[drez]

Quote:

 
I think that would hold true regardless of whether or not he hears a difference when using different cables.
 

 
But if I denounced audibility of jitter and importance of cabling I would no longer be crazy and therefore be better able to maintain my own hygeine

 

[liamstrain]

It is a wave form, but it is also 1s and 0s. There is a lot of variation in that wave that registers as a 1 or 0. Quite shocking amounts of flux, in fact. The spec itself accounts for a good bit of gap:
 
 
Quote:

Transmitted signal levels are 0.0 to 0.3 volts for low (0) and 2.8 to 3.6 volts for high (1) in full-bandwidth and low-bandwidth modes, and −10 to 10 mV for low (0) and 360 to 440 mV for high (1) in hi-bandwidth mode.

 
And from what I understand, most devices have even more slop built in. Which means you can have noise in the wave form, affecting the voltages at least up to 10% and still be within spec.
 
Clocking issues aside (and there is slop there too - Clock tolerance is 480.00 Mbit/s ±500 ppm, 12.000 Mbit/s ±2500 ppm, 1.50 Mbit/s ±15000 ppm) - since there is little agreement on whether even large amounts of jitter is audible - this allows for a properly encoded usb signal under even the ****tiest of conditions. And very little to suggest the cable makes much difference in this - especially a cable built to spec.

 

[proton007]

Quote:

 
As far as I know it is not that simple.  It depends how effective the digital input section is at eliminating jitter [and whether upsampling is used] - some good DAC's use asynchronous USB with galvanic isolation (controlled by it's own clocks), or otherwise SPDIF input with galvanic isolation or a ferrite choke to stop any noise being transmitted further downstream.  After this a DAC will typically have an SPDIF receiver such as WM8805 which uses an oscillator but to my knowledge is an adaptive process and does not reclock the signal, and this chip typically outputs I2S signal.  From here many DAC's will upsample the signal with various digital filters to shift the jitter distortion out of the audible spectrum (typically controlled by a separate clock.)  Some DAC's will output I2S directly from the USB input rather than outputting SPDIF to an SPDIF receiver chip which is more direct but skips a possible stage of jitter reduction.

 
Most chips have onboard USB support, with an on chip clock generator that is used for both the USB and Sampling clocks. 
They use a FIFO to buffer the data received, and then it is clocked by the DAC. An I2S is not required because the data has already been transferred by the USB.
Usually, an SPDIF/I2S out is implemented in case the data needs to be transferred further. 
Hence, if the data packet has been received, jitter on the USB should not affect the DAC sampling.

 

[drez]

Quote:

 
Most chips have onboard USB support, with an on chip clock generator that is used for both the USB and Sampling clocks. 
They use a FIFO to buffer the data received, and then it is clocked by the DAC. An I2S is not required because the data has already been transferred by the USB.
Usually, an SPDIF/I2S out is implemented in case the data needs to be transferred further. 
Hence, if the data packet has been received, jitter on the USB should not affect the DAC sampling.

 
I am not an electronics professional but to my understanding I have not heard of any DAC chips or designs that send USB signal directly to the DAC chip.  My limited understanding is that I2S is the usual format used to carry the PCM data from digital input receivers to the DAC chip.  I was under the (possible false) impression that a separate oscillator is used for sampling in most cases but I am not an expert here.
 
Either way most decent DAC's use more than one buffer/filter to reduce jitter, and some DAC's use more/superior jitter reduction techniques than others, therefore are less likely to be affected by jitter.  Due to my lack of expertise I am also unsure about the relationship between jitter audibility and the DAC conversion process eg is this jitter being fed directly to the DAC without any filtering or not.
 
Another possibility is that it is not jitter that is causing the problem after all and rather than some divices and cables transmit or pick up more noise which may or may not find its way to the DAC chip, eg whether the difference in performance or lack thereof depends entirely on whether a DAC or SPDIF interface uses galvanic isolation...

 

[proton007]

Quote:

 
I am not an electronics professional but to my understanding I have not heard of any DAC chips or designs that send USB signal directly to the DAC chip.  My limited understanding is that I2S is the usual format used to carry the PCM data from digital input receivers to the DAC chip.  I was under the (possible false) impression that a separate oscillator is used for sampling in most cases but I am not an expert here.
 

USB DAC is a pretty common integrated solution nowadays with either an On-Chip USB Controller. TI has a lot of those.
Another common way to do this is to use a USB (Audio Class) Controller chip -> I2S out -> DAC, if you want to use separate chips. The I2S is very low jitter as the bus carries a separate clock signal. The DAC will use this data for over-sampling, with its own clock as you said.
 
Quote:

Either way most decent DAC's use more than one buffer/filter to reduce jitter, and some DAC's use more/superior jitter reduction techniques than others, therefore are less likely to be affected by jitter.  Due to my lack of expertise I am also unsure about the relationship between jitter audibility and the DAC conversion process eg is this jitter being fed directly to the DAC without any filtering or not.

Usually the jitter will result after the filtering, and into the DAC stage.
 
Quote:

 
Another possibility is that it is not jitter that is causing the problem after all and rather than some divices and cables transmit or pick up more noise which may or may not find its way to the DAC chip, eg whether the difference in performance or lack thereof depends entirely on whether a DAC or SPDIF interface uses galvanic isolation...

I think it should depend on how effective the USB/SPDIF receiver is at rejecting a noisy signal. Unless the cable noise is so bad that the data gets corrupted, the issue is more about whether your device receives all the packets the host sends. It will not matter if you use an expensive usb cable as long as its USB certified, because chances are that packets might be dropped still by the device, and the host won't know.
So I guess the jitter is a characteristic of the host and the DAC design, rather than which cable is used as long as the cable is reliable.

 

[drez]

^Thanks for explaining that

 
EDIT: Do you think it might be possible that when we are talking about DAC designs for example USB DAC chip with on-chip USB receiver that the level of noise either picked up or transmitted by the cable could find it's way past the filtering stage if there is no galvanic isolation.  In the case of digital transports such as the one I am using noise still doesn't go much distance toward suggesting a possible mechanism for the influence of USB cables on a DAC.
 
If I am to understand Liamstrain posted in theory noise should not really affect the performance of an asynchronous USB receiver as there are large enough tolerances built in that there should not be any packet drops?  I guess though one would need to take into consideration the range of tolerances from the USB standard for the cable, to the USB standard for the USB controller, the level of noise on motherboard power supplies for the USB ports (or just measure the USB signal with an oscilloscope.)   "0.0 to 0.3 volts for low (0) and 2.8 to 3.6 volts for high (1) in full-bandwidth" does seem like quite a large tolerance and that if devices can tolerate up to 10% of these numbers in noise it would certainly seem pretty hard for the USB input to exceed these tolerances.  I am probably at risk of laboring over this point excessively but still I am curious.

 

[proton007]

No Problem.
Quote:

^Thanks for explaining that

 
EDIT: Do you think it might be possible that when we are talking about DAC designs for example USB DAC chip with on-chip USB receiver that the level of noise either picked up or transmitted by the cable could find it's way past the filtering stage if there is no galvanic isolation.  In the case of digital transports such as the one I am using noise still doesn't go much distance toward suggesting a possible mechanism for the influence of USB cables on a DAC.
 

Past the filtering stage is unlikely. Noise in digital signal transmission usually does not affect its interpretation, unless it is causing swings in the voltage level. Hence, any data, if corrupted by such noise, will be used as such by the receiver if not checked for error.
 
Quote:

If I am to understand Liamstrain posted in theory noise should not really affect the performance of an asynchronous USB receiver as there are large enough tolerances built in that there should not be any packet drops?  I guess though one would need to take into consideration the range of tolerances from the USB standard for the cable, to the USB standard for the USB controller, the level of noise on motherboard power supplies for the USB ports (or just measure the USB signal with an oscilloscope.)   "0.0 to 0.3 volts for low (0) and 2.8 to 3.6 volts for high (1) in full-bandwidth" does seem like quite a large tolerance and that if devices can tolerate up to 10% of these numbers in noise it would certainly seem pretty hard for the USB input to exceed these tolerances.  I am probably at risk of laboring over this point excessively but still I am curious.

Basically, the concept is this. All USB transfer types (Control, Bulk, Interrupt) except Isochronous, use a Token Packet -> Data Packet -> Handshake Packet during transfers.
Token packets transmit the state of the host/device (ready/not ready), Data packets carry data, and Handshake packets transmit a final response from host/device (acknowledge/negative acknowledge/stall) to end a transaction.
Isochronous transfers do not have the handshake packets, which means, the device will not acknowledge the receipt of data packets. Since timely transfer of data is more important for audio/video streaming, the host (your PC) may choose to drop certain packets if its not able to meet the bandwidth requirement (maybe busy doing other stuff, or the bus is too busy with other data).
However, this does not mean there's no possibility of error checking. If a CRC is transmitted along with the packet, the device can perform an error check on its own. Whether that is of consequence is another issue. Since the device won't re-request the frame from the host, it can just drop the packet.
So as you said, the tolerance for voltage is pretty high, so usually, the question is whether all packets are transferred reliably by the host, received by the device and not dropped. Windows can use ASIO for example, to minimize latency and give direct access to the sound card, bypassing the kernel mixer.

 

[drez]

I guess I'm stumped then - I cant really find a theoretical justification for the effect of USB cables on a DAC.

  I still admit that I have a preference for my better USB cable, but I will have to do some DBT before I can say with certainty that what I am hearing is a legitimate observation, but you can take my word that I am highly convinced by my preference such that even when faced with this overwhelming theoretical disproof I still can't help but trust what I seem to be hearing.  I am in a state of cognitive dissonance which at this stage I cannot resolve, but hopefully one of these days I will be able to resolve this either by conducting a DBT which produces positive results or otherwise.  This may be challenging though as the differences I have observed are that the generic cable has more grain and fatigue in the midrange, luckily one of them seems to hurt my ears while the other doesn't (I'm not kidding lol) so hopefully when if I get around to doing a DBT this observation and ear pain does not dissappear

 

[proton007]

Quote:

I guess I'm stumped then - I cant really find a theoretical justification for the effect of USB cables on a DAC.

  I still admit that I have a preference for my better USB cable, but I will have to do some DBT before I can say with certainty that what I am hearing is a legitimate observation, but you can take my word that I am highly convinced by my preference such that even when faced with this overwhelming theoretical disproof I still can't help but trust what I seem to be hearing.  I am in a state of cognitive dissonance which at this stage I cannot resolve, but hopefully one of these days I will be able to resolve this either by conducting a DBT which produces positive results or otherwise.  This may be challenging though as the differences I have observed are that the generic cable has more grain and fatigue in the midrange, luckily one of them seems to hurt my ears while the other doesn't (I'm not kidding lol) so hopefully when if I get around to doing a DBT this observation and ear pain does not dissappear

 
I think a DBT would be insightful. Maybe you can also chime in on the other thread about headphone cables. 

 

[LizardKing1]

Quote:

Basically, the concept is this. All USB transfer types (Control, Bulk, Interrupt) except Isochronous, use a Token Packet -> Data Packet -> Handshake Packet during transfers.
Token packets transmit the state of the host/device (ready/not ready), Data packets carry data, and Handshake packets transmit a final response from host/device (acknowledge/negative acknowledge/stall) to end a transaction.
Isochronous transfers do not have the handshake packets, which means, the device will not acknowledge the receipt of data packets. Since timely transfer of data is more important for audio/video streaming, the host (your PC) may choose to drop certain packets if its not able to meet the bandwidth requirement (maybe busy doing other stuff, or the bus is too busy with other data).
However, this does not mean there's no possibility of error checking. If a CRC is transmitted along with the packet, the device can perform an error check on its own. Whether that is of consequence is another issue. Since the device won't re-request the frame from the host, it can just drop the packet.
So as you said, the tolerance for voltage is pretty high, so usually, the question is whether all packets are transferred reliably by the host, received by the device and not dropped. Windows can use ASIO for example, to minimize latency and give direct access to the sound card, bypassing the kernel mixer.

 
WASAPI should have the same effect as ASIO, right? Since it also bypasses Windows mixing.
If I understood correctly, in audio  USB transfer it uses an Isochronous mode, which has no way to correct the error (even if it's detected as an error) since it won't request the packet again. So there's a chance we get pops and artifacts in songs played from a USB DAC due to the cable? 
Also what's this I hear about a CRC-able DAC?

 

[nicholars]

Quote:

I guess I'm stumped then - I cant really find a theoretical justification for the effect of USB cables on a DAC.

  I still admit that I have a preference for my better USB cable, but I will have to do some DBT before I can say with certainty that what I am hearing is a legitimate observation, but you can take my word that I am highly convinced by my preference such that even when faced with this overwhelming theoretical disproof I still can't help but trust what I seem to be hearing.  I am in a state of cognitive dissonance which at this stage I cannot resolve, but hopefully one of these days I will be able to resolve this either by conducting a DBT which produces positive results or otherwise.  This may be challenging though as the differences I have observed are that the generic cable has more grain and fatigue in the midrange, luckily one of them seems to hurt my ears while the other doesn't (I'm not kidding lol) so hopefully when if I get around to doing a DBT this observation and ear pain does not dissappear

 
LOL so after about 3 pages of arguing and calling me and other people stupid cynics etc. You have now basically admitted that you are probably wrong and my explanation is actually a lot more likely. Maybe you should do some more research before calling people names and making yourself look silly.... Basically my comments were aimed at steering people away from spending extortionate amounts of money on scam USB cables. I was giving a more logical arguement then your subjective, psuedo science and random opinion based claims.

 

[sridhar3]

Quote:

 
LOL so after about 3 pages of arguing and calling me and other people stupid cynics etc. You have now basically admitted that you are probably wrong and my explanation is actually a lot more likely. Maybe you should do some more research before calling people names and making yourself look silly.... Basically my comments were aimed at steering people away from spending extortionate amounts of money on scam USB cables. I was giving a more logical arguement then your subjective, psuedo science and random opinion based claims.

 
You completely missed the point AGAIN.  And you're talking about Drez making himself look silly... Sweet irony.

 

[Lorspeaker]

The usbcable i bought from screwtech sounded cleaner deeper clearer than the stock usb...from the first min..err ...secs.
Clear cut better. Be it on the centrance or hdp... there is no placebo or gazebo....my dog agreed wagging his tail.
 

 

 

[sridhar3]

Quote:

screwtech
 

 
I don't know where the hell you're buying your cables from, but it's a far more interesting place than where I'm buying my cables from.

 

[nicholars]

Quote:

 
You completely missed the point AGAIN.  And you're talking about Drez making himself look silly... Sweet irony.

How did I miss the point? I made a simple comment trying to stop people wasting money because those cable companies annoy me... He starts defending them based on psuedo science and random opinions... How did I miss the point?