DAC Clock mod

A

adamdea

Joined
May 18, 2006
Messages
36
Reaction score
0
I have a somewhat modded MF X DAC V3 (a few years ago- thank you Zanash) and have wondered whether it was worth having the clock upgraded .
I wondered whether other had any experience of this and whether they can comment on
1 how much of an improvement it might make
2 which of the many manufacturers (audiocom, tentlabs, trichord, sercal etc) would provide the best product. I don;t fancy messing around myself so i will probably get them to install.
Audiocom claim that the clock upgrade is more significant than any other upgrade, but then perhaps they would.

I use an MF cd player as transport, but mainly now use a squeezebox touch.
 
Last edited by a moderator:
DACs usually use a PLL to generate the clock. You can not drop a "clock mod" in, as one would do in a CDP. An option would be to use a VCXO, and a PLL, that has a slower loop than the main loop.

I do know there are some folks in the DIY crowd who do something that is frowned upon by those of us who build this sort of equipment. They use a clock that runs at a higher frequency than the bit-clock, and asynchronously re-clock the bit-clock signal. Invariably, this will lead to dropped bits. Whether one can hear when the drop-outs occur, is another subject. But, in theory, it uses a stable clock to generate the bit-clock, even if some bits are missed.

Any clock must have a supply that has very low noise, especially 1/f noise. Most clocks are spec'ed at jitter frequencies > 1 kHz, so there is no way to know the 1/f of any of them. Part of the reason 1/f is not specified is that it is higher than mid-band noise, and therefore looks worse on a data sheet. Another is that it is harder to measure. Plus, to get good numbers, you can only get them with a low-noise supply. There really isn't much incentive to go to all the bother of measuring 1/f performance, especially since no one else does. And the customer will in all likelihood get nowhere near the same performance.

Instead, the folks who make these clocks just stick a number of "x pSec of jitter!" on their product, and sit back and collect $. Since there is no easy way to compare competing units, there is a lot of room to pull the wool over the customer's eyes.

Having said all of that, I would be curious to know how these folks will tell you "Sure, we can stick one of our clocks in your DAC. No problem!"

Sorry: problem.
 
YOU!!!!!!!!!! I can't get away from you.

Just joking, folks.

Anything you think I should add, Keith?
 
Sorry Pat it must seem as though I am stalking you!
As to adding, well I would like to add your Legato to my 'portfolio' !
Very Best,Keith.
 
Well in the meantime I succumbed to temptation and got the mod by audiocom. I think the DAC might sound a bit better now, but the clock was not the only change. Perhaps you could ask Audiocom how they feel able to drop in a clock mod into a DAC.
The thing that really puts me off is Audiocom's promotion of bybee filters, which seem to me not to make any sense. Mind you they apprently purport to address the question of the dreaded 1/f noise, albeit incomprehensibly.
 
Outfits like Tent Labs make PLL/VCXO combos. Not the same as a stand-alone clock.

The Bybees are not supposed to be understandable. I am sure Jack makes all of that up, just to keep folks from figuring out what he is doing.
 
good to see that its still up and running !

most clock up grades are a clock module ..some are quite large and theres not a huge ammount of space left in the xdac case .

Most do what they claim ...though I've not found them to be quite as effective ,and certainly not as cost effective as four schotky diodes at a pound plus on the overal sound ..

personally the cost of a clock might be better put to a newer/higher perdormance dac ?
 
Hi Pete
apologies for delay in replying. Yes still running. I think you may be right. I have being seriously toying with buying a linn ds for some time. Otherwise the only affordable DAC which would think of buying would be a benchmark.
 
ar-t said:
Outfits like Tent Labs make PLL/VCXO combos. Not the same as a stand-alone clock.

The Bybees are not supposed to be understandable. I am sure Jack makes all of that up, just to keep folks from figuring out what he is doing.
Click to expand...

Forgive me for being obtuse, but do you mean that Jack "makes all of that up" to protect his intellectual property rights in a scientifically sound and practically effective audio innovation, or to conceal the fact that he is selling something which does nothing.
 
It is the former. The building block is used in submarines, we have reason to believe, to reduce 1/f noise on the battery supply lines. At least, that is the guess folks who know Jack, and have some understanding of what they are claim.

(OK, some feel it is some kind of microwave resonator. No, I have no clue how one would reduce 1/f noise. But then, I flunked quantum mechanics.)

Pat
 
ar-t said:
DACs usually use a PLL to generate the clock. You can not drop a "clock mod" in, as one would do in a CDP. An option would be to use a VCXO, and a PLL, that has a slower loop than the main loop.

I do know there are some folks in the DIY crowd who do something that is frowned upon by those of us who build this sort of equipment. They use a clock that runs at a higher frequency than the bit-clock, and asynchronously re-clock the bit-clock signal. Invariably, this will lead to dropped bits. Whether one can hear when the drop-outs occur, is another subject. But, in theory, it uses a stable clock to generate the bit-clock, even if some bits are missed.

Any clock must have a supply that has very low noise, especially 1/f noise. Most clocks are spec'ed at jitter frequencies > 1 kHz, so there is no way to know the 1/f of any of them. Part of the reason 1/f is not specified is that it is higher than mid-band noise, and therefore looks worse on a data sheet. Another is that it is harder to measure. Plus, to get good numbers, you can only get them with a low-noise supply. There really isn't much incentive to go to all the bother of measuring 1/f performance, especially since no one else does. And the customer will in all likelihood get nowhere near the same performance.

Instead, the folks who make these clocks just stick a number of "x pSec of jitter!" on their product, and sit back and collect $. Since there is no easy way to compare competing units, there is a lot of room to pull the wool over the customer's eyes.

Having said all of that, I would be curious to know how these folks will tell you "Sure, we can stick one of our clocks in your DAC. No problem!"

Sorry: problem.
Click to expand...

Hi,

The MF DAC contains an upsampler, hence can be clocked by a free running oscillator.

On the specs: As far as I know, Tentlabs is the only audio clock manufacturer actually publishing phase noise data:

http://www.tentlabs.com/Components/XO/assets/TentLabs_XO_33.8688.jpg

shows that we are at 1 ps rms jitter

all the best

Guido
 

Latest posts

Back
Top