Is there any logical reason to use a tube as the oscillator in a DAC?

[TimFox]

An interesting historical article about the transition at -hp- from vacuum-tube to solid-state crystal-controlled frequency reference generators.
https://www.hpmemoryproject.org/news/100e/hp100e_page_00.htm
At that time, there was little discussion of jitter:  the improvements concentrated on frequency stability, which was more a function of the crystal and its oven than the good electronics that followed it.

[vk4ffab]

However, the design of low-jitter clock circuits is a quantitative design problem.

True, and while I was being humorous, not snarky, I made a very real point, and that is, in what situation will jitter be an issue in audio recording or playback? Arguing about the technical merits of the amount of jitter in a system is one thing, in certain situations this can be a huge issue, but in audio recording, I am doubtful anyone is going to hear any degradation in audio quality due to clock jitter for the DAC.

So for a qualitative measurement of the detection threshold of jitter in digital audio, we are talking about 120ps random jitter on a  20khz tone with 16bit resolution for it to be accuralty reproduced. This is a value that is orders of magnitude over the jitter found in most consumer grade products.

So, if you cant hear it, arguing about 5ps or 100ps in the clock is pointless and a manufacturer arguing their 5ps tube clock makes better audio is nothing more than pseudoscience.

[coppice]

So for a qualitative measurement of the detection threshold of jitter in digital audio, we are talking about 120ps random jitter on a  20khz tone with 16bit resolution for it to be audibly noticeable. This is a value that is orders of magnitude over the jitter found in most consumer grade products.

So, if you cant hear it, arguing about 5ps or 100ps in the clock is pointless and a manufacturer arguing their 5ps tube clock makes better audio is nothing more than pseudoscience.

Low jitter isn't really about reproduction. Its about recording. If you want something close to real 24 bit recording, so you have lots of headroom, and don't have to be hyper sensitive about record levels, you need really low jitter. Once its all mixed down, the dynamic range is a lot lower, and it matters less.

[TimFox]

Of course this is just marketing. Bonus point if someone did a proper teardown and found out that the tube was actually not connected to anything.

Not even talking about the microphonic behavior of tubes and how even small vibrations could throw it all off. Maybe you need to mount this equipment on 10 tons of concrete or something, like the very high-end turntables.

If anyone here is able to design a tube-based oscillator that exhibits better frequency accuracy and lower jitter than a pure solid-state approach with modern ICs, I'll be happy to have a look at it.

Here is an example of a high-grade vacuum-tube based 1 MHz crystal oscillator design from 1960  (US Navy Bureau of ships article).  See page 19 for the 6AH6 pentode Meacham-bridge oscillator (with a transistor included in the thermostat for the crystal oven).
https://www.navy-radio.com/journal/journal-freq-synth-6012.pdf
Again, jitter in time units is not discussed, but the short-term stability is quoted as 2-3 x 10^-9 per day.  With the Meacham-bridge circuit, this should depend almost completely on the crystal itself and the passive circuit connected to it.  Therefore, using the pentode instead of a modern solid-state device is not the magic in the circuit.

[ataradov]

Low jitter isn't really about reproduction. Its about recording.

This topic is about consumer DAC though. And their own claim is "By using a tube, we have significantly lowered the amount of jitter and noise, resulting in superior detail retrieval." There is also a lot of BS superlatives.

They also claim "Here are a few real SuperTubeClock™ pictures of what it can do: experts will certainly like them. No other commercial clock on earth will produce a square wave as this one." This is accompanied by pretty mediocre looking scope captures.

They also confuse units "mHz" and "MHz". But such details are above them, I guess.

And their super clean (tm) clock still goes into the CMOS DAC where it has to pass though regular CMOS gates subject to the same variation in the edges.

[vk4ffab]

So for a qualitative measurement of the detection threshold of jitter in digital audio, we are talking about 120ps random jitter on a  20khz tone with 16bit resolution for it to be audibly noticeable. This is a value that is orders of magnitude over the jitter found in most consumer grade products.

So, if you cant hear it, arguing about 5ps or 100ps in the clock is pointless and a manufacturer arguing their 5ps tube clock makes better audio is nothing more than pseudoscience.

Low jitter isn't really about reproduction. Its about recording. If you want something close to real 24 bit recording, so you have lots of headroom, and don't have to be hyper sensitive about record levels, you need really low jitter. Once its all mixed down, the dynamic range is a lot lower, and it matters less.

True, but in playback, which is what we are talking about with this audio fool DAC, even trained listeners struggle to hear random jitter less than 500ns. If I remember rightly, even the high end recording sound cards back in the day had jitter levels in the low 1000's of ps and no one complains their favorite song from 00's has audible jitter on it.

[langwadt]

Low jitter isn't really about reproduction. Its about recording.

This topic is about consumer DAC though. And their own claim is "By using a tube, we have significantly lowered the amount of jitter and noise, resulting in superior detail retrieval." There is also a lot of BS superlatives.

They also claim "Here are a few real SuperTubeClock™ pictures of what it can do: experts will certainly like them. No other commercial clock on earth will produce a square wave as this one." This is accompanied by pretty mediocre looking scope captures.

a 10cent tiny logic gate will make sharper edges than that

[TimFox]

Low jitter isn't really about reproduction. Its about recording.

This topic is about consumer DAC though. And their own claim is "By using a tube, we have significantly lowered the amount of jitter and noise, resulting in superior detail retrieval." There is also a lot of BS superlatives.

They also claim "Here are a few real SuperTubeClock™ pictures of what it can do: experts will certainly like them. No other commercial clock on earth will produce a square wave as this one." This is accompanied by pretty mediocre looking scope captures.

a 10cent tiny logic gate will make sharper edges than that

Sharp edges are the easy part.  A good design of the crystal and its oscillator circuit is more important.
A sinusoidal-output crystal oscillator can have good jitter, but its edges are not sharp.

[PCB.Wiz]

https://www.nisshinbo-microdevices.co.jp/en/products/quartz-crystal-oscillator-ic/

 NJU6385
Phase noise   Phase noise (-103dBc/Hz(Typ.) @10Hz Offset  -163dBc/Hz(Typ.) @1kHz Offset
RMS Jitter   0.05psec(Typ.) @12kHz to 20MHz
Operating Current   9.0mA(Typ.) @CL=15pF.

Looking for alternative measurements on CMOS gate oscillators, I find this app note from TI.

https://www.ti.com/lit/an/scaa119/scaa119.pdf
That's old, but curves are similar but not quite as good as the NJU6385, tho I see NJU6385 specs a low 17.7 ohms crystal, with Q ~ 67,000

a 10cent tiny logic gate will make sharper edges than that

Buffers are easy,

TI info gives 3.3-V LVCMOS Clock Buffer Family additive jitter of  ~7.5 fs typical at VDD = 3.3 V (Sq in, Sq out)

[vk6zgo]

When "the rubber hits the road", we are perceiving the result of all this "wizardry" using sensors which had their last design upgrade  tens of millions of years ago & are subject to wide variability in individual characteristics.
On top of that, going back one step in the chain, the transducers which convert the electronic back to sound have characteristics which are much more variable than the electronic part of the chain.

Of course, to many the measurement of small increments of difference in various characteristics is an end in itself, somewhat like those religious scholars who spent years in debate as to how many angels could dance on the head of a pin.

Meanwhile, the audiophile "true believers" are happy, the scammers are happy, the wheels of commerce roll on, who are we to question?

[antenna]

Seeing a glowing filament next to an oscillator makes me immediately think of Jeri Ellsworth's video titled "Don't use lame parts".  Maybe not related, but thats what comes to mind.

[SiliconWizard]

Low jitter isn't really about reproduction. Its about recording.

This topic is about consumer DAC though. And their own claim is "By using a tube, we have significantly lowered the amount of jitter and noise, resulting in superior detail retrieval." There is also a lot of BS superlatives.

They also claim "Here are a few real SuperTubeClock™ pictures of what it can do: experts will certainly like them. No other commercial clock on earth will produce a square wave as this one." This is accompanied by pretty mediocre looking scope captures.

a 10cent tiny logic gate will make sharper edges than that

The tube can only get them worse results than a proper CMOS buffer here, including microphonic effects as I mentioned, which will add jitter unless again the whole thing was 100% vibration-free.

This is absolutely not the same as using tubes in analog stages in the audio signal path - while many find that questionable these days, tubes do add a particular type of distortion and frequency response that are often considered "pleasing" to the ear, it's not about high-fidelity, but twisting the signal in a specific way which has been known to "work" for listening to music for a pretty long time now.

Here they claim lowering jitter, and I do not buy that one second, I'm rather convinced of the opposite. So to answer vk6zgo, I personally rarely question this kind of things, people can well buy whatever they want at whatever price, even if they are just buying air. But if the overpriced designs actually make things *worse*, which I highly suspect here, then this is where I would personally put the line.

[bdunham7]

I came across a PrimaLuna DAC that uses a tube for the oscillator.
Why? is there any logical reason for this?

To give this guy something to talk about.

[ELS122]

Ah yes, I totally mentioned the pure sound of tubes and detailed highs in OFC cables as well as anti-static grounding making less buzz in my tweeters.

If most of you could put aside your anti-audiophile agendas, maybe we'd get somewhere.
right now there's those saying that "jitter does matter even with today's tech"
and the rest which are "this is marketing, have I mentioned about marketing. yes, it's that".
So it seems no one knows how a tube as the oscillator would compare to a solid state circuit.

[bdunham7]

So it seems no one knows how a tube as the oscillator would compare to a solid state circuit.

No, I think that was covered, in Reply #28 for example.  TLDR: having a tube doesn't offer any technical advantage, lower jitter or otherwise.

[vk4ffab]

So it seems no one knows how a tube as the oscillator would compare to a solid state circuit.

Its irrelevant when the jitter is less than 500ns because most normal people cannot hear it. The fact that the oscillator is in the 5ps range just means the 24bit dac is not reproducing the audio perfectly, for that it would require a jitter of 0.5ps. You know, math, 1 over 2 pi * frequency * 2 to the power of n or whatever it is for perfect digital reproduction. So either the manufacturer really does not know basic theory, or they are using a whole lot of spin to sell over priced tube things to the gullible.

The science is out there for anyone who can read and there have been plenty of studies on the audibility of jitter in digital audio. There are a lot of other real things that effect audio quality, but you have to design a pretty bad oscillator for jitter to actually matter. You could toss an SI5351a into that thing, save a few thousand bucks by having a tiny tube that does bugger all and still not hear its typical 35ps jitter in the recovered audio. You could probably use the first xtal oscillator I built on copper clad board and have great audio as well as the joy of oscillation on the scope.

Most of my electronics is based in the practical, not theory so 5ps or 100ps who cares when in practice its all wankery.

[Haenk]

I think all this has it's roots in early DAC implementation - which has indeed been mostly awful and was probably widely discussed (and is forever burnt as Eeeeevil into the brains of early enthusiasts and HiFi-magazine readers). Due to cost restrictions, most of the early CD players came with just *one* DAC (one DAC for both channels, that is). So the companies had the great idea to toggle the channels and to delay one output channel.
This wasn't an issue any more at the end of the 80s, but seems to be still a concern. Probably someone heard about it when still a kid...
Honestly, I think jitter is the smallest problem in a home audio setup. And so did all the engineers back then, obviously. Symmetrical construction for the audio path - yes, please. But those damned bits, some flying wires are good enough. And I guess, they were right...

[radiolistener]

I came across a PrimaLuna DAC that uses a tube for the oscillator.
Why? is there any logical reason for this? Is jitter of any concern at all with modern oscillators? Is there any feedtrough from the oscillator to the audio?

Yes, oscillator phase noise (jitter) is very important for high performance DAC. High phase noise leads to a dynamic range reduction for DAC.

I'm not sure if that tube oscillator has low phase noise, but if you're using top end DAC, it requires to use ultra low phase noise oscillator for clock. Such oscillators are pretty expensive (20-40 USD). Here is example of such oscillators: https://abracon.com/Precisiontiming/ABLNO.pdf

Good quality DAC requires oscillator with jitter about 0.1 ps or less. Most of Chinese TCXO cannot provide so low jitter.

But I din't hear that tube oscillator can provide ultra low phase noise, so I'm not sure if this is marketing thing or not.

True, but in playback, which is what we are talking about with this audio fool DAC, even trained listeners struggle to hear random jitter less than 500ns. If I remember rightly, even the high end recording sound cards back in the day had jitter levels in the low 1000's of ps and no one complains their favorite song from 00's has audible jitter on it.

Yes, in most cases the difference with usual good TCXO will not be noticeable, it requires to hear special sound patterns and have good ears to detect it.

[Kleinstein]

The typical audio data are not 24 bit quality and the DAC used is likely also not true 24 bits. The quoted 0.5 ps jitter requirement would only be needed for a FS sine signal near the top of the frequency range.  Actual audio has rather limited high frequency contend and the DAC is usually oversampling, which also reduces the requirements.

Low jitter helps, but it is usually not that critical and other points more be more important. The ear can anyway not hear that much of imperfections. If at all one may be able to measure with really high end gear.

I don't see a way why a normal tube should be low jitter or especially stable - more of the opposite: a tube is sensitive to magnetic fields and vibrations and not super fast. So one can expect rather high jitter from a tube based oscillator.   Alone having the tube and crystal separated that far apart and not well shielded makes it susceptible to EMI.
Chances are most of the 25 cent canned oscillators are better.

[coppice]

I think all this has it's roots in early DAC implementation - which has indeed been mostly awful and was probably widely discussed (and is forever burnt as Eeeeevil into the brains of early enthusiasts and HiFi-magazine readers). Due to cost restrictions, most of the early CD players came with just *one* DAC (one DAC for both channels, that is). So the companies had the great idea to toggle the channels and to delay one output channel.
This wasn't an issue any more at the end of the 80s, but seems to be still a concern. Probably someone heard about it when still a kid...
Honestly, I think jitter is the smallest problem in a home audio setup. And so did all the engineers back then, obviously. Symmetrical construction for the audio path - yes, please. But those damned bits, some flying wires are good enough. And I guess, they were right...

CD players produced in the first 3 or 4 years were quite abysmal. They eliminated the hiss, rumble and scratches we had to put up with from LPs or tape, but the actual sound quality was poor. Any complex music had the muddiness you associate with a lot of intermod products. The DACs they used were only about 13 or 14 bits linear. It took several years before they cleaned things up. It was about 1990 before CD reproduction was outclassing LPs on every front. This was also about the date when sales of CDs really started to accelerate. I expect that was largely coincidence.

[coppice]

Its irrelevant when the jitter is less than 500ns because most normal people cannot hear it.

I think you mean 500ps. Even the tone deaf would hear 500ns.

[coppice]

I came across a PrimaLuna DAC that uses a tube for the oscillator.
Why? is there any logical reason for this? Is jitter of any concern at all with modern oscillators? Is there any feedtrough from the oscillator to the audio?

Yes, oscillator phase noise (jitter) is very important for high performance DAC. High phase noise leads to a dynamic range reduction for DAC.

I'm not sure if that tube oscillator has low phase noise, but if you're using top end DAC, it requires to use ultra low phase noise oscillator for clock. Such oscillators are pretty expensive (20-40 USD). Here is example of such oscillators: https://abracon.com/Precisiontiming/ABLNO.pdf

Good quality DAC requires oscillator with jitter about 0.1 ps or less. Most of Chinese TCXO cannot provide so low jitter.

But I din't hear that tube oscillator can provide ultra low phase noise, so I'm not sure if this is marketing thing or not.

True, but in playback, which is what we are talking about with this audio fool DAC, even trained listeners struggle to hear random jitter less than 500ns. If I remember rightly, even the high end recording sound cards back in the day had jitter levels in the low 1000's of ps and no one complains their favorite song from 00's has audible jitter on it.

Yes, in most cases the difference with usual good TCXO will not be noticeable, it requires to hear special sound patterns and have good ears to detect it.

Anyone talking about the need for a TCXO or a jitter level of <0.1ps has seriously lost the plot. TCXOs, and rubidium clocks, have their place in recording, to allow multiple sources to be freely mixed and matched. They have no value for reproduction.

[tszaboo]

You see this below? That's how it is recorded. There are no tubes there, just a bunch of NE5532s, and usually a muppet that will compress  every single recording so it can be played back on a 3 dollar earphone sold by Apple for 200 dollars.

[CaptDon]

I can't imagine any tube oscillator being better than a cheap TCXO. Thinking back to the military tube type crystal controlled reference oscillators where the crystal, the oscillator tube and the buffer tubes were all contained within a double oven where the inner and outer oven had separate controls. Even the glow tube oscillator voltage reference was inside the oven and the double ovens also blocked anything except cosmic rays to obtain very low phase noise. If you pounded really hard on the rack (these beasts were rack mounted) you could produce phase/noise/jitter modulation but even the outer oven was shock mounted. Mine was built by C.R.L. or something like that for military / Wallops Island Goddard flight center. Navy gray beast of about 25 pounds and taking a couple days to achieve final stability (but still, it had a crystal aging factor). So the end observation, they were selling magic snake oil in a box to audiophools.
 

[floobydust]

From EVO 100 pics, you can see they use russian submini triode 6S6B, then AD8611 comparator to pick off the signal at the plate. I doubt the triode gets the ~120V, no idea about the source of clean HVDC for it.
It's a straightforward xtal osc, using the cathode drive. But it's wrong to have any DC across a quartz crystal to prevent ion migration. So I see design errors, drive likely high as well.

This is obviously for psycho-acoustics a $4,000 DAC needs something to stand out from the others. It also keeps the 8051 warm lol.