I had the 4x iv and 2x lpf opamp changed to 1612. It's a better dac for the changes. HF resolution is improved, the soundstage much more precisely delineated. Its just better all round.
Its now better than my rpi, fifopi, lifepo psu, 2x 9038qm dac, biselik transformer setup. It wasn't quite as good stock. Imo the smsl400 is just a tiny bit better than the modded a18 on balanced outs.
YMMV...
Its now better than my rpi, fifopi, lifepo psu, 2x 9038qm dac, biselik transformer setup. It wasn't quite as good stock. Imo the smsl400 is just a tiny bit better than the modded a18 on balanced outs.
YMMV...
Despite huuuge noise after 1 opamp, why is that schematic without output capacitors. Smsl400 and loxije d50, both have 4 green muses on xlr output. I measured 2,5v on output before capacitors and 0 after summing opamp for RCA output.
Why is feedback resistor connected to dac not to noinverted opamp input like filter capacitor? Blue values are in smsl and loxije dac. Did anybody tried to change resistor even lover ... to 100r to see how is level of noise on opamp output dropping? or adding more capacitance on feedback?
Why is feedback resistor connected to dac not to noinverted opamp input like filter capacitor? Blue values are in smsl and loxije dac. Did anybody tried to change resistor even lover ... to 100r to see how is level of noise on opamp output dropping? or adding more capacitance on feedback?
Attachments
There are only the passive filter output caps as shown on the schematic used on the AK4499 eval board. However, they used through hole resistors and SMD film caps for that. Maybe a little extra inductance while still effective capacitance at RF that way. Also, the film caps may be a little better for audio verses some alternatives.
That's normal since Vref is 5v (Vref being the same as what ESS calls AVCC). The dac designer has to choose how to remove the DC offset.
People have speculated about that. So far as I know AKM never said why. Maybe its to compensate for dac output resistance. Rohm says they used gold bond wires for the outputs of their dac. Don't know if it matters or not.
Topping ran D90 opamp rail voltages to +-11v. Sounded better to me that way when I tried it with the eval board. Also, I run the PCM at 22mA output rather rather than the higher default value. Thought it sounded a bit better and it matches the DSD level too that way. Could be tweaking the opamp resistor and or cap a little could help too, never got around to trying it.
Regarding the output noise, its probably mostly frequency shaped switching noise rather than feedback resistor noise. Up to the dac designer to decide how to deal with it. A common approach is to use an opamp multi-feedback differential summing stage to remove DC and do some more noise filtering at the same time. Handy that it all gets taken care of in one step. IMHO the downside is that SQ suffers. Topping D90 used differential summing, but no additional filtering after the passive filters following the I/V stage. Other people used caps to remove the DC. Choose your poison, I guess
Will say something else about the AK4499 eval board that I don't think I mentioned in the forum before. The Jung type regulator AKM used for Vref didn't sound all that great. As a quick fix while I looked at other dac issues, I found that an Allo Shanti dual 5v power supply sounded better to me for Vref use. Did have to figure out how to drop the output voltage to exactly 5v since Allo sets them a bit on the high side. Not too hard to do, PM if interested in how. Later found that the Shanti 3A output sounded better than the 1A rated output when used to supply Vref. Maybe not too surprising if a slightly different regulator design was used on the 1A output verses the 3A one, since Vref has a direct multiplicative effect on the analog output of the dac chip. After that I used two Shanti supplies for awhile, only the 3A outputs, until I designed my own discrete regulator. That one was the best found so far, but I keep thinking about how it might be done better.
Did you set Vref to 5 v ...did you test ak4499 with 4.75 or 5,25...5,5v?
SMSL and Loxije have 10R on power line and 10R to gnd so the voltage drop to 4,5V. I think this is a little to low. I bypased 10R on power line and left 10r to gnd. Because of 4400uf bas is tooo long and soft and undynamic. Whole sound was too slow, too polite. Now it is much faster. Any suggestion? Tomorow i will measure and compare SMSL with original values vs moded Loxyie. Right of the box SMSL is too tinn light sounded, to shy bass but clean ... Loxije is soft and rounded. Both don*t come close to moded Es9038pro in terms of dynamics and musicality. I don*t miss any detail on both AK4499 dacs but it is like i will sit in 10th row es9038pro is third and Audio Note dac4 is first row.
SMSL and Loxije have 10R on power line and 10R to gnd so the voltage drop to 4,5V. I think this is a little to low. I bypased 10R on power line and left 10r to gnd. Because of 4400uf bas is tooo long and soft and undynamic. Whole sound was too slow, too polite. Now it is much faster. Any suggestion? Tomorow i will measure and compare SMSL with original values vs moded Loxyie. Right of the box SMSL is too tinn light sounded, to shy bass but clean ... Loxije is soft and rounded. Both don*t come close to moded Es9038pro in terms of dynamics and musicality. I don*t miss any detail on both AK4499 dacs but it is like i will sit in 10th row es9038pro is third and Audio Note dac4 is first row.
Yes. Doesn't seem to make much difference other than changing the dac output volume level a little.
With the optional extra 10 or 20 ohms of resistance in the Vref circuit, the sound seems to be more dependent on the sound of the caps at the dac chip and less dependent on the sound of the voltage regulator. However, that isolating effect of the resistors is limited. As far as caps to use, one combination I kind of like are Panasonic FM or FC electrolytics in parallel with, say, maybe .1uf to .01uf polypropylene and foil film caps. There seemed to be a sweet spot for the electrolytic cap values, maybe around 470uf.
Ultimately, I got rid of both 10-ohm resistors and connected a custom discrete Vref regulator right at the dac chip. There are still some output filter caps though. The Vref power supply runs from its own transformer so the regulator circuitry is tied to ground only at the dac chip. That works well, but maybe it could work okay without the separate power transformer; that's just how it happened to work out when I tried it.
Other things that affected dac sound were changes and developments in the output stage area, and some things with clocking. PCM is also normally converted to DSD256.
Last edited: