Shunyata AES/EBU & S/PDIF Required Lenghts

[W9TR]

The 1.5 m length was specifically about USB cables since they are not impedance matched. The idea is to ensure that reflections from the transmitter to receiver and back again do not mix with the original signal and cause timing errors. Instead of being near the transitions, the reflections are moved to a part of the waveform where they don’t matter.

For everything else, it is impedance matched and so transmission line theory comes into play and the length does not matter at all.

That is if and only if the characteristic impedance of the transmitter and receiver are accurate. Given that most digital engineers are not RF engineers the implementation of these circuits is variable.

[SCAudiophile]

Originally Posted by W9TR

The 1.5 m length was specifically about USB cables since they are not impedance matched. The idea is to ensure that reflections from the transmitter to receiver and back again do not mix with the original signal and cause timing errors. Instead of being near the transitions, the reflections are moved to a part of the waveform where they don’t matter.

For everything else, it is impedance matched and so transmission line theory comes into play and the length does not matter at all.

That is if and only if the characteristic impedance of the transmitter and receiver are accurate. Given that most digital engineers are not RF engineers the implementation of these circuits is variable.

I've seen, many years ago, the same discussion regarding other digital cable types as well, SPDIF 75ohm, 50ohm, etc...I'll have to dig a bit with Google but I remember seeing the discussion back in 2005-2010 and after, long before USB became hugely popular.

[W9TR]

Sure, the same principles would apply to non-usb interfaces.

I doubt that most digital designers know how to execute a true 50, 75, or 110 ohm interface anyway. And manufacturers sure aren’t publishing any s- parameter data for their interfaces.

What is known about USB is the packet sizes and word lengths are standardized making an ‘ideal’ length cable possible.

With other formats the data rates are different so an ideal cable length that would keep reflections away from the critical transition zone would vary.

All this said I have identical WW Platinum Starlight usb cables - one is 1m the other 1.5m. Can’t tell the difference sonically.

[SCAudiophile]

Originally Posted by W9TR

Sure, the same principles would apply to non-usb interfaces.

I doubt that most digital designers know how to execute a true 50, 75, or 110 ohm interface anyway. And manufacturers sure aren’t publishing any s- parameter data for their interfaces.

What is known about USB is the packet sizes and word lengths are standardized making an ‘ideal’ length cable possible.

With other formats the data rates are different so an ideal cable length that would keep reflections away from the critical transition zone would vary.

All this said I have identical WW Platinum Starlight usb cables - one is 1m the other 1.5m. Can’t tell the difference sonically.

Good post!

[PHC1]

https://youtu.be/zXV513UC8Ms

[CGabriel]

There is no ideal length for a USB cable since it is designed to operate at multiple frequencies: 1.5 Mbit/s, 12 Mbit/s, 480 Mbit/s and 5 Gbit/s. The packet lengths are irrelevant to transmission line operation and more importantly to time domain reflections. It is the frequency of transmission that matters and the length of the cable and whether or not the cable has a characteristic impedance of 90 ohms or not.

USB cables have a MAXIMUM allowable length due to acknowledgement time requirements. This means that the maximum length at certain frequencies is 5 meters or 3 meters in the worst case. So effectively, the maximum length should be considered to be about 3 meters.

There are no minimum lengths stated in the standard or protocol. However, cables shorter than 1.5 meters are audibly inferior. And as the cable is shorter it gets progressive worse. So a .5 meter cable is significantly worse than a 1.5 meter. You don't need audiophile grade USB cables to verify this. Just go buy some Belkin cables of different lengths at the office supply house. And then compare them for yourselves. It will only cost you less than $70 bucks to learn this for yourself. Do your own testing.

[W9TR]

So what’s the mechanism that causes shorter cables to sound worse than 1.5 meter and longer cables?

[CGabriel]

As someone else mentioned: signal reflections primarily. No hardware interface has 'perfect' characteristic impedance and no cable has perfect impedance. So there is always a certain level of signal reflection. When a signal transitions from a low state to a high state, as it crosses the zero point it is highly susceptible to interference. A shorter cable causes a strong reflection back into the hardware driver circuit which skews the relative phase of the originating signal. In other words it increases temporal distortion (jitter).

[Puma Cat]

Originally Posted by CGabriel

As someone else mentioned: signal reflections primarily. No hardware interface has 'perfect' characteristic impedance and no cable has perfect impedance. So there is always a certain level of signal reflection. When a signal transitions from a low state to a high state, as it crosses the zero point it is highly susceptible to interference. A shorter cable causes a strong reflection back into the hardware driver circuit which skews the relative phase of the originating signal. In other words it increases temporal distortion (jitter).

[PHC1]

The 1.5m is indeed the universally accepted magic length, at least with current technology. . Or odd multiples of it. So 1.5, 4.5, 7.5m

It does in fact have to do with the imperfect world of the transmitter, cable and receiver impedance mismatches. If there was no mismatch, the signal would be perfect with no reflections back at the receiver which is sending the next signal.

The reflection can arrive in phase or out of phase. typically the reflections are not a big deal and are not large enough to screw up the zero point which will determine if it’s a 1 or a 0.

The reflection that screws up the “edge” of the pulses can and does supposedly affect the timing of the clock which is derived from the signal and that is jitter. Not great for the accuracy or the resulting music signal on the other end.

As it stands, imagine a pulsing laser that is transmitting precisely timed signals but the receiver is reflective and sends the laser back to the source. Now that reflection is either leading or lagging the edge and length of the next laser pulse but is essentially extending the “time” of the event.

A water hose spraying into another hose and that hose then squirts the water back into the the first house is the visual example for impedance mismatch.

This is not my area of expertise but I think I understand the basics here. Stick to 1.5m as it was determined to be the lesser of the inaccuracy evils.