What does DAC implementation mean?

In a lot of article people write it is not which DAC is being used but the “implemenation matters”. I come from computer science background and did some research at my end but could not get proper answer the following questions.
To simplify the questions I will only talk about DAPs, because everything is in one enclosure and self contained.

  1. [Baseline case] If there are 2 DAPs, DAP1 and DAP2 that have the same DAC chip and have identical input and output stages to the DAC can there be any implementaion differences between DAP1 and DAP2 from DACs perspective? If there are implementation differences will they be audible?
  2. [Realworld case] If there are 2 DAPs the have the same DAC chip, what are the “implementation differences” can be there?
1 Like

I think when people talk about implementation, they talk about designing the output stage. Everyone can use the same DAC chip, but they all design the output stage differently. You can consult this link for some of the examples: https://www.psaudio.com/pauls-posts/duties-output-stage/

For your question though, the missing piece is that in your premise has “identical output stages”. They don’t. Different DAPs have different output stages from the DAC chips even if they use the same chips.

5 Likes

I’m sure that @Torq can 'splain better. However many DAC chips permit different options regarding filters and setup. Some hardware permits user modification, some does not. Also the DAC is only one part of a larger system, and there can be profound differences in quality in the larger system.

For example, Audioquest builds the Dragonfly Cobalt " with a minimum-phase slow roll-off filter for more natural sound". The Dragonfly Red does not use a slow roll-off. This is an implementation choice, not dictated by the DAC.

OK, I’ve used up my buzzword quota for the day.

7 Likes

Is the output stage also part of the DAC chip itself? Then it would make sense that there is a large number of configurations and different products pick and choose different configurations.

An off the shelf DAC has several parts to it:

  1. The power supply
  2. The DAC portion (sometimes including digital volume control, DSP, upsampling, etc.)
  3. Analog Output stage (sometimes including a volume pot)

How each of these is designed and implemented can impact the final sound that you hear. Broadly speaking, there are a few main types of dacs (#2 above):

  1. Resistor ladder, or R2R (oldschool tech)
  2. FPGA (field programmable gate array) - Any chip is basically a large collection of transistors wired to do specific functions. And FPGA is one where the transistor functions can be changed on the fly using software.
  3. Chip-based - Like the ones made by ESS, AKM, etc. These make it so the actual DAC duties are generally the same, but not always. For example, the RME ADI-2 DAC fs uses a single AKM 4493 chip (which is also used by several other way cheaper DACs), but it adds a lot of additional features on top of it:
    a. Digital volume control
    b. DSP
    c. Perceived loudness correction (e.g. compensating for how human ears roll off high and low frequencies way harder than the midrange as you lower volume)
    d. Way more than can be listed here…

The power supply and analog output stages then play a big role in overall quality

6 Likes

Before the DAC chip gets the data, quite a bit has to happen. FLAC, mp3, and all the other formats have to be uncompressed and/or converted back into PCM. The data has to be measured and timed by clocks that control how much data over time the DAC chip gets. The processor may get involved to apply EQ or digital volume control. Many other things happen, I’m sure.

Then, DAC designers typically have choices on which built-in filters to use, because ESS, AKM, TI all have them built in. Or do they use custom filters? Do they oversample? If so, by how much? Do they enable DSD support? Lots of choices.

After the DAC outputs an analog signal, the amplification section is going to change the sound too. There are probably more ways to implement an amp section than you can shake a stick at.

So, when someone talks about implementation differences, it’s the quality and design of a boatload of different components before and after the chip that make a huge difference in your final sound. I can think of no better example than the Matrix X-Sabre Pro. There are a ton of DACs that use the ES9038 chip, but none of them sound as natural as the XSP. This is a chip that on every other DAC will usually have a steely glare to the treble and an over sharp sense to the music it produces. Matrix implemented that DAC differently, and you can hear it.

7 Likes

Some of the comparisons at Audio Science Review demonstrate different implementations of the same chips. For example, some ESS DACs demonstrate a “hump” that others do not. The explanation follows from the earlier replies to this thread – lots of different ways of executing the same function.

Note that ASR can be opinionated and controversial, but the measurements are usually consistent with other sources.

5 Likes

Thanks for so many pointers. This makes a lot of things clearer that i was confused about.

1 Like