Of course it would. Take a phase noise measurement with the Microsemi 5125 right before the DAC board and you'll get a reading. You will find it's much worse than the same grade of clock in 22Mhz.
Last edited:
As an example, look at the 10Hz phase noise difference between the 22Mhz and 100Mhz in the Pulsar OCXO. And remember every 3dB is double, and goes up exponentially every 3dB.
http://www.pulsarclock.com/ds/Pulsar_Clock.pdf
I can get far better performance from a $2 NDK NZ2520SD @ 22.579Mhz, than a 400 Euro Pulsar @ 100Mhz.
http://www.pulsarclock.com/ds/Pulsar_Clock.pdf
I can get far better performance from a $2 NDK NZ2520SD @ 22.579Mhz, than a 400 Euro Pulsar @ 100Mhz.
Ah, so now we should believe the theory on the spec sheet instead of real-world results?
Ah, so now we should believe the theory on the spec sheet instead of real-world results?
Real world results come with listening tests. Like the one here when compared with the $90000 MSB Select DAC 2 with 170dB dynamic range. Also note he had the worlds best dynamic headphones attached directly to the output of the DAC in the comparison. You won't get better than that. No bottlenecks in the signal path. That 170dB dynamic range directly into the headphones. My DAC is handicapped with a preamp and a set of interconnects in the signal path.
https://www.head-fi.org/f/threads/p...ne-rig-focal-utopia-mivera-superstack.838570/
Last edited:
Oh and here's the 15000 Euro combo that my 3500 Euro combo replaced.
https://www.head-fi.org/f/threads/fs-totaldac-d-1-dual-d-1-server-sold.822189/
https://www.head-fi.org/f/threads/fs-totaldac-d-1-dual-d-1-server-sold.822189/
I don't recall Bruno saying he upsampled all the way to 100MHz. The 100MHz clock in my understanding is giving 10nS quantization to his output pulse widths. So if his modulator were producing 8bit output words he'd have a maximum pulse width of 2570nS which suggests he might be only doing 8X OS.
Yes that's what he said it's equivalent to about 8x DSD. But the clock is 100Mhz. Which means it will have the phase noise of an 100Mhz clock. There's no doubt it's a great design. But as Bruno knows there's 1000 ways to skin a cat. I personally like to leverage the awesome power in the server everyone is using anyways for the digital processing. Then clock the data with lower phase noise clocks completely isolated from the digital processing with fiber.
Our next gen design has around 128dB dynamic range, but still 121dB will get you a noise floor lower than any human ear will be able to detect anyways. 128dB dynamic range matches the Mola Mola Kaluga amp. So a full Mola Mola setup will only have 128dB dynamic range anyways.
Well this thread has attracted twice as many views in 10 days than the Ncore DIY thread has in 3 months. What this tells me is the DIY community is pumped up about these Hypex plates. I'm just crossing my fingers Hypex decides to adopt AES67 into them.
I've already come up with the designs for 3 active kits for these plates. They will be the very highest standard stacked Baltic birch plywood, reinforced by steel rod. Pre-finished in many different high quality finish options. The only tools required to build will be an Hex key. Similar to when you order Ikea furniture.
I'll come up with 5 presets for crossovers with different voicing, and of course anyone will be able to tweak them how they want as well for their room acoustics, taste etc. If you mess it up no problem, just flash one of the original presets back onto the plate. I've looked very hard and can't find anything near this caliber available on the market today.
Look at the Dynaudio Focus XD series for examples of the 1st 3 models as far as how the cabinet shapes and driver configurations will be.
https://www.dynaudio.com/home-audio/focus-xd
Due to the extremely efficient way I'm going to have these manufactured, the price/performance is going to be a breakthrough, without resorting to Asian outsourcing. The cabinet grade is going to be up there with the best of the best commercial designs. Nobody will have a clue they weren't purchased from a high end retail store if you don't tell them. Best of all they will be so easy to build, anyone can easily do it in the comfort of their livingroom. And without even making a mess. If you can build an Ikea dresser, you'll be able to build these.
I've already come up with the designs for 3 active kits for these plates. They will be the very highest standard stacked Baltic birch plywood, reinforced by steel rod. Pre-finished in many different high quality finish options. The only tools required to build will be an Hex key. Similar to when you order Ikea furniture.
I'll come up with 5 presets for crossovers with different voicing, and of course anyone will be able to tweak them how they want as well for their room acoustics, taste etc. If you mess it up no problem, just flash one of the original presets back onto the plate. I've looked very hard and can't find anything near this caliber available on the market today.
Look at the Dynaudio Focus XD series for examples of the 1st 3 models as far as how the cabinet shapes and driver configurations will be.
https://www.dynaudio.com/home-audio/focus-xd
Due to the extremely efficient way I'm going to have these manufactured, the price/performance is going to be a breakthrough, without resorting to Asian outsourcing. The cabinet grade is going to be up there with the best of the best commercial designs. Nobody will have a clue they weren't purchased from a high end retail store if you don't tell them. Best of all they will be so easy to build, anyone can easily do it in the comfort of their livingroom. And without even making a mess. If you can build an Ikea dresser, you'll be able to build these.
Last edited:
A 170dB SNR? Does your neuronal metabolism have such a high dynamic range as 170dB? Empty brain maybe? Occupied brain has much less, toxins that reduce neuron SNR are produced when solving problems, and have to be cleaned up later.
Purpose of music is to provide emotional test and repair frequency patterns for metabolism activation and/or check. Can you hear a 110dB noise floor in a 250W amplifier with a 110dB/W horn tweeter? At 50cm this is about the limit to tell external signals from the own internal metabolic noise of living tissue.
Project ideas? Maybe a new type of mitochondria (celular sugar burner) with lower noise emissions has to be developed, or maybe a new type of blood vessel wall cell resulting in less flow noise, as when the metabolism is clean the noise floor of blood flowing through vessels can be heard, even in absolute rest.
Self negation results in inability to create good equipment.
Purpose of music is to provide emotional test and repair frequency patterns for metabolism activation and/or check. Can you hear a 110dB noise floor in a 250W amplifier with a 110dB/W horn tweeter? At 50cm this is about the limit to tell external signals from the own internal metabolic noise of living tissue.
Project ideas? Maybe a new type of mitochondria (celular sugar burner) with lower noise emissions has to be developed, or maybe a new type of blood vessel wall cell resulting in less flow noise, as when the metabolism is clean the noise floor of blood flowing through vessels can be heard, even in absolute rest.
Self negation results in inability to create good equipment.
Well this is what they claim. Actually 173dB dynamic range. 28.5 effective bits. But my client couldn't hear any difference in noise floor with Beryllium diaphragm headphones connected direct, than my 121dB dynamic range DAC.
An externally hosted image should be here but it was not working when we last tested it.
It depends on the clock. You can get near equivalent phase noise from a
100MHz clock to a 10MHz clock by using clock multiplication or phase locking.
It depends where you want your phase noise, using multiplication will throw
away HF phase noise floor for LF phase noise close to carrier. Phase locking
will get best of both worlds. I've seen phase locked 100MHz OCXO's with
better than -120dBc at 10Hz. This translates to a 10MHz clock with
better than -140dBc at 10Hz.
Bruno knows exactly what he is doing and you can guarantee that 100MHz
clock will have phase noise characteristics that are superb.
T
I checked that 173dB figure a while ago and going on the quoted voltage OP
level versus OP impedance (thermal noise) it was basically impossible.
I guess they are just an optimistic bunch of guys!
T
Unfortunately that is not the case. The Phase noise is referred to carrier so
for an equivalent jitter the phase noise of 20MHz clock will be 6 dB worse
than 10Mhz clock. And so a 100MHz clock will have 20dB worse phase noise
than 10Mhz clock - but same jitter.
Now for other reasons, generally most clocks degrade much more than this
6dB per doubling of frequency. So in fact the Pulsar 100MHz clock at -110dBc
at 10Hz offset is a very good spec and equiv to a 10MHz clock of -130dBc.
This is worth a read:
www.silabs.com/documents/public/white-papers/Clock-Division-WP.pdf
cheers
T
Clock prescaling or multiplying always adds major jitter, not reduces it. You can't get less phase noise from a clock than the clock has itself. All you can do is try to minimize more from getting into system downstream. I have no doubt he's using greats clocks. But he can't defy the laws of physics. Higher frequency clocks have higher phase noise. The same grade of clock in 22Mhz will have several times less phase noise.
Last edited:
In their example they are using a 10Mhz reference clock. And the damage is already done at all frequencies when you go through a divider or multiplier. No matter how you manipulate the signal. You will never get even close to as low of jitter using one of those units as clocking direct off the crystal. All crystals are unique and you will never get as uniform of phase noise spread at all frequencies as when you use a PLL generator and a reference clock. This is because the same clock is the reference for all frequencies.
In our Purestream we clock direct off the crystal. LRCLK is generated off the rising edge, and BCLK off the falling edge. We use an 22.579Mhz clock. So works out to a perfect 11.288Mhz for each without any prescaling or multiplication. It's impossible to do it any better than that, besides using lower jitter clocks.
Last edited:
I asked my friendly neighborhood clock guru his take on Terry's statements. Here's what I got back:
Statement: "It depends on the clock. You can get near equivalent phase noise"
Response : "Key word is "near equivalent". But, the way most folks look at things, they will come to a wrong conclusion."
Statement: "from a 100MHz clock to a 10MHz clock by using clock multiplication or phase locking. It depends where you want your phase noise, using multiplication will throw away HF phase noise floor for LF phase noise close to carrier."
Response: "True."
Statement: "Phase locking will get best of both worlds. I've seen phase locked 100MHz OCXO's with better than -120dBc at 10Hz. This translates to a 10MHz clock with
better than -140dBc at 10Hz."
Response: "Only if you start with a really good clock, to begin with. You will not be able to get better than the ref clock. Not sure his math is 100% correct, but it can be done. With a lot of work, and parts. But, not entirely out of the question.
Of course, people will point to SPDIF, and how noisy it is. It is, but only because it needs a wide capture range and fast lock-up time.
Part of the problem is that you can not pull really good crystals very far. Whoever makes the crystals better be good at what they are doing."
So my take is no matter what sort of trickery you do downstream, you will never get better than the reference clock itself.
Statement: "It depends on the clock. You can get near equivalent phase noise"
Response : "Key word is "near equivalent". But, the way most folks look at things, they will come to a wrong conclusion."
Statement: "from a 100MHz clock to a 10MHz clock by using clock multiplication or phase locking. It depends where you want your phase noise, using multiplication will throw away HF phase noise floor for LF phase noise close to carrier."
Response: "True."
Statement: "Phase locking will get best of both worlds. I've seen phase locked 100MHz OCXO's with better than -120dBc at 10Hz. This translates to a 10MHz clock with
better than -140dBc at 10Hz."
Response: "Only if you start with a really good clock, to begin with. You will not be able to get better than the ref clock. Not sure his math is 100% correct, but it can be done. With a lot of work, and parts. But, not entirely out of the question.
Of course, people will point to SPDIF, and how noisy it is. It is, but only because it needs a wide capture range and fast lock-up time.
Part of the problem is that you can not pull really good crystals very far. Whoever makes the crystals better be good at what they are doing."
So my take is no matter what sort of trickery you do downstream, you will never get better than the reference clock itself.