Maybe reading the tech manual/operation manual will help. But I did what I said and got what the scope and spectrum analyzer shows. If thats impossible then I dont know what to say.
I am sure if the confusion is solved we can all get on track again. Its the waveforms. Maybe it helps if you know the adc sampling is at 24/96 and converted with DSP to redbook spec. for output/burning.
But, the noise is there, never-the-less, on some CD and DVD machines regardless of understanding what happened in the machine regarding adc-dac thru-put tests of the analog output.
THx-RNMarsh
I am sure if the confusion is solved we can all get on track again. Its the waveforms. Maybe it helps if you know the adc sampling is at 24/96 and converted with DSP to redbook spec. for output/burning.
But, the noise is there, never-the-less, on some CD and DVD machines regardless of understanding what happened in the machine regarding adc-dac thru-put tests of the analog output.
THx-RNMarsh
Last edited:
Richard why don't you just post a few seconds (one second would do) of a 16/44.1 .wav file of your 1kHz square wave. Your scope photo violates Nyquist plain and simple, and you are fooling yourself in some way.
The comment and photo in post #76842 shows complete lack of understanding of the sampling theorem. The "nicer" photos using apodizing (windowing) in your reference allow aliases into the audible region this is why you only find this pushed by audiophiles. No precision instrument maker would ever do this.
BTW the high frequency noise from noise shaping has nothing to do with this at all.
Last edited:
How would you like me to do that? I used an HP 3012B generator for the input signal. A digital scope (2.5GHz sampling rate) TEK TDS 3032 scope (300MHz effective BW).
I am not sure what kind of format the scope waveform is recorded .... does onto a disk. Then disk into computer. and then?
There is one possibility regarding the filtering..... the impedance of the BenchMark isnt what mfr said it is and filter was designed for their number).
THx-RNMarsh
I am not sure what kind of format the scope waveform is recorded .... does onto a disk. Then disk into computer. and then?
There is one possibility regarding the filtering..... the impedance of the BenchMark isnt what mfr said it is and filter was designed for their number).
THx-RNMarsh
Last edited:
At DC the gain is unity, and at HF the gain approaches (1+220/200) = +2.1
The pole is at 1 / 2Pi x R2C1, and the zero is at 1 / 2Pi x (R1+R2)C1.
This HF boost counters the sinc roll off, to a point.
Last edited:
You said you are burning a Red Book CD. Grab a second or so in any audio editor and post it as a 44.1/16 wave file.
You misunderstand my meaning.... the waveform is 'correct' .I mean the ripple portion would not be allowed if analog without filtering.
As for the noise... I understand it isnt related. Two seperate things. Sorry to mix the two things together.
But, I really dont care what causes the HF noise in the processing. I cant change it. I can filter it though. And, I can test for distortion caused by it. I dont design this stuff and never will. So a certain amount of ignorance is accepted as fact. Clear it up and move on.
THx-RNMarsh
Last edited:
further clarifying -
The two waveforms are related to my discussion on filtering the HF noise. It affected both the HF and the ringing-like wave on the square wave. That is what I was showing. Both were affected with similar out-come. And, that same end result was achieved when input to ADC was filtered.
That is the connection I was showing and not that they are a manifestation of one same particular cause.
I agree with Demian and others the input to ADC and output of DAC should be filtered and with same filter characteristics. But.... This doesn't always happen, though. Unless there is an agreed on filter standard(s)....
THx-RNMarsh
The two waveforms are related to my discussion on filtering the HF noise. It affected both the HF and the ringing-like wave on the square wave. That is what I was showing. Both were affected with similar out-come. And, that same end result was achieved when input to ADC was filtered.
That is the connection I was showing and not that they are a manifestation of one same particular cause.
I agree with Demian and others the input to ADC and output of DAC should be filtered and with same filter characteristics. But.... This doesn't always happen, though. Unless there is an agreed on filter standard(s)....
THx-RNMarsh
Last edited:
Come on use Audacity or Audition, will someone help here?
I guess you're right PMA this borders on surreal.
Last edited:
Before your suggesting Audacity, I down loaded it and am looking at it wondering if I want to spend the time to do this. But, I will put in some time on it before falling asleep. I would like you to understand and explain it to me/us.
I can put the test disk in the computer and figure how to read it into Audacity et al. Any help would speed this up for making a wav file.
THx-RNMarsh
Dick you're suggesting the entire audio industry has been pulling the wool over our eyes for years and you can't figure out how to extract a .wav file.
I am not suggesting any such thing. Just see noise on instruments and filtered it.
The rest is an unknown to me. And, yes. I never bothered with creating/editing wav files before.
Never learned to cook, either and many other things.
I can send the disk to you, if you like. Easy for you and easy for me.
THx-RNMarsh
The rest is an unknown to me. And, yes. I never bothered with creating/editing wav files before.
Never learned to cook, either and many other things.
I can send the disk to you, if you like. Easy for you and easy for me.
THx-RNMarsh
Scott, if I had spend years of my life in editing sounds, for professional purposes, and even collaborated to the design of "virtual" editing machines, I don't remember to have done this kind of activity at home just for fun ;-)
What the need ? When i'm bored by a tune, I just push the "next button"...
Slowly making progress...... Audacity says they cant open it because it is an audio CD file and I have to convert it to wav first. Nice.
Good thing I am here alone for a week... freezing outside... snowing tomorrow. I have enough raw food stocked for a few more days...... maybe i wont have to learn how to cook.
My neighbors have taken pity on me and Patti or John make an extra plate and bring it over to me. [They know I cant cook.] ...
-RNM
Good thing I am here alone for a week... freezing outside... snowing tomorrow. I have enough raw food stocked for a few more days...... maybe i wont have to learn how to cook.
My neighbors have taken pity on me and Patti or John make an extra plate and bring it over to me. [They know I cant cook.] ...
-RNM
Last edited:
More clarity --- maybe >
I used a program called CDex to convert the disk I made (CD?) to wav and imported it into Audacy. Comes up with 44100, 32 b floating. 64 MB for the 6 minute recording. I can play it in a CD player and a DVD/CD player and hear the steady tone.
So, maybe its the filter shown isnt what it actually is? I used a resistor term for filter based upon mfr spec data. I will have to measure it while actually connected to Master recorder as a cross-check.
But here is what is also interesting. That Gebb phenomenon ripple occures when you have an input signal BW or Tr that is faster than the ADC/DAC system. When you filter the input transistion times down slower the "ripple" reduces. I can watch this happen because I can vary the Tr and Tf of the step transition of the sq wave via the HP-3012B controls.
When you have the min ripple you are looking for, i can then measure the Tr on the scope and compute the filter cut-off BW for a LPF design. .
THx-RNMarsh
I used a program called CDex to convert the disk I made (CD?) to wav and imported it into Audacy. Comes up with 44100, 32 b floating. 64 MB for the 6 minute recording. I can play it in a CD player and a DVD/CD player and hear the steady tone.
So, maybe its the filter shown isnt what it actually is? I used a resistor term for filter based upon mfr spec data. I will have to measure it while actually connected to Master recorder as a cross-check.
But here is what is also interesting. That Gebb phenomenon ripple occures when you have an input signal BW or Tr that is faster than the ADC/DAC system. When you filter the input transistion times down slower the "ripple" reduces. I can watch this happen because I can vary the Tr and Tf of the step transition of the sq wave via the HP-3012B controls.
When you have the min ripple you are looking for, i can then measure the Tr on the scope and compute the filter cut-off BW for a LPF design. .
THx-RNMarsh
Last edited:
I think you're just seeing that a band-limited system will not produce Gibbs phenomenon, could be wrong.
Oh and Richard, if you're looking for a DAC with PCM1794A you can also look for one with PCM1792A as well, it is the same chip but with I2C and SPI interfaces for control.
While PCM1794 is a good chip, I am not sure it's even close to the best when it comes to out-of-band noise performance. I have not seen independent measurements, but AKM seems to believe their AK4396 and later DACs have the lowest out of band noise of the current crop of DAC chips.
Oh and Richard, if you're looking for a DAC with PCM1794A you can also look for one with PCM1792A as well, it is the same chip but with I2C and SPI interfaces for control.
While PCM1794 is a good chip, I am not sure it's even close to the best when it comes to out-of-band noise performance. I have not seen independent measurements, but AKM seems to believe their AK4396 and later DACs have the lowest out of band noise of the current crop of DAC chips.
Last edited:
Another thing to keep in mind (although this is likely unrelated to Richard's observations) -
Most digital filters in audio DACs are implemented with an intentional flaw in the selection of the filter cutoff frequency. They actually allow aliasing to occur because they have not entered the "stopband" at Fs/2.
For example: The PCM1794A/PCM1792A datasheet shows the stopband begins at 0.546*Fs. That's 24.078kHz which violates Nyquist! They allow the passband to extend too far for what I guess is marketing reasons. The transition band is as narrow as they can afford to make it, but the ultimate marketing requirement that trumps Nyquist is that the DAC passband shall extend to 0.454*Fs which conveniently happens to be almost exactly 20kHz.
Wouldn't be able to sell any if they were losing "signal" at 18kHz I guess...
Most digital filters in audio DACs are implemented with an intentional flaw in the selection of the filter cutoff frequency. They actually allow aliasing to occur because they have not entered the "stopband" at Fs/2.
For example: The PCM1794A/PCM1792A datasheet shows the stopband begins at 0.546*Fs. That's 24.078kHz which violates Nyquist! They allow the passband to extend too far for what I guess is marketing reasons. The transition band is as narrow as they can afford to make it, but the ultimate marketing requirement that trumps Nyquist is that the DAC passband shall extend to 0.454*Fs which conveniently happens to be almost exactly 20kHz.
Wouldn't be able to sell any if they were losing "signal" at 18kHz I guess...
- Status
- Not open for further replies.