[teraguy]

You can not use other frequency for the TCXO. It should use 12M only. Otherewise, it will not work. ^-^

Quote:

Originally Posted by Fujak /img/forum/go_quote.gif Hi Thoppa, thanks for sharing your very interesting modding. Did you test several tcxo (means several frequencies)? It's a pity that it is so difficult to explain in an objective way what the increase of quality by a certain step of optimisation is. Given a scale of 1 to 100 and given the original state of the TeraLink is positioned at a value of let's say 75: at what position would you set the increase by modding your unit? I hope my question is clear enough (as you probably noticed I'm not a native speaker ). Fujak

 

[teraguy]

I can not speak of the price or any other business related issue. Otherwise, we may be deleted by the Admin.
If you are interested in the modding, please mail Michael directly.

Quote:

Originally Posted by Fujak /img/forum/go_quote.gif Yes, same to me!

 

[regal]

If you have a good upsampler on your computer a ood test would be to upsample the 44.1 material to 48khz, this would take the PLL out of the equation.

 

[thoppa]

Quote:

Originally Posted by regal /img/forum/go_quote.gif If you have a good upsampler on your computer a ood test would be to upsample the 44.1 material to 48khz, this would take the PLL out of the equation.

It might take the pll out but it would add in the quality of software upsampling. Besides this, I don't understand why it'd be necessary because the only things I can improve are the power and the jitter on the xo. Is there something I have missed ?

 

[regal]

Quote:

Originally Posted by thoppa /img/forum/go_quote.gif It might take the pll out but it would add in the quality of software upsampling. Besides this, I don't understand why it'd be necessary because the only things I can improve are the power and the jitter on the xo. Is there something I have missed ?

Yes when the TerralinX is downsampling 48 to 44.1 it is adding jitter, but when the computer upsamples it doesn't. This is why the Hiface has a separate 44.1 multiple clock.

 

[thoppa]

Quote:

Originally Posted by regal /img/forum/go_quote.gif Yes when the TerralinX is downsampling 48 to 44.1 it is adding jitter, but when the computer upsamples it doesn't. This is why the Hiface has a separate 44.1 multiple clock.

Hmmmm....do you know this thread is for the X2 (not the X) ?

 

[regal]

Quote:

Originally Posted by thoppa /img/forum/go_quote.gif Hmmmm....do you know this thread is for the X2 (not the X) ?

Yes, they both operate on a 12mhz clock (multiple of 48khz) instead of a multiple of 44.1khz like the Hiface and high end CDP's.

 

[thoppa]

Quote:

Originally Posted by regal /img/forum/go_quote.gif Yes, they both operate on a 12mhz clock (multiple of 48khz) instead of a multiple of 44.1khz like the Hiface and high end CDP's.

....um, I meant the X2 doesn't downsample....it can receive up to 96/24.

BTW I listen to a mix of native 44.1 and 96 files, which is why I got this. Does the hiface handle 96khz ? If so, then its xo won't be an exact multiple either, but maybe it has two xo's ?

So is there something I am missing about another avenue of modification available to me other than power and xo jitter ?

 

[regal]

Quote:

Originally Posted by thoppa /img/forum/go_quote.gif ....um, I meant the X2 doesn't downsample....it can receive up to 96/24. BTW I listen to a mix of native 44.1 and 96 files, which is why I got this. Does the hiface handle 96khz ? If so, then its xo won't be an exact multiple either, but maybe it has two xo's ? So is there something I am missing about another avenue of modification available to me other than power and xo jitter ?

Um, yes it does downsample if you send it a 44.1khz RBCD signal, it is recieved from the computer thru 48khz USB with the information for the PLL to put it back to 44.1. Its a great transport because it does this bit-perfect. Now if you send it 48khz or 96khz there is no downsampling. It is ideal for those formats. That's why I suggested a good upsampler for the computer so you avoid that extra step of jitterery PLL when playing RBCD 44.1khz.

And yes the Hiface has 2 Xo's.

 

[thoppa]

Quote:

Originally Posted by regal /img/forum/go_quote.gif Um, yes it does downsample if you send it a 44.1khz RBCD signal, it is recieved from the computer thru 48khz USB with the information for the PLL to put it back to 44.1. Its a great transport because it does this bit-perfect. Now if you send it 48khz or 96khz there is no downsampling. It is ideal for those formats. That's why I suggested a good upsampler for the computer so you avoid that extra step of jitterery PLL when playing RBCD 44.1khz. And yes the Hiface has 2 Xo's.

Haha...oh no it doesn't....oh yes it does....oh no it doesn't....we're gonna get into children's panto here.

I suggest you buy one, download the driver and try for yourself. and then compare 96/24 through both.

If you do know any way to modify this other than the xo and power, I hope you'll say.

 

[slim.a]

Quote:

Originally Posted by regal /img/forum/go_quote.gif Um, yes it does downsample if you send it a 44.1khz RBCD signal, it is recieved from the computer thru 48khz USB with the information for the PLL to put it back to 44.1. Its a great transport because it does this bit-perfect. Now if you send it 48khz or 96khz there is no downsampling. It is ideal for those formats. That's why I suggested a good upsampler for the computer so you avoid that extra step of jitterery PLL when playing RBCD 44.1khz. And yes the Hiface has 2 Xo's.

Quote:

Originally Posted by thoppa /img/forum/go_quote.gif Haha...oh no it doesn't....oh yes it does....oh no it doesn't....we're gonna get into children's panto here. I suggest you buy one, download the driver and try for yourself. and then compare 96/24 through both. If you do know any way to modify this other than the xo and power, I hope you'll say.

thoppa,

I think that regal didn't express himself clearly, so I am going to jump and try it another time.

The Teralink-X2 has only ONE clock. It is 12 mhz and so it is a perfect multiple of 48khz and 96khz.

When the Teralink X-2 receives 44.1 data, it has to derive/generate a 44.1 clock from the 12mhz clock. When doing so, it generates a lot of amount of jitter.

So let's suppose for one instant and that you got your hand on a rubidium clock that has a vanishingly low level of jitter. That accuracy of the clock would be pointless on the Teralink-X2 since the 44.1 clock (not data) has to be generated through a PLL. The jitter will be equal to that of the PLL. So whether you have a 0.0001 ps clock or a 1ps clock, most jitter will be generated by the PLL and not by the clock itself.
So the weakest link in the Teralink-X2 is probably the PLL and not the clock itself.

Also, keep in mind that since the Teralink-X2 is an adaptive usb model (and not async), its master clock is the one used in the computer and not the one inside the teralink. While putting a better clock might help. It is not the same as putting a higher quality clock on an async device where the devices use their internal clocks as master clocks.

So to sump up, Regal idea to upsample data to 48 or 96khz is an excellent idea (if the upsampler is transparent enough). It would get rid of the jittery generating of the 44.1 clock.
BTW, the Teralink X2 (and X) are not the only devices handicaped by using only one clock (instead of 2). If you read the Musiland thread, you will see that it was so poor at generating the 44.1 frequency (before drivers update) that some DACs couldn't lock at 44.1 (while they did at 48 and 96).

I hope this will clear up what regal said.

ps: I will be away from the computer for the next 1-2 days, so I won't be able to respond

 

[thoppa]

Thanks Slim. Do you know any way other than the power and xo ?

 

[slim.a]

Quote:

Originally Posted by thoppa /img/forum/go_quote.gif Thanks Slim. Do you know any way other than the power and xo ?

Personally, I would suggest to try with different usb cables, but that would launch another debate about jitter in usb cables ... So I won't

Beside trying with usb cables, I would rather suggest getting a hiface and modding it if you have the skills (I don't). A change of clocks in the hiface for example will provide 100% benefits since it isn't diluted by a PLL and by the noisy clocks of the computer (that serve as a master clock in the case of the teralink-x2).

ps: I probably won't be able to answer any new questions in the next day.

 

[thoppa]

Thanks - much appreciated. I'll continue with my efforts and let you all know if changing the xo does or doesn't make much difference. It's worth trying I feel as it's pretty easy to do.

I might consider the M2Tech (sadly it doesn't have i2s output though) but I have an SRC4392EVM to use as a hardware async and that has selectable xo's for input and output, so eventually I plan to use this to de-jitter and upsample.

Enjoy your trip !

 

[regal]

Quote:

Originally Posted by slim.a /img/forum/go_quote.gif thoppa, I think that regal didn't express himself clearly, so I am going to jump and try it another time. The Teralink-X2 has only ONE clock. It is 12 mhz and so it is a perfect multiple of 48khz and 96khz. When the Teralink X-2 receives 44.1 data, it has to derive/generate a 44.1 clock from the 12mhz clock. When doing so, it generates a lot of amount of jitter. So let's suppose for one instant and that you got your hand on a rubidium clock that has a vanishingly low level of jitter. That accuracy of the clock would be pointless on the Teralink-X2 since the 44.1 clock (not data) has to be generated through a PLL. The jitter will be equal to that of the PLL. So whether you have a 0.0001 ps clock or a 1ps clock, most jitter will be generated by the PLL and not by the clock itself. So the weakest link in the Teralink-X2 is probably the PLL and not the clock itself. Also, keep in mind that since the Teralink-X2 is an adaptive usb model (and not async), its master clock is the one used in the computer and not the one inside the teralink. While putting a better clock might help. It is not the same as putting a higher quality clock on an async device where the devices use their internal clocks as master clocks. So to sump up, Regal idea to upsample data to 48 or 96khz is an excellent idea (if the upsampler is transparent enough). It would get rid of the jittery generating of the 44.1 clock. BTW, the Teralink X2 (and X) are not the only devices handicaped by using only one clock (instead of 2). If you read the Musiland thread, you will see that it was so poor at generating the 44.1 frequency (before drivers update) that some DACs couldn't lock at 44.1 (while they did at 48 and 96). I hope this will clear up what regal said. ps: I will be away from the computer for the next 1-2 days, so I won't be able to respond

Slim you are a class act and understood fully what I was trying to communicate. I have the TerralinX1 and was frankly disappointed that they didn't go after the Hiface with the X2 design to be frank. The X1 is such a better build quality than the Hiface that it keeps up with it despite these limitations, if they had gone asynch with 2 clocks the X2 would have been a state of the art best computer transport. That said I haven't seen any X1 vs X2 comparisons so my wallet stays shut for the X2 for now.