HOLO Audio May Fully Discrete R2R Decoder DAC

May is divided into L1, L2 and L3.

The power capacitor of L1 is Mundorf's gold and silver oil immersed capacitor. The USB module is a "standard" module without a remote control

The power capacitor of L2 is a customized capacitor, and the USB module is an "enhanced" module with a remote controller. L2 in the photo

On the basis of L2, the DAC module of L3 is manually selected, and the capacitors of two power supply parts are upgraded. The internal power cord is made of pure silver, and the fuse is also upgraded.

The characteristics of May are as follows

1. The new generation of linear compensation technology better solves the accuracy error caused by resistance, switching elements and PCB, making the equivalent accuracy after compensation reach 0.00005%

2. The new generation of de glitch technology can provide full range de glitch without increasing the noise floor and other adverse effects

3. Based on this new generation of technology, "May" can provide>115db SINAD and>130db dynamic range, which represents the performance limit of today's cutting-edge R2R architecture DAC

4. The PLL+FIFO technology with extreme performance is adopted to provide the ability of restraining Jitter with 0.1Hz and third-order low-pass. High performance femtosecond VCXO is used as PLL clock source. Under the premise of almost immunizing the front end Jitter, it can also lock up up to 1.5us-2us@1KHz Signal with high Jitter of

5. The power supply part adopts independent chassis, and the power supply of the left and right channels is completely independent to provide better channel separation and sound field positioning

6. USB and I2S support the sampling rate up to DSD1024 and PCM1.536M

7. The USB interface uses self-developed firmware, with ultra-low latency, highly reliable data transmission, and performance 2-4 times higher than the official firmware

8. Two groups of independent HDMI-I2S input interfaces are provided. Each group of I2S has four independent receiving circuits, so that the clock signal of I2S is subject to lower interference. In addition, each group of I2S can be configured with line sequence, making it compatible with most HDMI-I2S digital devices on the market

May is a discrete resistor network decoder, commonly known as R2R. Unlike ordinary decoders that use chips to complete, May uses FPGA to control the switching of discrete resistor networks to convert digital signals to analog signals. This method can realize the algorithm of the designer in a more flexible way, thus breaking the limit of the chip.

The conversion of discrete resistor networks is very difficult to control, and the difference is thousands of miles. Therefore, if only for the purpose of separation, the performance is not as good as the chip. For example, the SINAD of the R2R chip PCM1704K (the ratio of signal to distortion plus noise, also known as THD+N) can reach 9xdb. However, many discrete R2R decoders on the market cannot achieve the performance of PCM1704K, even some have only 6xdb. This kind of discrete R2R decoding usually has a natural and smooth voice color unique to the discrete R2R, but its analysis, sound field, separation and tri frequency density cannot reach the ideal, which is called "paste"

As a well-known Buddhist brand in the industry, we will never take this as a marketing tool for separation. When our first generation discrete resistor network decoder "Spring" was introduced four years ago, its SINAD could reach about 105 db. As a result, many users will find that Spring is not only non paste, but also quite transparent in tone and has a large amount of information. The "high-quality" R2R experience can only be experienced on Spring.

The new generation of may has increased SINAD to more than 115 db and DNR (dynamic range) to more than 130 db. This is the highest performance of R2R audio decoder in the world. If you have a full understanding of audio decoder technology, you will know what this number means. Therefore, we are proud to announce that "may" represents the highest level in the field of resistive network audio decoder.

Each channel of may has an independent DAC module. Each module has 6 groups of networks on the front and the same 6 groups of networks on the back. A total of 12 groups form a fully balanced structure. Among them, four groups of networks constitute DSD decoding, which we call "vector stepping resistance network", four groups of networks constitute R2R structure for PCM decoding, and four groups of networks are used for linear compensation of R2R network

Therefore, the DSD decoding of may does not need to be converted to PCM, but is directly performed by the "vector stepping resistor network". This technology was implemented on Quanquan four years ago. At that time, no one could understand such black technology, because conventional R2R could not be used for DSD decoding. This is also the original creation of Spring, and now some other manufacturers have learned from it. However, our algorithm of "vector stepping resistance network" actually has some tricks that are difficult to copy. Basically, except for the inherent noise base of its DSD, no harmonic distortion can be observed. Therefore, those manufacturers who learn from our practices can only copy a shape but can't copy to God.

PCM decoding of may is completed by four groups of R2R resistor networks. In addition, there are another 4 groups of resistor networks to compensate the main R2R network linearly. For example, an ideal value is 32768. Due to resistance error, the actual value is 32700. The auxiliary resistance network generates a 68, which is added with the main resistance network to form the ideal value of 32768. Through this technology, we can make the equivalent resistance precision after compensation reach an incredible 0.00005%.

We must know that ready-made resistors cannot achieve 0.00005% accuracy. This is why so many discrete R2R decoders cannot achieve high performance. Because precision determines linearity, and linearity determines harmonic distortion, IMD, and almost all AC performance. This is why the performance of so many discrete R2R decoders measured is only 6xdb, 7xdb, and at most 8xdb. However, PCM1704K can reach 9xdb, because PCM1704K gets higher precision by trimming the resistance with laser. Spring reached 105 db because of linear compensation technology. However, the reason why May has reached 115db is the new generation of linear compensation technology.

Another black technology of may is its powerful PLL+FIFO technology. We all know the importance of jitter. May adopts a conventional technology, PLL (phase-locked loop). However, this is not an ordinary PLL, but an extreme performance, which is close to the perfect PLL. This PLL provides 0.1Hz, third-order low-pass jitter suppression capability, and uses femtosecond VCXO as the clock source of the PLL. In the test, we analog injected a 1.5us, 1KHz jitter into the digital input. Please note that the unit is us, not ps, ns. 1.5us=150000ns=1500000ps=15000000000fs, which is an exaggerated jitter. In the actual measurement, we can see that it is completely removed. It has no effect on the output.