Signal integrity & it's possible influence on digital audio

[sbgk]

magnets were for educational purposes only

does the theory account for the lack of digital noise when silence is played ie all the instructions are running, only 0s are being copied ? the digital noise is only heard when there is a payload and seems to be variable with the frequency. Does a constant tone have noise issues or is it where frequencies change ?

I know the magnets were a test for the doubters to try but I doubt any of them will :) I've never tried it, btw - must give it a test.

The concept of signal integrity applies to all signals including digital silence - the better the signal quality, the less work being done at the receiver. But I can surmise that if the digital audio source is a dynamic system (PC) then the quality of the signal (the waveforms) can be varying - more varying under load than when not under load. This varying waveform signal will then be transmitted to the receiver & cause another dynamic load adjustment to deal with the dynamic waveform being encountered. The complex relationship that is communicated by this waveform disturbance is pretty difficult to predict unless some real-time waveform capturing at the DAC could be achieved with monitoring between the two systems?

I guess the magnets are having the same effect as when pc internals are overly damped by emf absorbing material or certain instructions/settings in the code are used. I had the same effect when applied to a lan cable feeding a squeezebox, which was isolated by transformers on the receiving end. Ferrite chokes also affect things. What does that tell you ?

[jkeny]

magnets were for educational purposes only

does the theory account for the lack of digital noise when silence is played ie all the instructions are running, only 0s are being copied ? the digital noise is only heard when there is a payload and seems to be variable with the frequency. Does a constant tone have noise issues or is it where frequencies change ?

I know the magnets were a test for the doubters to try but I doubt any of them will :) I've never tried it, btw - must give it a test.

The concept of signal integrity applies to all signals including digital silence - the better the signal quality, the less work being done at the receiver. But I can surmise that if the digital audio source is a dynamic system (PC) then the quality of the signal (the waveforms) can be varying - more varying under load than when not under load. This varying waveform signal will then be transmitted to the receiver & cause another dynamic load adjustment to deal with the dynamic waveform being encountered. The complex relationship that is communicated by this waveform disturbance is pretty difficult to predict unless some real-time waveform capturing at the DAC could be achieved with monitoring between the two systems?

I guess the magnets are having the same effect as when pc internals are overly damped by emf absorbing material or certain instructions/settings in the code are used. I had the same effect when applied to a lan cable feeding a squeezebox, which was isolated by transformers on the receiving end. Ferrite chokes also affect things. What does that tell you ?

Yea, I had noticed the very audibly detrimental effect (some or all?) ferrites had on USB cables. I always thought that it interfered with the waveform by squashing the higher frequencies needed for well-formed square waves. I believe the rule of thumb is along the lines that the 5th harmonic is needed for a well formed squarewave. USB high speed is 480Mbit/sec - does this equate to 480MHz? If so 5 times 480MHz is 2.4GHz - is this right? We would then need a cable bandwidth of > 2.4GHz to ensure well formed USB square waves. I seem to remember Light Harmonic settled on 2GHz bandwidth cable

[DaveF]

Also variations in the square wave waveform structure, rise, fall times may well be of significance

Never seen the waveform structure vary at all regardless of what load was on the system and I've scope out all sorts of boards over the years. The PCB designers generally do extensive testing on the signals or if not they really should. If rise/fall times/skew dont seem to be correct then component changes usually sort that out. But variances, never seen it. The exception to that is jitter.

Dave, did you look at the two files I linked to at the beginning of the thread showing eye diagrams for the same signal transported through different USB cables? For those who don't know what an eye diagram is - it's simply an oscilloscope shot of a USB square waveform overlaid with the next shot, overlaid with the next shot, etc up to an agreed number of shots. The resulting image looks like an eye & the ideal eye diagram would be a single pixel-depth line that traces the image. This is not possible as some fluctuation in the waveform always happen giving a bit of a fuzzy line. The greater the fuzziness, the more fluctuations there are & the worse the signal integrity is. USB standard will allow a certain degree of fuzziness. This works fine for digital but is it fine for digital audio (for the reasons stated in previous post)?

Dave, I know USB is a far lower speed than you are handling & you are probably not testing cable transmission but rather pcb transmission line traces so your experience may be different but what do you think of those USB eye diagrams & the differences between them?

I've seen plenty of eye diagrams far worse than that and yes from the end of cable as well on a PCB trace. As you said the USB standard and other standards like HDMI allow for a certain eye closure yet still being able to receive the data perfectly. I've never seen the usual noise on a digital cable affect anything on an analog stage in an RX board and we've got some pretty complex ones here with very very sensitive electronics and optics. (one exception to that which turned out to be a very thin USB cable with a very poor power line)
Good proper PCB layout and partitioning is far more important in maintaining low noise performance in the final analog stage. If I were designing a new DAC board this is where more of my effort would be spent.

[jkeny]

I've seen plenty of eye diagrams far worse than that and yes from the end of cable as well on a PCB trace. As you said the USB standard and other standards like HDMI allow for a certain eye closure yet still being able to receive the data perfectly. I've never seen the usual noise on a digital cable affect anything on an analog stage in an RX board and we've got some pretty complex ones here with very very sensitive electronics and optics. (one exception to that which turned out to be a very thin USB cable with a very poor power line)
Good proper PCB layout and partitioning is far more important in maintaining low noise performance in the final analog stage. If I were designing a new DAC board this is where more of my effort would be spent.

Good points, Dave, completely agree about spending a major portion of effort in trying to ameliorate the influence of external noise issues on the analogue side but the shape of the waveform (signal integrity) isn't just about noise, I don't believe?

The point of Swenson's treatise was about the self-generated noise emanating from the digital receiver chips (because of differences in the signal quality being received - all with spec, though) & this noise having an influence downstream. The way I read it is that there is not just one point of origin for this self-generated noise - it is a consideration for every chip that receives a high speed digital signal, no? So let's say that we have controlled all the noise that is emanating from the chips on the board, ultimately, the signal then hits the D/A chip (let's consider it is a standalone chip), the quality of the signal arriving here could well cause varying power draw which leads to varying ground currents. This ground is a shared ground between digital & analogue sections of the DAC chip so we can't do much about separation here (at least with this configuration)

[DaveF]

...... the shape of the waveform (signal integrity) isn't just about noise, I don't believe?

Generally down to the driver of the signal be it an FPGA, buffer, opamp etc and the bandwidth of the components in the signal path. Dont forget that an open or high bandwidth circuit will let more noise through. Restricting that to whats just required you can shut out noise above a certain cut off point.
Piss poor or non optimimal layout can give crosstalk between signals too.

The point of Swenson's treatise was about the self-generated noise emanating from the digital receiver chips (because of differences in the signal quality being received - all with spec, though) & this noise having an influence downstream. The way I read it is that there is not just one point of origin for this self-generated noise - it is a consideration for every chip that receives a high speed digital signal, no? So let's say that we have controlled all the noise that is emanating from the chips on the board, ultimately, the signal then hits the D/A chip (let's consider it is a standalone chip), the quality of the signal arriving here could well cause varying power draw which leads to varying ground currents. This ground is a shared ground between digital & analogue sections of the DAC chip so we can't do much about separation here (at least with this configuration)

Yes I've seen ground bounce (caused by many fast switching signals at once, usually going from high to low, i.e sinking current) screw up digital signals before but again that was down to a poor layout and not enough layers on the board, ie the stackup as its called. For example separating signal planes with ground planes and so on. Of course there are plenty of not so good PCB designers out there so its well possible that some DAC boards are less than optimal.

In my opinion you'd really have to examine each DAC board in detail to see how its noise performance is.

[jkeny]

I just came across this & haven't read it all yet but it seems very relevant to the thread topic & I wanted to post it here as much as a placemark as anything. Feel free to comment though
http://www.sabritec.com/technotes/PDF/H ... torial.pdf

[Sligolad]

Another converted "bits are bits" naysayer on CA.
Linear PS all the way...........

http://www.computeraudiophile.com/f8-ge ... post360364

[jkeny]

Yea, interesting, isn't it?

For those wishing to delve into an understanding of what is needed for real-world DAC chip implementations & the major contributors to errors and distortion in modern switched current DACs - should read this http://wiki.analog.com/university/cours ... dent_noise

In it you will find excerpts " due to the mixed-signal nature of a DAC, digital data activity on the die will cause interference in the analog and clock sections of the device." The premise of this thread is that even with a digital signal, the signal waveform itself can have be an influencing factor on the final analogue output i.e sound quality.