as for me, Im not buying it 
I cant even imagine how the alpa "de ringing" works. I tought its about SM5845 filter trying to look below noise floor
http://src.infinitewave.ca/
look, these are the algorithms they use during recording phase. The one called Izotope is used most, 96khz -> 44.1 downconversion rings .
as for "slow / fast " roll-offs / tradeoffs , I like this opinion:
http://www.ayre.com/PDF/Ayre_MP_White_Paper.pdf
I cant even imagine how the alpa "de ringing" works. I tought its about SM5845 filter trying to look below noise floor
http://src.infinitewave.ca/
look, these are the algorithms they use during recording phase. The one called Izotope is used most, 96khz -> 44.1 downconversion rings .
as for "slow / fast " roll-offs / tradeoffs
, I like this opinion:http://www.ayre.com/PDF/Ayre_MP_White_Paper.pdf
interesting one:
“Apodizing” Digital Filter, -200+ dB at 22,050 Hz, No Pre-Ringing, ~20 Cycles of Post-Ringing
used by Meridian. The insane attenuation above 20.000 hz supposed to deal with various forms of ringing, like the 96k -> 44.1 used at recording studios. I like this approach and Ayre 's.
“Apodizing” Digital Filter, -200+ dB at 22,050 Hz, No Pre-Ringing, ~20 Cycles of Post-Ringing
used by Meridian. The insane attenuation above 20.000 hz supposed to deal with various forms of ringing, like the 96k -> 44.1 used at recording studios. I like this approach and Ayre 's.
That impulse waveform was play back TEST CD
My question.
This impulse was playing TEST CD not Music CD.
I think Music CD was made by A/D Converter with pre-filtering.
If we can impulse waveform by A/D Converter unit.
Does that inpulse wavefome add ringing, because pre-filtering process?
>http://www.ayre.com/PDF/Ayre_MP_White_Paper.pdf
That impulse looks like threw IIR filter, looks like first generation CDP with non FIR DF, with heavy analog LPF.
Many NOS DAC freeks don't like analog LPF for good interpolation, but NOS DACs have terirble high frequency waveform.
And the people who thinking DF technic is nice for get low order analog LPF, also they don't like high order analog LPF.
For such a reason, there are few people, like 13th order heavy analog LPF sound .
It is a question here.
How do you think about Ayre's IIR filtering with digital domain?
I remember AIWA portable CD had IIR digital filter like Ayre's.
My question.
This impulse was playing TEST CD not Music CD.
I think Music CD was made by A/D Converter with pre-filtering.
If we can impulse waveform by A/D Converter unit.
Does that inpulse wavefome add ringing, because pre-filtering process?
>http://www.ayre.com/PDF/Ayre_MP_White_Paper.pdf
That impulse looks like threw IIR filter, looks like first generation CDP with non FIR DF, with heavy analog LPF.
Many NOS DAC freeks don't like analog LPF for good interpolation, but NOS DACs have terirble high frequency waveform.
And the people who thinking DF technic is nice for get low order analog LPF, also they don't like high order analog LPF.
For such a reason, there are few people, like 13th order heavy analog LPF sound .
It is a question here.
How do you think about Ayre's IIR filtering with digital domain?
I remember AIWA portable CD had IIR digital filter like Ayre's.
im testing such slow roll off 192k minimum phase filter right now, results:
a 44.1k bandlimited "handclap" signal can reach -125 dB at 23khz, but its again up to -74 dB at 26.000 hz and dissapears completely again above 30khz. Clearly this is a kind of leakage ( I doubt -74 dB "leak" at 26khz is a problem though) .
a 44.1k bandlimited "handclap" signal can reach -125 dB at 23khz, but its again up to -74 dB at 26.000 hz and dissapears completely again above 30khz. Clearly this is a kind of leakage ( I doubt -74 dB "leak" at 26khz is a problem though) .
) . Now you tell me , IS denon ALPHA ver3 , switching to Fast roll-off on hi-hats ? Re: Analog LPF when we use 8fs DF DF1706 and 96kHz DIR DIR9001.
I chose 4th order analog filter since the goal is act like a true brick wall filter. My analog filter is a butterworth with Fc of 50Khz. This gives about .04dB of attenuation at 20Khz (negligible) with very small phase shift relative to a Bessel ( linear phase ) at this frequency. At 332.8Khz the response is -65dB. More attenuation here is theoretically better to remove the sampling artifacts that were not in the original source recording.
So far I have only considerd 44.1K sample rate audio CD, but I think the filter would be the same for 48 or 96Khz.
It is true that many DAC manufacturers recommend a 2nd or 3rd order filter after the DAC. Maybe for cost reduction ? I have also seen up to 8th order Bessel filters recommended in "high end" application notes. Provided the op-amps used have enough open loop gain to work properly at 8xFs (352Khz ), I would think there is no problem.
I have added some additional bench test data here
http://home.comcast.net/~saudiodac24/site/ ( link at bottom ) showing the output of the I-V converter as well as analog filter output. Do you have similiar plots for your design ? I also see no 20Khz modulation at the filtered output using DF1706 in 8x mode ( sharp roll off ).
Hello nagaesan,nagaesan said:
I chose 4th order analog filter since the goal is act like a true brick wall filter. My analog filter is a butterworth with Fc of 50Khz. This gives about .04dB of attenuation at 20Khz (negligible) with very small phase shift relative to a Bessel ( linear phase ) at this frequency. At 332.8Khz the response is -65dB. More attenuation here is theoretically better to remove the sampling artifacts that were not in the original source recording.
So far I have only considerd 44.1K sample rate audio CD, but I think the filter would be the same for 48 or 96Khz.
It is true that many DAC manufacturers recommend a 2nd or 3rd order filter after the DAC. Maybe for cost reduction ? I have also seen up to 8th order Bessel filters recommended in "high end" application notes. Provided the op-amps used have enough open loop gain to work properly at 8xFs (352Khz ), I would think there is no problem.
I have added some additional bench test data here
http://home.comcast.net/~saudiodac24/site/ ( link at bottom ) showing the output of the I-V converter as well as analog filter output. Do you have similiar plots for your design ? I also see no 20Khz modulation at the filtered output using DF1706 in 8x mode ( sharp roll off ).
AttackKing sound --> Slow Rolloff mode
This is a catalogue of DENON DCD-SA11 with slow and sharp rolloff automatic changing new ALPHA system.
I read this figure, DENON seems like to choice slow rolloff if play back attackking sound like attack hi-hats.
I don't understand why they don't choice sharp rolloff that good high frequency responce.
And my biggest suffering is, does not a problem occur even if changing rolloff curve of long TAP FIR filter, when playback music.
This is a catalogue of DENON DCD-SA11 with slow and sharp rolloff automatic changing new ALPHA system.
I read this figure, DENON seems like to choice slow rolloff if play back attackking sound like attack hi-hats.
I don't understand why they don't choice sharp rolloff that good high frequency responce.
And my biggest suffering is, does not a problem occur even if changing rolloff curve of long TAP FIR filter, when playback music.
Attachments
Pre-filtering curve of UpSampler
tritosine,
If fs=44.1kHz data to fs=192kHz upsampling, using FIR filter interpolation. When this process, we must pre-filtering for prevent aliasing noise.
In this upsampling process, do not you think of impulse data was band limitting looks like 4.35fs over sampling DF?
I think in this system, it is not problem if you choice 2nd order analog LPF.
tritosine,
If fs=44.1kHz data to fs=192kHz upsampling, using FIR filter interpolation. When this process, we must pre-filtering for prevent aliasing noise.
In this upsampling process, do not you think of impulse data was band limitting looks like 4.35fs over sampling DF?
I think in this system, it is not problem if you choice 2nd order analog LPF.
Transversal summed multiport analog delayline DAC
Hell rudeesan, I clealy understood why you choice 4th order analog LPF. Thank you.
But many people wants to "No Analog Filter" or "more low order LPF". Because precision analog LPF needs precision circuit with precision device like 0.1%register and cap. It needs high cost than high speed sampling oversampling digital technology.
Now my opinion is "need low order LPF but don't using fs conversion"
There are many method of low order analog LPF, one is oversampling or upsampling, and second one is like WADIA's.
So I think that WADIA's patented "64fs Over Sampling" Transversal summed multiport analog delayline DAC is very nice idea.
Using 4DAC/ch, input data 16fs using delay line, to 64fs with moving averaging filter. This method, analog LPF devices are not necessary.
How do you think about WADIA's Transversal summed multiport analog delayline DAC?
Hell rudeesan, I clealy understood why you choice 4th order analog LPF. Thank you.
But many people wants to "No Analog Filter" or "more low order LPF". Because precision analog LPF needs precision circuit with precision device like 0.1%register and cap. It needs high cost than high speed sampling oversampling digital technology.
Now my opinion is "need low order LPF but don't using fs conversion"
There are many method of low order analog LPF, one is oversampling or upsampling, and second one is like WADIA's.
So I think that WADIA's patented "64fs Over Sampling" Transversal summed multiport analog delayline DAC is very nice idea.
Using 4DAC/ch, input data 16fs using delay line, to 64fs with moving averaging filter. This method, analog LPF devices are not necessary.
How do you think about WADIA's Transversal summed multiport analog delayline DAC?
4 or 2 times the DAC chips is not exactly a cheap way to implement a low pass filter I think they also did it because of glitch. Now they use only 2 dac last time I saw (2x pcm1702 inherently glitch proof ) .
Can we average out DC offset somehow? Now that would be something
// On the low ringing filter I simulate here: I retract my comment "leakage looks bad", Its Very-Very program dependent... Im averaging some more now I even had some HF artifacts in my first file //
I think they also did it because of glitch. Now they use only 2 dac last time I saw (2x pcm1702 inherently glitch proof ) . Can we average out DC offset somehow? Now that would be something
// On the low ringing filter I simulate here: I retract my comment "leakage looks bad", Its Very-Very program dependent... Im averaging some more now I even had some HF artifacts in my first file
//Re: That impulse waveform was play back TEST CD
Ayre CDs are supposed to sound good, but I have never auditioned one.
I think the lower end model uses BB pcm1792, one or two chips, same as the Weiss Minerva/DAC2 converters.
Me being one of them. I really dont like analog filters after the D/A conversion.
This is why I'm pursuing to find the "best" digital (read software) oversampling algorithm.
Wish I could listen to it (no distributor in italy), but I probably won't like it
nagaesan said:
Ayre CDs are supposed to sound good, but I have never auditioned one.
I think the lower end model uses BB pcm1792, one or two chips, same as the Weiss Minerva/DAC2 converters.
Me being one of them. I really dont like analog filters after the D/A conversion.
This is why I'm pursuing to find the "best" digital (read software) oversampling algorithm.
Wish I could listen to it (no distributor in italy), but I probably won't like it
Re: Transversal summed multiport analog delayline DAC
They called it time-delayed something, but it looks like linear interpolation (and they have a patent somewhere).
Well, on principle, I like it very much. Wadia 9 series (which uses 8 mono dacs per channel) lacked a transport to be paired with, so it wont be easy to audition it at its best.
nagaesan said:
They called it time-delayed something, but it looks like linear interpolation (and they have a patent somewhere).
Well, on principle, I like it very much. Wadia 9 series (which uses 8 mono dacs per channel) lacked a transport to be paired with, so it wont be easy to audition it at its best.
United States Patent 5075880
Hi! tritosine
This is the US Patent of WADIA's (Using Moving Averaged filter) DAC. They called this Transversal summed multiport analog delayline DAC
United States Patent 5075880
http://www.freepatentsonline.com/5075880.html
Several time, I met WADIA people 1995 - 2001 with Japanese importer AXIS people.
Then I introduced them, a certain Japanese maker imitated WADAI's patented "Transversal summed multiport analog delayline DAC".
And also I introduced them, "But in JAPAN, the patent holder of this method is PANASONIC".
Hi! tritosine
This is the US Patent of WADIA's (Using Moving Averaged filter) DAC. They called this Transversal summed multiport analog delayline DAC
United States Patent 5075880
http://www.freepatentsonline.com/5075880.html
Several time, I met WADIA people 1995 - 2001 with Japanese importer AXIS people.
Then I introduced them, a certain Japanese maker imitated WADAI's patented "Transversal summed multiport analog delayline DAC".
And also I introduced them, "But in JAPAN, the patent holder of this method is PANASONIC".
Analog LPF devices are not necessary but....
Using 4DAC/ch, input data 16fs using delay line, to 64fs with moving averaging filter.This method, analog LPF devices are not necessary.
But this method, this filter's attenuation characteristic is not good, because this filter is moving averaging filter, not FIR, and very low TAP.
Using 4DAC/ch, input data 16fs using delay line, to 64fs with moving averaging filter.This method, analog LPF devices are not necessary.
But this method, this filter's attenuation characteristic is not good, because this filter is moving averaging filter, not FIR, and very low TAP.
schiller said:
In terms of _measured_ performance, the newer delta sigma dacs are going to give you better performance in most of the standard sorts of tests (e.g. unitone or duotone systems where you're measuring basic HD/IMD performance). It's considerably easier to get <-100dB THD+N with a PCM1794 than it is a PCM1704 for the common 20-20KHz 22KBW test, though it can be accomplished with both (at least, I've been able to with the PCM1704). For more music-centric tests like a complex multitone system, I've gotten mixed results but I guess I favour the distribution of products I get out of the PCM1704. In terms of subjective performance, I prefer the 1704 to pretty much anything, although the Sabre looks like it might be a promising alternative.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.