I had some time to spare with my beloved O2 this weekend, so while trying to find what's my O2's max. non-clipping output voltage I was able to get a decent output voltage of 7.4V RMS (with dummy load of 500ohms resistors + original Beyers cable serialized, instead of headphones) before starting to clip and without any sort of strange spikes or oscillations.
So, while my 3 years old kid was spilling water inside a power stripe's outlet (thank's God of the 30mA differential safety) and he was unscrewing all the screws from a wooden IKEA chair (for all these screws I applied a 10-12nM of torque when I built this chair; I had no idea that kids today have such a force to unscrew'em) I did few mods during this weekend:
- 1) Changed the gain resistors from infinite/1Kohm (corresponding to 1X/2.5X) to 2.5Kohm/516ohms (corresponding to 1.6X/3.78X). I did some "custom resistors" of about 516ohms by serializing 274ohms + 242ohms resistors that I had in stock; this corresponds to a real gain of about 3.78X and an output voltage of about 7.3V RMS (a bit less than max. non-clipping voltage of 7.4V RMS, just to be sure clippings will never occur when ODAC is connected, even with loud and highly compressed music). Perhaps some of you might still say that a gain of 3.78X is too much and voltage gain will clip, but with my ODAC XL connected to O2 and 0dB sinewaves applied (taken from audiocheck.net) there's no clipping at all. Looks like my ODAC can output 1.921V RMS when sinewaves are applied; I have tried 20Hz, 32Hz, 1KHz, 16KHz, 20KHz and I get no clipping, no roll-offs, no oscillations, no strange FFT harmonics, no THD >0.1% (checked with my PicoScope 2204A).
- 2) Added 2 x 1uF film capacitors immediately after the 2 regulating diodes from power input plug. These 2 caps should decrease any possible noise coming from the mains.
- 3) Added 2 x 10uF/25V AVX tantalums on output regulators, to further decrease the ripple.
- 4) Added 2 x 47uF/35V Nichicon KA and 2 x WIMA 0.1uF/100V for decoupling the input stage OPAMP (LME49720).
- 5) Added 2 x 47uF/35V Nichicon KA for each of the 2 output buffers for decoupling purposes (NJM4556A).
Did I gained anything?
- Besides a lot of fun, I was able to decrease ripple & noise from 2.3mV p-p to 0.747mV p-p, measured directly on each regulator output and also on OPAMPs +/-V rails. That's a good ripple decrease, though it should probably not be noticed on headphones out, because O2's OPAMPs already have a good ripple reduction inside (PSRR & CMRR of about 90-100dB).
- Soundstage increased a little bit, though this is a subjective impression that I'm going to test it soon by A/B'ing.
- OPAMPs rolling should be easier now with additional decoupling caps used.
Note: The tested O2 has LME49720 on input stage and 2 x NJM4556A as buffers. All 3 OPAMPs have heatsinks on top, glued with Arctic Alumina compound. There are 2 power plugs in front and back, 2 headphone jack: 3.5mm and 6.3mm on front and RCA outputs on the back. I'm also using original 9V/300mAh batteries and original 15V/500mAh power adapter.
P.S.: Looks like the way I arranged one of tantalums was not good, because I couldn't close the case, so I need to move it a few mm away from the edge of the PCB.
| |