"The number of taps, more fully known as FIR taps, is an indication of the amount of memory required to implement the filter and the amount of filtering the filter can do. "Off-the-shelf DACs contain only a limited number of taps. A filter has a long delay line featuring a bunch of taps along it. If you want to reconstruct the original waveform, you need lots of taps. There’s a theory that says that, if you want to perfectly re-create the original recording, you have to have an infinite tap-length filter. That’s what the math says. I looked at off-the-shelf systems. These DACs were using around 100 taps. I could see that this would create severe problem in timing."
So Watts started experimenting, using his punkified FPGA chip with a longer tap length and algorithms to try to improve the ability of the filter to reconstruct the original timing. Hence, on the QuteHD, you will find 10,000 taps (the older Chord DAC64 had 1000, while a generic, standard, off-the-shelf DAC features 128). The more taps you feature on a DAC, though, the more memory you need, which adds to the expense.
So, if taps are so important, why do the common chips that populate most of the world’s DACs get by with so few? Reportedly, well-known manufacturers don’t see the low tap count as being a major problem. "They don’t even do listening tests," said Watts. "It doesn’t feature on their event horizon. Design is all done mathematically. They make assumptions and then they just do it." According to Watts, technocrat graduates populate these companies, showing a greater interest in mathematics than sound."
