A Synthetic Review
- High power.
- Great construction.
- Very elegant design.
- High quality, powerful sound with great properties.
- Simple functionality.
- Heavy weight and large size for portability.
- The relatively high output impedance may not be to the liking of all enthusiasts, as the impact on low impedance hybrid headphones is unpredictable.
- It is not a neutral DAC/AMP because its Zout modifies the FR of some headphones, especially hybrids.
- The high gain mode is realised from a volume step. At this point, the power changes are high.
- The potentiometer is not too precise, too much voltage for less than one turn. It is also too soft, I would have liked it to have a harder travel.
- The mirror finish will reveal all fingerprints.
- The extreme simplicity of the functions makes it a bit short in handling possibilities.
Link to the Store
Thanks to the European Tour organised by Wendy Li, I had the great opportunity to test the TRI TK-2 DAC-AMP. As specified on the side of the device itself, the TK-2 is the result of a collaboration between TRI and KAEI. TRI is like the high-end brand of KBEAR. KAEI, up to this point, was unknown to me. But it seems to be a classic brand, known for its amplifiers. The HiFi USB DAC Amplifier is the first product of its kind that TRI has brought to the market. It’s certainly a very powerful set-up, also thanks to the hand of KAEI. The presence of the device is almost colossal. I didn’t expect it to be so big and relatively heavy. That gives you an idea of its power, as well as its impeccable appearance.
This review is not going to be a classic DAC-AMP review like the ones I have written so far. Admired reviewers have participated in this tour. I believe that a humble servant can say little about the sound of this device that they have not already published. In fact, I think there will be more reviewers who will write their opinion about this barbaric device after me. So, as is also usual, I will talk about this device in terms of its relation to some IEMS in my collection. It will be like a comparative account of measurements and frequency responses. I hope you like it. Let’s get on with it.
- Outputs: SE 3.5mm, Balanced 2.5mm and 4.4mm.
- Compatibility with 8Ω to 600Ω headphones.
- Frequency format decoding: 44.1K, 48K, 88.2K, 96K, 176.4K, 192K, 352.8K 384K/DSD 64-256 (Native & DOP).
- DAC: Dual 9038Q2M
- Output Amplification: Class A and B.
- THD+N: 0.005% (1k, dsd128).
- Frequency response: 10-40kHz (<1dB).
- Maximum output power: 1250mW (32Ω).
- Fixed LO output: 2V
- Battery life: about 9 hours (5000mah).
- Dimensions: 130x68x24mm.
- Approximate weight: 280g
- Charging interface: USB TypeC. QC 3.0 fast charge.
- Decoding interface: USB TypeC (OTG cable required to connect to mobile phone).
The TK-2 comes in an eminently elongated box. On the main side, at the top, is the TRI logo, underneath it is the model name, and even smaller and next to it, the product description. All this in white letters. At the bottom of this side, there is an actual photo of the front edge of the device. On the back side are the specifications, in several languages, including English. After lifting the lid, you can see the device encased in a dense foam mould. A lighter layer of foam is glued under the lid, to protect the device even more. Underneath the device are the rest of the accessories. The complete contents are as follows:
- TRI TK-2.
- A cleaning cloth.
- User guide.
- One long USB to TypeC cable.
- One short TypeC to TypeC cable.
- Two rubber bands for fastening.
- Adapter for Apple devices.
There is little to complain about in this section, it has everything you would expect. Perhaps it could come with a carrying bag, like the one that comes with the Earmen TR-AMP, for example.
Construction and Design
The revised model has a metallic/mirror finish that is a magnet for our little fingers. If only the CSI guys could make all surfaces so fingerprint-revealing. The mirror finish is amazing, but it’s impossible to keep clean. The top face is dark, it looks like a plate of some kind of hardened glass. The rest of the sides are metallic. The device has a bezel on the top face. The corners are rounded. On the top face, in the upper left corner, is the brand logo, model name and description, in the same form as on the product box. On the lower edge, next to the main front, there is a series of green LEDS, which indicate the sample rate of the decoded file. The combination of some of them implies the highest frequency. There are also LED indicators for DSD. On the right, on a border next to the large golden volume knob, you can read the word «VOL». On the front side, there are the audio outputs. From left to right: 2.5mm BAL, 4.4mm BAL (in the same golden triangular section with rounded corners), 3.5mm SE (in a simple round golden section). A power LED (green) and low battery (red). Finally, the large golden potentiometer. Off is at approximately 7 o’clock.
On the left side you can read «Co-Presented by TRI & KAEI».
On the back edge are the USB connections inside an oval slot. First on the left is the USB TypeC connector for charging, then there is a selector for line out or headphones, finally the USB TypeC connector for connection to a source. On the left, outside this recessed oval, is a green LED, which flashes when the device is charging.
The base is a single piece of CNC-machined aluminium alloy.
Internally, it uses two 9038Q2M DACs, in a true, fully balanced design. It uses Texas Instruments Dual OPA2131 + Dual TPA6120A2 operational amplifiers for class AB amplification. All internal circuitry is gold-plated.
The battery is 5000mah and provides a battery life of 9 hours of use. Supports QC3.0 and PD charging (5V, 9V and 12V). It does not have a gain selector.
As I said, I had not imagined it so big. It is true that it weighs a bit. All this limits its transport and I understand that it is intended for portable use, but on a table.
The construction is excellent and there is a model completely in black. The design is beautiful, but the mirror finish is impractical as it is a fingerprint magnet.
It can be connected to Android devices via OTG cable. It is compatible with Apple devices and also with PCs. Windows 10 does not require drivers, but Windows 7 does. I looked for the ASIO drivers but I couldn’t find them, I didn’t ask Wendy for them because when I was messing around with the drivers installed on my system, I saw that I can use the ASIO BRAVO-HD which is used, for example, by the xDuoo Link2 BAL.
It decodes PCM up to 384K and DSD256 natively and Dop.
Very simple, USB TypeC for charging, USB TypeC for connection to a source, output selector, charging indicator, operation indicator, resolution frequency indicator, three audio outputs (BAL 2.5mm, BAL 4.4mm, SE 3.5mm). One potentiometer for power on and volume control. Switched off at 7h, it starts sounding at 9h, gain goes down until 3h (1.5V for SE), then there is a small volume drop (it seems to go into another amplification mode) and it reaches a maximum voltage close to 7V at no load (BAL) and almost 3.5V (SE).
I have taken measurements with my small oscilloscope, with pure tones of 20Hz, 100Hz, 1kHz, 5kHz, 10kHz and 20kHz. These tones have been created with Audacity software, with amplitude 1, 96kHz FLAC and 24bits.
The procedure is to turn the volume up to the maximum until distortion of the waveform is observed, then the volume is turned down until the waveforms are clean, trying to obtain the maximum RMS voltage without observing visible distortion.
No Load SE
The output goes up to about 3.5V just before the limit, then the signal is clipped.
At 1kHz, the voltage is 1.74V, the power is 200mW, the current is 120mA, a barbaric amount.
At 1kHz, the voltage is 2.29V, the power is 160mW, the current drops to 70mA. If it were 120mA, the power would be 480mW, but it is not. Above this voltage there is clear distortion.
At 1kHz, the voltage is 3.1V, the power is 96.1mW, the current is 31mA.
No Load BAL
The output goes up to almost 6.7V just before the limit, then the signal is clipped.
At 1kHz, the voltage is 2.53V, the power is 430mW, which implies a current of 170mA, the most powerful amplifier I have ever measured in a portable device.
At 1kHz, the voltage is 4.11V, the power is 510mW, the current is 120mA. Actually, it is not the specified power of 1250mW, but just over half a watt is very large.
At 1kHz, the voltage is 5.63V, the power is 320mW, the current is 56.3mA.
The response is very flat from 5Hz to 40kHz, at all volume steps. Only a very slight crosstalk is noticeable.
The Zout of the SE output is close to 10Ω, while the Zout of the balanced output is almost 20Ω.
Enthusiasts who have ever read one of my reviews of a source or DAC/AMP know that I am critical of the output impedance. But really, my red line is the frequency response: if a source or DAC/AMP does not have a flat frequency response, I will probably discard it, regardless of its final sound. There are more or less major flaws in this respect. A frequency response that has a slight dip above 10kHz is not as critical to me as a slight dip below 100Hz.
On another note, a source or DAC/AMP can have a flat frequency response and an output impedance of more than 1Ω. In this sense, this would most likely suggest a modification of the frequency response of the connected headphones.
After this introduction, I will tell you about my actual experience with the TRI TK-2. When it arrived, I tried it with the TRI Meteors and the Grace-S cable, via the balanced 4.4mm output. I was immediately captivated by the sound of the pair. My frequency response of the Meteor revealed a resemblance to the Moondrop Blessing 2 Dusk, some IEMS that a colleague left me and I wasn’t too keen on. I’m also an avowed bass lover and found the Dusks boring. The Meteor’s FR promised a bit more fun, but not as much as I heard after connecting it to the TK-2. That’s how I spent the afternoon. But something told me that this sound didn’t fit with the FR I had taken. I connected the Meteors to other sources and noticed that the powerful and noticeable low end was lost. One of the qualities I had read about the TRI TK-2 was its neutrality. Clearly, I was not hearing neutrality. Mostly, I use IEMS and earbuds, I don’t have many distinguished large-format headphones. So my headphones tend to be low impedance. I quickly thought that these differences in sound were probably due to the effect of a relatively high output impedance of the TRI TK-2. With my measurements I have been able to observe that the output impedance of the balanced connection is almost 20Ω and of 10Ω, the SE output. These are not strange figures, very similar to those of a great sounding DAP, the ACMEE MF02s.
To find out, I graphed some IEMS using the TRI TK-2 as source and my reference DAC/AMP Burson Audio Playmate with OpAmps V6. The first result is obvious.
The TRI Meteors gain 4dB in the low end and 2dB from the high mids, making them IEMS with a more V-shaped profile, therefore more fun at first glance. The explanation was served. The strangest thing, though, is that the Meteor’s FR variation is the most exaggerated of the few IEMS I’ve graphed. And I never thought it would affect low frequencies so much. It is clear that a higher or lower Zout does not act in the same way for all IEMS and it also depends a lot on the impedance of the headphones. The Meteor has an impedance of 12Ω, while the Zout of the TK-2 is 20Ω.
I know that it affects IEMS with dynamic drivers much less, if at all. While BA or other drivers suffer, to a greater or lesser extent, from the consequences of the Zout.
The next test was with the Dunu Titan S (32Ω), through the SE output of TRI TK-2. Although the graphs show a small difference, it is barely audible. The TK-2 shows its true heart with these headphones, displaying the famous neutrality the other reviewers talk about, embellished by a lushness of detail and nuance.
I repeated the experience with another Dunu, the Falcon Pro with the «Atmospheric Immersion» filter (26Ω), for the balanced 2.5mm output. Who can say that these IEMS are boring? Certainly, the 20Ω Zout of the TK-2 doesn’t affect the FR of these Dunu’s at all. But the combination suits them perfectly. The warm sound of the Atmospheric filter is less noticeable and the result is vibrant and dynamic, with a sparkle that enhances the sound of one of the best IEMS in this price range. Highly musical, detailed, vivid, euphonic and enjoyable. A pleasure.
During the week, the new NiceHCK M5 (16Ω), a hybrid IEMS with 1DDD+4BA, arrived. I tested them with the blue bass filter on the 3.5mm SE output. With this configuration I expected the differences in their FR to be large. However, the curve is almost identical in both sizes. The M5s are IEMS on the analytical side, which is strongly influenced by their filters, which have a strong influence on their low end. With the blue filter they become IEMS with an unabashedly pronounced W-profile, lots of sparkle and a punchy sub-bass. The TK-2 gets an extremely clean sound out of them, achieving a high level of transparency, precise treble and dry, deep bass.
Stranged by the low incidence in the FR of the M5s, I looked for a similar configuration in my collection. The Rose QT9 MK2 (10Ω), one of my favourite IEMS, is one of the ones I am most fond of. On this occasion, I used the 2.5mm balanced output. The result was that the Zout incidence is superior and audible. The TK-2 describes a steeper W curve, increasing the peaks and moving them away from the valleys. It is noticeable in the upper midrange and early treble, making the QT9s more resolving IEMS and losing some of the magic I get with other quieter sources. Despite the stronger vocals, the sound becomes more synthetic and colder, with less present vocals, less descriptive power and texture. It doesn’t always affect all headphones in a positive way.
Finally, I wanted to test the result on low impedance tribrids like the Geek Wold GK80 (8Ω), through the 3.5mm SE output. Really, brighter, with a clear incidence, but less than expected.
I could test many more IEMS from my collection and keep looking for the 7 differences, but my aim was to explain, in a more or less synthetic way, the effect of the DAC/AMP in some headphones, through the incidence of the output impedance, on the frequency response of each one of them.
And how does the TRI TK-2 sound, independently of the output impedance? Undoubtedly, it is good, very good, so good that I have even thought about buying the sample unit. I like its sound, its colour, its resolving capability and its power. Experiencing how it alters the sound of my headphones no longer becomes such a pejorative for me. It can become fun and beneficial, as in the case of the TRI Meteor. As long as the sound quality is maintained, the sound variations obtained by a relatively high Zout may have a higher appeal than I had previously thought. And when the FR is not altered, the result is better.
The TRI TK-2 is a DAC/AMP of undoubted power. It has a stunning, brilliant beauty, a truly premium finish, with a sleek and gorgeous build. Its power capability is up there with the best, it is one of the highest voltage and current delivery portable amplifiers I have ever tested.
The sound is truly premium. But because of its relatively high output impedance, it is capable of modifying the frequency response of the connected headphones, for better or worse. The result, where the difference exists, will be unpredictable and until you try it, you won’t be able to tell if it’s an improvement or a step backwards. In this sense, the experience can be a rewarding adventure. Do you want a great, very powerful and portable DAC/AMP that offers a truly different profile to your headphone collection? Don’t hesitate any longer, it’s the TRI TK-2.
Earphones and Sources Used During Analysis
- HiBy R3 Pro.
- Xiaomi Redmi Note 8 Pro.
- PC Win10.
- TRI Meteor.
- Dunu Titan S.
- Dunu Falcon Pro.
- Rose QT9 MK2.
- NiceHCK M5.
- Geek Wold GK80.
- Shuoer S12.