Audio-gd NFB-7 - Page 8

Thank You axw. I will stay tuned for Your comments.

PSU's are important.

Dac's in general are sensitive for the quality of the supporting PSU's.. and the ESS-chip is known to be even trickier than other brands dac's to do it's best. If You have looked at the Buffalo-dac based on the ESS-chip, You know that alot of efforts are put in that area with regulators and filtering.

/Jan

Here's a summary of main differences between NFB-10ES and NFB-7 (with 80 Mhz TCXO). Both have passed 2 weeks of continuous playback. For comparisons I used HiFace Evo and a pair of active Adam S3X-V speakers, both DACs tested via balanced XLR outputs only. Most of the differences are probably attributable to improved power supply, I guess.

Try to get the NFB-10 to have the TCXO upgrade and the "blur and fuzziness" will improve, also the sound stage will increase in depth and instruments will seperate better. This is what I found after the doing the TCXO upgrade on my NFB-1. The width of the sound stage would remains though.

The higher end audio-gd would always have better SNR and soundstage. But like you have said it depends on whether the amp and the source equipment can take advantage of it.

Thank You axw for taking the time writing and sharing.

I have also ordered one of these to serve in my main-rig. I will use it with the DI.

/Jan

Quote:

Originally Posted by borrego 

Try to get the NFB-10 to have the TCXO upgrade and the "blur and fuzziness" will improve, also the sound stage will increase in depth and instruments will seperate better. This is what I found after the doing the TCXO upgrade on my NFB-1. The width of the sound stage would remains though.

That's a universal benefit in any source -- the better the performance of the oscillator, the clearer and more focused your music will sound.  And it isn't all about the jitter levels, of which there are different kinds: phase jitter and period jitter, so if someone just gives you a jitter measurement you need to ask which one and understand the differences between the two.  More importantly is the stability of the oscillator rated in PPM (parts per million).

A good TCXO should be rated at an error rate of 1ppm per year.  That means that the frequency of the oscillator will be off by about 30 seconds per year.  How this translates to audio is in the pitch (which can also effect perceived soundstage).  As you can test with some soundcards, if you change the sample rate lock during music thats playing, you will hear the music either speed up or slow down.  This is an example of drastic deviation, but it illustrates the effect.

Let's say you have a clock with perfect accuracy that had 0 ppm error (this doesn't exist, the closest one is a rubidium oscillator made by Antelope Audio that is off by 0.03 parts per BILLION, which is 1 second every 1000 years).  With a 0ppm error rate, if you were listening to a continuous tone at A440 (a standard musicians tuning fork tone is note A at 440 Hz) then you would hear exactly 440 Hz (assuming it was digitally generated and recorded at exactly 440Hz).  But with a 1ppm error rate you would be hearing 439.999999 Hz.  The difference isn't much, but over time it will get slightly worse.  And it can get even worse if the temperature stability isn't very stable.  These reasons are why cheap oscillators can make your music sound like crap.  If you are not using a TCXO and there are large thermal differentials inside your component, this will effect the stability of the clock.

Most TCXO are rated between 0.5ppm - 2ppm.  Implemented properly in a temperature stabilized circuit such as the Dexa Neutron Star clock (which you can add to your source), the error rate can be brought down to 0.02ppm but the clock will set you back about $700.  What most manufacturers won't tell you is that even these higher quality TCXO oscillators will increase in error by about 2ppm per year, so someone who is really anal about their audio might replace the oscillators every few years. This could also explain why when people hear new gear they think it sounds better than their older gear and want to upgrade.  In many cases the upgrade bug can be avoided by replacing those older oscillators with newer ones, assuming that the difference of tenths, hundredths, thousandths, ten thousandths, etc. of a Hz is detectable, or rather uncomfortable to listen to.

Hi Ipod,

you certainly put it in a way very easy to understand. another way of looking looking at it, if one were to play normal CD music with high PPM clock, one will lose a sample between an hour or between a song or within a song ? all I know is that with the antelope clock, you will only lose a sample every ??? days.

Is it true that all Sabre based comes with the TCXO chip as per the website? It say since Feb.

I'm also a bit confused as Kingwa is not confident with regards to the 24/192 support for coaxial and 24/96 for optical. i thought the Sabre can support up to 32/192 without any issues?

I will be using the macbook through the digital optical. 

Quote:

Originally Posted by ccschua 

Hi Ipod,

 

you certainly put it in a way very easy to understand. another way of looking looking at it, if one were to play normal CD music with high PPM clock, one will lose a sample between an hour or between a song or within a song ? all I know is that with the antelope clock, you will only lose a sample every ??? days.

I don't think you're losing a sample, but the timing of the sampling is not 100% accurate; it's off by a fractional amount so the end result is a tiny shift in pitch.  But those external master word clocks such as the Antelope Audio gear are beneficial in studio settings where you need to synchronize the sample rate between many pieces of gear.  This is not the case in a home audio system.

There was an article floating around that I read from an online magazine (if you don't have a subscription you have to purchase it) that did a review of all the top notch clocks by the big name companies.  In every case where only a transport was hooked up to a DAC, they actually performed worse than the stock oscillators inside the source components themselves.  The reason for this is the relatively long length of cable connecting the components.  Even though the master clocks themselves were more stable than the stock oscillators, the jitter introduced by an external BNC cable when used for the purpose of word clock output>input vs. the very short signal paths inside the transports and DACs resulted in worse measurements.  I think there was one word clock unit that performed just about on par with the source components themselves, but why spend all that money to end up with the same results?  If you want increased performance, install a high-performance clock inside your component.

If you are a musician and using multiple effects processors, ADCs, DACs and recording equipment then you will net a benefit by using a master clock.

Quote:

Originally Posted by snip3r77 

Is it true that all Sabre based comes with the TCXO chip as per the website? It say since Feb.

 

I'm also a bit confused as Kingwa is not confident with regards to the 24/192 support for coaxial and 24/96 for optical. i thought the Sabre can support up to 32/192 without any issues?

I will be using the macbook through the digital optical. 

I can't speak for Audio-gd's Sabre series, but you should try some other sources instead of your macbook digital output.  There's a very high quality, cheap Chinese audio player made by QLS Hi-Fi and called the QA-350 v2.  I am using a Mac Mini's optical output temporarily until I receive my Transporter back, and I can tell you that there is very significant difference in performance.

I know that the Ref 7.1 has a TCXO that the DSP-1 module is slaved to.  If I really wanted to take the Ref 7.1 up to another level of performance I would disconnect the TCXO and hook up a Dexa Neutron Star (or second choice would be an Audiocom Ultraclock mk.2).  But since I don't have that extra money, I have not done that in my modified Logitech Transporter.  I have installed an Audio-gd low noise power supply and two Audio-gd JZ1 clocks which use TCXO oscillators.  The Transporter requires two oscillators to output all the resolutions from 8 bit to 24 bit and all the sample rates from 32kHz to 96kHz.  Since adding the clocks, power supply, twin torroids, lower impedance caps, and least significantly better quality wire the two units are a match made in heaven, and this is most likely because both units are using the same TCXO oscillators.

Quote:

Originally Posted by borrego 

Try to get the NFB-10 to have the TCXO upgrade and the "blur and fuzziness" will improve, also the sound stage will increase in depth and instruments will seperate better. This is what I found after the doing the TCXO upgrade on my NFB-1. The width of the sound stage would remains though.

 

The higher end audio-gd would always have better SNR and soundstage.

Very likely you are right. If that is the case, then what is the effect of extra r-cores and where does better SNR hide?
 

Quote:

Originally Posted by IPodPJ 

I know that the Ref 7.1 has a TCXO that the DSP-1 module is slaved to.  If I really wanted to take the Ref 7.1 up to another level of performance I would disconnect the TCXO and hook up a Dexa Neutron Star (or second choice would be an Audiocom Ultraclock mk.2).  But since I don't have that extra money, I have not done that in my modified Logitech Transporter.  I have installed an Audio-gd low noise power supply and two Audio-gd JZ1 clocks which use TCXO oscillators.  The Transporter requires two oscillators to output all the resolutions from 8 bit to 24 bit and all the sample rates from 32kHz to 96kHz.  Since adding the clocks, power supply, twin torroids, lower impedance caps, and least significantly better quality wire the two units are a match made in heaven, and this is most likely because both units are using the same TCXO oscillators.

Interesting. So the TCXO feeds not PCM1704 chips but DSP-1 board? This reminds that NFB sabre dacs do not have a DSP and any de-jittering that takes place is only in the Sabre chip. I got Audiophilleo2 today and must admit (again) that relying solely on Sabre for de-jittering is not quite the best idea. Every transport upgrade so far brings a lot of improvement. Btw, it's interesting to compare Transporter to Audiophilleo..  Stock Transporter supposedly measures "11ps at oscillator, 17ps at DAC and 35ps at S/PDIF receiver" -- how does that compare to Audiophilleo's 2.5 ps / 8 ps RMS phase jitter and < 5 ps RMS period jitter ? Does that mean that Audiophilleo bests stock Transporter? I am not sure what is meaant by "at DAC" and "at oscillator"..

Would the NFB-7 be a good DAC to pair with the Apex Peak/Volcano and Audeze LCD-2?

My source would be my PC. I have SPDIF out but would I be better off running SPDIF directly to the DAC or PC > USB > USB to SPDIF converter > DAC?

My NFB-7 should be arriving in the next day or two.

NFB 7.1 is up at the Audio-gd website. Uses the new output stage from NFB-9 it seems (the one without the module boxes). Nice to see Kingwa keep this DAC in production.

NFB7.1 is already terminated!!

I just place dorder 1 item and waiting to play with