Thought experiments are good, I use them all the time, especially in my work in loudspeaker design.
Take this one: You invite a subject into a room, you show him a desk and on the desk is a large piece of paper with a very intricate continuous line, with a starting point and an end. You give the subject a pen and ask him to trace that complex line and tell him not to hurry, even if a clock is going while he is doing the task, and he can have all the time he needs, and he simply needs concentrate on getting every detail right.
The you tell him to start as you click the clock on. At first he is doing really well and then he starts to sense something odd. You see the room is actually on springs and it is starting to move about erratically from forces he cannot see (is not allowed to). It is soon obvious that his task has become much harder than he ever thought and that now he will make many mistakes.
See the point? If those intricate lines to be traced represents high frequencies, then what happens at even ultra-low frequencies will affect the end result.
This is a thought experiment that helps us to understand the nature of stability and that ultimately it is related to Terra Firma, that which is beneath our feet. If it moves, you are up the creek.
Phase noise ("jitter") is commonly measured down to about one Hertz.
What if we compare two clock/oscillators, and the one with inferior phase noise above one Hertz sounds better than one that is superior in phase noise, i.e. lower above one Hertz?
This is the question raised in post #169 http://www.diyaudio.com/forums/digital-source/222596-oppo-s-bdp105-discussions-upgrading-mods-17.html#post3294742
What is happening below that one Hertz?
And since now we are talking in stability in time that have equivalent frequency way below where there is musical content, how can that be? The above thought experiment may offer an explanation.
Greater than 10 seconds, 100 seconds, a THOUSAND seconds?
If the SAW sounds better than the Crystek - as Coris says - then what is going on? Above one Hertz the Crystek is better. So it is supposed to sound better.
Hence it is now important for others to do the SAW versus Crystek 957 comparison.
If there is a great consensus that the SAW is superior, then we have an incentive to find out what is going on!
Cheers, Joe R.
Last edited:
I want to express her my opinion that I consider as enough unfortunate that it seems that I`m yet the only one who is cited as above...
The SAW oscillators are not expensive. That because it will be very fortunate if many DIY-er will get the opportunity to try it, tested, used in DACs and so on. As experiment. And not at least, come here and publish their impressions, conclusions, results, measurements. Only then we can have an quite objective conclusion: better or not than those very expensive oscillators, with very good jitter figures, etc.
I may correct a little bit the above assertion about my appreciation of the sound using an SAW oscillator vs a Crystek one: my opinion is that an SAW 125Mhz sounds better than an 100Mhz Crystek on ESS9018. I did not heard yet the resulted sound out of this DAC chip, when clocking with an 125Mhz Crystek, and with an 125Mhz SAW oscillator. Such comparing it may be the right one.
Last edited:
About this above subject, I may say that I forgot something quite important.
Before doing any modification here, I will want to measure first the amplitude of the oscillations on those two clock pins of the main processor. I just suppose that on this new model of Oppo processor it may be changing of the needed clock amplitude signal, comparing with the precedent processor (used in 95). This aspect may be important when modding in this area.
It is enough important to know what kind of oscillator it may be used here: 3,3V, 1,8V or something else.
Then, I will want to cut the traces which goes to the processor, isolate the oscillator components, and measure those components. Then remake the connections, remove the passive components around the crystal, and at last connect an external (custom) oscillator, which will deliver an clock signal with the same (measured) amplitude as the original one.
In my opinion this way it may be a safe one...
I could never noticed any noise using either 125Mhz clock or 122Mhz or something else over 100Mhz. Just very good sound...
I`ve got only sparkling like noises in silence/in between tracks, when I used 133Mhz oscillator. I have my own analogue circuit design after DAC, without any filtering (32mVpp residual HF noise on final opamp output, for an max 28Vpp audio signal).
I`ve got only sparkling like noises in silence/in between tracks, when I used 133Mhz oscillator. I have my own analogue circuit design after DAC, without any filtering (32mVpp residual HF noise on final opamp output, for an max 28Vpp audio signal).
It clearly varies according to the player. I have my own at 125MHz right now and I get some slight noises. I did another player after that and at 125Mhz was very much more noisy, this was fixed when 100MHz SAW was fitted. So since I am doing, not a single player but players, I have to stay safe. I have now done 100MHz on a number of players and they don't sound much difference to 125MHz.
Cheers, Joe R.
54MHz SAW Oscillator:
On Digi-Key website they call this a SAW Oscillator and available in 54MHz:
KC5032A54.0000C10E00 AVX Corp/Kyocera Corp | 1253-1054-1-ND | DigiKey
The Datasheet does not mention "SAW" but does say "High Frequency" and the range starts at a low 50MHz (up to 170MHz), so this sounds like a SAW unit.
I am sending AVX/Kyocera and email, to see if they can confirm SAW.
The 54MHz should be convertible to 27MHz using the 74HC4040N divide-by-two I posted earlier in:
http://www.diyaudio.com/forums/digi...iscussions-upgrading-mods-18.html#post3295011
It is the smaller form factor of 5x3.2mm - but that's OK - and 1.6V to 3.63V DC range.
Make sure 4040 chip is the NXP 98 MHz version as other 4040 don't have sufficient bandwidth. Element14 part #380880.
Cheers, Joe R.
PS: Alas Digi-Key does not have stock and quoting Feb 20th.
On Digi-Key website they call this a SAW Oscillator and available in 54MHz:
KC5032A54.0000C10E00 AVX Corp/Kyocera Corp | 1253-1054-1-ND | DigiKey
The Datasheet does not mention "SAW" but does say "High Frequency" and the range starts at a low 50MHz (up to 170MHz), so this sounds like a SAW unit.
I am sending AVX/Kyocera and email, to see if they can confirm SAW.
The 54MHz should be convertible to 27MHz using the 74HC4040N divide-by-two I posted earlier in:
http://www.diyaudio.com/forums/digi...iscussions-upgrading-mods-18.html#post3295011
It is the smaller form factor of 5x3.2mm - but that's OK - and 1.6V to 3.63V DC range.
Make sure 4040 chip is the NXP 98 MHz version as other 4040 don't have sufficient bandwidth. Element14 part #380880.
Cheers, Joe R.
PS: Alas Digi-Key does not have stock and quoting Feb 20th.
Last edited:
It may be easier to get 53.125MHz SAW than 54MHz. Divide by 2 gets us 26.565MHz.
I wonder how critical 27MHz actually is - 26.565MHz is 1.6% lower in frequency.
Is the Crystek 957 available in 27MHz?
ALSO - is this a SAW oscillator from Crystek - when the range starts at 50MHz and upwards, then there is a chance it's a SAW:
http://www.crystek.com/crystal/spec-sheets/clock/CCHD-575.pdf
Cheers, Joe R.
PS: Crystek makes another oscillator called "CVS575S-500 - a 500MHz Voltage Controlled SAW (surface acoustic wave)
Oscillator, or VCSO. SAW crystal technology provides low-noise and low-jitter performance with true sinewave output." Note
that they have 575 in common.
Makes you think?
I wonder how critical 27MHz actually is - 26.565MHz is 1.6% lower in frequency.
Is the Crystek 957 available in 27MHz?
ALSO - is this a SAW oscillator from Crystek - when the range starts at 50MHz and upwards, then there is a chance it's a SAW:
http://www.crystek.com/crystal/spec-sheets/clock/CCHD-575.pdf
Cheers, Joe R.
PS: Crystek makes another oscillator called "CVS575S-500 - a 500MHz Voltage Controlled SAW (surface acoustic wave)
Oscillator, or VCSO. SAW crystal technology provides low-noise and low-jitter performance with true sinewave output." Note
that they have 575 in common.
Makes you think?
Last edited:
I wonder actually about the same: how critic is the 27Mhz to clocking the main processor...
I will want to see what is happen if this clock is just double f. ex. then try lower frequency, but higher than 27Mhz.
At last this clock it may drive an PLL or an master clock distribution circuit inside the processor. This circuit it must deliver also clock signal for the devices which were clocked by 25Mhz oscillator in the earlier model of this processor...
I think one may try to find an (well) working (if possible another) frequency for this clock, using an standard one oscillator, then find the best quality oscillator. I personally want/plan to experiment a little bit more in this area... I already ordered that so called SAW from Digikey, and waiting to get it...
I will want to see what is happen if this clock is just double f. ex. then try lower frequency, but higher than 27Mhz.
At last this clock it may drive an PLL or an master clock distribution circuit inside the processor. This circuit it must deliver also clock signal for the devices which were clocked by 25Mhz oscillator in the earlier model of this processor...
I think one may try to find an (well) working (if possible another) frequency for this clock, using an standard one oscillator, then find the best quality oscillator. I personally want/plan to experiment a little bit more in this area... I already ordered that so called SAW from Digikey, and waiting to get it...
I have been in contact with Epson Toyocom office in Singapore. Hopefully will hear from them next week about getting 54MHz SAW made - then I can divide by two and get 27MHz.
The problem of using other than 27MHz is that it might disable other functions of the player, video or sound capabilities. In my case that is not an option. Likewise with Ric, I'm sure.
But I will be trying 50Mhz SAW divide by two = 25MHz (a la '95), and maybe the PLL(?) might work with that just fine and still synthesise all the correct frequencies. We shall find out.
Cheers, Joe R.
It may be a good solution if Epson can produce/deliver 54Mhz SAW oscillator, and it will sell it in reasonable small quantities...
In my opinion there are not "most effective" mods in this case. A sum of many it may be in one way "effective"...
Is quite difficult to make a list, because are enough many, and those mods are connected in between: is not big deal to use better opamps without to improve the power system, f. ex. Is not so big deal to improve the clock oscillator without put in place a ultra low noise regulator for it, and so on.
It is a really nice work and big reward to do step by step the mods, and experience every time an increasing of the results/improvements...
For those who might want to see it, I uploaded a Service Manual for the "audiolab" 8200DQ Service Manual.
What makes it interesting is that here we can see the ES9018 in the context of a Service Manual and full Schematics.
This the link I created, it with be active till the 25th of January:
http://www.yousendit.com/download/UW14OGNYTWNwcFZsQXNUQw?cid=tx-02002207340200000000&s=19102
Cheers, Joe R.
What makes it interesting is that here we can see the ES9018 in the context of a Service Manual and full Schematics.
This the link I created, it with be active till the 25th of January:
http://www.yousendit.com/download/UW14OGNYTWNwcFZsQXNUQw?cid=tx-02002207340200000000&s=19102
Cheers, Joe R.
Nobody say or think to reproduce that schematic... At last Audiolab precise quite clear that the schematic used in the manual is an initial (prototype) design of that device, which for sure it were modified afterwords for the final product.
One may expect that such schematics are not published so that another concurrent company will reproduce it and sell then the same product with the same parameters...
I couldn`t see many schematics where ESS9018 is involved, and this it were the main point (as I understood) when Joe uploaded this manual. This upload is for sure not meant as an apology of the Audiolab design...
It have been published an information here and this is to be appreciated at last...
One may expect that such schematics are not published so that another concurrent company will reproduce it and sell then the same product with the same parameters...
I couldn`t see many schematics where ESS9018 is involved, and this it were the main point (as I understood) when Joe uploaded this manual. This upload is for sure not meant as an apology of the Audiolab design...
It have been published an information here and this is to be appreciated at last...
Last edited: