TC Impact Twin as firewire to spdif interface

[Mintzar]

Both Delta Sigma and Ladder DACs have resistors. In the case of ladder DACs you have 16 resistors that all have a voltage applied in order to determine the value of each bit. In Delta sigma you have a voltage applied to a single resistor multiple times. Timing is an issue with delta sigma because any timing error in application of voltage creates distortion. Extrapolated: Applying the wrong voltage to the wrong bit.

 

But this still goes back to the voltage that's actually being applied. If proper voltage is applied the external master clock is less important because the problem isn't as much the timing, it's the cleanliness of the voltage being applied. Stock clocks have poor power, but their jitter is usually quite low (as I said, jitter is almost always WELL below the audible spectrum). The better clocks have a better power supply, which is why they tend to improve the sound in similar ways to power supply upgrades.

 

 

 

 

 

[Panelhead]

I have no interests in Core Audio. Just interested in whether a better supply will improve a TC Electronic Impact Twin.

Got some doubts after looking it over. But have not tried it. The parts are there in the Impact Twin (maybe the AK4620 is a weak link) but I do not see the AK4395 used in the MH 8 series being a SOTA chip either.

Never used the Impact Twin as a format converter, FireWire works just fine for now.

 

 

George

 

[esldude]

Upstream in this thread, Chris posted:

 

"Ryan (Mintzar), Posting about your products, including testimonials, is not allowed here."

 

Yeah Chris did say that. In the most recent post by Mintzar he was describing what upgraded power supplies sound like on the TCIT. Something he should know if he makes them. Something pertinent to include in his reply. I suppose he could have left out the contact me and I will give you testimonials.

 

For what it is worth, the TCIT I own was purchased from Mintzar who was fair and honest to deal with. I didn't in the end get one of his power supplies. It was discussed and sounds like a good idea. I just decided to upgrade an interface rather than polish the one I was using (TCIT). So I cannot speak to his products, but I don't think the guy is trying to put one over on anyone.

 

 

[louawalters]

I have lurked on CA for sometime, and figured it was time to give back. I have owned a TC Electronic IT for a couple of years. Replacing the stock switching supply with an after market linear makes a very noticable improvement. The TC accepts from 12 - 18 volt DC. I used the Nuforce Linear Power Supply (LPS-1), with a custom made cable (as the polarity on the TC is reversed from what the NuForce stock cable outputs). Improvement was heard even on the SPDIF out.

 

Big mistake I made - as I was moving cables around was that I unplugged (both firewire and power) without turning off the unit... So my TC Impact got fried...

 

I replaced it with a TC Studio Connect 48, which sounds just as good. Power Supply is built in, so replacing with a nice after market cord made a noticable difference.

 

[Miska]

In Delta sigma you have a voltage applied to a single resistor multiple times.

 

No, you apply voltage to a capacitor through a switch.

 

Extrapolated: Applying the wrong voltage to the wrong bit.

 

Contribution of the voltage is the other 50% of the overall result. Higher the frequency, lower the contribution of single bit.

 

For example ESS DAC runs at 40 MHz, so duration of a single bit is 25 nanoseconds.

 

Stock clocks have poor power, but their jitter is usually quite low (as I said, jitter is almost always WELL below the audible spectrum).

 

Jitter problem is not so much in the inherent jitter of the clock itself, even though the frequency properties of phase noise still matters. Part of the jitter is caused by noise leakage from PSU, part is caused by leakage of other signals to the clock lines.

 

If you want to reach 192 kHz 24-bit resolution, timing accuracy has to be 0.3 picoseconds. 192 kHz 16-bit requires 79 picoseconds accuracy.

 

 

[Panelhead]

I have heard of hot plugging the FW can kill the receiver. Never experienced this, but it is advertised as an issue. I have plugged and unplugged an KRK Ergo a lot with any damage.

This may be related to not having power on the FW cable. I willbe more careful about disconnecting the Impact Twin if it can be damaged this way.

I did learn about rolling back drivers. Was forced to use a much older Mac after normal one died. Thought it would be good to roll back the drivers to match the OS. System said current drivers were more recent than install. Updated anyway and the unit never worked again. KRK replaced it free of charge. Very seldom that someone fixes my screwups for free.

I hear of people rolling back drivers all the time. First time I tried this it was catastrophic. Thought it was a 350.00 lesson.

 

[Panelhead]

Generic clocking can sound fine at 16/44.1. I have used Audiocom, Kwak, and Tent clocks as up grades.The Audiocom was very close to stock, two versions of Kwak were unreliable, and the Tent modules seemed an improvement at first. Extended listening to another showed I could not tell the stock clock from the Tent. Had to look inside.

But this was redbook. The Audiocom and Tent may have been a big improvement with 24/96 or 24/192. What level of timing accuracy is required for 32/384?

 

George

 

[louawalters]

Basically I disconnected both the power connector and FireWire without powering off the TC. Then I connected everything back up - with the power switch not turned off. I believe the inrush of DC current and FireWire power smoked my IT. And yeah the instructions do say not to do this. This is what happens when you are in a hurry and don't pay attention.

 

[Mintzar]

At 44.1KHz a D/A converter reads a new digital word every 22.6 microseconds (.0000226 seconds).

 

If you 8X over sample, the same converter reads a new digital word every 2.8 microseconds (.0000028 seconds).

 

If you are up sampling to 192KHz and then over sampling 16X (very common), the converter reads a new digital word every 300 nanoseconds (.0000003 seconds).

 

In addition to the faster clock speeds required for up and over sampling, a 32-bit D/A converter would have to read twice the number of bits as a 16-bit D/A converter within the same period of time. So you're talking anywhere from 70-450x the clock precision required in a Delta Sigma DAC compared to Multi-bit ladders.

 

Multi-bit DACs have 16 latches and 16 data paths, one per bit. Delta sigma type DACs only have a single latch and a single data path. In Delta Sigma DACs there are all kinds of filters and algorithmic estimations that occur because the technology doesn't really exist to process data that quickly.

 

There was a white paper written by Vishay and TI about high-resolution digital playback. The part datasheets SAY 24 or 32bit, but there actually isn't enough tolerance in the parts to achieve that resolution in real world scenarios. Even with .0005% tolerance resistors from Vishay I only was able to achieve 20bits accurately -- and that's in a discrete DAC. One of the reasons why a good 16/44.1 NOS DAC sounds arguably better is because it is a true 16 bit reproduction. 99% of Delta sigma DACs don't achieve more than 12-14bit resolution in real world scenarios.

 

It's one of the reasons I always preach that high rez is impossible, it's also why a low-rez low-distortion file will always sound more organic than a high-rez file. In a high-distortion system the high rez file sounds better because the pieces of the pie that are distorted are proportionately smaller than the pieces of the pie in the low-rez. But you have to have really low distortion to begin to hear this.

 

 

 

[Panelhead]

This is something I did not know. I will not hot plug the TC.

 

 

George

 

[ar-t]

Wow.

 

I'll let the readers read my mind, and try to figure out that cryptic message.

 

[Miska]

What level of timing accuracy is required for 32/384?

 

That would be already 0.6 femtoseconds. Quite far from 346 picoseconds needed for 44.1/16.

 

 

[Miska]

Delta sigma type DACs only have a single latch and a single data path. In Delta Sigma DACs there are all kinds of filters and algorithmic estimations that occur because the technology doesn't really exist to process data that quickly.

 

Well, it's not a problem to process data that quickly. "Quick" is generally not a problem in audio at all. Problem with multi-bit DACs is that even a tiny inaccuracy (such as temperature variation) in most significant bits ends up in huge non-linearity and can overwhelm least significant bits. One-bit delta-sigma (DSD) is very linear because there are only two level and all bits are equally significant. Higher the delta-sigma frequency, finer levels you can present.

 

It is nice because mathematics much easier to make accurate than physical world. So physical world side is kept as simple is possible and accuracy and complexity is handled mathematically.

 

ESS Sabre for example is multi-bit SDM running at 40 MHz and six bits. This allows it to use fairly simple low-order modulators.

 

Even with .0005% tolerance resistors from Vishay I only was able to achieve 20bits accurately

 

That's not enough for 20-bits. It would need 0.000095%. Multibit DAC-chips use matched laser-trimmed resistors on the silicon die.

 

One of the reasons why a good 16/44.1 NOS DAC sounds arguably better is because it is a true 16 bit reproduction.

 

...with huge distortion, or missing 10-20 kHz band...

Reconstruction filter for that is just not practical or attempt to create such analog beast would have horrible phase response.

 

99% of Delta sigma DACs don't achieve more than 12-14bit resolution in real world scenarios.

 

That's not true at all, all top class chips can achieve 20-bits worth, with possibility of more with couple of tricks. Only the very cheapest ones have roughly 12-14-bit, but that's usually due to shortcomings in their digital filters rather than the DAC stage itself. In many cases it can be overcome by running the chip at maximum rate with external upsampling.

 

 

[hionhifi]

The TCIT looks interesting. How many digital inputs does it accept? I see from the back panel, toslink, firewire, spdif (coax).

 

Problem is I have three digital sources, all w/toslink or coax.

 

Question is can I use a converter to go from spdif to firewire?

 

What I'm trying to do is get the digital signal (toslink or coax) from either one of my sources Squeezebox Touch, Playstation 3, and Directv into the TCIT.

 

Any suggestions?

 

My current DAC only has 2 inputs which leaves me one short.

 

 

 

[esldude]

You are right in that it takes Toslink and Coax digital inputs. So it only has two effectively.

 

One option would be to put your DirectTV into the analog line input. Even as an analog pre-amp the Imptact Twin is pretty high quality. All you need are some TRS to rca phono adapters. Available at Radio Shack for a few bucks.

 

[hionhifi]

@esldude,

Another forum member made an excellent suggestion which is direction I'm going.

 

To deal with my input shortage he suggested I use a 4X1 HDMI® Switcher w/ Toslink & Digital Coaxial Port (Rev.2) w/ 3D support.

 

http://www.monoprice.com/products/product.asp?c_id=101&cp_id=10110&cs_id=1011002&p_id=5557&seq=1&format=2

 

It has toslink and coax outputs, has remote control input switching and other benefits making it a perfect choice. Thanks for the suggestions.

 

 

[jaynyc]

Resurfacing an old thread here... would an upgraded or 3rd party power supply improve the sound quality of a TC Impact Twin or TC Konnekt Live *as a SPDIF transport* feeding an external DAC? Or would an upgraded power supply only aide the analog line outs?

[pj]

Resurfacing an old thread here... would an upgraded or 3rd party power supply improve the sound quality of a TC Impact Twin or TC Konnekt Live *as a SPDIF transport* feeding an external DAC? Or would an upgraded power supply only aide the analog line outs?

 

Yes, it does improve the SPDIF output. I've done some testing on a Konnekt 8 with a Welborne Labs linear supply and there are measurable improvements over the same interface powered from Firewire. The Konnekt 8 is pretty clean with firewire powering but the LPSU does improve it further.

 

This shows the spectrum analysis of a 16/44 JTEST.wav sent Mac -> Konnekt8 -> SPDIF -> D1V3 (Pass Labs D1 clone) -> Konnekt8 mic inputs.

 

 

And this is same setup with Konnekt8 powered by LPSU...

 

 

The LPSU is probably helping somewhat with the input/ADC on the Konnekt8, but you can see the effect on the spread at the base of the main tone. The height and width of the spread is related to low frequency jitter, so the LPSU is clearly improving this aspect of the SPDIF output. On the other hand there is virtually not difference in the jitter side spurs so the Konnekt8 isn't doing much to attenuate jitter on the SPDIF output.

 

A Stello U3 is clearly better in terms of jitter attenuation, but not as effective as one might be inclined to believe after reading some of the audiophile press reviews. You can also see the Stello U3 has a significant amount of spread in the main tone.

 

 

If your DAC has decent jitter attenuation you might be better off with the TC Konnekt 8/Live/Impact Twin + LSPU than a Stello U3.

 

As an example, I've been experimenting with a home brew secondary PLL in my DAC and in conjunction with the Konnekt 8 its cleaner than the same DAC setup + Stello U3.

 

Stello U3 + D1V3 +PLL/VCXO Jitter reduction

 

Konnekt8 (FW powered) + D1V3 +PLL/VCXO Jitter reduction

 

cheers

Paul

[Panelhead]

The measured reduction in jitter using a LPS is impressive. Have you tested a wallwart with the TC? The USB, FW, or Thunderbolt bus power should be as clean as a wallwart.

I may have to open up my Impact Twin and do some more work. One easy PS I use is either a big wallwart or cheap regulated power supply. Then insert a large value 4 pole Jensen cap to remove the hash. Have a few of the 47,000 ufd model. There is enough inductance to remove a lot of the grain from the power supplies. With some decoupling close the the internal regs and active devices the sound smoothes and gets much warmer. Usually decouple with a 0.0047 ufd Seimens stacked film and a sub ohm resistor to "snub" ringing. The low value resistor, 0.27 ohm or so makes a huge difference.

[ginetto61]

... One of the reasons why a good 16/44.1 NOS DAC sounds arguably better is because it is a true 16 bit reproduction. 99% of Delta sigma DACs don't achieve more than 12-14bit resolution in real world scenarios.

It's one of the reasons I always preach that high rez is impossible, it's also why a low-rez low-distortion file will always sound more organic than a high-rez file. In a high-distortion system the high rez file sounds better because the pieces of the pie that are distorted are proportionately smaller than the pieces of the pie in the low-rez. But you have to have really low distortion to begin to hear this.

 

Hi ! sorry to resurrect and old thread but I have seen it only now.

I find the quoted statements extremely interesting and even shocking if I think that many high end converters have delta-sigma dacs inside.

I wonder if there is an agreement about this.

Thanks a lot, gino

[esldude]

Hi ! sorry to resurrect and old thread but I have seen it only now.

I find the quoted statements extremely interesting and even shocking if I think that many high end converters have delta-sigma dacs inside.

I wonder if there is an agreement about this.

Thanks a lot, gino

 

Yes, it is true the overwhelming majority of DACs high end and otherwise have sigma-delta DAC chips in them. Everything else in the quote is baloney. These DACs achieve 18-20 bit resolution, and that is because thermal noise limits you to about that much. There are some few that do a little better than that.

[ginetto61]

Yes, it is true the overwhelming majority of DACs high end and otherwise have sigma-delta DAC chips in them. Everything else in the quote is baloney. These DACs achieve 18-20 bit resolution, and that is because thermal noise limits you to about that much. There are some few that do a little better than that.

 

Hi ! thanks a lot for the very helpful reply.

I was worried because I have only sigma-delta dacs ...

By the way I see a return of multibit dacs.

Maybe there are just good and bad examples of both types.

Thanks again, gino