I believe what's needed is a differential measurement of dynamic noise profile under dynamic load not just the noise measurement at fixed frequencies usually given for voltage regulators
+1
I'm not a guru in electronics, rather an enthusiast.
Several times I've discussed power supply for DSC2 with Pavel. His opinion - built-in in LDO ADP7118 (2.5.2 / 2.6.2) with a futher filtering on the DSC board is a very good approach.
I've experimented with his older DSC2.5 version (LT3042, LT3045, TPS7A4700, Salas Reflector D) w/o any substantial improvements (IMO).
Honestly I am very interested in your approach. I'll try.
1. Connecting ultracaps to DSC2 directly (built in LDO disabled)
2. Connecting ultracaps to DSC2 with some resistor in series to increase impedance up (built in LDO disabled).
Built in LDO on DSC 2.5.2/2.6.2 can be disabled easily by shorting two jumpers.
I think no substantial improvements might be because of couple of reasons:
1. All those regs are based on some better/worse current feedback design
2. Overal perfromance might be dominated by LC filter on DSC2
3. In audio all counts and final perferomance depands on the whole audio chain. The system is as good as it worst part.
Sure thing. All you say is a valid point. I am also curious about measurements.
Got mine "running" at least according to the scope.
Some modifications here:
5v to 3.3v regulator on the DSC2 board.
Mount the 3pin headers so I can stack the board and detach DAC board and output board.
Using Cronus to reclock inputs from Amanero. They are connected via u.fl cables through a header I made.
Mute pin is taken directly from Amanero but isolated by a TLP222A.
DSDON is not connected.
To-do:
Manage heat from Reflector-D. the 1.5Ohm resistor is getting to hot to run long.
Finish output board.
Some modifications here:
5v to 3.3v regulator on the DSC2 board.
Mount the 3pin headers so I can stack the board and detach DAC board and output board.
Using Cronus to reclock inputs from Amanero. They are connected via u.fl cables through a header I made.
Mute pin is taken directly from Amanero but isolated by a TLP222A.
DSDON is not connected.
To-do:
Manage heat from Reflector-D. the 1.5Ohm resistor is getting to hot to run long.
Finish output board.
Attachments
How are you going to cope with the cooling? Big radiator?
I've tried active cooling with one 12cm Noctua fan and got stable 38 - 40 C (3 Reflector D, 2Ohm, 2Ohm, 1Ohm). But there were small chasis vibrations that I did not like.
There are two Reflector-Ds in the picture. One is supplying cronus with 1.5Ohm resistor. Another is supplying both channel of DSC with 1Ohm.
Right now I'm measuring 1.9v across the 1.5Ohm resistor. That's about 2.5W I need to get rid of. Not a big deal, I can connect it to the base plate like what I did for the 1Ohm one.
For the 1Ohm one, transistors are getting too hot too. So they are mounted on the base plate. For the 1.5Ohm one it's not that bad. They will stay with on board heatsink.
Last edited:
u.fl adapters...
Hi, are your u.fl adapters for Cronus, etc, available from Oshpark? I would be interested.
Hi, are your u.fl adapters for Cronus, etc, available from Oshpark? I would be interested.
I'd say it's more for DSC.
It exposes D1,D2 and CLK. Use TLP222A to isolate DSD_ON and MUTE.
I did some tweaks after the first version. I will publish and share the new one later on here and via oshpark.
Has anyone ever tried building a DSC-1/2 using super high quality resistors? Say Z-foil? I see Denafrips appears to use .005% thin films on their top model, and there are claims as to sound quality benefits. As the signal is basically analog at this point, i would think resistor sound quality would matter.
Just thinking out loud here and wondering. I know somewhere on this thread I did see some measurements which appeared to show higher precision resistors giving better measured performance... Thoughts?
Just thinking out loud here and wondering. I know somewhere on this thread I did see some measurements which appeared to show higher precision resistors giving better measured performance... Thoughts?
High precision resistors are required in the DAC R2R type.
DSC uses the Rectangular Window FIR Filters. Therefore, there is no need for very high precision resistors.
According to my measurements, there is no difference between 1% and 0.1% resistors.
Last edited:
How do you handle grounding among your PSUs? I'm guessing you are using 1 Reflector-D per channel and 1 for the digital source?
Yes...
Pavel, I am aware that super precision resistors are not required for DSC-1 technical performance vs. PCM/R2R DACs... I am curious though as to subjective sound quality. At some point someone will build a DSC style DAC and use Z-foils just to see what the sound quality might be.
Just curious...
Pavel, I am aware that super precision resistors are not required for DSC-1 technical performance vs. PCM/R2R DACs... I am curious though as to subjective sound quality. At some point someone will build a DSC style DAC and use Z-foils just to see what the sound quality might be.
Just curious...
I'am using Reflector D for another DAC (Soekris dam1121 - Left, Right output line power, Didital line power), not for DSC. Reflector's earth connected to dam1121 earth. All of them are placed on nylon spacers.
But I have 50Hz harmonics with Reflector and can't eliminate them (have look at http://www.diyaudio.com/forums/power-supplies/261031-reflektor-builds-122.html#post5146329)
I think I too have it: Imgur: The magic of the Internet
This is a plot from years ago with DSC1 with Reflector-D, although it's far superior than LM1086: Stock DSC1 - Imgur
I remember when I was supplying Reflector-D with a bench top PSU I no longer have noise from AC lines. When my DSC2 is running and stable, I will see if the same applies for DSC2.