Презентация на тему DAVE. Technology Presentation

to Analogue Veritas in Extremis
DAVE’s development was centered

upon a question: why was Hugo so musical?
Where was Hugo’s sound quality performance coming from in technical terms?
DAVE has an FPGA ten times the capacity of Hugo
This gave opportunities to further improve performance:
Improved time domain (transient timing accuracy)
Improved noise-shaper performance
DAVE has much more advanced analogue electronics

is to reproduce the un-sampled continuous analogue waveform from

sampled digital data
Conventional DACs do a poor job of reproducing the original continuous analogue signal, with timing errors on transients
Increasing tap-length of FIR filters gives better time domain accuracy in terms of timing of transients
The ear/brain is extremely sensitive to very small timing errors
Timing accuracy upsets the perception of the start and stopping of notes
It also degrades the ability to perceive instrument timbre and power
It degrades soundstage precision

The WTA algorithm was subjectively optimised and improved to

suit the longer tap lengths
WTA filtering is now up to 256 FS – no other DAC has ever FIR filtered at such a high rate
DAVE has massively parallel processing with 166 DSP cores
Further advanced filtering to 2048 FS
This means DAVE more accurately retrieves the original continuous analogue un-sampled signal

FS data and converts to 5 bits
It also

creates the 20-element Pulse Array outputs, so is the heart of the DAC
Initially, Hugo-standard noise-shapers were employed
But increased FPGA capacity and 20-element operation allows better performance
Over 3 months of continuous listening tests and redesign pushing to improve sound stage depth perception – I (Rob) would not stop until performance would no longer increase
Constantly pushing for better sound stage depth-perception led to world’s most advanced and complex noise-shaper
It employs 17th order noise-shaping, with a total of 46 integrators; the design of the noise shaper alone would not fit into Hugo’s FPGA.

analogue noise-shaper for output stage – this gives ultra-high

HF linearity and no increase in distortion with difficult loads
Still employs single global feedback path with equivalent of simple 2 resistors and two polypropylene capacitors in direct signal path
Ultra-low-noise sub-milli-ohm power planes for Pulse Array element flip-flops
Digital DC servo
Headphone drive 6V and 0.5A RMS OP capability

the reference voltage dBA, measured using APx 555
THD

and noise at 5V RMS 1kHz -124 dBA A wt
THD and noise at 2.5V RMS 1kHz - 127 dBA A wt (-124 dBA into 33 ohms)
THD 1kHz 2.5V RMS 0.000015%
Dynamic Range at -60 dBFS 1kHz -127 dBA A wt
No measurable noise floor modulation, no anharmonic distortion
Analogue distortion characteristic – no distortion for small signals

signal zero noise floor modulation. Noise floor at -178

dBA, distortion peaks -150 dBA, no anharmonic distortion.

at -152 dB
I suspect that the APX555 audio

analyzer is adding more distortion than DAVE
Elevated noise floor is the APX555

similar left/right channels are

are Extremely low noise

Redefines DAC measured performance
In my view, it sets new

DAC sound quality and musicality performance