Noir (formerly: T2) Headphone Amp
Update 2019/10/14: It was brought to my attention that diyAudio brand had been used on the T2 without the consent of the brand owner. At the request of the brand owner, I have edited the text to reflect this.
Update 2019/10/07: The product name has changed from T2 to Noir to avoid a potential naming or trademark conflict. The circuit remains the same.
Also, pricing is now available: $29 for a set of PCBs (amplifier PCB, front and rear panels). PCBs + chassis will set you back $69. The parts will set you back an additional $85 (not including the volume knob), so a total build should come in around $165, not including shipping fees. The PCB and chassis kits are available through the diyAudio Store. I have edited the text below to reflect these prices.
One of my clients was among the early adopters of the T2 headphone amplifier designed by Mark Johnson, a member of diyAudio. After my client completed his build, he expressed interest in having the performance of the amplifier measured. His builds tend to be of top notch quality, and I was intrigued to see the amp's performance as well, so I agreed. What followed was a very thorough exploration of the capabilities and quirks of this DIY headphone amp.
- Conflicts of Interest
- Kit Overview
- Build Instructions
- Summary of Key Performance Parameters
- Performance Measurements
- Listening Impressions
Some may argue that I am inherently biased in my review, as I am a circuit designer and vendor who operates in the DIY audio market. Thus, they may argue, that I have an interest in presenting competitors' products in a negative light. However, I don't derive any satisfaction from exaggerating the negative aspects of a competitor's product; in particular as any savvy reader will be able to see through such exaggerations pretty easily.
I should also point out that many of my customers and clients find me through diyAudio, and I wish to remain in good standing with them. I, thereby, also have an incentive to downplay any negative aspects and highlight the positive aspects of the reviewed product.
I firmly believe that honesty will win in the end. An honest review will result in a better informed consumer, which is beneficial for everybody involved. Those who would rather be without my subjective commentary can always skip ahead to the Summary of Key Performance Parameters section and allow the measurements to speak for themselves.
After a week of measuring and characterizing the T2, I have yet to find one facet of this amplifier where it really shines. With clipping starting to set in at 55 mW (20 Ω), 75 mW (300 Ω), and 90 mW (32 Ω), I find it quite easy to drive this amp to clipping on the peaks – even with my easy-to-drive Focal Elex headphones (80 Ω, 104 dB SPL @ 1 mW). Those with even more power hungry headphones, such as many of the headphones with planar magnetic transducers, will find the T2 to be severely lacking in output power. Unfortunately, the T2 does not make up for this elsewhere. It produces significant distortion – in particular intermodulation distortion – resulting in an abysmal 66 dB SINAD (A-weighted) and SMPTE IMD exceeding 0.1 %. Furthermore, the T2 appears to have some difficulty driving reactive loads, and, thus, shows a propensity for ringing on the transient response even at modest cable capacitance. Also note that the T2 produces significant voltage transients (thumps) during power-on and power-off. While I cannot say for sure that these thumps (which reach 7 V peak!) are enough to destroy a pair of headphones, I do recommend that you leave your headphones unplugged when you turn on the T2 and unplug the phones before you turn off the T2. Similarly, I recommend those who plan to use the T2 as a preamplifier develop a strong habit of turning the T2 on before the power amplifier, and turning the T2 off after turning off the power amplifier. The line output on the T2 is simply a switched version of the headphone output, thus, failure to follow this power up/down sequence will definitely send your speaker cones for a wild ride.
After spending a few evenings listening to the T2 on my Focal Elex and Sennheiser HD-650 headphones I remain unimpressed. Then again, I'm comparing the T2 to a precision headphone amp of my own design, which delivers a performance that is significantly better than what my Audio Precision APx525 audio analyzer can measure. That said, it's not like the T2 is unlistenable. In some aspects, it's quite pleasant sounding. But during critical listening, I find the T2 lacking in terms of precision in the rendition. This is especially evident on the pronunciation of unvoiced consonants, and on more complex sounds, such as heavily distorted electric guitar, snare drum hits, and the metallic resonance of cymbals. I also find that the T2 seems to compress the sound stage and create more of an "in my head" feeling. Later in my listening sessions, I generally found myself reverting to my precision headphone amp, as I found the T2 to result in listening fatigue.
Where the T2 does shine is on the build cost and complexity. With a total build cost of approximately $165, it is a bargain of a DIY kit. Furthermore, the T2 circuitry is about as simple as it gets. This combined with the thorough build instructions should allow even a relatively inexperienced builder to complete a T2 within a weekend. If the builder opts to use the provided font and rear chassis panels, the only mechanical work will be to drill and de-burr four holes in the chassis bottom for mounting the circuit board. That's pretty attractive and – performance aside – would make the T2 a good candidate for a first DIY electronics project.
The T2 amplifier circuit contains only two transistors, hence its name. Each amplifier channel consists of a PNP darlington transistor driving an N-channel MOSFET output device biased in Class A at a quiescent current of 150 mA. It's a quite simple circuit, which is reminiscent of the vacuum tube circuits of yesteryear in its topology. You can find the schematic for the T2 thread on diyAudio.
Another legacy feature of this circuit is that it uses a positive ground. I.e., the system ground reference is actually the +24 V output of the wall wart that powers the amplifier. Unfortunately, this imposes some limitations on the power supplies that are compatible with this amplifier. The main limitation is that the output of the power supply must be galvanically isolated from the mains input, including the mains protective earth. This also means that the T2 is not compatible with many laptop chargers as these often connect the negative supply output to the mains protective earth. Using such a laptop charger with the T2 would short circuit the charger. That's a bit unfortunate, as many of us likely have a few such chargers and wall warts in our collections.
Interestingly, the T2 could easily have been designed with a conventional negative ground. As 'ground' really is an arbitrary designation, all that would be required, is for two resistors to be connected differently and the ground symbol moved to the negative rail. That appears to be a design oversight.
The T2 is implemented in a 1U 230×170 mm Modushop Galaxy chassis ($44 from the diyAudio Store). This chassis is quite nice. It features powder coated steel panels for the top and bottom, which add quite a bit of heft to the chassis. The front and rear panels are made from aluminum. The sides are extruded aluminum profiles. To my knowledge, there is no prefabricated chassis option for the T2 kit, thus, builders who wish to use the aluminum front and rear panels will have to drill the holes themselves. This is straight-forward for those of us who have the tools. For those who would rather not drill into the panels, the designer of the T2 offers front and rear panels made from FR-4 fibre glass circuit board material. These panels are 2.0 mm thick and feature black solder mask to match the chassis. The T2 I received was built using these ready-made front and rear panels as shown below.
While the chassis is unlikely to win any design prizes, it does do a decent job of making the T2 look like a finished product, albeit, a product with a somewhat distinct "I built this myself" look. The front panel features the headphone output (1/4" jack), volume control, and power switch and power-on indicator LED.
Note that after the name change from T2 to Noir, the front panel text was cleaned up considerably. You can find pictures of the Noir chassis in the Noir thread on diyAudio.
The rear panel presents a bit more professionally as seen below.
The rear panel features the power input (barrel connector) and two pairs of RCA connectors. One pair of RCAs are line inputs. The other pair can provide output to a power amplifier, thus, allow the T2 to be used as a preamplifier. The output signal is automatically switched off when the headphones are plugged in.
While the use of PCB fibre glass material for the front and rear panels is certainly cheaper than having the aluminum panels custom drilled and printed, the fibre glass panels do have a couple of disadvantages. Fibre glass is fairly flexible. Thus, even with their 2.0 mm thickness, the panels have a bit of bow to them as shown below.
Light clearly shines through all the way on the inside of the front panel. The panel is further bowed by a washer between the front panel and the bottom left mounting screw as seen in detail below.
Even though this amplifier came fitted with a very large volume knob, the thin front panel does not allow the knob to sit flush with the panel. Rather a 5-6 mm gap is left between the knob and the front panel, which further adds to the home made look of the amp.
The power switch features a washer designed to prevent the switch from rotating in its mounting hole. However, for this washer to work, its locking tab needs to engage with a hole in the front panel. Unfortunately, this locating hole was not provided in the panel, which appears to be a design oversight.
One nice feature of the T2 is that it features a dimmable power-on indicator. This is accomplished by inserting a multi-turn trimpot in series with the power-on indicator LED, thereby allowing its brightness to be adjusted somewhat. Unfortunately, adjusting the brightness of an LED by adjusting its series resistance is not all that effective. This is especially true for the blue and white LEDs. Thus, builders who use these colours of LEDs will likely find them to be too bright, even with the brightness trimpot turned all the way to minimum brightness. A more reliable approach would be to pulse-width modulate the LED. Pulse-width modulation would work with any type of LED and can be implemented with a small handful of components. What the T2 lacks in craftsmanship on the outside, it makes up for on the inside.
The insides of the T2 show an entirely different level of attention to detail.
All the components except the power resistors are mounted tight against the circuit board and the soldering is impeccable. The power resistors are elevated slightly above the PCB to allow for better air flow, and the PCB is equipped with ventilation holes under the resistors. Excellent!
Peeking inside the T2 also reveals that the washer between the chassis side and the inside of the front panel is actually a grounding lug (connected by the purple wire in the upper right of the picture above). A similar grounding wire and grounding lug grounds the rear panel. That's an interesting design choice given that the front and rear panels are PCBs. One could have implemented this connection by leaving a solder pad exposed on the inside of the panels (which is a solid copper pour) and soldering the grounding wires to these pads.
The amplifier PCB is mounted on metal standoffs, and the builder made sure to remove the powder coating below the standoffs to ensure a good electrical connection between amplifier ground and the chassis bottom.
The bill-of-materials is shared in the form of an Excel spreadsheet and associated project set up with Mouser Electronics. The BOM appears to contain all the parts needed to assemble the amplifier with the exception of the M3 machine screws, nuts, and washers needed to attach the PCB to the chassis. Interestingly, the M3 standoffs for mounting the PCB are included on the BOM. Thankfully, metric hardware is commonly available at hardware stores, even in the US. The BOM also does not include the volume knob. The total parts cost comes to just shy of $90, plus the cost of the additional hardware and volume knob. The cost of the PCBs has yet to be determined.
The build instructions that accompany the T2 are a bit rough and appear incomplete. For example, there is no table of contents, introduction, or even a title page. That said, the instructions do contain enough information to allow most builders to assemble a T2, and although they don't seem to follow a step-by-step format, they are quite thorough.
The first task in the build process is to create a drill template for mounting the PCB in the chassis. I find it interesting that the author chose to spend two pages describing how to make a drill template rather than just providing the template. Creating a drill template would take 15 minutes in a CAD tool. Many builders are likely to request one anyway. Either way, the builder should be prepared to drill four holes in the bottom panel of the chassis. This panel is made of steel, so a sharp HSS bit, cutting fluid (or acid-free oil), and some patience will be needed. If you are using the ready-made front and rear panels, drilling the four mounting holes for the PCB will be the only mechanical work necessary.
Overall, the instructions are written in a quite lighthearted language. The language is a bit too colloquial for my taste, and in some cases the instructions come across as condescending. The preface to the description of how to mount the power resistors provides a good example of this:
"The 68 ohm, 3 watt resistors get hot. I mean HOT. This is a class A amplifier and those resistors are running at high current without a heat sink. Mommy's little favorite, the MOSFET, gets a big huge aluminum heatsink with tall fins and lots of radiator area. These resistors get nothing, nada, bupkis."
I'm not sure what the author's intent is here. The 68 Ω resistors are pairwise in parallel, thus, the 150 mA bias current is shared equally between the two resistors in each pair. This means the power dissipation in each resistor is 383 mW. The resistors are rated for nearly ten times that, thus, even after 12 hours of operation, the resistors only reach 55 ºC. The Vishay PR03 power resistors used in the T2 are rated for operation at a maximum temperature of 250 ºC. While I would certainly never run them that hot, the resistors in the T2 are operating well within their thermal limits. Thus, the dramatic language seems a bit out of touch with reality.
The MOSFET heat sink is a generic PCB-mounted heat sink. It stands about 25 mm (1") tall. That's not big or huge or even "big huge". After 12 hours of operation, the heat sink reaches 45 ºC, so regardless of the dramatic language, it certainly provides plenty of cooling for the MOSFET.
I do support the author's recommendation to elevate the power resistors some 5-6 mm above the PCB surface. Similarly, I commend the choice of heat sink for the MOSFET. However, the dramatic language and repeated mention of the power dissipated in this Class A amplifier and the needs for adequate ventilation throughout the documentation make it sound like the amplifier is a powder keg ready to explode. This seems completely out of proportion given that the T2 is actually a very conservative design, at least in terms of power dissipation in the resistors and cooling of the output device.
I should note that the builder of this amp was one of the early adopters, thus, it is possible that the documentation has been cleaned up since his copy was printed.
The key performance parameters for the T2 are tabulated below.
|Output Power||55 mW||20 Ω, THD+N < 0.1 %|
|Output Power||90 mW||32 Ω, THD+N < 0.1 %|
|Output Power||75 mW||300 Ω, THD+N < 0.1 %|
|THD||-85 dB (0.0056 %)||1 kHz, 1 mW, 32 Ω|
|THD||-65 dB (0.056 %)||1 kHz, 50 mW, 32 Ω|
|THD||-91 dB (0.0028 %)||1 kHz, 1 mW, 300 Ω|
|THD||-68 dB (0.040 %)||1 kHz, 50 mW, 300 Ω|
|THD+N||-66 dB (0.052 %)||1 kHz, 50 mW, 32 Ω|
|THD+N||-69 dB (0.035 %)||1 kHz, 50 mW, 300 Ω|
|IMD: SMPTE 60 Hz + 7 kHz @ 4:1||-55 dB (0.171 %)||50 mW, 32 Ω|
|IMD: DFD 18 kHz + 19 kHz @ 1:1||-66 dB (0.049 %)||50 mW, 32 Ω|
|IMD: DFD 917 Hz + 5.5 kHz @ 1:1||-78 dB (0.012 %)||1 mW, 32 Ω|
|IMD: SMPTE 60 Hz + 7 kHz @ 4:1||-59 dB (0.111 %)||50 mW, 300 Ω|
|IMD: DFD 18 kHz + 19 kHz @ 1:1||-68 dB (0.039 %)||50 mW, 300 Ω|
|IMD: DFD 917 Hz + 5.5 kHz @ 1:1||-85 dB (0.0057 %)||1 mW, 300 Ω|
|Multi-Tone IMD Residual||< -102 dB Ref. 50 mW||AP 32-tone, 50 mW, 300 Ω|
|Channel Separation||70 dB||1 kHz|
|Channel Separation||> 62 dB||20 Hz - 20 kHz|
|Gain||6 dB||1 kHz, 50 mW, 300 Ω|
|Gain Variation||±0.3 dB||20 Hz - 20 kHz, 50 mW, 300 Ω|
|Input Sensitivity||1.9 V RMS||50 mW, 300 Ω|
|Bandwidth||2.9 Hz - 630 kHz||50 mW, 20 Ω|
|Slew Rate, rising edge||3.2 V/µs||300 Ω || 220 pF load|
|Slew Rate, falling edge||33 V/µs||300 Ω || 220 pF load|
|Total Integrated Noise and Residual Mains Hum||1.8 µV RMS||20 Hz - 20 kHz, A-weighted, min. volume|
|Total Integrated Noise and Residual Mains Hum||2.7 µV RMS||20 Hz - 20 kHz, Unweighted, min. volume|
|Output Impedance||0.19 Ω||1 kHz|
|Dynamic Range (AES17)||114 dB||50 mW, 32 Ω, CCIR-2k weighted|
|Signal to Noise and Distortion (SINAD)||66 dB||50 mW, 32 Ω, 20 Hz - 20 kHz, A-weighted|
|Signal to Noise and Distortion (SINAD)||65 dB||50 mW, 32 Ω, 20 Hz - 20 kHz, unweighted|
|All parameters were measured after a minimum of three hours of warm-up and at the maximum setting of the volume control unless otherwise noted.|
The T2 provides remarkably low output power and delivers an overall level of performance similar to that of a mid-fi vacuum tube amp. This is not entirely surprising as the T2 uses the same circuit topology as some vacuum tube based amplifiers.
Observant readers will note the highly asymmetric slew rate. The T2 is a single-ended design rather than a push-pull design, thus, its current sinking capability is vastly better than its current sourcing capability. As result, its slewing behaviour is asymmetrical, which results in the tenfold difference in rising edge versus falling edge slew rate.
The T2 uses a fixed bias, i.e. the application of a constant voltage to a bias node within the circuit. In the T2, the bias voltage is formed by a variable resistive divider and applied to the base of a PNP darlington driver transistor, which then sets the operating point of the output N-channel MOSFET. Thus, the bias of the circuit will depend on the device characteristics of both the PNP darlington and the NMOS output device. The measured bias voltage will also depend on the temperature coefficient of the power resistors loading the MOSFET drain. These characteristics are dependent on the device temperature. Thus, given the thermal mass of the heat sinks and power resistors, the bias of the T2 should be adjusted or verified periodically after the amplifier is first turned on.
The build instructions recommend setting the bias when the amp is initially turned on and repeating this procedure after five minutes. I find this recommendation to be inadequate. As shown in the measurement below, the bias takes well over an hour to settle to its final value of 5.10 V, and takes nearly five hours before it is fully settled. Thus, I recommend verifying the amplifier bias periodically for the first few hours following the first power-up, and readjusting as necessary.
For the measurement below, I allowed the amplifier to warm up for over 12 hours and adjusted the bias periodically during this time. I then allowed the amplifier to cool overnight. The following day, I performed the measurement below, which shows the bias voltage for the first six hours following power-up.
The vast majority of the following performance measurements were performed after this six-hour warmup period.
The T2 is an AC coupled design. I.e. its inputs and outputs are capacitively coupled to the source and load, respectively. This also means that when the amplifier is turned on, it produces a significant voltage transient (thump) in the headphones as its output capacitor charges. Similarly, the amplifier will produce a large voltage transient when it is turned off as its output capacitor is discharged. As seen below, these transients reach 7 V peak. While I cannot definitively say that these thumps would destroy a pair headphones, I certainly made sure to unplug my Focals before I turned the amp on and off. Those who are into circuit modifications, might want to consider adding an output relay – or better yet a MOSFET-based protection circuit such as my Guardian-86. This is especially true for those who plan to use the T2 as a preamp as the line output on the T2 is simply a switched version of the headphone output.
The measurement below shows the THD+N vs output voltage swing for the T2 at various load impedances. The output voltage droops significantly as the load impedance is lowered. At 20 Ω load, the amplifier is able to provide slightly over 1.0 V RMS before the distortion rises above 0.1 % and the amplifier starts to clip.
In all the measurements below, Channel 1 (Ch1) was connected to the left channel of the T2, and Channel 2 (Ch2) was connected to the right channel. The left channels consistently shows worse performance that the right. This is likely due to part-to-part variation between transistors in the two amplifier channels.
The measurement of the THD+N vs output power for 32 Ω load is shown below. The distortion starts increasing already at 500 µW into 32 Ω. The onset of clipping is fairly soft, although not quite as soft as the soft clipping exhibited by a typical vacuum tube based amplifier. The THD+N of the T2 crosses 0.1 % at 90 mW into 32 Ω.
The results of the THD+N vs output power measurement with 300 Ω load are shown below. The amplifier provides 75 mW into 300 Ω at the onset of clipping (0.1 % THD+N).
The THD+N vs frequency measurements for 50 mW into 32 Ω and 300 Ω, respectively, are illustrated below. Note that the measurement bandwidth has been increased to 60 kHz for this measurement. Thus, these measurements include a significant proportion of the switching noise from the wall wart switching power supply (more on that below).
Due to the amount of ultrasonic noise present in the output of the T2, a more accurate picture of the THD+N vs frequency can be deduced from the measurement of the THD+N vs output power at various frequencies. This type of measurement is common for measurements of Class D amplifiers due to their ultrasonic emissions. The result of such a measurement for the T2 is shown below for 32 Ω load and 300 Ω load, respectively.
In addition to the THD+N numbers, the spectral composition of the distortion products are also quite telling of the sound quality of an audio amplifier. The harmonic spectrum of a 1 kHz sine wave reproduced by the T2 are shown below for 50 mW into 32 Ω and 300 Ω, respectively. The spectral composition is quite typical for a low loop gain solid state design, and shows a smattering of both even and odd order distortion products.
Casual listening is more likely to occur at significantly lower power levels, however. Thus, I repeated the measurement at 1 mW output power into 32 Ω and 300 Ω as shown below. This is likely more reflective of the performance you will experience at casual listening levels.
At 1 mW output power, the distortion products are mostly comprised of the second and third harmonic. Curiously, a significant amount of mains hum is present and appears to have caused a large number of intermodulation products.
The intermodulation distortion of the T2 is quite high, as shown below. This test uses two test tones, 60 Hz and 7 kHz, at an amplitude ratio of 4:1 following the SMPTE standard. High SMPTE IMD can often tease out thermal issues and poor power supply rejection within the amplifier circuit. In the case of the T2, I suspect the latter.
The performance improves a bit when the two test tones are 18 kHz and 19 kHz at an amplitude ratio of 1:1, although significant IMD products are still present.
Siegfried Linkwitz argues that the 1 kHz + 5.5 kHz intermodulation distortion (IMD) measurement is one of the measurements which is more indicative of the perceived sound quality. He bases this argument on the fact that IMD products in this measurement fall in the frequency range where the ear is the most sensitive (see the Fletcher-Munson curves for more detail). I think this argument carries a good amount of weight, so I measured the T2 accordingly. The measurement is shown below. Note that due to a limitation in the MOD IMD source of the APx525, the frequencies used must be an integer multiple of each other. Thus, I measured at 917 Hz (5500/6) + 5.5 kHz. I performed this measurement at 1.0 mW into 32 Ω.
Another reliable indicator of good sound quality is low multi-tone IMD. This measurement is as close to music reproduction as I can get with a deterministic test signal. The test signal is comprised of 32 tones, of equal amplitude, logarithmically spaced in frequency and sounds a bit like an out-of-tune pipe organ. An ideal amplifier would show just the 32 tones. However, as shown below, the T2 adds a plethora of intermodulation products (the 'grass' between the tones).
Another important measurement is that of the residual hum and noise of the amplifier. The graph below shows the output noise spectrum of the T2 with its inputs shorted. The plot shows quite a bit of mains hum and harmonics, which extend into the low kHz range. In addition, a significant amount of switching noise is present in the 40-50 kHz range. That is a bit disappointing considering that the circuit designer went through the effort of including a power supply filter in the T2. Unfortunately, however, the designer appears to have forgotten to measure the efficacy of the filter following its implementation.
One may be tempted to chalk this up to, "Well! It's just DIY. If you don't like it, use a different power supply!" I suggest resisting this temptation, though. In my view a project should work well when assembled according to the instructions. This is especially true of a kit, as electronics kits tend to attract the less experienced builders. Thus, the onus is on the kit designer to ensure that the circuit works well without modifications.
That said, it is possible to clean up the noise by using a well-regulated power supply. The measurement below was performed with the T2 power by an HP6237B laboratory power supply.
For completeness, I measured the amplitude response (gain) and gain flatness as shown below.
The measurement below shows the output impedance of the T2. At 0.19 Ω (1 kHz), the output impedance is quite a bit higher than I would have expected from a solid-state circuit. That said, it's certainly low enough to allow the T2 to drive even very low impedance headphones.
Finally, as the T2 is a stereo amplifier, I measured the channel separation. The result is shown below. The channel separation is not stellar. For comparison note that I measured channel separation better than 85 dB on the JDS Labs O2, over 90 dB on the Sjöström QRV08, and 115 dB on my HP-1. I suspect the two channels within the T2 are coupling though the power supply.
The measurements below show the output voltage vs time for the T2 as it is driven into clipping. As seen below, the T2 clips pretty cleanly with no signs of oscillation as it enters or exits clipping.
The same is true when the T2 clips hard as seen below. The T2 exhibits slight rail sticking (i.e. the little downward step as the output recovers from clipping), which is likely caused by the PNP darlington driver exiting saturation. While the rail sticking is generally frowned upon, it is quite common and relatively harmless. Advanced builders can experiment with the addition of saturation clamps should they desire to get rid of the rail sticking. Such clamps can be as simple as a diode in the right spot. That said, the T2 shows no signs of instability as it enters or exits clipping, which is the main point.
In addition to providing a clean output and well-behaved clipping response, a headphone amplifier must be able to drive significant capacitance. Many – myself included – use a headphone amplifier as part of a home stereo setup, thus need to connect to the amplifier with a relatively long cable. The cable capacitance varies by cable geometry, but a reasonable figure is around 100-150 pF/m. Thus, a 4-5 meter extension cable will contribute around 500-600 pF of capacitance. In addition to the cable capacitance, the amplifier will also need to drive the reactance of the headphone driver itself. In my designs, I generally err on the side of caution and design the amplifiers such that they can drive 10 nF cleanly.
The graphs below shows the transient response of the T2 when driving 470 pF, 1 nF, 4.7 nF, and 10 nF in parallel with 300 Ω. As seen below, the T2 shows a slight buzz on the falling edge even with 470 pF load capacitance. 470 pF is a quite reasonable cable capacitance, so I am a bit disappointed that the T2 cannot drive such a load cleanly. The ringing increases with increasing capacitance, and with 10 nF in parallel with the 300 Ω load resistor, the T2 shows significant ringing.
I must say that I find it very hard to contain my disappointment in the technical performance of this amplifier and I remain decidedly unimpressed by it. I suppose one could argue that with a build cost under $200, this is what you get. However, for $200 you could buy two (2) JDS Atom headphone amps and have a few bucks left over for a cup of coffee. The Atom provides vastly better performance than the T2, but then you don't get the satisfaction of building it yourself. Tradeoffs, tradeoffs...
Okay. So what does it sound like? Actually... Not bad... But coming from an ultra-clean headphone amp (think THD below -130 dB), I find the T2 lacking in clarity. At low to moderate volume levels with relatively simple music, the T2 performs quite well. However, even at these volume levels, I find, that the T2 renders the voices of some singers (such as Mark Knopfler on Dire Straits "Brothers in Arms" and Chris Stapleton on "From a Room Volume 2") with a bit of fuzziness, in particular on unvoiced (hard) consonants. On more complex or louder passages, the lack of precision in the T2 shines through. Thus, distorted electric guitar is rendered with a little extra grunge, snare drums get muddied, and cymbals are rendered slightly less metallic and slightly more fizzy than those instruments sound like live. If I crank it loud, I also find it quite easy to get the T2 to distort on the peaks.
I find the notion of sound stage to be a bit odd when it comes to headphone amplifiers. That said, I do find that the sound stage of the T2 seems a bit compressed or more "in my head" than that of a precision headphone amp. I also find recordings which feature a sense of space, such as many of the tracks on Pink Floyd "A Momentary Lapse of Reason", to be rendered more spatially compressed on the T2.
Overall, I spent four evenings listening to the T2, occasionally switching back to my precision headphone amp for reference and comparison. I mainly used my Focal Elex headphones, but also spent a few hours listening to the T2 using my Sennheiser HD-650. In addition to the albums mentioned above, my test tracks included tracks from Eric Clapton "Slowhand at 70", John Mayer "Where the Light is", Celine Dion "Falling Into You", Dire Straits "On Every Street", many of Mark Knopfler's solo albums, along with several others.
You can find the conclusion at the top of this page.