I myself have listened to many different dacs and all of them have about the same specs but the expensive ones tend to sound clearer with less distortion. Is there some other way to measure a dac quality other than price?
I myself have listened to many different dacs and all of them have about the same specs but the expensive ones tend to sound clearer with less distortion. Is there some other way to measure a dac quality other than price?
Most better and unfortunately more expensive DACs don’t use cheap delta/sigma, mass produced chipsets that make more economically priced DACs affordable. Prior to the introduction of delta/sigma DAC chipsets, the only DAC conversion chipsets available were those used in high resolution medical imaging and radar systems that required bit-perfect precision.
Delta/sigma DACs employ shortcuts such as interpolation to cut down on the massive processing and power requirements that bit-perfect conversion requires and for most listeners the small trade-offs aren’t all that apparent. Mind you, the first few generations of delta/sigma DAC chipsets were far from ideal. In fact, what’s often referred to as digital glare was because of early delta/sigma DAC chipsets and flaws that Phillips and Sony had made in the decoding process. Thankfully, the latest crop of delta/sigma chipsets used in current DACs have all but eliminated the flaws that plagued early digital conversion and some truly excellent DACs use them.
Brands like dCS, Mola Mola, PS Audio DirectStream, Chord, MSB and Denafrips rely on either bit-perfect, non-interpolating DAC chipsets used in high resolution medical imaging, programmable FPGA chips running proprietary digital conversion software to DACs like mine that uses a discrete R2R resistor ladder network to convert digital files to the analog waveforms we recognize as music. There are literally over a thousand, individual, laser trimmed surface mount resistors on the DAC board. My previous DAC was the internal DAC in my Oppo UDP-205 4K Universal Disc Player which uses the ESS Sabre 9038 Pro chipset. It’s excellent and ruthlessly revealing but my current Denafrips Hermes DDC/Pontus II DAC combo simply sounds far more tangible and real. However, if the recording is shitty or the recording engineer got sloppy, both DACs make it painfully obvious.
Cool thing is you can get a discrete R2R ladder DAC from Denafrips. Both the Ares II and Enyo be had for under $1000. Reviewers lost their minds over them.
Wrong on nearly every point.
Prior to delta-sigma designs there were DACs such as the TDA1640 from the very early 80s marketed as for audio conversion, displays, electron beam recording (I’m reading from the data sheet). These weren’t repurposed medical imaging or radar DACs, they were relatively cheap devices intended for consumer electronics. Delta-sigma massively improved the quality and reduced the price of DACs when it became feasible to build high speed semiconductors cheaply and they displaced pretty much all other DAC designs for audio applications by the early 90s.
‘Digital glare’ typically came from these early multibit DACs which had either limited bith depth (the Phillips TDA1640 is a 14 bit DAC), had poor tolerances in the resistor networks to generate the reference voltages or required very aggressive filtering near the Nyquist frequency which either deadened the sound or allowed high frequency aliases back into the audio spectrum. Or, typically, all of these problems together. These problems pretty much went away for consumer gear with the advent of delta-sigma converters.
If you’re not a fan of interpolation, I have bad news for you. You can infer a bit about how the dCS Ring DAC that you like works by reading their website and looking at the circuit board pictures. They talk about their DAC running at between 2.822 and 6.14MHz which corresponds to 64x oversampling of a 44.1-96kHz signal, and they also talk about it getting fed a 5 bit oversampled PCM from the frontend. Since there are a couple of Xilinx FPGAs in front of the resistor network, it’s a dead cert that this oversampling and the required
interpolationfiltering is happening in software on those chips. The dCS blurb talks about having 48 current sources, which you can see on the photos of the boards (each current source needing two resistors, all doubled up for stereo) which is the resistor and latch network that’s laid out in a 12x16 grid. It absolutely is doing interpolation and it absolutely isn’t using repurposed medical imaging DACs and the dCS (at least) isn’t using thousands of indivdually trimmed resitors.