Sound Quality Vs. Measurements

Status
Not open for further replies.
All oscillators have jitter, because all oscillators are at heart a very narrow filter fed with noise. The noise is unavoidable and essential - without it the oscillator will never start. The issue is to have as little as possible, given the technology being used, and with the least unhelpful statistics. Whole textbooks are written about the theory of low-noise oscillators.

Even if you had a perfect oscillator, then as soon as you do something with the signal you can introduce jitter because all circuits have noise.
 
Could someone say how jitter is caused . I read about it so many times . Not sure anyone has gone inside the problem . People speak of the oscillator working with high Q . I remember the in port receiving a distorted sine wave from the crystal when I had a look . I doubt the scope was corrupting it greatly . I will try to recreate that tomorrow if I can , my new scope is very high impedance and floating which must help . If of any value I will post it . Seems to me a sine wave is a good starting point . It passes to a Schmidt trigger next I guess .
Actually, that's almost there - now add a little noise to the sine wave, and you can see that the Schmidt trigger will trigger a little earlier or later on each cycle, depending on whether the instantaneous noise value adds to or subtracts from the sine wave near the trigger voltage. This variation in timing is, of course, jitter.
 
Opamps are not optimal for I/V conversion, but until someone designs an integrated optimized device they are about the best we have, if we are constrained by size and cost. Barrie Gilbert once remarked to the effect that an opamp was the worst thing one could use for this task!

The problem is that opamp input Z, open loop, is high, but for fast and accurate I/V conversion we want to start low and reduce further. Patrick has shown an approach using common-gate FETs which is a good start. I augmented it a bit by adding a loop around the input device to reduce the input impedance, and additional enhancements are desirable.

If so-called "current output" DACs had outputs, as the name suggests, that were high impedance, the problems would be alleviated to begin with. But "current output" is almost a misnomer. A better description is "output with a code-dependent output impedance and limited voltage swing capability, which has to be terminated in something that behaves like a much lower impedance." :D

Brad

Fair enough, Brad, but if so, does it not make sense that for the op amp, which we know or assume to be the weakest link, we should pick one of the better ones?

Beside exchanging those two op amps, I didn't touch anything else in my player. And changing it was immediately heard as a positive difference - not radical, not in-yer-face, but easily heard, and in my view, very beneficial.

AS is out of the box, that player sounded not only unlike most other players, not only definitely analog with no trace of the usual digital shortcomings, in fact it sounded a little too warm, bordering on syrupy. While this can be enchanting for the first half hour, after a day or so, it begins to sound a little false. Much warmt, easy flowing sound, but I still had a feeling it was missing out somewhere.

The op amp change brought about a sastisfying effect - it lost none of it very analog sound, but gained in definition and focuis, a bit more detail and definitely more ambience. It shifted the tonal balance and brought in more detail.

Remember, the base unit price was €800, or about $1,100 or so, meaning this was a mid price device, far removed from any High End. Yet, it managed to box well above its price range. Modified, in my view, it became the one to beat at that price point.

And that was 10 years ago.

But the real big deal was the subsequently purchased real time DAC, that one kicked the whole game at least a notch, and I'd say even two notches.
 
I do wonder if many cloaks that worked well just had better power supplies ? One thing I will try is improving the power supply all around the DAC . That is a spectrum analyzer job so can be measurement based . I remember doing this years ago and getting all sorts of nasty things . In the end I measured across the decoupling caps , I feel that is reliable .

I think this is what makes my player so good, as ever, it's the power supplies at work.

Most uncommonly for its price class, it has two separate and in looks hefty power transformers, one for the digital section and drive, the other for the analog section. This is followed by discrete full wave bridge rectifiers, each feeding two 6,800 uF capacitors by Matsushita. Thereafter, there is some local voltage regulation for individual circuits.

You don't see power supplies like that every day, and frankly, the Japanese industry has used half of that to peddle 2x50...60W stereo amps. The point being, somebody took their sweet time over this model. To the best of my knowledge, in this price class, there was nobody to even approach it for build quality and quantity.
 
If so-called "current output" DACs had outputs, as the name suggests, that were high impedance, the problems would be alleviated to begin with. But "current output" is almost a misnomer. A better description is "output with a code-dependent output impedance and limited voltage swing capability, which has to be terminated in something that behaves like a much lower impedance." :D

Looks like you're hanging out with the wrong chips. Try TDA1387, nice compliance range, decently high output impedance. :)
 
H42f5.jpg


My tests with crystals proved difficult . Here is a 3.2768 MHz taken down to 100 Hz for a strobe via 2 x 74HC4060 . There is an unfiltered 7805 regulator in addition . If ever a better example of how not to design an audio device this is it . Great for a strobe .

Thinking about what I saw in the past . It was my old oscilloscope doing Fourier . If it was 10 MHz and the device 3.2768 MHz it was working as at the very least a first order 10MHz filter . Hence a distorted sine-wave ( 1/3 F3 = 33 % distortion ) . Knowing how things more easily do square-waves than sine the crystal in such an oscillator is a square-wave . How nice I hope to find out soon .

Keeping that square wave clean seems critical long before wondering if it wobbles . The point about zero crossing voltage is excellent . How well the chip can recognize that point must be in question , and the nature of the signal etc being a source of unavoidable jitter . Knowing how sensitive even NE 555's are to instantaneous voltage fluctuations I have to think keeping voltage extremely stable and clean must be the major source of jitter avoided . DF 96 what you said seems the best first attack , trust what I already have . Making a PSU clean is nice gentle work . I said to my friend it will take a month , seem about right .
 
diyAudio Member RIP
Joined 2005
Looks like you're hanging out with the wrong chips. Try TDA1387, nice compliance range, decently high output impedance. :)

That is an unusually high compliance and output resistance, although many will sneer at its "mere" 16 bit resolution.

Try some of the ESS parts these days by comparison :eek:. Patrick's preferred old part is about 1k, and quite constrained on output voltage swing.
 
diyAudio Member RIP
Joined 2005
Fair enough, Brad, but if so, does it not make sense that for the op amp, which we know or assume to be the weakest link, we should pick one of the better ones?
Oh sure. And we need to be careful about what constitutes "better" in this application. One suspects that the recent material presented on differential nonlinearity of opamps will be applicable, for example.

My point is that most are blinkered about the approach to I-V conversion. Arbel had a similar problem trying to persuade people that an analog differentiator was best done by prefacing an opamp with a common-base stage, then applying feedback around the whole shebang.

I may use the development of optimized I-V converters as an excuse to buy an Oppo player :D Even since an ancient Sony died I've been without decent silver disk playback.
 
diyAudio Member RIP
Joined 2005
Could we use a discreet I to V ? I wish Wave would say one using EF184 or something .

EUVL/Patrick has presented some in here, and the variant I suggested is shown and discussed here: http://www.diyaudio.com/forums/digital-source/218038-insense-current-conveyor.html

I like Patrick's use of floating power supplies, but the circuits can be adapted without a lot of fuss to more conventional supplies.

Using vacuum tubes optimally is tougher as the resistance at the cathode of a "grounded-grid" stage is not all that low. What some do is simply terminate the DAC output(s) in a low passive resistance, then drive a tube. Not optimal in terms of signal-to-noise, to say the least.
 
I have to admit I had next to zero understanding of this when asking my question . The I to V function always intrigue me .

I had thought of a VAS type circuit running varying current levels . The idea being to fine tune the input impedance at the base by current ( perhaps by ear ) . Even better now reading your reply to use the same circuit in common base . Linearity might just be by using high voltage as a start . Then an output stage .

Lets say this . In all the years I have been reading about digital this seems the most interesting . My ears tell me that digital is not the problem , making it work is a big one .

I have to relate a very strange part of this story . My friend who wants the Rotel CD player improved showed me his new Arcam wireless DAC . I have been told it is an abysmal device . Using Youtube it certainly wasn't . As a music lover the music if of OK sound quality comes first . Youtube was great for listening to T Rex and the earlier version of the group . When people talk of Queen why not T Rex ? Anyway . I was very close to thinking I want an Arcam wireless DAC . Then using CD , abysmal ( harsh and just horrible ) . The point being that although detail was very poor , musicality was high with Youtube ( via Apple ) . My head now says less to process , less to go wrong . One thing occurs to me , could we process the signal to have less detail and more musicality ? Throw away that which is unpleasing ? The DAC unit could have preference choices . If preferring it a certain way it could ask when playing preference or standard ? The choice of transports was large enough to say CD not well liked , Youtube was a musical adventure ending in Louis Armstrong Summertime .

Ella Fitzgerald and Louis Armstrong - Summertime - YouTube
 
Lets say this . In all the years I have been reading about digital this seems the most interesting . My ears tell me that digital is not the problem , making it work is a big one .
Spot on. Digital is fussy, pure and simple: it's mighty easy these days to get pretty close, but it's getting the last 5% or so right that's all important. Though, I have to say there has been a distinctive improvement in the last few years, my ears are telling me that there is less of a last major step to be taken than ever before ...
I have to relate a very strange part of this story . My friend who wants the Rotel CD player improved showed me his new Arcam wireless DAC . I have been told it is an abysmal device . Using Youtube it certainly wasn't . As a music lover the music if of OK sound quality comes first . Youtube was great for listening to T Rex and the earlier version of the group . When people talk of Queen why not T Rex ? Anyway . I was very close to thinking I want an Arcam wireless DAC . Then using CD , abysmal ( harsh and just horrible ) . The point being that although detail was very poor , musicality was high with Youtube ( via Apple ) . My head now says less to process , less to go wrong . One thing occurs to me , could we process the signal to have less detail and more musicality ? Throw away that which is unpleasing ? The DAC unit could have preference choices
Not strange at all. This tussle between detail and musicality has been an apparent burden for digital since it first emerged from the womb, but it's a false dichotomy. The detail has always been, still is, embedded in digital recordings but frequently a hazy or unpleasant distortion is injected, mixed in during playback. So far some of the solutions are to either soften the distortion, discard the extra detail, or improve playback so that this fine detail is reproduced cleanly, musically. The last is hardest to do, is frequently the most expensive way at the moment. But it is still the correct way, so should be pursued with full vigour ...

Frank
 
I think we can have our cake and eat it too . The idea of having digital options is good . These are not in the old fashioned ways filters . Tamla Mowtown is said to have digitized the entire library using Sony PCMF1 then dumped the master tapes . If true how sad . Still , all is not lost if we look at how we might predict the errors of a Sony PCMF1 encoder .

Video digital is interesting . As I have said before I bought a cheap and cheerful Panasonic 37 inch TV for much less than $500 . To be honest I was very disappointed . I then bought a PVR also by pure chance Panasonic as it was on sale for $250 (half price and top spec ) . Subtly it dawned on me that everything is better via PVR . Even SD is better than TV HD . The upshot of this was some Ray Davis 1969 on SD was almost the best picture I have ever have seen . I was told that very early PAL video was of a high standard ( vidicon -plumbicon or whatever ? , apparently short lived , 1990's certainly not as good ). The little green shadows under the eyes I now remember . A quality which isn't film and isn't modern . I would say my TV is now excellent and cheaper than TV's of this standard , the PVR being free of charge taking the combined price . Ivor Tiefenbrun was right FRIGO , feed rubbish in , get rubbish out . The TV is innocent if ignoring it's front end . I am told HD is typically 2 million pixels . My friend wont beleive me that a phone needs less than the 8 million she based her buying decision on . Super HD TV is 4 million ?
 
On a just slightly different track, Nigel's notion of having "digital options" is not new, even though it's very intriguing, and for those who do have them here and there, it quickly becomes a must.

Let me remind you all that some better quality CD players, notably by Marantz (but there were others as well), did offer some limited digital filter options in the past of their top flight models. I found them to be subtle but still audible, and I clearly preferred Option 1 to 2 and 3. Others might feel differently.

My now old Denon DVD 3910 was, in its time some say 6 or 7 years ago all the rage, being hailed by just about every (European) audio and video magazine I picked up. The audio side, which is good but hardly ground breaking (2+2 BB DACs in dual differential mode, another 4 BB DACs for surround channels), was fine, but on the video side, it was one of the first, if not THE first, to use the then brand new Faroudja video chipset. This one is distingushed by allowing the user to fool around with the picture IN THE DIGITAL DOMAIN.

Even though it has been surpassed since then, even in this let's say rudimentary form, the effects are stunning. I would say at least 99.9% of all possible users woud be able to get the picture they want, and actually at home achieve a near studio quality picture on a reasonably recent and higher quality TV set.

I'm not a video freak, not even a fan, but I do like my picture clear, clean and with lots of detail. The scope of these controls lets me get that from everything I ever put into it (DVD - can't do Blue Ray, unfortunately, job relegated to another Sony Blue Ray player), but more importantly, the quality of those changes far surpasses anything analog I have ever seen, and I have seen and used professional video equipment for the three years of editing in TV studios while I had my weekly TV show.

I've often wondered how nice it would be to have similar controls on my music CD player, even in basic form (balance, treble, mid and bass, a bit like tone onctrols), but also in full digital domain.
 
...

Ivor Tiefenbrun was right FRIGO , feed rubbish in , get rubbish out . The TV is innocent if ignoring it's front end . I am told HD is typically 2 million pixels . My friend wont beleive me that a phone needs less than the 8 million she based her buying decision on . Super HD TV is 4 million ?

Nige, it's not all in the pixels, just like sound is not all in THD figures.

A simple example would be digital cameras, widely available. Many 12 and 14 million pixel cameras can't portray the texture of colour my now ancient 4 megapixel Olympus does. My son has a Fujitsu camera with 12 MP and its actual pictture doesn't look half as good as mine, even it does have 3 times more pixels than mine.

He has to rework every shot in Photoshop to get the right colour balance, whereas I don't, the only weakness of my take on colour being some shades of lilac, which it botches up. And which is, thankfully, extremely rare in real life, usually a fabric someone is wearing.

Not to even mention his nominally present but actually almost useless macro facility, while mine is brilliant (and is one of two key reasons why I bought that camera in the first place).

So, while it may look great on paper, in real life, quite a few cold showers to be had.
 
DVV . My friend Paul Stewart ex of JVC I speak of is what is called a Snapper or press photographer . He says that 1.5 MP is fine for normal photographs . I was reading about the old Canon professional cameras of about 2 MP . I dream to own one if cheap enough . So spot on DVV . The best photos I have taken were on an old Olympus 1.5 MP . It was as good as my old Yashica / Contax for everyday things ( as was an Olympus Trip ) . It broke and nothing I have owned since has been better and perhaps as good .

I was thinking about this high stability oscillator . Perhaps having a device at resonance longer makes it more accurate ? I have doubts about that . I could imagine a clipped output of a Wein bridge to be more predicable . My little mind says a crystal flapping about seems less likely to be accurate in the nS range than poly-whatever ( paper even ) cap . An oven to be sure it stays on frequency . What gets me is people will say in production this will not work . I have trained people to build such things .Usually 10 minutes training which includes all the strange topics the engineers discuss for light relief ( you don't want to know ) is all it takes .


I can see how high Q would be adventurous for a clock oscillator reference . I just don't trust that it works in practice . My friend John who makes precision magnetometers does not always use crystals . If used with care film caps are better . He does not use COG either as they are 3rd choice devices . Thinking about it what makes a film cap chaotic on a micro level ? A crystal seems to me more likely to be chaotic ? I am sure research has proved this wrong years ago . What sticks in my mind is some Harwell engineers saying to me that RC oscillators were more stable than crystals if enough care taken and a warm up period allowed ( a day I seem to remember , we hi fi people don't mind stand by , I would rather have good sound if I remember to switch it on than OK sound with instant start up ) . We can have LC also . My mind says it is the zero cross voltage that matters , not the shape of the wave . Also a sine-wave seems to have a better defined zero point than a square-wave ? A triangle-wave would let you see how good more easily . I use triangle-waves a lot . They often tell me more than sine or square .


I met a lady called Lucy Reeves who archives BBC material ( she has a Garrard 401 ) . She said the best she ever saw was NTSC mater-tapes circa 1968 . Having seen the Ray Davis of 1969 I remember seeing colour TV for the first time and being astonished . Then less so with time . I now realize what I saw was very high grade initially in 1968. As soon as money was involved that changed . I suspect also when seeing how many people set up their TV's with orange faces the producers thought why bother . I hope we will soon see these early tapes transcribed . Soon as I doubt they will last much longer . This sometimes happened with 78's . The master stampers had been kept and vinyl versions could be made and transcribed . It is now known late 78's were cut to 32 kHz ( FFRR ) the typical 10 kHz due to shellac not cutting ( 5 kHz after one playing it is said ) . I made some equipment for a Decca cutting engineer who told me the whole story . The FFRR ( later FFRR LP with different cutting stylus and lather screw ) was a Sonar device that Decca made for submarines . They knew it could be adapted when accepting the contract . This guy worked for Parlophone . I made him a universal EQ box with dial in EQ to any desired combination . All capacitors polystyrene and resistors Russian with sliver wires ( no idea where my brother got them ) .
 
Last edited:
nigel pearson said:
I could imagine a clipped output of a Wein bridge to be more predicable . My little mind says a crystal flapping about seems less likely to be accurate in the nS range than poly-whatever ( paper even ) cap . . . . I can see how high Q would be adventurous for a clock oscillator reference . I just don't trust that it works in practice .
Nigel, find a good book on low-noise oscillator theory and read it. Come back with your intuition realigned to physical reality. Then you will know why high Q (such as a quartz crystal) is essential for good short-term stability (jitter). It happens to be true that quartz is also good for long-term stability, but that is different. One arises from high mechanical Q, the other from good thermal stability and ageing processes. Quartz just happens to combine both requirements in one material.

If you can find C and R components with excellent thermal and mechanical stability then you might be able to build an oscillator with excellent long-term stability but it will still have lousy jitter. No good for digital audio.
 
Status
Not open for further replies.