It's notable how often the "missing information" and other misunderstandings are repeated. While I have no doubt that most analog fans are operating in the best of faith and truly feel analog is "better", some of what falls out has a certain connection with Brandolini's law. In particular, I'm thinking of Bastiat's early formulation:
I'm grateful to Monty for making a video to spare interested parties from having to repeat George Horne's "thirty pages to answer." 🙂
With the recent revelation that Mobile Fidelity has been digitally processing a large number of their pressings for years, I think it's another good time for analog aficionados to consider whether what they like has more to do with what analog adds to the signal rather than what it preserves. There's nothing wrong with this, but if it were me, I'd want to be sure that I've identified the causes of my preferences correctly.
I was using it it Europe, so maybe 44.1? It was quite a chore getting to CD back then. Don't remember the whole process.
Isn't the issue with this that you need effectively infinite frequency resolution? You start with something that's conceptually similar to a discrete Fourier transform and end up at full (i.e., integral) Fourier because of the length of music. You could take overlapping "snapshots" of fixed periods, but now you're implementing a lossy process.
I am not dissatisfied with the digitation of sound including songs and music. Enough perfection/quality has been reached regarding the reproduced sound. In my posts, specifically the later posts, I was after an analysis which could show that if any frequencies are distorted, they are at the high extreme and almost all individuals do not have the hearing resolution to notice anything.
Why always with the DACs with you, do you sell them? This thread is about recording. Maybe he's not so let him say so, stop putting words in other peoples mouths.
Last edited:
If you are interested in the mathematics behind it, you could read Shannon's 1948 article: https://www.cs.ucf.edu/~dcm/Teaching/COP5611-Spring2012/Shannon48-MathTheoryComm.pdf
The sampling theorem is not the main subject of the article, just something Shannon needed to prove to come to his theorems about information and channel capacity.
No.
As I explained earlier in the thread, someone may judge digital recording by the digital playback they have heard. In that context I believe there are some very well digitized recordings. The problem with digital audio as most people experience it tends to be on the dac side. Why? For one thing, sigma delta modulation doesn't put RF into ADC analog preamplifiers in the same way it puts RF into the DAC output stage amplifiers. RF in output stage amplifiers can cause some problems that don't show up very well on typical audio FFTs. Some people believe APx555 FFTs tell all that can be known about device performance. I would disagree.
A sigma-delta ADC has a feedback DAC that injects the sigma-delta modulate into the analogue amplifier at the ADC input.
IIRC a digital zero-order hold inside the ADC can serve as the DAC. Consider the ESS application diagrams attached below. Also the block diagram from MathWorks (used here for educational purposes). Where is the analog feedback?
Attachments
Last edited:
I had been doing analog organ recordings back in the 70's using better end consumer
recorders, dual process Dolby B and aligning the machine for the specific reel of tape
to be used. Response was a fraction of a dB from dead flat and noise floor was very
low. Flipping the source/tape switch could ALWAYS tell tape vs E-E. My first digital
machine was in September 1982. The machine did not have source tape switch but
when played back it was absolutely indistinguishable for the first time ever. There is a
LONG list of the defects in analog recording. Wow and flutter causes phase shifts vs
standing wave when listening to speakers in a room. That's why you notice almost no
speed variatoins in headphones. The tape 'wobbles' through the tape path. This causes
interchannel phase variations. You can see this on a X-Y dsiplay where you get a wobbling
oval rather than a straight line. There are HF response changes due to to particles passing
between the tape and the heads. I can show you some recordings on London Decca with
this issus and they are one of the best labels out there. These look like level variations
worse at higher frequencies. So would you set the response for the peaks (most of the
time) or the average which is what you would see on a meter vs a scope. If you hear ONE
good recording in digital, you know the medium is OK. Many Telarc releases are worth a
listen and might even change your mind. My favorite are LP folks who prefer the 'musicality'
of the LP vs digital. THEN they will tell you the CD they made is just as musical as the LP.
That tells me they are used to the LP defects and prefer them. That's fine. If they can nake
a copy to digital that sounds as good as the LP, doesn'that suggest that recoding the CD direct
from the digital source will be just as perfect a copy as the CD of the LP recording?
End of rant
G²
I'm glad someone posted this. I have watched it several times over the years and still
find some new things.
G²
I live in LA area and the FM processing is as 'good' BAD as it gets. Louder stations
get better ratings and they make more money God bless 'em. I NEVER listen to FM.
A friend called me and told me a buddy of his was running a pirate radion station
near me and I should be able to pick it up. My tuner at the time was an
H-K Citation 15, quite good in its day. It was the BEST FM reception I have ever
heard. The pirate had his CD player connected directly to the transmitter with no
audio processing. WHAT A DIFFERENCE. His taste in music was pretty nice and I
listened the whole night. Good audio, no processing, no commercials, no work on
my part. Nothing better.
G²
The 44.1 vs 44.056 is B/W vs color. 15750 line rate B/W, 15734 in color. The PCM-F1
locks to the frame time (V sync) for the PLL. It should not have surprised me to hear
the pitch change when I turned the tracking control.
G²
In the application diagram, it is somewhere inside the ADC chip. Note that the recommended filter circuit is designed to have a low output impedance, also far out of the audio band. That suggests there is a considerable amount of high-frequency current coming out of the ADC inputs.
For the other case, see the orange line:
Yet I am not convinced that opamps driving the ADC inputs in experiencing the same type of distortion artifacts, and or to the same degree, as those in a dac output stage. It would be interesting to see if there are distortion artifacts such as those shown below at the output of the ADC input opamps.
EDIT: To look at the situation another way, if its easy enough to put a high performance dac inside an ADC then why not make high performance voltage out dac chips that don't require external I/V conversion, just a buffer instead? Nobody seems to be able to do it that way and still achieve SOA impressive measurements.
Attachments
Last edited:
The op-amp that has to handle all the quantization noise coming out of the feedback DAC while producing almost no distortion is the op-amp at the input of the loop filter inside the ADC. There are two differences compared to the first op-amp after the DAC chip in an audio DAC:
1. The first loop filter op-amp in the ADC only needs to handle the difference between the signal and the feedback, so there is much less low-frequency content. You can therefore use much smaller integration capacitors, making it much less expensive in terms of chip area.
2. It only needs to provide the right output voltages at the moments when the quantizer samples its output. You can therefore use it in a switched capacitor style, where slew rate limiting doesn't matter as long as it settles in time.
If the op-amp is used as a switched-capacitor integrator, its input signal needs to be sampled on a capacitor. That's probably done straight at the input, which is why the buffer driving the input needs to be very low impedance up to high frequencies.
That seems like a significant difference from an opamp that needs to provide the right output voltage at all times. Also, the internal opamp may not need to be able to swing as much output voltage?
Beyond that we know that swapping output stage opamps changes the sound, when in principle ether opamp should work fine. Say for example, OPA1612 verses OPA1656. Which one if either of them gives the correct sound? What about the sound of the internal opamp?
Last edited: