I know nothing about their digital 'secret sauce' but it wouldn't be fair to attribute the whole of the success of this design to DSP. The layout also is worth studying carefully - note that they don't follow the traditional grounding approach of a single (sometimes partitioned) groundplane and take mains noise seriously through avoiding the fashionable choice of toroidal trafos.
Aren't those U-cores, not toroids?
PeterSt is already doing this with his Phasure NOS1 DAC, which in his case requires the use of his XXXHighend software (where the number crunching is done). It is still very much a hand built device, (not that there is anything wrong with that), and unless we get some DIY software building blocks, it will take a lot of work to replicate what he has done..
Well, it sounds like he was pointing out that they avoided toroids, so I'd say he's on the same page.
Obviously no one is meaning skipping the interface. The concept is simple as the previous poster said its being done with the phasure, but could be open with the widget being interface with firmware such that the output format is flexible for different formats.
As you said the MCU it incredible complex, one day it will all be programmed for flexibility instead of proprietary.
Do you understand the concept yet?
Agree here though with the proviso that cheap DAC plus DSP might well be the more cost-effective way to go than super-dooper DAC and no DSP. The function of DSP should be to help make the DAC chip's (and any following filter's) jobs easier. If you have a perfect DAC then DSP isn't required and probably shouldn't be used. There does seem to be a very common perception amongst people with little or no DSP experience that its some kind of 'magic bullet' for the sound and marketing departments do appear to play up to that. Those who work with DSP and produce marketable designs generally know better
Agree here though with the proviso that cheap DAC plus DSP might well be the more cost-effective way to go than super-dooper DAC and no DSP. The function of DSP should be to help make the DAC chip's (and any following filter's) jobs easier. If you have a perfect DAC then DSP isn't required and probably shouldn't be used. There does seem to be a very common perception amongst people with little or no DSP experience that its some kind of 'magic bullet' for the sound and marketing departments do appear to play up to that. Those who work with DSP and produce marketable designs generally know better
DSP has been standard in digital audio reproduction for 20+ years. Oversampling is DSP, period from the first 2xOS CDP.
If you don't want DSP then you build an NOS DAC. The concept of moving the digital side of a converter to the computer is obviously hard to grasp forget I brought it up.
That's to make the reconstruction filter's job easier - already covered that point
It does though have the downside of making the DAC chip's job harder.Interestingly enough I am building an NOS DAC but it does have a DSP too. Go figure
There is nothing proprietary about an open-source USB Audio Class firmware. It works with any operating system that has full support for UAC (1&2), and it can talk to any DAC that you can conceive. I fully understand the concept and have worked on such designs (hardware & firmware), but none of the free ones are complete and none of the complete ones are free. Such is the curse of open source, at least sometimes.
Just because everyone does not agree with you does not mean that they do not understand the concepts. Perhaps your wording was misleading, but you gave the impression that the obvious incompatibility between USB port hardware and DAC inputs was something that you did not understand.
There are also interface designs which transmit the unaltered audio over the interface and then perform DSP outside the computer with SHARC or TMS320 chips that are more efficient and higher quality than general computer processors. The SHARC can handle 80-bit DSP, which is beyond the capabilities of today's computer, at least in real time.
Last edited:
That's to make the reconstruction filter's job easier - already covered that point
Interestingly enough I am building an NOS DAC but it does have a DSP too. Go figure
Oh I can figure, compensation for the treble roll-off.
You guys have really added very little but play words games. But I guess its entertaining (to some one)
Its one possible use for it, for people who value flat frequency response yes. But its not the primary reason - that's to convert between 64fs I2S and 32fs I2S as the latter sounds better
Oh and also to control the WM8805 which doesn't have much flexibility in hardware mode.That's an interesting claim - evidence to support it is going to be forthcoming later?
But I guess its entertaining (to some one)
Here's an interesting quote you might like which I found only the other day:
Anyone who tries to make a distinction between education and entertainment doesn’t know the first thing about either. - Marshall McLuhan
there's a thread on digital line level (I think) for my spdif switch. it has some discussion and progress on what's I've tried over time and what I plan to do for the eventual product. it will be controlled by my preamp, which you can see on my place-holder website, www.sercona.net. the preamp, the spdif switch and a remote smart power strip will all be part of the same controller-based system and will work together.
Hi oneoclock et al,
Today I installed the two Demian/oneoclock PSU boards in my AB1.1. One board is for the XO and one board is for the AVCC of the ES9023. Thanks to Demian and oneoclock
There is subjective SQ improvement in that voices sound more natural and smooth. I will have to do more listening tests with instrumental music and various sample rates to confirm that there are audible improvements over the ADP151's
It is a good thing that the PSU will be incorporated in the AB1.2
Alex
Today I installed the two Demian/oneoclock PSU boards in my AB1.1. One board is for the XO and one board is for the AVCC of the ES9023. Thanks to Demian and oneoclock
There is subjective SQ improvement in that voices sound more natural and smooth. I will have to do more listening tests with instrumental music and various sample rates to confirm that there are audible improvements over the ADP151's
It is a good thing that the PSU will be incorporated in the AB1.2
Alex
I was doing some testing of the PCM4222EVM from TI that I'm now using for a capture system for jitter and distortion measurements. In qualifying it I was running from an external ultra low noise 10 MHz oscillator from Wenzel so I can check on the internal noise floor. What struck me was that the on board oscillators which are enabled and disabled with the outputs connected together were still seriously modulating the clocks when using the external clock through an on board buffer. Disconnecting them completely cleaned things up.
The reason I'm mentioning this is that the same method is being used on the AB1.1 and its possible that the same problem could be present on that implementation.
The good news is that the EVK seems to be able to deliver state of the art audio capture with really low residual phase noise when the clock is good enough. It can be had for $160 from Mouser. With a really good external oscillator the skirts are 1/16 Hz wide at -120 dB which is at the level of insignificant in any real application. I think the Crystek oscillators would work. The power supply on board is not good enough for them.
Maybe now I can get reasonable accurate jitter measurements.
The reason I'm mentioning this is that the same method is being used on the AB1.1 and its possible that the same problem could be present on that implementation.
The good news is that the EVK seems to be able to deliver state of the art audio capture with really low residual phase noise when the clock is good enough. It can be had for $160 from Mouser. With a really good external oscillator the skirts are 1/16 Hz wide at -120 dB which is at the level of insignificant in any real application. I think the Crystek oscillators would work. The power supply on board is not good enough for them.
Maybe now I can get reasonable accurate jitter measurements.
ive been looking at the wenzel units for my dac, but i'm scared to email about the price. i was looking at the streamline or ULN series can you give me some hint? under 500? seems something like this would make a good investment that should last some time and be able to move from dac to dac
Benign Digital Dictator
Here is my suggested pinout for I2S over CAT5. One thought has struck me: Anyone who is capable of patching this on the PCB is also capable of making a custom CAT5 cable.
So what I will do is fix the CAT5 pinout on the AB-1.12 and leave people to build their own custom cables. That means out goes the "patch panel" of this schematic.
Børge
Here is my suggested pinout for I2S over CAT5. One thought has struck me: Anyone who is capable of patching this on the PCB is also capable of making a custom CAT5 cable.
So what I will do is fix the CAT5 pinout on the AB-1.12 and leave people to build their own custom cables. That means out goes the "patch panel" of this schematic.
Børge
Attachments
Actually they are more than 3X that number, with something like a 2 month lead time. They are also a controlled product and can't be shipped outside of the US. The lower performance oscillators can be shipped and they are in the $500 range. Somehow it still doesn't scream group buy.
I think that they are in fact overkill with performance that cannot be exploited by ADC's and DAC's made with current technology. But that is what I'm exploring.
Why not use the EIA/TIA pairing for the CAT5 connection? I know its screwy but it is reall common. You can get 10G ethernet through it so it must work OK. If we could use the magnetics we would have isolation as well. I don't know if we can get word clock and data over the magnetic interfaces but it would be a worthwhile experiment.
the controlled product thing I assumed, but i have other such things =) there are ways around it; mostly border patrol is after more illicit controlled goods. hmm still worth considering, i too was thinking of using with ADC for measurement as well. by lower performance you mean the streamline model i mentioned?
of course its overkill, glorious overkill, most of what we do here is. do i need Rogers PTFE substrate for my dac? no, did i need to build one per speaker and a balanced power amp per driver? no, do i need 8 x 14awg pin ITT Cannon milspec umbilical connectors for my multichannel speaker cables? no (these cannot normally be exported either)
i just figure like you did, that its worth exploring and i also think its a long term investment. i could probably even write it off as a business expense as measurement gear for my business.
500 I can justify, its inline with the rest of the overkill parts i'm using and I probably will, 1500 I cant, particularly as i need 2, though i'm likely to just get one for the 22.1x integers.
thanks
Jeremy
of course its overkill, glorious overkill, most of what we do here is. do i need Rogers PTFE substrate for my dac? no, did i need to build one per speaker and a balanced power amp per driver? no, do i need 8 x 14awg pin ITT Cannon milspec umbilical connectors for my multichannel speaker cables? no (these cannot normally be exported either)
i just figure like you did, that its worth exploring and i also think its a long term investment. i could probably even write it off as a business expense as measurement gear for my business.
500 I can justify, its inline with the rest of the overkill parts i'm using and I probably will, 1500 I cant, particularly as i need 2, though i'm likely to just get one for the 22.1x integers.
thanks
Jeremy