[DACLadder]

Kingwa checked out my Singularity digital board that I sent back and found no issue. The issue I experienced was when installed in my old Master 7 chassis only 1 in 10 power cycle boots had good audio on 44.1Khz sample files. Otherwise sound was muffled with loss of highs, slightly lower gain, and grainy vocals. It was immediately noticeable when bad.

Kingwa offered to refund my money but I asked Kingwa to ship back a different board for a 2nd try. I am happy to report the new board works much better. On a cold start up the audio is always good and remains OK until power is switched off. I can restart the M7S in the first hour and still have good audio. After several hours of operation with Master 7S good and warm does the same problem show up. But shows up less frequently than previously experienced with the first board. The new Singularity digital board started up OK 7 days straight.

I feel like this is a victory as I have an upgraded M7 that works. Rarely do I reboot a warm DAC. It gets turned on early in the morning and left on until no longer needed and shut off overnight. Here's a video (below) of my experience with the latest digital board. The video captures the audio of my warmed up system and is broken into three segments. The first segment, where audio is normal, is followed by two power cycles of the M7S. You will notice less highs (muffled) and gain after the two warm power cycle restarts.

https://1drv.ms/v/s!AqFmDQboxG1jgWsiar5k7E5GSwd_

 

[PeterCraig]

Your post proves that a video is worth a thousand words.

What happens if you power off for a full minute and then power on?

 

[DACLadder]

Your post proves that a video is worth a thousand words.

What happens if you power off for a full minute and then power on?

I have to let the new board cool off a while before it will restart OK. The day of the video was an extremely hot day with indoor ambient temps about 80 deg F. But if it starts OK it will remain working fine until power is again cycled.

The 1st board was totally random experience. Warmed up or cold the M7S would fail to start OK 90% of the time.

 

[LarsHP]

I have the Reference 7.1 which basically is an older version of Master 7, both using the PCM1704UK chips.

One of the things that I have been telling myself when comparing the specs of these DACs to the Yggdrasil specs is that the Audio-gd machines (contrary to the Schiit machine) uses a proper 24-bit DAC chip. Guess what? I was wrong!

The PCM1704 chip will receive a 24-bit signal, but only output 20-bit...

Check the specs (three online documents saying the same) :
http://html.alldatasheet.com/html-pdf/525802/TI/PCM1704U-K/27/1/PCM1704U-K.html
http://www.datasheetlib.com/datasheet/46809/pcm1704u-k-2k_ti-texas-instruments.html
http://elcodis.com/parts/515443/PCM1704U-K.html#datasheet

The claim is: typical signal-to-noise is 120 dB

... and compare to this:
https://en.wikipedia.org/wiki/Audio_bit_depth

... which says that 20-bits => 120.41 dB signal-to-noise.

All this may (partly) explain why Kingwa uses four chips per channel in balanced configuration and that the standard setting is 8x oversampling.

 

[DACLadder]

Page 4 of the PCM1704 datasheet with the measured o'scope pictures has always impressed me. Look at the page 4 bottom right picture (-120dB, 20-bit operatIon). You are flat line and in the noise floor with only 20-bits. No signal can be less than the number of bits of the DAC. The picture to the immediate left (bottom) is the PCM1704 in 24-bit operation. You can clearly see a sine wave above the noise floor at -120dB. So I would argue you are getting better than -120dB in 24-bit mode. At these very low levels non-linearity goes through the roof adding to the signal-to-noise ratio and fuzziness seen in the pictures at -120dB. So to measure a sine wave at -120dB in 24-bit mode is remarkable.

The Yggy is a good DAC. I owned one and listened to it exclusively for 6-months. But I always like the dynamic sound of the the AGD 7 series DACs. I believe Kingwa increased the DAC chip number to eight to improve dynamics. Analog output of a PCM1704 is just current with a miserly +/-1.2mA full swing. Paralleling outputs provides more current and easier to drive stray capacitance of the circuit providing faster transients from sample to sample. Could be other benefits like improved linearity and yes they may add to the overall signal-to-noise improvement.

And it is all about the sound for me. The PCM1704 DACs may no be perfect but they always sound like music.

 

[FredA]

Page 4 of the PCM1704 datasheet with the measured o'scope pictures has always impressed me. Look at the page 4 bottom right picture (-120dB, 20-bit operatIon). You are flat line and in the noise floor with only 20-bits. No signal can be less than the number of bits of the DAC. The picture to the immediate left (bottom) is the PCM1704 in 24-bit operation. You can clearly see a sine wave above the noise floor at -120dB. So I would argue you are getting better than -120dB in 24-bit mode. At these very low levels non-linearity goes through the roof adding to the signal-to-noise ratio and fuzziness seen in the pictures at -120dB. So to measure a sine wave at -120dB in 24-bit mode is remarkable.

The Yggy is a good DAC. I owned one and listened to it exclusively for 6-months. But I always like the dynamic sound of the the AGD 7 series DACs. I believe Kingwa increased the DAC chip number to eight to improve dynamics. Analog output of a PCM1704 is just current with a miserly +/-1.2mA full swing. Paralleling outputs provides more current and easier to drive stray capacitance of the circuit providing faster transients from sample to sample. Could be other benefits like improved linearity and yes they may add to the overall signal-to-noise improvement.

And it is all about the sound for me. The PCM1704 DACs may no be perfect but they always sound like music.

It sure is great, no doubt.

 

[Articnoise]

I have the Reference 7.1 which basically is an older version of Master 7, both using the PCM1704UK chips.

One of the things that I have been telling myself when comparing the specs of these DACs to the Yggdrasil specs is that the Audio-gd machines (contrary to the Schiit machine) uses a proper 24-bit DAC chip. Guess what? I was wrong!

The PCM1704 chip will receive a 24-bit signal, but only output 20-bit...

Check the specs (three online documents saying the same) :
http://html.alldatasheet.com/html-pdf/525802/TI/PCM1704U-K/27/1/PCM1704U-K.html
http://www.datasheetlib.com/datasheet/46809/pcm1704u-k-2k_ti-texas-instruments.html
http://elcodis.com/parts/515443/PCM1704U-K.html#datasheet

The claim is: typical signal-to-noise is 120 dB

... and compare to this:
https://en.wikipedia.org/wiki/Audio_bit_depth

... which says that 20-bits => 120.41 dB signal-to-noise.

All this may (partly) explain why Kingwa uses four chips per channel in balanced configuration and that the standard setting is 8x oversampling.

The number of bits in a DAC chip is not the same as effective resolution or signal-to-noise rate. Both can be explained in bits but mean different things. A 32 bit DAC chip doesn’t have more resolution or better S/N than 24 bit.

20 or 21 bits actual resolution is great.

 

[LarsHP]

So this could indicate that a HDCD signal should sound just at good as a 24-bit 44.1kHz I expect?

I am not sure that I would be able to hear the difference between a 20-bit and 24-bit bit signal anyway...
Difference between a 16-bit and a 20-bit signal is another story though.

 

[DACLadder]

I am not sure that I would be able to hear the difference between a 20-bit and 24-bit bit signal anyway...
Difference between a 16-bit and a 20-bit signal is another story though.

Yup, I agree with that statement. -120dB at 20-bits, if linear, is outside my level of detection. My amp has -95dN s/n ratio and it is quiet as a church mouse when holding an ear next to the speakers. But even so I can detect more detail with additional bits over the standard 16-bits of a CD.... in most cases.

My first CD player was a Sony with only 14-bit resolution and even it sounded very good in its day (mid 80s). Featured only one DAC chip that multiplexed between the left and right channels causing a 180 deg. phase shift at 20 kHz. From there seems sound quality went down hill fast in the 80/90s with introduction of the much more affordable 1-bit DAC players.

 

[FredA]

Yup, I agree with that statement. -120dB at 20-bits, if linear, is outside my level of detection. My amp has -95dN s/n ratio and it is quiet as a church mouse when holding an ear next to the speakers. But even so I can detect more detail with additional bits over the standard 16-bits of a CD.... in most cases.

My first CD player was a Sony with only 14-bit resolution and even it sounded very good in its day (mid 80s). Featured only one DAC chip that multiplexed between the left and right channels causing a 180 deg. phase shift at 20 kHz. From there seems sound quality went down hill fast in the 80/90s with introduction of the much more affordable 1-bit DAC players.

Around or beyond 20 bits, you will notice a difference with classical music. Any background info recorded at low/very low level will be more distinct. Same for ambience noise in live recordings. But only is the rest of your gears can get this above the noise threshold. Usually, an amp can do it. You mention your amp doing 95dB, but amps are usually specified at minimum listening level, meaning you have much more than the spec at normal listening levels.

 

[LarsHP]

Around or beyond 20 bits, you will notice a difference with classical music. Any background info recorded at low/very low level will be more distinct. Same for ambience noise in live recordings. But only is the rest of your gears can get this above the noise threshold. Usually, an amp can do it. You mention your amp doing 95dB, but amps are usually specified at minimum listening level, meaning you have much more than the spec at normal listening levels.

S/N in an amplifier is AFAIK specified at maximum output, not minimum.
Where did you see that it's at minimum?
And what is minimum listening level in an amp with analog volume control? For this very reason I find it hard to believe it's normal to spec at minimum level.

 

[DACLadder]

S/N in an amplifier is AFAIK specified at maximum output, not minimum.
Where did you see that it's at minimum?
And what is minimum listening level in an amp with analog volume control? For this very reason I find it hard to believe it's normal to spec at minimum level.

You are correct.... S/N is typically the ratio of maximum output to minimum or the background noise. Cranking the analog volume control down will reduce maximum output but not necessarily the noise floor. S/N would drop if the noise level remains the same as the control is adjusted. If noise level drops linearly then S/N is identical (but highly unlikely).

Dynamic range in a digital device is the ratio of maximum output to resolution of its lower bits acceptable above the noise floor.

 

[FredA]

S/N in an amplifier is AFAIK specified at maximum output, not minimum.
Where did you see that it's at minimum?
And what is minimum listening level in an amp with analog volume control? For this very reason I find it hard to believe it's normal to spec at minimum level.

Depends on the manufacturers. But i admit you are right in general. I was doing wishful thinking.

For instance, audio-gd states >115dB (presumably at full power, not mentioned), I found an instance of Simaudio specifying S/N at full power and Benchmark specs its state-of-the-art amp this way:

NOISE RELATIVE TO 2.83 VRMS
Output noise relative to 2.83 Vrms, inputs shorted

• -112 dB, A-weighted, Stereo Mode
• -109 dB, A-weighted, Mono Mode

So it depends really. Probably worth upgrading your amp of it is limited to 95dB at full power if audio-weighted. You need more than 130dB at full power to take full advantage of a dac good for 120 dB. Audio-weighted specs are a good thing to have in hand cause the noise at med/high frequency has more impact on definition. It's also important to note that It's common that a manufacturer do better than what is specified.

 

[DACLadder]

My old, tired ears have a s/n ratio of about 60dB LOL.... Any upgrade may be useless. Plus with speakers ambient noise hovers around -60dB. And I don't think I have any program material that is much below 16-bits and detectable. But in general, with higher bit recordings, I can notice better separation of instruments and improvement with echo and reverb tails.

As a test I re-normalized the classic Star Wars theme to -60dB. So peeks are at -60dB with average levels around -70dB or so. At -60dB I can barely hear audio when fully cranked up. So to have the ability to resolved down to -120dB would be remarkable in normal playback. If you want to try download these files from my One Drive location. Don't forget to return volume back to normal after trial.

Star Wars Theme 0dB .flac
https://1drv.ms/u/s!AqFmDQboxG1jgW1Rh5Ex89TVy_M3

Star Wars Theme -60dB .flac
https://1drv.ms/u/s!AqFmDQboxG1jgWxVP3YbhwHFebr3

 

[FredA]

My old, tired ears have a s/n ratio of about 60dB LOL.... Any upgrade may be useless. Plus with speakers ambient noise hovers around -60dB. And I don't think I have any program material that is much below 16-bits and detectable. But in general, with higher bit recordings, I can notice better separation of instruments and improvement with echo and reverb tails.

As a test I re-normalized the classic Star Wars theme to -60dB. So peeks are at -60dB with average levels around -70dB or so. At -60dB I can barely hear audio when fully cranked up. So to have the ability to resolved down to -120dB would be remarkable in normal playback. If you want to try download these files from my One Drive location. Don't forget to return volume back to normal after trial.

Star Wars Theme 0dB .flac
https://1drv.ms/u/s!AqFmDQboxG1jgW1Rh5Ex89TVy_M3

Star Wars Theme -60dB .flac
https://1drv.ms/u/s!AqFmDQboxG1jgWxVP3YbhwHFebr3

Good points.

The specs only tell a small part of the story. In my case, I'd say the power grid and ambient noise are the limiting factors to sound quality. Not the gears themselves. And my hearing, which is not always consistent.