[ketcham]

Greetings!

After being a 35-year audio hobbyist I founded Kevalin Audio with the goal to help promote the already established Allnic Electronics and introduce to North America two other Korean based companies, Allnic ZL cable line and Waversa Systems Inc.

Over the years I have enjoyed Audio Circle and gain much insight from the contributors here.  Three industry participant who I adore and respect are Steve Nugent of Empirical Audio, Lou Hinkley of Daedalus Speakers and Dave of Zenwave -- all who produce amazing products which I have used over the years and enjoy today.

This baby industry thread is not with the intention of promoting product, specials or create sales.  Instead, I wish to establish a more central location providing information on the current and newly released products.  All three manufacturers are on the cusp of releasing a series of new and interesting products.  I believe Audiocirle would be the best forum to provide a short summary and discussion when items are released.  This thread would be a great place to discuss manufacturer related questions.

If there is any opposition, please PM me and these concerns will be considered.

Sincerely,

John Ketcham
Kevalin Audio LLC
Portland, OR, USA

[ketcham]

Please note the following information is not intended to be promotional with the intention of sales. Instead, informational only and learn if other manufacturers also have a similar or unique approach to digital processing. 

All information on this thread is a collaborative effort by the following dealers worldwide:

www.kevalinaudio.com - North America
https://www.audiotra.de/waversa - Most of Europe
www.waversasystems.com - Japan
www.hificlub.co.kr - Korea
http://www.waversa.com.cn/col.jsp?id=117 - China

With over 30 years in both valve base circuit design and IT technologies, Collin Shin founded Waversa Systems in 2012.  Over the next eight years, he integrated technology used in telecommunications, military, medical and security applications and applied them to novel approaches to the various aspects of digital audio design with most software and hardware, including chip design and manufacturing, with cutting edge in house technologies used in both Waversa products and partner companies.

Digital technology is rapidly evolving and Waversa is constantly changing and evolving, pushing the envelope in unique ways.  As such, each post will cover one novel approach with periodic updates with the goal to provide a central go-to informational reference thread.

WAP (Waversa Audio Processor)

Hardware method with very low jitter rate and accurate timing
A proprietary solution from digital input to processor
Dual WAP captures all signals on the optimized route
Centralized clock management in the signal path
Technology to minimize Quantization Error
Processes huge information up to 32Bit 1.5MHz
Highest level signal restoration algorithm by applying medical technology

Generally, upsampling uses an FIR filter (Finite Impulse Response, fixed-form impulse response filter). Interpolation (interpolation) is performed during the upsampling process to correct the sound quality. FIR Interpolation is a mathematical calculation that estimates and fills between digital points.

In the case of the Saber ES90x8 series, such FIR filters are basically built-in and easily implemented by DAC manufacturers.  WaversaSystems uses a proprietary WAP instead of the basic FIR filter included in the DA chip. Waversa Audio Processor (WAP) is designed in house with a multi-stage structure including a number of proprietary FIR filters and IIR filters (Infinite Impulse Response Filter) and performs high-resolution upsampling. This high-resolution upsampled signal is processed through thousands of iterative calculations using state-of-the-art source estimation algorithms to reduce digital errors and create a waveform that is as identical as possible to the original analog signal. This is the role of WAP.





The software processing method has elements that cause errors because the processing is sequential and complicated. On the other hand, in the hardware processing method, since signal processing is performed one at the same time, there is no room for jitter, and the timing of the audio signal is exactly the same.

WaversaSystems digital processors are processed in hardware-based digital processing, so they are processed simultaneously in a simple, optimized path resulting in fewer errors, and the timing of the audio signal is handled more accurately.

Dual WAP captures all signals In order to dramatically increase processing speed and efficiently manage signal processing. The first WAP (P1) processes all input signals, and the second WAP (P2) receives signals processed by P1 on I2S and decodes signals with clocks according to sampling rates. In this way, by assigning a unique role to each digital processor, signal interference is minimized, more sophisticated clock management is performed, and sound quality benefits from accurate signals that do not compromise digital signal integrity. The heart of WAP lies in signal restoration. 32Bit 1.5 MHz, processing huge information.





The newly developed WAP has greatly increased the amount of data that can be processed. This new WAP has succeeded its predecessor (24Bit / 368kHz) with 32Bit 1.5MHz data processing. In comparison,16Bit, 65,532 digital signal processes are performed, and in 32Bit, there are 4 billion digital signal processes, and the sophistication of signal processing is dramatically increased. This has made it possible to generate digital signals that are closer to analog, and the subsequent digital restoration technology has become more accurate.






To put it simply, the improved internal processing speed of 32Bit 1.5MHz is similar to upgrading from DVD quality to 4K quality. This dramatically increases the detail and dynamic range that can be expressed.

The higher the WAP level, the more sophisticated the processing per unit time.  This improves the sound image and detail, overall producing a more natural sound. The tone of the instruments is also flexible, organic and easy to hear.  Each Waversa product has different processing capabilities that when linked creates a cumulative benefit.

[FullRangeMan]

This 32Bit/1.5MHz processor are really great, it could be useful a upsampler unit to enhance 16/44 Red Book CD playback.

[Freo-1]

This 32Bit/1.5MHz processor are really great, it could be useful a upsampler unit to enhance 16/44 Red Book CD playback.

It sure could.  I'm using the Sony DSEE-HX function for 2 channel playback, and it definitely works quite well.  The DSEE-HX works on on PCM up to 96 KHz, and DSD/PCM up to 176.4 KHz.   

[ketcham]

This chip also allows for two other processes that are unique and quite innovative, both were released in as many months:




WAP / X (Waversa Audio Processor Extension) that reproduces the sound of vacuum tubes

The characteristics of a vacuum tube amplifier that has been deeply loved for its warm tone are reproduced using digital technology.  These overtone characteristics of WAP / X were developed focusing on the 1940's WE300B.

The technology adds a sense of overtones in WAP / X and WAP, realizing the sound of a transparent and warm vacuum tube, digitally reproduced technology designed to be upgradeable in future firmware updates.

High-end audio processing is a fairly complicated process involving a huge amount of data in real-time.  Waversa addresses this in two parts.  The most recent breakthrough is to task the new generation field-programmable gate array, which unknown to most of us, including myself, is an exceedingly complex fast logic circuit that has proven to handle these resource-intensive algorithms.  By tasking the FPGA to take on the overtone processing algorithm [WAP/X], this processing can be done accurately to many harmonics in real-time.   

Some say that analog sounds are bright and warm, while digital sounds are cool and crisp.  It is also thought that harmonics are masked by jitter and electrical noise inherent in digital equipment.  All DAC manufacturers have understandably focused on minimizing these issues.  Waversa over the years was able to characterize the harmonics of a classic 1942 Western Electric 300b and digitally replicate using a hardware-based algorithm integrated into both the new generation FPGA and Waversa Audio Processor (WAP).  As a result, you can feel warm overtones without the vacuum tube.  If you want to hear higher harmonics, the only way to get them is virtual because there is a limit no matter how good a vacuum tube is.  Currently, the end-user can select this type of processing or bypass it.  In future firmware updates, the degree of harmonics will also be adjustable.

The second process is Native DSD:

Both USB and Ethernet use the same mechanism to transmit data internally and the USB Audio path is fundamentally a PCM path.  DSD over PCM (DoP) uses a repeating 8-bit marker to indicate DSD processing of the digital stream.  This standardization was established with 24-bit capable processing chips, whereby the 24-bit data stream consisted of an 8-bit marker, 16-bit information, and when applied to a modern 32-bit processor, the last 8-bit segment is ignored.  With the advent of 32-bit processing, DoP is limited to only half of the capable bandwidth.    Native DSD uses a non-standard PCM path in USB at a 32-bit stream without the need for a marker and the full 32-bit stream is dedicated to digital music. Therefore, DSD64 can be transmitted in PCM 88.2kHz format, DSD128 can be transmitted at 176.4kHz, and DSD256 can be transmitted at 352.8kHz.  A PCM352.8kHz compatible system that supports Native DSD can stream DSD256 in its native format.

[FullRangeMan]

Great feature, another useful audiophile attribute to a click of a button

[ketcham]

What I appreciate about Waversa products overall is how often firmware updates happen with notable changes in sonic qualities and improved versatility.  The firmware update process is very easy for the end-user.  The WAP / X will also undergo further development with no further FPGA reprogramming, which is a somewhat more involved process. We were proactive in updating all known Waversa Dacs in North America.  If anyone owns a unit that was sourced elsewhere please reach out to me by PM to discuss this update.

[ketcham]

The last technical aspect of the Waversa Audio Processor I would like to discuss:

Channel separation technology for sound range and (WMLET – Waversa Multi-Layer Energy Transfer)

Waversa Systems has introduced two unique technologies that make optimal use of the ES9038PRO’s 32-bit 8-channel Digital to Analog conversion capabilities.  One is a source estimation algorithm that operates at 32Bit 1.5MHz. The other is Range Separation Technology (WMLET).







Rather than using the commercially available ES9038PRO chip as designed, This D/A conversion is the only function of this chip in this setting.  Instead, the circuits of the input stage and the oversampling filter stage are bypassed and a fundamentally different processing algorithm is applied in the in-house designed Waversa Audio Processor. This data stream is then passed on to the ES9038PRO by another key processing technique:  Range channel separation technology (WMLET-Waversa Multi-Layer Energy Transfer)

As the name implies, the auditory frequency band is divided into high, mid-high, mid-low, and low bands at the digital stage for each of the left and right channels then passed through all eight channels of ES9038Pro for processing.  Instead of 8 channels, in conventional signal processing only the left and right channels are processed together and more energy is concentrated in the mid-band. To prevent this, WMLET regulates the energy distribution evenly across four channels per frequency band and creates a band balance that cannot be felt with conventional DACs.  Distortion caused by energy bias is reduced revealing subtle musical elements that are otherwise unrealized.  This approach to digital processing is unique only to Waversa products.

The energy concentrated in the mid-band is evenly distributed, and the band balance is excellent.
The resolution increases with each band, taking advantage of the nuances of band separation creating a more natural presentation of music.
Implementation of a wider music signal dynamic range and broadband frequency reproduction.
A natural balance of the sound can be maintained, and the originality of the tone is maintained.

[ketcham]

In addition to hardware-based digital processing and fundamentally unique interpolation algorithms, Waversa Systems is developing proprietary codecs and protocols.  One Ethernet protocol that has been established in their product line I wish to discuss here.  Like ROON RAAT, I anticipate once proven beneficial these codecs and protocols will be licensed to other manufacturers who may have an interest. 
 
The biggest problem with streaming widely used networks using DLNA is that the standard does not consider continuous audio streaming at all, it is just a standard for packet-based data transmission allowing multitasking Ethernet switches to function seamlessly while maintaining accuracy. In this transmission method, noise is generated during the buffering process during signal transmission resulting in extremely poor sound quality. WNDR is a more advanced proprietary protocol developed by WaversaSystems to solve these problems. It has been developed to minimize delays in communication and provide excellent noise immunity by design. In addition, the WNDR method is a dedicated protocol, so there is no need for a conversion process between protocols, and thus improving sound quality.

WNDR (Waversa Network Direct Rendering) dedicated protocol

Transmission scheme optimized for continuous signal processing without buffering
Protocol with strong immunity to noise
Protocol without conversion process by dedicated route
Increased WAP level due to interconnection (improved sound quality)

WNDR is a protocol dedicated to the transfer of pure music with little and if selected no buffering thus creating a special advantage whereby there is almost no noise during signal transmission and eliminating any notable sound quality distortion.

Noise generation by network transmission:







 

When reading buffered data as shown in the figure, the process of reading and transmitting data with a renderer board occurs intermittently, which causes electrical noise. The stable transfer of continuous information without buffering will be the core technology of the future, and since media integration tends to be parallel to communication, stable clock synchronization without buffering will be achieved. The technology that does this is WNDR, which is a great sound quality advantage.


You can tell if there is buffering by unplugging the network cable.  In the case of DLNA, music is played continuously for 30 seconds or more, and in the case of ROON RAAT, music is played for several seconds. This means that buffered music data is being played. However, in the case of WNDR, the music is interrupted almost immediately when the network cable is unplugged. This is because there is no buffering.




WNDR, which operates with WaversaSystems DAC, is the only product in the world that operates with a proprietary music data transfer protocol. WNDR produces a synergistic effect when all devices are integrated, and combines with WAP to achieve even better sound quality. When WCORE, WRouter, and WDAC3C are all connected by WNDR as shown in the figure, they will have clear sound quality benefits. In addition to that, it has been raised to the WAP level, achieving higher-end sound quality with higher dimensions.  Below is a summary of one customer review on WNDR:

“Given what I’m used to, the greatest degree of improvement, though perhaps for some what would be the most subtle, is in terms of the realism, as I experience it, of the tone and timbre of voices and instruments. I can’t get over it; it’s a revelation, and the effect is greatest using Waversa’s alternative to Roon Ready, WNDR. You need to have at least one other Waversa device besides the DAC3c to use WNDR. I have the Wcore set up to see the Wrouter as its WNDR target, and the Wrouter seeing the DAC3c as its WNDR target. There is also the ability to select from 5 settings for the amount of buffering the Wcore and the Wrouter use in WNDR. The Waversa default is “4”, with the most buffering at “5” and the least at “1”. I have both set at “2”. If your network is unstable, it’s suggested you use the higher settings. I lost signal a couple of times on the “1” setting, though it came back quickly. There’s been no dropout with “2”. To my ears, there appear to be small differences in the audibility of low-level detail and “air” with going up and down among these settings, with transparency and clarity improving as buffering decreases. I did note that, in my system, the 4 and 5 defaults provided a little less of that “air” and, perhaps as a result, more of a sense of “warmth” and solidity. My system has plenty of that (all tubes, including the Allnic L-8000 DHT preamp, and/or long cable runs), so I’m liking a bit more transparency and air. “2” works best for me; others might be looking for more warmth and solidity.

I’ve gone back and forth between using the DAC3c as a Roon Ready DAC and as a WNDR processor (they appear as two different players – audio zones - in Roon on the iPad). In my system and according to my druthers, the WNDR is superior. But this is a place where one’s preference could be determined by personal taste and system factors, including one’s hearing, and even the physical space in which you listen. In my room, though with a more prominent top end, Roon Ready seems to me less organic, somewhat “disarticulated” by comparison; sonic images are slightly less lifelike, high frequencies are not as clear (cymbals have more “sizzle” but less accurate timbre, for example), and bass is as deep as it is with WNDR but not as articulate and varied, meaning that with WNDR some bass “rolls” along, under and through the floor; some swells and washes like a wave; some blossoms to fill the whole space; some bounces and thrums; some digs right into your body cavity and bones; and sometimes two or more of these at once, depending on the recording/instrument(s).”