiophile myths which I've always expected Audioholics to take leadership and fight against, not copy and present in some front-page review.
I noticed that there are no objective measurements in this review and that's OK so long as the write-up is good and factually researched. I would give the benefit of a doubt to THX for making a decent headphone amp/DAC so a positive review is just fine. But please, let's leave out the MQA crap. Sure, mention MQA as a feature if you have to (was this review sponsored by the way?) but let's stay real and ignore Beekhuyzen for example and show a little critical thinking when it comes to that diagram above. Shame on you THX-Razer for creating that diagram in the first place.
One more thing. I hope Wayde Robson has had a closer look at his example of a good sounding MQA song - Andra Day's "God Bless the Child" (from The United States vs. Billie Holiday). Yes, it is a lovely tune, but let's think about the nature of the data and how it relates to MQA. Here's the actual 24/96 which the MQA version would have been derived from:
It's quite loud, audibly dynamically compressed for an intimate vocal jazz recording, and measures a paltry DR7 for dynamic range. Dithered down to 16-bits would be fine IMO. The noise floor isn't particularly good either. There's basically no content above 26-27kHz so you'll lose nothing by going with a 48kHz downsample (24kHz bandwidth):
So with a recording like this, what in the world do you think MQA is doing that might in any way be good? When Robson says of the MQA playback:
"It seemed to open-up, bringing air to the little details inside the song, like the little near-silences when Day draws a breath between lines. The dynamics even brought space in-between each instrument. While it didn't exactly a widen the soundstage, the dynamic range gave the illusion of spaciousness to the performance. In that moment of discovering how MQA really sounds, Andra Day's voice convinced me I wasn’t listening to microchips."
What in that statement can really be attributable to MQA? Are the little "breath between lines" not heard in a basic 16/44.1 or 16/48 downsample? What "dynamics" is he talking about in the context of a compressed recording that in no way needed high-resolution 24-bit reproduction (which MQA is incapable of anyways) in the first place?
For a track like this, 24/96 already is a waste of space, and now adding nonsensical MQA encoding / decoding / rendering is also a waste of time, energy and money (all for an advertising feature checkbox and colored LEDs?). In fact, I would argue that MQA with its bizarre filter settings is technically worse than using orthodox high quality filtered playback (like Chord, HQPlayer, or roll your own).
As usual, people can say anything they want and create "attribution theories" however they please (as if the "spaciousness" of this song actually might have anything to do with "how MQA really sounds"). Uninformed, purely speculative subjectivity is a problem that audiophiles should be acutely aware of by now and apply reasonable critical thinking against - whether it's MQA or fantastic "mythical creatures" AudioQuest cables!
[Addendum: I noticed for the Audioholics review how the THX Onyx "First Impressions" rating jumped from "Pretty Cool!'" to "Gotta Have It!" over the last week. Interesting how arbitrary subjective ratings can be, eh?]
Please music lovers and audiophiles. MQA was poorly conceived from the start. And within its DNA, it did not truly, honestly, desire to benefit music lovers IMO. Even if there were aspirations to be beneficial at one point, the days of needing lossy compression for hi-fi audio streams are well over. After 6 years of hype, living in the nether regions of respectability as a zombie "format" by anyone who seriously looks at what this thing does, and with Tidal being the sole source of MQA-encoded content of any significance, not only would it be merciful to let MQA die, it would also be the right thing to do.
The easiest thing for consumers to do is to simply not spend money on anything with the MQA feature on it knowing that perversely, dollars spent on this scheme will drive the continued meaningless corruption of musical fidelity. Given the competitiveness for hardware sales and streaming subscriptions, change in consumer habits of even a few percent probably is enough to make a significant difference to the bottom line for companies. They will take notice when there are dollars on the line.
For those who invested money into MQA (including ESS - this is not how you serve audiophiles), it's time to look elsewhere for ways to actually benefit the consumer.
Psssst... Here's an amazing video on MQA...