
WM8726 Production Data
w PD Rev 4.6 May 2008
11
DEVICE DESCRIPTION
GENERAL INTRODUCTION
The W M8726 is a high performance DAC designed for digital consumer audio applications. The
range of features ma ke it ideal ly suited for use in DVD players, AV recei vers and other consumer
audio equipment.
The W M8726 is a com plete 2-c hannel stereo audio digital -to-analogue converter, inc luding digital
interpolation filter, multi-bit sigma delta with dither, and switched capacitor multi-bit stereo DAC
and output smoothing filters. It is fully compatible and an ideal partner for a range of industry
standard microprocessors, controllers and DSPs. A novel multi bit sigma-delta DAC design is
used, utilising a 128x oversampling rate, to optimise signal to noise performance and offer
increased clock jitter tolerance. (In ‘high-rate’ operation, the oversam pling ratio is 64x for system
clock s of 128fs or 192fs)
Control of internal func tionalit y of the device i s provided by hardware control (pin programm ed) .
Operation using master clocks of 256fs, 384fs, 512fs or 768fs is provided, selection between
clock rates being automatically controlled. Sample rates (fs) from less than 8kHz to 96kHz are
allowed, provided the appropriat e system cl ock is input. Support is als o provided for up to 192kHz
using a master clock of 128fs or 192fs.
The audio data interf ace supports 16-bit right jus tifi ed or 16-24-bit I2S (Phil ips left j us ti fi ed, one bi t
delayed) int erface f orm ats. A DSP interf ace is al so suppor ted, enhanc ing t he interface options for
the user.
A single 3.0-5.5V supply may be used, the output amplitude scaling with absolute supply level.
Low supply volt age operation and low current cons umpt ion c ombined with the low pin count sm all
package m ake the WM8726 attrac ti ve for many cons umer applications.
The device is packaged in a small 14-pi n SOIC.
DAC CIRCUIT DESCRIPTION
The WM8726 DAC is designed to allow playback of 24-bit PCM audio or similar data with high
resolution and low noise and distortion. Sample rates up to 192kHz may be used, with much lower
sam ple rat es ac ceptable provided that the ratio of sample rat e ( LRCIN) t o mas ter c l oc k (MCLK) is
maintained at one of the required rates.
The two DACs on the WM8726 are implemented using sigma-delta oversampled conversion
techniques . T hese requi re that t he PC M sampl es are digi tally fil tered and interpolated to generat e
a set of sam ples at a m uch higher rate than the up to 192kHz input rate. This sample stream is
then digitally modulated to generate a digital pulse stream that is then converted to analogue
signals in a s witched c apac i tor DA C. The advantage of this tec hnique is that t he DAC is linearised
using noise shaping techniques, allowing the 24-bit resolution to be met using non-critical
analogue com ponents. A further advantage is that the high s ampl e rate at the DAC output means
that smoot hing filt ers on the output of the DAC need only have fairly crude charact eristi cs in order
to rem ove the c haract eris ti c steps, or i m ages on the output of the DAC. To ensure that generati on
of tones characteristic to sigma-delta convertors is not a problem, dithering is used in the digital
modulator along with a higher order modulat or. The multi -bit s wit ched c ap ac it or tec hnique us ed i n
the DAC reduces sensitiv ity to cl ock jitt er, and dramat ically reduces out of band nois e compared
to switched c urrent or single bit techniques us ed in other im pl em ent ati ons .
The voltage on the CAP pin is used as the ref erence for the DACs. Therefore the amplit ude of the
signals at the DAC outputs will scal e with the ampli tude of the voltage at the CAP pin. An external
reference c ould be used t o drive int o the CA P pi n if desired, with a value typically of about m idrail
ideal for optimum perf ormance.
The outputs of the 2 DACs are buffered out of the device by buffer amplifiers. These amplifiers
will s ource load cur rents of several mA and si nk current up t o 1.5m A allowing s ignificant loads t o
be driven. The output source is active and the sink is Class A, i.e. fixed value, so greater loads
might be driven if an external ‘pull-down’ resis to r is c onnected at the output.