Real or fake PCM63?

[spzzzzkt]

After some listening to various combinations I've currently got 3 and 6 installed as a pair. 7 is on the other channel so not an option.

After spending time listening I'm convinced the HD figures only come into play at quite high levels. 5 wasn't an obviously bad chip just very fatiguing to listen to. 4 looked like it should be the pick, but there was something not quite right about the sound - a slight "ear pressure" similar to what I experienced with the K's. As can be seen in the spectrum the noise floor on this chip is elevated from the fundamental out to 4Khz or so when compared with the last 2 chips.

My take is that the noise floor is the key determinant in a chips sound providing that the harmonic distortion levels aren't extremely high. From listening to the amped clips of the chips, I can't detect definite harmonics, except in the case of 5. There might be faint harmonics audible in 3, but if they are there they are only at the margins of audibility. Based on the worst harmonic of 3 this would suggest that somewhere between -115 and -116dB is the limit at which the harmonic peaks have cease to have any possible audible significance. This is about the same limit Bob Katz claims for a properly dithered CD.

Bob Katz' comments that dithering helps us hear beneath the noise floor might go some way to answering why the noise floor seems to be relatively important? The noise floor overlays the quietest harmonics so differences in noise floor are going to be clearly apparent (relatively speaking!!).

 

[PA0SU]

Spencer, IY,

found some interesting comments on Transport jitter here:

http://members.chello.nl/~m.heijligers/DAChtml/jittermeas/jitter.html

the second to last conclusion is that:

Look at:

http://www.tentlabs.com/Products/DACupgrades/XODAC/index.html

 

[PA0SU]

IY,

the transport recovers bits from the disc and converts to a a digital signal transmitted by spdif. If you are relying on recovered clock, any degradation of the transmitted signal will increase the level of jitter recovered by the receiver chip which is then passed on to digital filter and to the dac chips. So if you are relying on recovered clock changes to the transport and cable which give a cleaner digital signal, and result in reduced jitter in the recovered clock will be audible.

If the dac uses a secondary PLL to generate a low jitter clock, the influence of the recovered clock is significantly reduced. I wouldn't suggest that the transport has no effect, but it will be far less significant.

Paul

A mutch better aproach is to put the low jitter XO near the reclocking circuit between digital filter and DAC and lead the clock signal back to the transport. In this way YOU NEED NO PLL AND VCXO which gives always more jitter than a descent XO!

 

[PA0SU]

I should add that those comments regarding reducing transport dependence come from the team that designed the precursor of the Tent DAC. The reclocking circuit Herb posted is almost identical - only resistor values are different - to the circuit in their DIY dac.

http://members.chello.nl/~m.heijligers/DAChtml/dig_r2b.pdf

http://members.chello.nl/~m.heijligers/DAChtml/us.htm

http://members.chello.nl/~m.heijligers/DAChtml/dactop.htm

In my postings 141 and 140 I gave you on overview of the diagrams of this TentDAC. Here is NO PLL and NO VCXO and the XO is outside this diagram.

 

[irgendjemand]

A mutch better aproach is to put the low jitter XO near the reclocking circuit between digital filter and DAC and lead the clock signal back to the transport. In this way YOU NEED NO PLL AND VCXO which gives always more jitter than a descent XO!

Two questions, Herb:

1. If so, this is different then "Tent-Link", which uses XO and VCXO.

2. Is this possible to do also when DAC and Transport having different Quartz frequencies ?

Regards,

 

[PA0SU]

My take is that the noise floor is the key determinant in a chips sound providing that the harmonic distortion levels aren't extremely high.
Bob Katz' comments that dithering helps us hear beneath the noise floor might go some way to answering why the noise floor seems to be relatively important? The noise floor overlays the quietest harmonics so differences in noise floor are going to be clearly apparent (relatively speaking!!).

Because of dithering the noise floor rises from -135 to -125 dB with my analyser and nearly all my PCM63's.
I can not believe that a noise floor of, say, -132 dB is audible worse than a noise floor of -135 dB.

 

[spzzzzkt]

The noise floor you see depends on the size of the FFT - it isn't the actual noise floor of the DAC.

Doubling the number of sample points lowers the displayed noise floor by 3dB. The spectra I've posted made using 262,144 points vs the 4096 on your analyzer.

CD's still have a signal to noise ratio of 96dB, it's just that you can hear down to at least -115dB with dithering.

I really don't any point in continuing with this thread - it's basically turned into a promotional forum for tentlabs.

 

[PA0SU]

Two questions, Herb:

1. If so, this is different then "Tent-Link", which uses XO and VCXO.

2. Is this possible to do also when DAC and Transport having different Quartz frequencies ?

Regards,

Ans.1 My solution is 'dangerous'. If you forget to switch on the DAC (with the XO in it) the transport does not have a clock signal. Most transports do not like this. At least the CD-spindle will turn very fast! In other words, this is not a commercial solution, but for a serious DIYer it works fine. Just switch on transport and DAC with one and the same switch....

Ans.2 In this case you need a PLL and VCXO in you transport to translate the one frequency to the other.....
My solution in such a case is: I buy myself another transport from 'Marktplaats' for a 30 euro. The CD624's I use for my expriments are plenty in the Philips town Eindhoven........

 

[PA0SU]

The noise floor you see depends on the size of the FFT - it isn't the actual noise floor of the DAC.

Doubling the number of sample points lowers the displayed noise floor by 3dB. The spectra I've posted made using 262,144 points vs the 4096 on your analyzer.

Okay, okay, but the differences between the 'bare noise' and the noise enhanced by the dithering wil be the same apart from the method of measuring.

 

[irgendjemand]

I really don't any point in continuing with this thread - it's basically turned into a promotional forum for tentlabs.

Don't quit yet, Paul!

Believe me, I wanted to quit / unsubscribe at least 5 times.... but, I am at least waiting to know that you enjoying the sound of your D1V3 + "Y". I am also waiting to hear from our dear diyaudio member IMS about his conclusion, same from other members that are seemingly "stuck" in the process.

I think neither you nor I ever bought a Tent Clock, so it is not of a really great promotion for them yet . I am interested to hear your final decision however, also about the DAC re(c)locking. I remember that you already did something similar, without using a Tent.

And still, several people thinks that a clock is not needed, for example in my configuration.

Last, but fairly: Herb is not 100% doing Tent re-clocking. He is for the “dangerous” variant, which is at least slightly different, meant “for diy” (see his Post #288 ).

But yes, this thread already went long and most subjects have been discussed.

Where is IMS ??

Greetings!

 

[spencer]

IMS is working hard to make the DAC sound. He asked me few questions few days back.

 

[ims]

D1V3 in Progress...

Yes, indeed. I am working hard to make the DAC sounds. I just connect the power supply cables at the bottom D1V3 pcb when I read your post, IY. Many thanks to remember me; your posts always got me interested. I am not far away from the hot iron

I am always following this thread everyday with great attention. I want to learn more from you guys, and hope Paul will not quit since his contributions were very interested and reading to his opinions is absolutely worthy. Even with his efforts to give all of us the measurements of his equipment, give his best to this thread.

Paul, after I read your post at http://www.diyaudio.com/forums/showthread.php?postid=1477744#post1477744 and http://www.diyaudio.com/forums/showthread.php?postid=1466231#post1466231

actually make me interested to buy tentlabs product also. And I know that you are not promoting for this particular product when you posted the threads. So I will try it someday and of course the other method suggested by Spencer using DIR9001. I will decide what the most preferable to my ears, no matter that method was promoting or not by someone.

So please don't quit Paul, we are still looking for your post, hopefully in the near time. Thanks to all of you guys

-ims-

 

[ims]

PCM63PY 0403 G4 045

Yes, I got this chips from A'af also.

-ims-

 

[irgendjemand]

Ans.1 My solution is 'dangerous'. If you forget to switch on the DAC (with the XO in it) the transport does not have a clock signal. Most transports do not like this. At least the CD-spindle will turn very fast! In other words, this is not a commercial solution, but for a serious DIYer it works fine. Just switch on transport and DAC with one and the same switch....

Ans.2 In this case you need a PLL and VCXO in you transport to translate the one frequency to the other.....
My solution in such a case is: I buy myself another transport from 'Marktplaats' for a 30 euro. The CD624's I use for my expriments are plenty in the Philips town Eindhoven........

Thanks for the explanation, Herb.

Let's see: I bought my modified Marantz CD-10 for about 350 Euro. Modifications (specially done for me) includes (I don't have the terminology in English; this must sounds a bit funny now...) :

1. New Quartz, sitting in "Silicon-pack".
2. Copper isolation of the Quartz (electric).
3. A bit bigger Capacitors (Oscon) on the separate PSU for the control-plat of the motor & mechanic Unit.
4. A bit bigger capacitors around the Quartz-Chip.
5. Some new capacitors in the Digital section (even though it is being used as a Transport) and mainly on the Digital-Out unit.
6. New Gold-Fuses.
7. Rhodium coated power cable.
8. Absorber Materials (against vibrations)
9. Both sides of the reading "eye" where cleaned…
10. The Tray got a new gear-wheel and gummies.

Well, changes of the above nrs. 1-8 makes a big difference in the sound. If you mean that I could save all this money & work by simply implementing of a clock on a 30.- Euro used Philips CD player (transport) - I can't say anything: I don't know. The Muslims are always saying: "Allaaaa Achbar" = GOD IS BIG!

My personal thoughts? A clock/re-clock/re(c)lock is welcome as a “stabilizer” but not instead of all the above.

Greetings!

p.s. A re(c)locking of the digital signal (with XO and VCXO, with sending the signal from the DAC back to the transport & using a separate cable, separate PSUs, etc.) will cost me, as a non-diy- person, about 800 Euro... Can anybody here guarantee me an adaquate improvement for this money, especially when my recovery-unit is already doing a f_ _ _ ing great job? I wonder...
AND: This is NOT about the money. This is about the sound!

 

[irgendjemand]

Re: D1V3 in Progress...

IMS,

good to have you here! OK, I will be waiting, of course...

Also many thanks for supporting Paul with his amazing energy here in diyaudio, may I say - in behalf the different threads.
I didn't know about the
http://www.diyaudio.com/forums/showthread.php?postid=1477744#post1477744 and http://www.diyaudio.com/forums/showthread.php?postid=1466231#post1466231. This means that Paul is already using Tentlabs products and only I'm being left behind

One more question guys: With all this terrific work, when do you find time to listen to... live music?

Yours, IY.

 

[Bernhard]

Re: Re: D1V3 in Progress...

My take is that the noise floor is the key determinant in a chips sound providing that the harmonic distortion levels aren't extremely high.

When I tested the PCM56 with MSB adjust, I always made a rough adjustment by listening to the -60dB signal with headphones and it was always obvious from listening wether the chip was good or bad or even very good before watching the signal on the analyzer.
I think I would not need the analyzer anymore to test chips....
And that would be a good test how audible the distortion is: Adjust the MSB on a PCM56 only by listening to the test signal. The uncertainty starts around -62 dB relative / -122 dB absolute.

Noise is just noise.

 

[spzzzzkt]

Noise is just noise.

I don't think so. Several of the PCM63K's I have showed significantly lower HD than several of the PCM63P-Y I had received in the first batch. According to your theory they should have sounded better, but they sounded much, much worse. The clear difference from the specturm analysis was a higher, "grittier" noise. Not a huge difference in noise floor, maybe only 3dB but the effect was clearly audible.

Similarly with the current PCM63P-Y I've been testing and listening to a couple that measure better in terms of HD but sound worse and the only thing I can see in the spectrum analysis is low level junk in and just above the noise floor of the FFT.

I don't understand what the junk and grit are revealing about the inner workings of the dac, but based on what I've heard the effect is audible. I'll take some small comfort in the knowledge that Pass Labs were selecting PCM63P-K for their D1 DAC's on the basis of noise and linearity, not just THD figures.

 

[PA0SU]

Re: Re: D1V3 in Progress...

Well, changes of the above nrs. 1-8 makes a big difference in the sound. If you mean that I could save all this money & work by simply implementing of a clock on a 30.- Euro used Philips CD player (transport) - I can't say anything: I don't know. The Muslims are always saying: "Allaaaa Achbar" = GOD IS BIG!

My personal thoughts? A clock/re-clock/re(c)lock is welcome as a “stabilizer” but not instead of all the above.

Greetings!

p.s. A re(c)locking of the digital signal (with XO and VCXO, with sending the signal from the DAC back to the transport & using a separate cable, separate PSUs, etc.) will cost me, as a non-diy- person, about 800 Euro... Can anybody here guarantee me an adaquate improvement for this money, especially when my recovery-unit is already doing a f_ _ _ ing great job? I wonder...
AND: This is NOT about the money. This is about the sound!

One more question guys: With all this terrific work, when do you find time to listen to... live music?

Yours, IY. [/B]

First: The bare CD624 is a rather good player. Don't ask me why.... So I had a good starting point!
All the 8 changes before are minor important.
You wrote before that the distance between your home and mine is only 200 km or so...... We could arange something or just send me the player....
I'm retired and could do the job for you, only to hear on this thread how excellent reclocking with a real good clock sounds.

I think to know which experiance you have with music.

About your last question: I do not only LISTEN to live music, I'm a(n amateur) singer. Look at the and of my web site.

 

[irgendjemand]

Re: Re: Re: D1V3 in Progress...

All the 8 changes before are minor important

Might be, but if so, how come there is such a big sound-difference between my CD72-SE and CD10? Both are being used as a transport only, and mostly important: Both got the same treatment!!

We could arrange something or just send me the player.... I'm retired and could do the job for you, only to hear on this thread how excellent reclocking with a real good clock sounds

This is very kind of you Herb! Great!!! We will have to arrange it - and I will come. !!!!!!!

Needless to say, that I will post my remarks on this thread - just like I have done once with the “Y”. Well, I assume that you know what you starting with this…

By the way - did I understand it right? We are about to modify the CD Transport, but NOT the DAC?


Greeings!

 

[Bernhard]

The first 5 chips are from the batch A'af sent last week, and last two from the first four I received.

I measured the worst of the first six harmonics for each chip and got the following:

1. 53.4 dB
2. 55.1 dB
3. 52.9 dB
4. 55.9 dB
5. 47.0 dB
6. 54.0 dB
7. 57.2 dB

I missed a whole page of the thread...

If this is a dithered -60dB signal, all chips are garbage.

My best PCM56 are -67dB with 8x os, and -62 dB non os.