This review page is supported in part by the sponsors whose ad banners are displayed below

"The Pass explanation does apply to headphones and our HPA-21 just as suggested yet with 'phones, current-mode amplification has another big advantage. Most headphone drive units show an almost flat impedance across the audio spectrum to result in more stability and better response. Our battery supply for it doesn't incorporate the BPS-02 but EQA-11R circuit. That runs a dual-voltage supply on higher-voltage batteries just as you predicted. Here we use four cells in series to generate 16.8V. That's what Samsung SDI recommends as the max reliable working limit for their batteries.

"This battery control circuit is as complex as the BPS-02's. Once again in the analog domain it must control the likely dissimilar behavior of two batteries simultaneously. To make this circuit non-stop would require a total of four Li-ion batteries plus the complex power-cycling control circuit of the BPS-02. This overreaches on size and budget for how we've targeted the HPA-21. This we shall save for the later DAC and another source model. [Akira Nagai's bench in Japan at right.]

HPA-21 passing a 100kHz sine wave

"During our HPA-21 prototyping we already achieved up to 7 hours of playtime despite its rich class A bias. This we consider sufficient for headphone listening. The automated recharging process cuts off during playback but starts as soon as the power switch is depressed. On finish we will consider a silver option but keeping stock in two colors can be difficult for distributors and dealers alike. We must first solicit their feedback.

First silver prototype case

"Our planned roll-out of 2013 products will be the HPA-21 in the 2 Series; the 50wpc AMP-51 and DAC-51 in the 5 Series; and finally the 200-watt AMP-91 monos in the 9 Series. Each series will feature a different industrial design yet share the same company identity as indicated by the HPA-21 above which incorporates key elements from the 1 and 3 Series (hence 2 Series). We will most likely apply the same idea for all new models.

"One exception could be the AMP-91. This will be a cost-no-object design outside and in. For example it will have 8 power transistors, 16 drive transistors and more than 200 micro transistors for calibration and bias per channel. Hence we are planning on an equally impressive aesthetic and mechanical design to match this very ambitious circuit.

"The AMP-91 will be very important for us and represent a monumental effort to sum up all of Nagai-San's circuit development history whilst introducing new circuits which over time will affect our future products along the way."
World premiere at the Highend Mässan 2013 Stockholm. Swedish importer Klutz Design and Soo In Chae both share my admiration for how well Bakoon and Audez'e combine.

Headphone amp from Bakoon Japan
  Designer Akira Nagai on the HPA-21: "Speaker and headphone drivers produce sound by emitting vibration caused by electrical current flowing through a magnetic voice coil. Here the drive force of the voice coil F can be represented by this formula: F = I x B (force = current x magnetic field). As we see, the driving force is generated by current, not voltage. However, when driven by a voltage source with an impedance of Z, we use Ohm’s Law and substitute the above formula with this: I = E/Z (current = voltage/impedance) and F = E/Z x B.

"When driven by a voltage source, now impedance affects the driving force. The value of Z is comprised of voice coil reactance, capacitance, DC resistance and contact resistance all of which exhibit non-linear behavior to cause distortion errors in the driving force.

"As the HPA-21 drives headphones with current, impedance doesn't factor into the equation. We can expect far lower distortion. Such has been studied by Dr. Yoshiaki Muda of the Department of Electrical Technology at the Nara University of Education in his 1996 paper Transient and frequency response of the moving-coil loudspeaker current-drive and voltage-drive method*. During his experiment he found that when a speaker was driven from a current source, he could achieve up to 1/10th its distortion. The
HPA21's current output achieves the exact same behavior. The major drawback of current drive for loudspeakers is that whilst there are continuous impedance changes across all audio frequencies, the current-driving force remains fixed. Therefore the unit’s impedance curve becomes the frequency response of the driver. Because of this it is not appropriate to use current drive for most loudspeakers whose impedance varies too greatly.

Mechanical rendering

* The abstract of this paper states that the "frequency response of the moving-coil loudspeaker in the current-drive and voltage-drive methods has been investigated. It is shown that for current drive the low cut-off frequency is f0=(k/m)1/2π and the high-frequency response is limited only by the cut-off frequency fA of the acoustic impedance of air. In voltage drive the frequency response is far more complicated. It has another high-frequency cut-off at ωRL=R/L along with fA where R and L are the voice-coil resistance and inductance respectively. For voltage drive when R increases, ωRL increases but Q decreases; when Bl increases, efficiency increases but Q decreases. In current drive the reduction of pistonic radius increases the HF cut-off frequency but any reduction of pistonic radius doesn't necessarily improve the HF response for voltage drive."

This first sample of the power supply PCB had a few mistakes to show some resistors and a capacitor above the PCB which final production corrected.

"Back to headphones, using a single driver of mostly constant impedance, we can safely adopt current drive with great results both on the bench and at the ear. The HPA-21 offers 6.3mm voltage and current outputs which may be used simultaneously. Then their current signal level becomes the same. Therefore even when driving a headphone with greater impedance variation, connecting it to the voltage output results in a variation of the output current (high impedance = less current). Due to this one can expect to automatically compensate the output level of the current-output socket when running two headphones at once.

Amplification board showing the four output devices per channel bolted directly to the undercarriage as monolithic casing heatsink and the two SATRI EX current-mode gain modules.

"Damping factor with a headphone amplifier is of less significance especially in current-drive mode. The formula for calculating damping factor is D=R/Z (damping factor = resistance of headphone/amp's output impedance). Since the HPA-21’s voltage output is about 1Ω, a 10Ω headphone encounters a damping factor of 10 whilst a 100Ω headphone sees 100. As we see the damping factor varied by a factor of 10. The HPA21’s current output is about 2.5MΩ. This renders damping factor effectively zero. Now the music signal itself becomes the drive current and it is unaffected by any other variables to generate much higher and more precise output drive."

Soo In: "We started production this third week of March. We redesigned the power supply board by meticulously studying all PCB patterns and planes to create an even lower noise floor. It's surprising how performance became even better than it had been at the Stockholm show. This took us a while but I'm very pleased with the outcome. Above is the final PCB from actual production. You'll notice quite a difference from the prior version. To the right are the final mother boards for the gain circuit. The other review unit out of two besides yours is currently in the USA with one of the organizers for the NYC headfi meet. He finds it excellent with the Fostex TH900 too and prefers this combo over the LCD-3. (The unit will go to Tyll at Inner Fidelity after the meet.) I've not yet heard the TH900 but like Denon 'phones based on Fostex drivers. Perhaps you'll have a chance to hear the Fostex TH900?"

final production