That is totally wrong, and beside you contradicting your self.
I don't even believe this in connection with jugging power amps.
Naturally I'am to for the advancement of science, and being on the cutting edge of development, but somewhere it must be a limit for the need of the better.
May I ask you what kind of equipment you using to reach your listening conclusion?
Also, how is your hearing, how old you are?
Human senses have they threshold, what is your personal experience of your jitter threshold?
Did you ever run on your play back equipment the Dunn Jitter test, available here on Diyaudio ,from Wakibaki?
This would give a visual feedback about performance, and a chance to compare players with different clocks.
Thanks.
I think the issue of JTest came up here before. Its not difficult and the necessary ADC need not be expensive. Pretty much any of the EMU USB interfaces are adequate. i can provide details to get plots that match the stereophile plots for reference.
Its way easier to do that testing the clock. And even getting the clock to a very high level doesn't mean that its still good inside the DAC. Really, implementation may be more important that exceptional clock quality.
I agree that listening tests are far from definitive, especially when not genuinely blind.
Its way easier to do that testing the clock. And even getting the clock to a very high level doesn't mean that its still good inside the DAC. Really, implementation may be more important that exceptional clock quality.
I agree that listening tests are far from definitive, especially when not genuinely blind.
Thank you Andrea for the reference.
Interesting article. I'll read it and get back to you.
As far as close-in PN is concerned, I think it would be possible to achieve close to that with AT crystals (not the noise floor).
In the mean time, I feel I have to get some experimenting done.
I am sending you an updated xtal orderform.
Cheers.
I fully agree on that...
You can try to eradicate close in noise completely, but at what cost? We listen to audio and are not running a science lab. I think it's good engineering practice trying to achieve the best cost effective solution.
NB: The TWTMC as is, is in my opinion that "good engineering practice". Trying to improve on that with, say 10db, would drive up cost considerably.
Last edited:
The audio chain was:
- modified Aune X5 SD player as the source to compare the oscillators
- DIY NOS AD1865 DAC, passive I to V and 5687 tube buffer
- modified Music Angel MA 845 (SET 845 triode ampli)
- Cambridge Audio floor 2 way speaker
During the listening test I never change any devices, obviously apart the oscillators.
My ears (and my friends ears) was the same along all the session, regardless of my age (that is 55).
With the same ears I usually like to listen to unplagged music (grand piano is the best instruments to do sonic test IMHO).
My listening impressions are obviously subjective, but are anyway coherents along the listening session.
I personally prefer the Pierce and the Driscoll oscillators with Laptech HC-43/U crystal, as well I don't like the sonic performance of the Crystek.
Again, these are my subjective impression and I invite everyone to do his own test.
As I said, at the end we build custom audio device to listen to music. And I would never listen to music with the Cristek crystal in my audio system.
@Andrea: Hi Andrea ... I am finalizing the Driscoll oscillator with a 6.144 MHz crystal yet have a slight "challenge" with the MMBT5179s I bought (from Mouser) relative to the specs you list for this transistor.
In the TWTMC-D document you note that the Hfe should be >200 for this transistor. The ones I've bought have an Hfe that is around 160 - 165 (at ~5 mAs) and I reckon that if they are run at a higher current the Hfe may be even a little lower. To this end: Can I ask you where you sourced your MMBT5179s? Or maybe if you have an equally good alternative that I can use instead ... with the consistency of the transistors I bought I reckon I may have to buy many before I find one with an Hfe>200 ...
Cheers & thanks,
Jesper
In the TWTMC-D document you note that the Hfe should be >200 for this transistor. The ones I've bought have an Hfe that is around 160 - 165 (at ~5 mAs) and I reckon that if they are run at a higher current the Hfe may be even a little lower. To this end: Can I ask you where you sourced your MMBT5179s? Or maybe if you have an equally good alternative that I can use instead ... with the consistency of the transistors I bought I reckon I may have to buy many before I find one with an Hfe>200 ...
Cheers & thanks,
Jesper
Hi Jesper,
I suggest hfe > 200 for crystals with high ESR such as the 90/98 MHz (ESR around 40 ohm), in your case the ESR of the crystal is very low (a few ohm), so you can use your MMBT5179 with hfe around 160 without any problem.
Andrea
Thanks for your answer.
I still like to see some J test run on them.
Also find odd that the ca. -50dB or less THD of your power amp doesn't bother you, but the minuit differences in the clocks registers in the listening tests.
This is not the right thread to talk about THD, but I like to listen to music with this chinese ampli at 1% or more THD. After I replaced the original 845 with a pair of Psvane WE845 it sounds very balanced and smooth.
In the early 90's I designed an integrated amplifier with mosfets. THD was something like 0.00....%, also with limited negative feedback. It got a "Best Buy" from (if I remember correctly) Hi-Fi Choice.
It sounds enough good, but I prefer the DHT chinese amp with much higher THD.
At the end of the listening session with my friend (we designed together the mostet ampli) we did switch to the solid state ampli to repeat the test for a pair of oscillators: the Crystek and the Pierce with the AT-Cut HC-43/U crystal.
We agreed that the results was worse than the tube amp, but the differences between the two oscillators were practically the same, the Pierce detailed and smooth while the Cristek confused and harsh.
Obviously, I look forward to measure the phase noise of all the oscillators to confirm the listening impressions, but at this moment I can't afford an Agilent E5052, so I have to wait until I will access the University Lab to do the measurement. In the meantime I will try to get the suitable stuff to do the JTest.
About ampli and THD, surely Demian can say much more than me.
Attachments
I think this is a valid test method.
That's a sweet setup by the way Andrea ;-)
I have done similar tests myself and it's surprising every time how certain components
so predominantly affect the reproduced sound.
I do agree with miklos...
...when a design is undertaken in the industry, so much resources are spent on test and verification.
In audio however, many (small) manufacturers charge astronomic prices and hardly
every present any measured data. once they do it usually is not good.
of all the audio clocks I know there are a few companies where measured data is presented:
Tentlabs and RutgerS and a few more....sad.
Cheers
Last edited:
I still haven't had time to read the document in detail yet :-(
Just a few pointers:-------------------
The Driscoll oscillator is very different from what we use here.
This one is a single stage emitter coupled oscillator. You wont get the same functionality here.
In theory, the xtal Q is a very effective ”dirt cleaner”.
In practice it seems there are other ”evils” at play that negatively affect the close-in PN beyond xtals ability to clean up.
The excellent PN measurements presented in the article are not solely due to the application of the Driscoll circuit.
The oscillator and peripheral circuitry is battery driven and very optimized.
Every design can be taken into the stratosphere, but I have learned that simplicity is a highly desired commodity here.
I still don't think the SC cuts will improve the close-in PN figures of the oscillators here,
just look at the graphs of post 1371, presented by the dude himself: Mr Driscoll.
Cheers
Last edited:
Yes, it's a sweet setup, but while with the Pierce oscillator it remains sweet, with the Cristek it becomes harsh.
NOS AD1865 with passive I to V (200R Vishay Z-Foil resistor) and 5687 tube buffer (NOS Tung Sol tubes).
No, it's a DIY design by Andrea Ciuffoli
http://www.diyaudio.com/forums/quan...-dac-end-best-dac-ad1865-updat-2011-a-18.html post #177
One more notation about the listening session.
At the end we changed the speakers replacing the Cambridge Audio with a pair of Linn Isobarik S.A.R.A.: nothing changed, the same impressions, with the Pierce that sounds warm and detailed, while with the Crystek the sound becomes harsh and confused.
Finally we replaced the source and the dac with a Naim Cd 3 player (TDA1541A S1). The sound was similar to the previous listening with the source equipped with the Pierce oscillator, maybe a little darker, but still clear, detailed and smooth. Our impression was that changing the oscillator the differences was huge and clearly audible, while changing the other components of the audio chain there was a differences, but not so huge, not more than shades.
Duty cycle option & TWTMC - D&D.
Hi all,
So now the Driscoll oscillator is up and running at 6.144 MHz and without knowing how it performs phase-noise-wise it otherwise looks as if it's utterly stable and also oscillates exactly at 6.144 MHz (according to my oscilloscope). Kudos to Andrea for this
However, for those of you considering combining any of Andrea's oscillators - while using the squarer's duty cycle option - with the TWTMC - D&D it may be of interest to know that it won't work ... Since the D&D is based on positive edge triggered FFs the FFs will "not notice" the negative going edge and thus will omit the duty cycle adjustment. The output from a divider FF will essentially be 50/50 percent duty cycle ... Didn't think about it and just learned it the hard way testing my PCB layout
Which leads me to the question of whether or not a FF can be used as a squarer with similarly fine phase noise figures .... ? The structure then could be the oscillator - followed by a FF divider sequence - ended by the inverter setup/other buffer structure ...
I know I've asked this question before but since it's really important to my application to be able to adjust the duty cycle I hope one of you may be able to help with answering this - appreciation is due Or maybe if you have a suggestion for a different duty cycle adjustment structure that works with FF dividers - and maybe also with the Driscoll squarer as it is - that would be very fine as well.
Have a good evening,
Jesper
Hi all,
So now the Driscoll oscillator is up and running at 6.144 MHz and without knowing how it performs phase-noise-wise it otherwise looks as if it's utterly stable and also oscillates exactly at 6.144 MHz (according to my oscilloscope). Kudos to Andrea for this
However, for those of you considering combining any of Andrea's oscillators - while using the squarer's duty cycle option - with the TWTMC - D&D it may be of interest to know that it won't work ... Since the D&D is based on positive edge triggered FFs the FFs will "not notice" the negative going edge and thus will omit the duty cycle adjustment. The output from a divider FF will essentially be 50/50 percent duty cycle ... Didn't think about it and just learned it the hard way testing my PCB layout
Which leads me to the question of whether or not a FF can be used as a squarer with similarly fine phase noise figures .... ? The structure then could be the oscillator - followed by a FF divider sequence - ended by the inverter setup/other buffer structure ...
I know I've asked this question before but since it's really important to my application to be able to adjust the duty cycle I hope one of you may be able to help with answering this - appreciation is due
Or maybe if you have a suggestion for a different duty cycle adjustment structure that works with FF dividers - and maybe also with the Driscoll squarer as it is - that would be very fine as well. Have a good evening,
Jesper
Last edited:
- Status
- Not open for further replies.