Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[Barry Hufker]

I agree if it sounds right it is right.  I just wonder how much glitz figures into the equation.

[Guillermo Piccolini]

WOW... nice talk about digital in the tape thread... LOL.

I wonder if mgod can explain which problems he has with the bad batch of ATR tape, because I had read about some people having shedding problems with RMG and was interesting in ATR. I´m also interested in the sonic differences between BASF/Emtec 900 and 911 formulation. I understand one being +6 and the other +9, but in the 90s, while still using tape I liked the 456 better than the 499 and felt the GP9 nicer than the 499 but never get really sure if I liked it better than the 456. That was on semi-pro equipment: Tascam 38 (8 track 1/2 inch 15ips) Now I have a more interesting studio than then and plan to mix to a Telefunken M15 1/4 15ips. My first tests with 911 are good, but the tape was not new and still waiting the test tape to calibrate the machine.

[Larrchild]

I think that what I am trying to say, is that, if the clock sounds really good, theory suggests that the long-term stability is not the reason.

So what I'm looking for is the plausible explanation for the improvement. This "It just does" stuff is fine, and I deal in subjective nuances I can't quite quantify, all day, so I understand "It just does". But I still try to know why.

[mgod]

I disagree that its subjective. Its observational - and observations lead to scientific exploration. Yes, it does sound better. Now, "Why?" is a valid response, as opposed to "The math says it doesn't".

Our problem with ATR is a flaw in the actual manufacture that results in one channel being about 5dB down. Shedding not a problem. I've heard about that shedding problem too with RMG, although the producer renting my 1" machine for the recent Lucinda Williams record has used a lot of it these last 3 months with no problem. I hope that the next batch of ATR will be flawless because its sounds great, although I'm not sure when I'll be in a position to take that risk again. Definitely cost us.

DS

[Larrchild]

Ok, Dan. Maybe I'm just unable to communicate this. I'm getting Brian's "Learn to Stop Worrying and Love the Bomb" angle, and your "It's observational, not subjective" detail.
Which is fine also.

Still all heat and no light though. So, I'll just find out what is actually better about it, since the Rubidium part doesn't pass the Lavry/Putzney analysis. Trust me, we are all on the same side here.

A good answer would please everyone.

[Andrew Hamilton]

mgod wrote on Sun, 04 May 2008 12:42

Its all in the listening Barry -  if you can't or don't hear it its not worth a cent. For me on this project it was worth anything, and our rental fee wasn't cheap, but a bargain nonetheless. All the debate, all the alleged "fact", all the so-called scientific positioning re: PLL doesn't mean anything at all if the results sound more like actual music. Internal clock, lead vocal is flat like a cartoon on a plasma screen. Antelope rubidium clock in, vocal pops out in front of the speakers and has depth to it. Sounds closer to a real voice in front of you. (C-12/EAR 824M/660/PT9[192kHz]).... DS

OK.  First of all, you might want to record and reproduce
at something lower than 192k Fs, but that's maybe because of PT converters?  Yes, for a filter of a given quality, it normally sounds better to put it at a higher frequency wrt Nyquist, but - clock speed piss match rights aside -  the chips presently in your DAC really need to relax between pulses for longer than 192k Fs allows  in order for the most linear response, especially the low end (that is, for the good garbage in to equal the same garbage out)...   That notwithstanding, if the vocals sound more present with the Rubidium clock engaged, it is most likely because there is more jitter in the monitoring chain.  Since vocal presence is in the 1k-6k zone, and if you add jitter to a monitoring chain, those frequencies will sound more "forward" (for one), I fear that, like all external clocking scenarios, the Rubidium is able to dial in an exciting bulge of stridence that really shouldn't be there (during monitoring) and won't be there, either way, after the clock has been removed from the chain - or else it has done nothing, anyway.  Internal clock is King for all time - sorry.  And 4x Fs is great for some DSP, but not for conversion to or from analog - until chips get faster and/or more robust.


Andrew

[mgod]

Larrchild wrote on Sun, 04 May 2008 17:50

A good answer would please everyone.

Especially a cheap one.

I never thought we weren't on the same side. But I'm sure you're well aware of how often the accusation of "subjective" has been used to dismiss valid observation. This approach, in my opinion and experience, stands in the way of forward motion.

This argument of all heat and no light is of course completely subjective and totally self-referential. For me, the math means nothing; the sound means everything - so its all light for me. Something has made this mediocre system sound better. Practical, useful illumination.

Andrew Hamilton wrote on Sun, 04 May 2008 19:02

That notwithstanding, if the vocals sound more present with the Rubidium clock engaged, it is most likely because there is more jitter in the monitoring chain.  Since vocal presence is in the 1k-6k zone, and if you add jitter to a monitoring chain, those frequencies will sound more "forward" (for one),... Andrew

I never wrote anything like that. If anything, there is less "presence" with these clocks.

So don't fear. Listen.

DS

[Barry Hufker]

I'd listen but I'm not paying somewhere around 8K for the privilege.

[mgod]

Yeah, well...there's the radiation burn. I had to convince the local rep to rent me the pair.

DS

[Barry Hufker]

It seems to me one could receive the 70 MHz time signal from the Naval Observatory and then bump that down to something useful and then amplify that as a clock (pardon me for being so technical in my explanation).  That is far more accurate than Antelope's box.  After all, the Navy has got to keep the exact right time all the time.

But even if one is using 10 MHz or 70 MHZ, isn't it the circuit dividing the signal down to the sampling rate that makes all the difference?  Is that a Phase-Locked Loop?  And if so, aren't they notorious?

[bruno putzeys]

A rubidium cell runs at a completely odd frequency which is then converted to 10MHz using a slow VCXO based PLL. Due to the operation principle of rubidium frequency standards (trying to lock a modulation side band into a spectral absorption line), the jitter requirement to this VCXO is very strict. So what you're seeing in terms of jitter is the quality of the crystal oscillator, not that of the rubidium cell. The rubidium only acts as very-long-term stabilisation.
Anyhow, rubidium time standards are largish tin boxes that one buys off-the-shelf from specialised suppliers. Someone buying those for inclusion in an end-user device may be easily fooled into believing that the rubidium is responsible for the jitter performance (the original manufacturer would know better). However, since it isn't, it is more effective to make just a crystal oscillator running directly at a digital audio related frequency. This makes life easier on the PLL of the audio device being synched.

[PaulyD]

Anyone here ever listen to the 3D Audio ADC CD (it's available from Mercenary Audio)? It's a two-CD set where Lynn Fuston took twenty nine A/D converters and ran three pieces of music through them. He ran a fourth test where he took any of the units with wordclock inputs and clocked them with a Lucid SSG-192.

I was startled by the external word clock test. The sound of some of the most expensive converters in the test clearly improved. This is not subjective. I've experienced this myself by clocking an old MOTU 2408 MkII/PCI-424 with the Apogee C777 clock. The 2408 MkII sounds better this way. Period.

Now is this to say that any and all AD/DA converters could be improved with external clocking? No, of course not. You must do what DS did; listen and decide.

I'm all for the technical understanding of how things work. I really am. But this is music we're talking about. If art and science conflict here, side with art.

As an aside, you may not need something so expensive to test external clocking. A Lucid GENx192 has a street price of about US$700.

Apologies to  Guillermo for the off topic post.

Paul

[bruno putzeys]

Some converters improve markedly when clocked by an external clock,  some get worse, some hardly change at all. This is not black magic. Presuming that the phase comparator of the PLL is not too noisy, the jitter spectrum at the output of a PLL is identical to that of the input clock at low jitter frequencies. At high jitter frequencies it's identical to that of the PLL's own oscillator. If the external clock is more stable than the internal clock, the low-frequency spectrum of the converter clock will be cleaner in slave mode than in master mode (while the high-frequency spectrum may be worse). But here's the rub: many PLL's have corner frequencies up to 10kHz, so "low frequency" might still be most of the audio band, which amounts to a wholesale improvement.

From this we learn that converters with a very good internal clock have nothing to gain from external clocking, while converters with a low-frequency PLL won't gain from external clocking. The ones that do respond with an improvement are those with a "garden variety" internal clock and a fast PLL. That is the majority. Unfortunately, the majority of word clock generators have a garden variety clock generator too... Very often the jitter performance is overhyped by specifying "nnn picoseconds of jitter",  which can mean anything. It's really simple to get 2ps jitter if you allow yourself to read the period-to-period jitter. The only useful way of specifying jitter is as a spectrum (example).

[Guillermo Piccolini]

PaulyD wrote on Mon, 05 May 2008 05:42

Apologies to  Guillermo for the off topic post. Paul

...well, I´m glad to see my topic so alive and interesting... even if it has nothing to do with the original Q....  






...BUT: can somebody explain me the sonic differences between 911 and 900 tape formulation?. Hiss doesn´t seem to be a big problem with 911... will 900 sound better in any way other than 3db more output?.



.. now you can keep talking about jitter and picoseconds.

[mgod]

Andrew Hamilton wrote on Sun, 04 May 2008 19:02

Internal clock is King for all time - sorry. Andrew

Bruno Putzeys wrote on Mon, 05 May 2008 04:32

From this we learn that converters with a very good internal clock have nothing to gain from external clocking, while converters with a low-frequency PLL won't gain from external clocking. The ones that do respond with an improvement are those with a "garden variety" internal clock and a fast PLL. That is the majority. Unfortunately, the majority of word clock generators have a garden variety clock generator too...