MB86060 16-bit 100MSa/s Interpolating DAC

2.1: What is MB86060 used for?

2.2: Why use Interpolation Filtering?

2.3: Should Noise Shaping be On or Off?

2.4: Programmable Clock Multiplier

2.5: Programmable Dither

2.6: What Drives the Interpolating DAC?

2.1: What is MB86060 used for?

MB86060 is designed as a simple solution to the problems of analog signal synthesis. In general, a better quality analog output can be generated by a faster DAC rate. This is largely due to smaller converter step sizes, but ultimately speed limited. However, driving a DAC faster can place a significant load on the digital pre-processing. This is where the MB86060's interpolation filtering provides a systems advantage by only having to supply data at a rate FDATA, while driving the DAC core at a rate x2 or x4 FDATA. If the system were capable of providing data at 2.FDATA or 4.FDATA then approximately the same level of performance would be achieved, but at the expense of faster pre-processing.

2.2: Why use Interpolation Filtering?

First and foremost, as described under 'What is MB86060 used for?', interpolation filtering enables data to be supplied at a convenient rate while benefiting from the DAC core running at x2 or x4 to achieve improved performance. In addition low-pass digital filtering is used to suppress image products significantly reducing the requirements on the analog output reconstruction filter. Finally the effect due to the inherent Sin.x/x roll-off caused by the converter output stage is significantly reduced from -4 to -0.22dB at the data input Nyquist rate. Note that there are images of the generated band around the DAC clock frequency. Trying to implement this degree of interpolation filtering using any kind of general purpose DSP prior to another DAC would be extremely inefficient and, in all practicalities, virtually impossible since the interpolation filters would require the equivalent of several Giga-operations per second processing.

2.3: Should Noise Shaping be On or Off?

MB86060 features a control line to select noise shaping on or off. In all cases where interpolation is used this would normally be set to on. The reason for this is that the digital pre-processing [input data and interpolation filters] uses 16 bit data while the DAC core only accepts 12 bits. Not using noise shaping would cause the 4 least significant bits to be truncated and results in significantly increased spurious products. However noise shaping shifts the quantisation noise to a band centred on the DAC Nyquist frequency. Second order noise shaping is used because it provides more SNR improvement for x4 interpolation than first order, and is less prone to generating spurious tones.

2.4: Programmable Clock Multiplier

The inclusion of a programmable clock multiplier would appear to provide an easy solution to clocking the DAC, for example distributing a relatively low frequency system clock to the device which is then multiplied up to the required DAC frequency. In reality, for applications requiring the highest performance it is recommended that the clock multiplier be bypassed and that an appropriate low jitter DAC clock be supplied.

2.5: Programmable Dither

Programmable amplitude, digital high-pass filtered dither is incorporated into the device. Using dither improves small signal linearity performance, typically for signal levels below -20dBFS. Please read in conjunction with section 1.8, 'What Advantage is Segment Shuffling?'

It is worth noting that dither only provides an advantage to un-modulated small-signal performance, which is not typical of the target applications. Most applications use large signals as well as some form of modulation, which itself acts as a dither function.

2.6: What Drives the Interpolating DAC?

Even with the advantage of interpolation filtering significant processing is required prior to the DAC in order to create the required wide-band, multi channel transmit signal. As with the interpolation filtering this is unlikely to be implemented using general purpose DSP, requiring either dedicated standard products or an ASIC.

Examples of standard products are

HSP50215: Digital Up Converter, Intersil, www.intersil.com/data/FN/FN4/FN4346/FN4346.pdf

GC4114, Quad Digital Up Converter, Graychip

Unfortunately Fujitsu do not currently offer any digital up converter ASSP solutions and there are no immediate plans to develop these.