In Reply to: The 4 ways to decode digital posted by J. Phelan on October 13, 2015 at 14:11:32:
Hi,The "Chip" or "No Chip" debate is bogus.
Even "Chipless DAC's" actually have Chip's, just not dedicated DAC Chips. Example, the DIY Soekris R2R DAC shown here:
In fact, it has a Xilinx Spartan 6 FPGA and 16 "Latch" Chip's (LVC series 74XX595 to be precise) that drive the R2R ladder. And they need as good a power-supply as any other DAC, given it is single-ended it actually needs a superb PSU as it has minimal noise rejection on the power-supply. And if they have mismatched switch impedance (and they sure have between different IC's and even on IC) you will get extra distortion, as you will if your resistors have any reasonable tolerance.
Incidentally, you cannot really use an FPGA as DAC, you get way too much Jitter. What you can do with an FPGA is to implement DSP, be it digital filters, volume control etc. as well as memory buffers and so on.
Of course, you can implement the same functions with the exact same results in a dedicated DSP Chip or even in a general purpose microprocessor.
One Product that prominently advertises the use of an FPGA as "DAC" in fact uses per channel four "tiny logic" latches as bitswitches, with suitable resistors. Yup, that is in effect a 4-Bit discrete with DSP in a FPGA Chip. But that companies marketing people clearly felt that "FPGA DAC" would sell better. This one has a pretty good implementation incidentally, with a very low noise regulator per channel powering the four chips, but again, as this DAC is single ended it is needed.
The Soekris DIY DAC pictured above has the same kind of structure, but with many more Bits and people are writing now all sorts of DSP Software for it now. It is quite an interesting development.
Incidentally, the Soekris DAC does not measure better than the PCM1704 or TDA1541 (never mind PCM63 or AD1862). Full Scale THD & N are -84dBFS (and the harmonics are not of a "nice" tube style type, unlike those on TDA1541 or Burr Brown DAC's) and the dynamic range comes out at around 112dB. That is all around worse than the ES9023 DAC Chip which sells for 4 USD in singles and less than halve that in full reels these days.
BTW, I am not knocking the Soekris DAC, I find it a fascinating (and positive) development. But it illustrates VERY WELL the challenges with such an approach.
In fact, non of the widely published and promoted "DIY discrete DAC's" (be they R2R or DS based) measure anywhere near as well when it comes to noisefloor and THD as any number of rather cheap dedicated Audio DAC Chip. There are a number of reasons for that, but it gets highly technical.
I'd rather not comment on commercial efforts by others, you can always look at Stereophile's site for measurements of many DAC's with many chips and others.
Of course, measured numbers rarely translate straight into what we hear, so which sounds better is a totally different question.
So, my suggestion, loose less sleep over FPGA vs. Discrete vs. Audio DAC Chip vs. Non-Audio DAC Chip. Any of these can in my experience be made to sound excellent, if care is taken in implementation. They can also sound totally horrible.
Beyond a certain level of objective performance I find that fine details of implementation often matter much more than what technology is used. Power supplies, analogue stages, clocks, digital filters and so on all have disproportionate levels if impact on sound quality. One thing you get with a Surface Mounted Chip DAC is a structure that usually allows very efficient power supply decoupling, something which is major challenge with circuitry distributed over larger spaces.
So listen carefully, using correct level matching in the process and select what you find to give you the sound quality you want, whatever the technology, rather than being blinded by a certain tech.
Thor
At 20 bits, you are on the verge of dynamic range covering fly-farts-at-20-feet to intolerable pain. Really, what more could we need?
This post is made possible by the generous support of people like you and our sponsors:
Follow Ups
- RE: The 4 ways to decode digital - Thorsten 10/13/1519:33:17 10/13/15 (11)
- RE: The 4 ways to decode digital - J. Phelan 20:26:51 10/13/15 (10)
- RE: The 4 ways to decode digital - Thorsten 23:53:21 10/13/15 (9)
- RE: The 4 ways to decode digital - J. Phelan 05:26:15 10/14/15 (4)
- RE: The 4 ways to decode digital - Thorsten 06:12:29 10/14/15 (3)
- RE: The 4 ways to decode digital - J. Phelan 08:25:26 10/14/15 (2)
- RE: The 4 ways to decode digital - Thorsten 18:09:28 10/14/15 (1)
- RE: The 4 ways to decode digital - J. Phelan 18:20:30 10/14/15 (0)
- BB PCM4104 - jusbe 00:37:13 10/14/15 (3)
- RE: BB PCM4104 - Thorsten 00:45:49 10/14/15 (2)
- RE: BB PCM4104 - jusbe 01:03:50 10/14/15 (1)
- RE: BB PCM4104 - Thorsten 02:09:13 10/14/15 (0)