frankl's stereo pages

[Overview] [Player] [RACE] [LoCo] [DRC]

Description of my stereo setup

The system is designed as a combination of a fairly minimalistic chain of hardware components and digital room- and system-correction software. The goal is to get optimal sound for one person in optimal listening position (the "sweet spot").

All the music is stored on computer hard disks in lossless formats. The collection was generated from a CD collection (which took about one year) and since several years I buy most music online (lossless formats only, at least CD quality or higher resolution).

The hardware signal chain

• Network storage and local hard disk and SSD holding the music files.
• A computer that converts the music files to a data stream that is sent via ethernet to the next computer for playback (see the "Software" section below). I use an Odroid XU with the fan substituted by a passive cooler "Zalman ZM-NBF47". This provides amazing computational power with an electric power consumption of less than 6 watts (5V DC). Operating system is Arch Linux.
• A computer that reads a playback stream from an ethernet connection and sends it to a DAC via USB. Here I use an Aria G25 board and soldered an SD-card holder, an ethernet jack, and data wires of a USB cable to it. Operating system is Debian Linux. Power consumption is below 1 watt.
• The DAC (digital-analogue-converter) is an Ayre QB-9 DSD. Its only input is USB, it supports asynchronous USB audio class 2, and accepts sample rates up to 192k. The input signal is regenerated via an Uptone Audio ISO-Regen.
• Both computers, an SSD connected via USB3 to the Odroid, and the ISO-Regen (both power inputs) get very clean power from three Uptone Audio UltraCap LPS-1.
• The symmetric output of the DAC goes directly into two Hypex NCore 400 mono power amp modules with SMPS600 power supply. (It is not true that all amps sound the same!)
• The output of the amps goes to DIY speakers, these have a single true fullrange driver Fostex Fe206E. The enclosure is a BIB, about 33x21x200cm, top is open and they need placement close to rear wall or in corners (picture). No subwoofer or tweeter needed! No attempt is made to improve the (not optimal) frequency response by an electronic circuit, this is better done by software.

Some tuning

• The mains supply for the whole setup comes via an AHP Klangmodul III with AHP 10x38mm fuse, a dedicated shielded power cable and an IsoTek GII power filter.
• Almost all cabling is now DIY. All DC connection from the three LPS-1's, connections of Hypex SMPS600 to Ncore modules, and the ethernet connection between the two mini-computers are made from Neotech UP-OCC solidcore copper wire. (Reports with pictures here and here.) Speakers are currently connected by solidcore CAT7 ethernet cable. Connections to and from the NCore modules are direct (without extra sockets and plugs). Extra wires connect the ground of the Ncore modules and of the DAC, and the minus-poles of the LPS-1 inputs with the earth terminal of the IsoTek mains supply.
• The listening room is about 5x3.2m and speakers are placed along a long wall, facing to the listening position. Sound reflections (above 300 Hz) are considerably reduced by spots of acoustic foam behind the listening position and on the ceiling.
• The fuses in the DAC and amp modules are substituted by Hifi-tuning Supreme³ fuses.

Software

During playback the music is manipulated by several software filters:

• Before playback music files are copied to an SSD with my bufhrt/improvefile program (discussion here). An album to be played is copied to RAM such that there is no disc access during playback.
• Upsampling to 192000 samples per second: All music is first upsampled in high quality using my resample_soxr program. Experiments have shown that the filters I want to apply give better results for upsampled music, and my DAC seems to be optimized for the higher sample rate.
• LoCo - a localisation correction, see here.
• RACE - cross talk cancelling, see here.
• DRC - room-, speaker- and system-correction, see here.
• To send the filtered music from my convolver computer to the audio computer which is connected to the DAC I use my bufhrt program and on the audio computer the music is received and played with my playhrt program. These programs do not change the content of the music stream but just improve the timing of writing the data to the sound device.
  All mentioned programs can be found here.