I just bought a NG98 on eBay. They are in China and I'm in the US, so now the long 3-4 week wait. 
This amp seemed like a steal at $99 including shipping.
I'm in the process of building a headphone amp (no DAC) based on LME49710 op amps with LME49600 buffers. This amp grabbed by attention when I saw it used BUF634 chips. They are about 90% similar to the LME49600s. The LME49720 dual op amps are pin compatible with the OPA2604s used in the NG98 so I'm hoping I can pop in the LMEs. Depends on what HA Info is doing with the power supply - the OPAs go up to +/- 24v while the LMEs only do +/- 17v. The adapter on the NG98 is 24 volts but I see what looks like a 5 pin LDO regulator in the photos. I'm hoping they are regulating the 24 down a few volts, then splitting it using a virtual ground design. The net result might be something like +/- 11v into the op amps and buffer chips. Plus they have to have a 5v regulator in there somewhere for the digital.
I'm using Shure SRH840 headphones which clock in at 44 ohms, hence my interest in current buffer chips. The op amps in CMOY type units typically only can do around 20-25 mA. The BUF634s and LME49600s can go up to 250mA. At 80mw headphone amp output that would be about 42mA RMS into a 44 ohm load. I also like the higher output voltage rating on this one if I want to use it with higher impedance headphones some day. At 44 ohms it would only take about 1.9V rms to push the 42mA. But at 300 ohms it would take about 5 volts to push 80mW. Or at 44ohms 4.5 volts - which is still just 100mA through the 250mA buffers - would likely blow my eardrums all the way through my head and out the other ear at 460mW. Curiously enough, the Shures at rated for that - 1000mW, which I assume means 500mW per side. Maybe spread the ear pads apart, crank up the NG98, and use them as bookshelf speakers? 
Speaking of power output, I'm going to open it up and map out the circuit for fun after I get them. To buz who started the thread - I'll see if I can figure out which resistor(s) have to be changed to reduce the gain. With lower gain the volume pot will work further in its range, of course, and solve low-volume balance issues and high volume surprises. I'll have to see what they have done with DC coupling between the stages. On the one I'm building the design has unity gain (just the headphone amp circuit given in the datasheet for the LME49600 chip) but I'm going to add a switch and resistors to give me a choice of unity or 3dB. That design also uses a DC servo circuit to make sure the output DC component is always around zero volts DC. Will be interesting to see what HA Info is doing there. I hope they are not using output coupling caps! 
Edited by agdr - 10/7/10 at 10:11pm
)? Don't think I see a socket in the pic...
