John Curl's Blowtorch preamplifier

Status
Not open for further replies.
Jump to Latest
#11,686
1audio said:


However when we talk about audio now the DACs are expected to be accurate to at least 1/2 LSB. That means the timing of the clock must be better still to meet a minimum accuracy, or better than 1 part in 2 to the 17th of the sample rate. (or less than 100 dB down). It really needs to be much better.
Click to expand...


Thank you for your input, Demian.

Let me ask you this:
In 80 MHz clock, what accuracy in PPM would be considered as good?
 
#11,690
Joshua_G said:


Let me ask you this:
In 80 MHz clock, what accuracy in PPM would be considered as good?
Click to expand...

It all depends on what matters. To get your cell phone to sync with the base stations 5E10-11 seems to be required.

To get a lock between a source (CD/Music server) and a dac the clock must be in range for the DAC. .05% should be good enough for the pickyest dac (the Apogee was the narrowest DAC I have played with).

However Jitter is usually measured in nanoseconds or picoseconds. and then usually expressed in rms values. Last I heard 10 pS was considered the target but lower values would be better. It can also be expressed as phase noise, a measure of the close in sidebands generated by the clock (something similar to FM modulation). The kind of numbers you will see are dB below carrier and X frequency from the carrier. Numbers like -105 dB at 100 Hz are pretty good However its possible to get as low as -130 dB at 10 Hz and -170 dB at 1 KHz. But you will pay. (The oscillator with those numbers is around $2500. And their obsession with low noise is like John Curl with details on the Blowtorch.)

And its not clear that jitter on the link is the dominant issue. There could be many other issues that affect the sound more than the clocks. Even noise coupling through the connections could affect the jitter of the internal clock more than the jitter in the source.
 
#11,692
1audio said:
Numbers like -105 dB at 100 Hz are pretty good However its possible to get as low as -130 dB at 10 Hz and -170 dB at 1 KHz. But you will pay. (The oscillator with those numbers is around $2500. And their obsession with low noise is like John Curl with details on the Blowtorch.)
Click to expand...

$2500 ?!

Less than -130 dB @ 10 Hz phase noise is available in a good OCXO and it would cost you probably under $500. And you get a very high stability as well. Perhaps somebody can organise a group buy for this kind of clock for CD? The price could be very reasonable compared with 2.5K 😀 .

Alex

P.S. -105 dB at 100 Hz is a pretty standard performance for a good TXCO for $20 .
 
#11,694
Originally posted by 1audio

To get a lock between a source (CD/Music server) and a dac the clock must be in range for the DAC. .05% should be good enough for the pickyest dac (the Apogee was the narrowest DAC I have played with).
Click to expand...


So, does it matter whether the crystal clock has accuracy of +/-100ppm or +/-25ppm?


Originally posted by 1audio

However Jitter is usually measured in nanoseconds or picoseconds. and then usually expressed in rms values. Last I heard 10 pS was considered the target but lower values would be better. It can also be expressed as phase noise, a measure of the close in sidebands generated by the clock (something similar to FM modulation). The kind of numbers you will see are dB below carrier and X frequency from the carrier. Numbers like -105 dB at 100 Hz are pretty good However its possible to get as low as -130 dB at 10 Hz and -170 dB at 1 KHz. But you will pay. (The oscillator with those numbers is around $2500. And their obsession with low noise is like John Curl with details on the Blowtorch.)
Click to expand...


Well, I am obsessed about sound quality, however, I don't have the finances to go as far as I'd like to … 🙂

There are two 80MHz crystal oscillators I'm considering, both have accuracy of +/- 25 ppm and Jitter RMS of 1 psec Max. One costs $3.62 and doesn't mention phase noise, while the second costs $29.08 with phase noise of about -105dBc/Hz at 100 Hz and -131 dBc/Hz at I KHz.

I believe the second one is preferred and is good enough. However, is there a crystal oscillator costing about $100 with a better noise floor that will justify the extra cost?


Originally posted by 1audio

And its not clear that jitter on the link is the dominant issue. There could be many other issues that affect the sound more than the clocks. Even noise coupling through the connections could affect the jitter of the internal clock more than the jitter in the source.
Click to expand...


Well, each jitter source should be considered and treated by itself. I believe that the fact that there is jitter on the link isn't an excuse to use $1 oscillator.
 
#11,698
Frankly some of this stuff reads like taking apart a watch to find out the time. Why not run Linux on a small profile one card PC and use an external pro grade Firewire DAC?

I know there's more to it but something like "world's best ripper" piped to Firewire out driver.

I forgot to mention, if you go the other way, instrumentation quality preterminated cables are available from several vendors. Ironically they are often cheaper than "audiophile" products.
 
#11,699
john curl said:
This is like a group of teenagers, when we talk about digital. Lots of opinion, without substance.
Click to expand...

Please speak for yourself. When the CD was introduced it was an marvel of making technology affordable and a lot of good engineers put in a lot of work into pulling it off. I knew some of them.

Optimization issues turned up and more people moved in to solved them.

The jabber of all the marketing bozos was just a distraction in the process.
 
#11,700
1audio said:



S/PDIF is from a family of self clocking data links. They do work quite well if your tolerance for timing errors is pretty wide. The data in HDMI is passed the same way, but its timing tolerance is much higher- approx 30% jitter of the bit rate will get the bits decoded properly and it will end up lighting the correct pixel on the LCD (how can it miss?).
Click to expand...

HDMI passes an explicit clock, it is not the same as S/PDIF. Four pair differential LVDS, r, g ,b , and clock. The low speed bookkeeping channels, I don't know about.

(Lineup - LVDS is low voltage differential signal)
 
Status
Not open for further replies.
Top Bottom