Hi John,
What about my question regarding how the output common mode DC level is established? Was that a DC servo that I saw for that?
Cheers,
Bob
No Scott
That was your misunderstanding. Charge propagates at close to c. You also as I recall got wrong that instantaneous charge changes will occur.
But you did get right that Ohm's law does not consider time and assumes c is infinite.
When you include c as finite it does change some of what you like to call "First Principles."
Now why you keep confusing charge with electrons is a different issue.
Demian,
Interesting chip. I don't feel too embarrassed about missing it as it seems to have been released just a bit after I did my design work.
Worth noting is that the lock range is +/- 3% of standard data frequencies.
The jitter spec shows typical 100 pS RMS with I presume an external crystal clock reference. Although that detail wasn't quite clear to me. It also is listed to be able to use a resonator. So it seems they stuffed enough options into the chip to cover just about every application.
One other detail wasn't clear as they just show in the block dia. data going through an "I/F" for conversion.
So I suspect this chip is now the method of choice for many of the Japanese manufacturers. It also may explain why I couldn't get my CD player to sync properly with the rest of 48k DSP system.
Thanks
ES
Interesting chip. I don't feel too embarrassed about missing it as it seems to have been released just a bit after I did my design work.
Worth noting is that the lock range is +/- 3% of standard data frequencies.
The jitter spec shows typical 100 pS RMS with I presume an external crystal clock reference. Although that detail wasn't quite clear to me. It also is listed to be able to use a resonator. So it seems they stuffed enough options into the chip to cover just about every application.
One other detail wasn't clear as they just show in the block dia. data going through an "I/F" for conversion.
So I suspect this chip is now the method of choice for many of the Japanese manufacturers. It also may explain why I couldn't get my CD player to sync properly with the rest of 48k DSP system.
Thanks
ES
Turntable 33.3 rpm wow is generally a smooth (sinewave) frequency deviation, with a period of approx 1.8 seconds.
Turntable flutter is of much shorter period, and usually related to drive motor mechanical eccentricities, but still sine-ish in nature.
Oscillator frequency instability is however 1/f/erratic in nature, due to the crystal noise itself, and also 1/f noise of the sustaining amplifier stage, and both noise sources interact.
1/f noise is an insidious error source, and distinctly heard once learned and noted.
Try comparing playback of pink noise on two systems, one with source of 'normal' 1/f noise, and one that is very low in intrinsic 1/f noise....the reproduced resultant can be quite different, and less 1/f noise is disarmingly better/accurate.
The VLF deviation adds a 1/f noise modulation to the decoded audio, and is the statistically major component of jitter.
No arguments from me.
Dan.
George, Dick,
I do have a rubidium clock. Fairly fast to get a lock in the PLL locked crystal oscillator included. But phase noise is higher than my best reference oscillators. But I just couldn't pass up the chance to get a piece of equipment that used to be almost unobtainable.
Now there are some that can be programmed to be used as a CD clock. But a good crystal oscillator should be better in that application.
I do have a rubidium clock. Fairly fast to get a lock in the PLL locked crystal oscillator included. But phase noise is higher than my best reference oscillators. But I just couldn't pass up the chance to get a piece of equipment that used to be almost unobtainable.
Now there are some that can be programmed to be used as a CD clock. But a good crystal oscillator should be better in that application.
Hi Simon,
Yes, I was trying to draw attention to that. A fractional divider should be stable though. Just nothing discipled as corrections mean jumps in frequency. However, if you stick the oscillator in holdover mode for the listening session you should be okay.
Sadly, all the overkill won't make any difference to digital playback as long as all the gear is synced together with a common clock. For straight consumer playback - zero difference. I only brought it up to show the crazy limits a person could reach as a pointless exercise.
I use a GPS, 10 MHz system (T-Bolt) to sync all my bench instruments. I'll bet you have done the same or similar. Need a bigger, better distribution amp now (the HP 5087A does not have an isolated output common).
-Chris
Yes, I was trying to draw attention to that. A fractional divider should be stable though. Just nothing discipled as corrections mean jumps in frequency. However, if you stick the oscillator in holdover mode for the listening session you should be okay.
Sadly, all the overkill won't make any difference to digital playback as long as all the gear is synced together with a common clock. For straight consumer playback - zero difference. I only brought it up to show the crazy limits a person could reach as a pointless exercise.
I use a GPS, 10 MHz system (T-Bolt) to sync all my bench instruments. I'll bet you have done the same or similar. Need a bigger, better distribution amp now (the HP 5087A does not have an isolated output common).
-Chris
OK. I've been convinced..... so who makes the best (lowest jitter) Crystals for adc/dac?
The MasterLink uses 3... all a brand called FARGO.
It also uses 5532 for all analog I/O which i can live with but analog input is applied to a polar electrolytic coupling cap !!! Same on output.
Some devices used are: ADSP-21065L (SHARK) with 30MHz xtal; AK4393V; CS8404A; CS8414 -cs
CD-RW drive made by LITE-ON IT Corp, model SOHR-5239V
THx-RNMarsh
The MasterLink uses 3... all a brand called FARGO.
It also uses 5532 for all analog I/O which i can live with but analog input is applied to a polar electrolytic coupling cap !!! Same on output.
Some devices used are: ADSP-21065L (SHARK) with 30MHz xtal; AK4393V; CS8404A; CS8414 -cs
CD-RW drive made by LITE-ON IT Corp, model SOHR-5239V
THx-RNMarsh
Last edited:
Hi Richard,
For the purposes of audio playback, a crystal oscillator with a proper drive level. Ovenized oscillators are better than you need, but they do make compensated oscillators that run less money. Those make the entire assembly a module, so you just have to drive the chip with the output. Most can be bought with a sine wave output, or CMOS for a standard square wave output. The chips often expect a sine wave, so if the area is electrically quiet you don't need to be concerned about circuit noise generated jitter.
-Chris
For the purposes of audio playback, a crystal oscillator with a proper drive level. Ovenized oscillators are better than you need, but they do make compensated oscillators that run less money. Those make the entire assembly a module, so you just have to drive the chip with the output. Most can be bought with a sine wave output, or CMOS for a standard square wave output. The chips often expect a sine wave, so if the area is electrically quiet you don't need to be concerned about circuit noise generated jitter.
-Chris
Hi Richard,
I'm not much for brand names. The actual technology is the important thing to worry about. Anything sold for this purpose will probably be over-priced with embellished claims to performance.
In a nutshell. Find out what frequency you need, buy the oscillator at that frequency. Look to make certain of the voltage levels, like a 3.3 V chip set. If the original uses a crystal, I would use a sine wave output type, output to x-tal input (not the drive side).
It really is that easy.
-Chris
I'm not much for brand names. The actual technology is the important thing to worry about. Anything sold for this purpose will probably be over-priced with embellished claims to performance.
In a nutshell. Find out what frequency you need, buy the oscillator at that frequency. Look to make certain of the voltage levels, like a 3.3 V chip set. If the original uses a crystal, I would use a sine wave output type, output to x-tal input (not the drive side).
It really is that easy.
-Chris
Check out Demian's post 30 in this: http://www.diyaudio.com/forums/digi...-phase-noise-jitter-crystal-oscillator-3.html
That's as far as I got in the thread, and one of the links appears not to work now. But the HP article by Karlquist is most informative.
DM, from the post:
"In the US you can go to Wenzel/Croven to get crystals. It will be slow and expensive but they are as good as you can get."
That's as far as I got in the thread, and one of the links appears not to work now. But the HP article by Karlquist is most informative.
DM, from the post:
"In the US you can go to Wenzel/Croven to get crystals. It will be slow and expensive but they are as good as you can get."
some people here have looked: H I F I D U I N O: Clock in Buffalo II DAC
but they didn't plot one I found a while ago:
but they didn't plot one I found a while ago:
The crystal thing is a confuser. Its there to provide a sample rate detection for downstream stuff. The chip doesn't need it. Its internal VCO can get to 100 pS without external help.
The 3% lock range is necessary to accommodate lousy sources. I had a PC motherboard that ran about 2% low and lots of stuff would not work with it. A consumer will just return the external device (AV receiver) not realizing that the PC is crap. The only fix is to make really tolerant products. There is no option in mass market for returns.
I posed a jitter plot recently http://www.diyaudio.com/forums/loun...ch-preamplifier-part-ii-6816.html#post4327006 that was this chip driving an AK4490 DAC with no special magic running spdif. The sidebands are all below -125 dBC.
I have measured the spectrum of the phase modulation of a turntable. I used a JVC CD4 test record with a 30 KHz unmodulated carrier and it had the same spectrum as you showed. I measured this on a Linn and on a Panasonic SP10. On the SP10 there were two distinct peaks on either side of the target frequency. They were sharper than the Linn but seemed to be a deadband in the servo.
I still do not see how the distribution of phase noise in the low frequencies even added to the high frequency phase noise would be more objectionable than the deterministic wow and flutter + random phase noise of a turntable. A good turntable will have a flutter level of .05% or around -70 dB and this is weighted with a bandpass from 1 Hz to about 20 Hz (CCIR). That compared to the -125 dB I measured with a digital link. I would think that a 50 dB lower phase modulation regardless of its spectrum would be way less audible.
Some of the audiophile claims for low close in phase noise are fantasy and not achievable with real hardware.
Same thing it does in a non-superconducting ring, 'cept it doesn't bump its head.
se
- Status
- Not open for further replies.