- 8,856 Posts. Joined 10/2001
- Location: Hong Kong
- Select All Posts By This User
Tyll Hersten/Innerfidelity's 21 Amp Measurements... - Page 2
One more thing. I didn't mention THD+N yet.
This plot is different in that the x-axis is output voltage and the y-axis is THD+N (%), which means total harmonic distortion plus noise.
While the harmonic spectrum shows harmonics and noise in great detail, it only shows this at a fixed output voltage and load. In the THD+N plot this spectrum is converted into a single number and that is done over and over again with rising voltage and different loads.
The way it is calculated is by summing the power of the harmonics and noise (everything else) and dividing it by the power of the fundamental. The resulting percentages can also be converted into dB:
10% = -20 dB
1% = -40 dB
0.1% = -60 dB
0.05% = -66 dB
0.01% = -80 dB
Since higher load impedance means the amp has to provide less current you'll generally see lower THD+N with such loads.
At low voltages noise starts to dominate, at high voltage the amp will reach its clipping point at some point (where the line shoots up). If this point, for example, is at 2 V then the amp can output about 250 mW (= 2*2/16) into 16 ohms.
Well, it seems to have problems with nonlinear distortion. What's weird is that in the measurement with 150 ohms it has higher THD+N than with more difficult loads. It seems to be quite a bit happier with 16 ohm IEMs. But keep in mind that you won't need high voltages to reach ear-splitting levels with IEMs and at low voltages the THD+N doesn't look so good.
Edited by xnor - 7/5/13 at 7:58am
It looks as if it has crossover distortion (high order THD products, and distortion - it is clearly not noise - increasing with lower output voltage), or some other problem. It may also be a "bad" sample/QC issue, though. There are some RMAA results around to compare against, it would be interesting to see if those also show the distortion. However, if the distortion is caused by a badly designed virtual ground, then an RMAA measurement with a sound card that does not have differential inputs might not show the problem correctly.
Edited by stv014 - 7/5/13 at 8:39am
Here is one with a 33 Ω load. The distortion looks much better than at InnerFidelity, although this was apparently tested at a higher level (1.8 Vrms); even when that is taken into account, however, there is clearly a difference. Perhaps if the distortion is caused by an AMB Mini3 style virtual ground, then the sound card loopback may hide it, because it references the output signal to the ground of the sound card, rather than the "third channel" in the amplifier.
The O2 has a bit lower distortion and noise floor, which should only be noticeable if you turn up the volume a lot and have sensitive headphones.
Qogir is a bit louder with low impedance headphones (up to about 80 ohms). I doubt you'll need the extra volume with low impedance headphones though.
O2 has lower output impedance (0.5 vs 10 ohms) so I'd say it is a better match for low impedance (again up to about 80 ohms) headphones. If you drive something like a PX100 with the Qogir you will get a bit looser and slightly boosted bass.