then it should be damn easy to prove right? please try it and post us the results.
this would be really interesting. I also would take a submultiple of FS. Of course the measurement would be only valid for that DAC, but still would give a nice indication what is happening - please do this test !
Last edited:
andrea_mori,
Do you have any ball park figures for tempture stablity of your 90M/98M XO designs?
I'm measuring 20ppb per DegC from say 30DegC TO 50 DegC - using an uncompensated 5th overtone AT cut crystal...
It be good to see how this compares with your experance?
Its been my experance that the L-R sound stage of my designs "opens up" after atleast a 30 minute warm up - I wonder how much of this can be put down to the Clock / XO reaching tempruture equalibruim?
Do you have any ball park figures for tempture stablity of your 90M/98M XO designs?
I'm measuring 20ppb per DegC from say 30DegC TO 50 DegC - using an uncompensated 5th overtone AT cut crystal...
It be good to see how this compares with your experance?
Its been my experance that the L-R sound stage of my designs "opens up" after atleast a 30 minute warm up - I wonder how much of this can be put down to the Clock / XO reaching tempruture equalibruim?
Last edited:
As far as I know, this thread is all about close-in phase noise, that is, about random frequency variations. Clock frequency variations also cause output signal frequency variations, so you do get phase noise skirts around the output signal.
Measuring them may be a bit of a challenge, though. When the output frequency is 1000 times smaller than the clock frequency, the frequency deviation is also 1000 times smaller while the modulating frequency is the same. As a result, the close-in phase noise is 60 dB smaller, so for example, -120 dBc/Hz at 10 Hz distance from a 20 MHz clock translates to -180 dBc/Hz at 10 Hz distance from a 20 kHz output signal.
If you have a bad enough phase noise floor, it will eventually worsen the audio noise floor. That only happens when you play back a signal (particularly a high dv/dt signal) in the case of a multibit DAC with no noise shaping. In the case of a sigma-delta DAC, it will even happen during silence, so it should be relatively easy to measure without fancy equipment.
The close-in phase noise that this thread appears to be dedicated to will produce phase noise sidebands (skirts) around the spectral peaks of the desired signal, independent of the type of DAC you use. Their level is so low I wouldn't worry about them (see post #4218), but objectively, they are an artefact that is smaller with Andrea's oscillators than with the usual crystal oscillators. Measuring them will be difficult.
To me it's like designing an amplifier for < 1 ppm distortion: objectively it's better than 100 ppm distortion, but I doubt very much if anyone will really hear the difference (*).
(*): Actually that's not a good criterion, because ideally the distortion and other artefacts of the entire signal chain have to be inaudible. In the case of clock phase noise, that usually means the ADC and DAC phase noise combined have to be small enough.
Last edited:
JohnW, FWIW, I leave my clock on 7/24. Also the entire digital chain is on 7/24.
Even with that, I experience the same sense of "opening up" after 30 minutes. The rest of the output and the amps are triode. I let them warm up for a while and still they open up after playing.
While clock warming up would have an impact, I suspect the rest of the system is key to passing the low level information that results in some of the audio pleasure some of us value.
I also have a switch on each amp that allows selection of more or less global feedback. Higher feedback can also maim the illusion.
I suspect that some of our measurement obsessed engineers run amps that measure very well but fail to pass some of the low level information. Perhaps this whole dichotomy of experience is as simple as with measurement driven design comes design traits like massive global feed back producing sound devoid of emotion. Then we hear - I am smart. I designed an amp with zero distortion. I hear no change based on clock/capacitor/wire/etc. Those that do are fools. Perhaps it is just ignorance. Once you hear "air" in your system, you don't want one that lacks it.
thanks Marcel, if i read you correctly:
- you will need an insane amount of jitter for audio noise floor to be affected
- the jitter induced audio noise is not likely be measurable with Andrea's clock or those audio grade Crysteks, using a soundcard
right?
Hello Andrea ,
Good idea!
Maybe use your brain and equipment to spread some more knowledge about how to reduce the impact of vibrations on your clock boards! From what i read so far the impact is bigger than expected..
We know that each board has to be in its own aluminium container. The once i have chosen are die cast aluminium and consist of just two parts a box and a lid. The less parts the better will be the screening. It can be closed almost like airtight. I think better than the boxes you propose lol
Probably there are other parts than just the clock that is sensitive to vibrations on the board. Like i wrote before if the boards are connected to the chassis by means of the outer ring of the SMA part on the circuit vibrations captured by the box will be transferred right on to the clock. THIS is were the two tin wires will do a nice job.
BUT we need to know which vibrations we need to stop before they reach the clock ( and the boards) . Should we consider the clock like an mc cartridge/ tone arm combination that can malfunction because it picks up more information besides the one in the groove?
There are six aluminium boxes that are " interconnected " by cables. No matter how flexible the cables are , if one box picks up a vibration because it touches a sidewall the vibration will be transferred to the other ones. During transmission there will be energy loss for sure.
Because of this interconnection my idea is to consider the 6 boxes together as one " unit". If you increase the weight of a single box with some sand and maybe some lead sheet it still wont be that heavy. Unless you are going to do some lead moulding you wont create a lot of weight. So joining them will give you the advantage of creating a 20 kilogrammes or more " container".
Maybe i will weld a big metal box that will allow me to surround the six aluminium boxes inside with a reasonable amount/ weight of dry sand and/or lead shot. My clock " vault " will be just above ground level so i can make it rather heavy without running into trouble. Using the inflatable cushions to lift it up, attach it with 3 heavy duty strings to the lowest in my audio rack and deflate and it will be free from most vibrations i hope.
Dont know if a clock during functioning creates its own vibrations?
Greetings,Eduard
Good idea!
Maybe use your brain and equipment to spread some more knowledge about how to reduce the impact of vibrations on your clock boards! From what i read so far the impact is bigger than expected..
We know that each board has to be in its own aluminium container. The once i have chosen are die cast aluminium and consist of just two parts a box and a lid. The less parts the better will be the screening. It can be closed almost like airtight. I think better than the boxes you propose lol
Probably there are other parts than just the clock that is sensitive to vibrations on the board. Like i wrote before if the boards are connected to the chassis by means of the outer ring of the SMA part on the circuit vibrations captured by the box will be transferred right on to the clock. THIS is were the two tin wires will do a nice job.
BUT we need to know which vibrations we need to stop before they reach the clock ( and the boards) . Should we consider the clock like an mc cartridge/ tone arm combination that can malfunction because it picks up more information besides the one in the groove?
There are six aluminium boxes that are " interconnected " by cables. No matter how flexible the cables are , if one box picks up a vibration because it touches a sidewall the vibration will be transferred to the other ones. During transmission there will be energy loss for sure.
Because of this interconnection my idea is to consider the 6 boxes together as one " unit". If you increase the weight of a single box with some sand and maybe some lead sheet it still wont be that heavy. Unless you are going to do some lead moulding you wont create a lot of weight. So joining them will give you the advantage of creating a 20 kilogrammes or more " container".
Maybe i will weld a big metal box that will allow me to surround the six aluminium boxes inside with a reasonable amount/ weight of dry sand and/or lead shot. My clock " vault " will be just above ground level so i can make it rather heavy without running into trouble. Using the inflatable cushions to lift it up, attach it with 3 heavy duty strings to the lowest in my audio rack and deflate and it will be free from most vibrations i hope.
Dont know if a clock during functioning creates its own vibrations?
Greetings,Eduard
No, it is not about close-in phase noise, is about clock and it’s effect on audio reproduction. Did I write anywhere in your quote about phase noise, or do I read “jitter”? Jitter to phase noise stands like Probability to the Probability Density Function.
No objections. Though, I was talking about measurable effects in audio, not about effects of the order of magnitude of the background radiation noise of the Universe.
It's Andrea's thread and I had a discussion with him about close-in phase noise and the phase noise floor many posts ago. From that discussion I understood that he's mainly interested in close-in phase noise and multibit DACs without noise shaping. Hence my claim that this thread is about close-in phase noise.
Hi Eduard,
we have already measured the sensitivity of the DRIXO oscillator to the vibrations.
We have shielded the oscillator into the Hammond enclosure as in the manual and we have compared the measured phase noise of the XO in two conditions:
- far form the speakers
- placed at something like 30-50cm from the bass speaker driver (I don't remember exactly the distance)
There was not much difference.
I'm sure I have published the phase noise plot comparison but I don't remember in which thread and the specific posts.
I have to do a search.
If someone remember the above posts I have pointed out, please provide the link.
Andrea
we have already measured the sensitivity of the DRIXO oscillator to the vibrations.
We have shielded the oscillator into the Hammond enclosure as in the manual and we have compared the measured phase noise of the XO in two conditions:
- far form the speakers
- placed at something like 30-50cm from the bass speaker driver (I don't remember exactly the distance)
There was not much difference.
I'm sure I have published the phase noise plot comparison but I don't remember in which thread and the specific posts.
I have to do a search.
If someone remember the above posts I have pointed out, please provide the link.
Andrea
Last edited:
Hi John,
I assume you mean the old emitter coupled oscillator since there are no 90/98 MHz option for the new Driscoll (up to 25 MHz).
And moreover we have not implemented the oven in the new Driscoll design.
About the old Driscoll design I don't remember, but I'm sure some members have compared the oscillator with and without the oven in listening session and if I remember correctly they did prefer the oven option.
Andrea
https://www.diyaudio.com/forums/dig...itter-crystal-oscillator-308.html#post6390899
Without any damping, and extreme conditions, worst case some spurious around -150dB. Otherwise said, nothing, zip, nada, jack ****.
Hello Andrea,
Regarding the clocks we are constantly talking about noise that most people cannot even measure so if spending 100 Euro on lead to be moulded just gives a tiny difference it would be a good investment.
Because my aluminium boxes are a bit bigger it wont be difficult to mould some " elements " to be glued on the inside.
Because of the position of the connectors and because there is no space to position them vertically i cannot completely surround them with lead shot.
I have to make a real dimension " box " to see what will be possible in real life.
Greetings,Eduard
Regarding the clocks we are constantly talking about noise that most people cannot even measure so if spending 100 Euro on lead to be moulded just gives a tiny difference it would be a good investment.
Because my aluminium boxes are a bit bigger it wont be difficult to mould some " elements " to be glued on the inside.
Because of the position of the connectors and because there is no space to position them vertically i cannot completely surround them with lead shot.
I have to make a real dimension " box " to see what will be possible in real life.
Greetings,Eduard
Hi Marcel,
yes, our target are R2R multibit DAC and low frequency oscillators since our DACs will run with 5/6 MHz master clock.
Many members did ask for higher frequencies so we have extended the range up to 90/98 MHz (basic oscillators up to 25MHz, above 25MHz with frequency doublers).
And yes, the title of the thread is "Building a low phase noise/jitter crystal oscillator", so we have reached our target with the DRIXO oscillators at 5/6 MHz (phase noise around -150dBc at 10Hz from the carrier and jitter around 40fs).
From our experience, we believe the close in phase noise is important in digital to analog conversion so we are following our own way: very low phase noise oscillators and FIFO buffer which really isolates the source from the DAC, operating in different time domain.
That's our way regardless of this thread and also regardless of any diy audio forum.
Everyone is free to think we are wasting time and money, no problem, we respect different opinions but we disagree and so we will keep following our way.
Me and my co-developer Roberto we have already discussed about a possible measurements at the output of the DAC.
We think the only way is using an up converter as I wrote in a previous post.
This way we could see the noise spectrum at the DAC output using audio frequencies such as 20Hz, 1kHz, 10kHz, 20kHz and so on.
The great problem is that it's not very simple removing the harmonics image generated in the up conversion.
A simple low pass filter will not work, the harmonics are too close to the audio signal.
Maybe we will resume this project later, now we have no time for this, too much projects have to be still completed.
Andrea
True, here's an excerpt:
Nope, it does not, as discussed then, and now. Unless Andrea and his happy customers team believe -180dB is audible. With at least 3 orders of magnitude worse source material, and in the presence of a strong masking effect.
Andrea is selling an illusion, which would not be a crime if it would not be misleadingly advertised "that's affects heavily the sonic result".
P.S. And again:
And is not about radar or missile guidance systems, but Audio. Nothing to support these bold statements.
Last edited:
Again, I'm not selling anything, we are developing the top devices for ourselves and some other devices asked from many members.
I have shared my experience like other members on this forum.
But everyone is free to think whatever he wants, no one is obliged to buy anything.
Just as we cannot afford to give away what we build.
So I repeat what I have already wrote in a previous post:
"From our experience, we believe the close in phase noise is important in digital to analog conversion so we are following our own way: very low phase noise oscillators and FIFO buffer which really isolates the source from the DAC, operating in different time domain.
That's our way regardless of this thread and also regardless of any diy audio forum.
Everyone is free to think we are wasting time and money, no problem, we respect different opinions but we disagree and so we will keep following our way."
That's all, the is no way to change our minds.
I have shared my experience like other members on this forum.
But everyone is free to think whatever he wants, no one is obliged to buy anything.
Just as we cannot afford to give away what we build.
So I repeat what I have already wrote in a previous post:
"From our experience, we believe the close in phase noise is important in digital to analog conversion so we are following our own way: very low phase noise oscillators and FIFO buffer which really isolates the source from the DAC, operating in different time domain.
That's our way regardless of this thread and also regardless of any diy audio forum.
Everyone is free to think we are wasting time and money, no problem, we respect different opinions but we disagree and so we will keep following our way."
That's all, the is no way to change our minds.
I'm afraid it is not you and your DIY partner(s) wasting time and money.
- Status
- Not open for further replies.