Do you believe this Schiit? I don't.

[Empirical Audio]

Longer cable lengths typically help with reflections as they will introduce higher propagation losses - which improves the return loss. The timing of the reflections as to being on an edge or not is a periodic function that can happen at short or long lengths - it's just that the delay of the reflection has to be within a certain multiple of the symbol time of the waveform.

Cheers, Joe

Return loss difference between a 0.5m and 1.5m cable is insignificant. It's the timing of the second reflection that matters.

Steve N.
Empirical Audio

 

[morricab]

The length of cable will change the time pattern of the reflections, but also their amplitude in cables having losses. Matching impedance in nuclear electronics is a trivial matter - sometimes we needed to measure times in the range of the picosecond, reflections are not allowed!

Have you designed a mass spectrometer with a dual polarity floating source? I doubt many have so take care in what you describe as "trivial". Matching the impedance of a flat plate and a scope input through a floating power supply is anything but trivial...and took months of work to optimize...in the end elimination of all ringing was impossible but was minimized.

Assuming the cables are constructed identically, the losses in 0.5 and 1.5meter will be trivial and the impedance should be essentially the same. There will always be some reflection, which is why the length probably matters...the question is what is it about that particular length that is special? Or is it only approximate? Maybe 1.45 m would be even better or 1.6? Has anyone experimented outside standard available lengths?

 

[morricab]

Return loss difference between a 0.5m and 1.5m cable is insignificant. It's the timing of the second reflection that matters.

Steve N.
Empirical Audio

Have you looked at this timing on a scope with discrete signal inputs to clearly see the ringing? With a continuous signal I guess this could be extracted with an FFT but it would be more interesting to see it discretely in time. Perhaps it is not possible to create a short enough pulse with such a short cable to see the effects directly. We were using several meters of cable and the pulses were ion packets that were a few ns wide.

 

[morricab]

Here's a very important point in Steve's article, at the very end, which people may gloss over:



As the article says, and as morricab pointed out, impedance matching all along the way is really the key here.

This doesn't quite make sense to me...the cable itself is often the problem with impedance matching unless it is a true 75 ohms. Therefore, the cable and it's length will always both matter.

 

[morricab]

There will ALWAYS be a reflection. No cable or connector is perfect. The question is how much of a reflection and does it occur when it does damage to the destination edge that is audible?

The reflection is less important if it occurs at a time at the destination when it's a "don't care". This is why the longer cable is advantageous.

Steve N.
Empirical Audio

In that case, why not go with 5 meters or 10? The signal loss in the cable itself is trivial...a bigger problem is the connectors and this is likely where impedance is changing as the geometry is nothing like the wire itself.

 

[morricab]

Longer cable lengths typically help with reflections as they will introduce higher propagation losses - which improves the return loss. The timing of the reflections as to being on an edge or not is a periodic function that can happen at short or long lengths - it's just that the delay of the reflection has to be within a certain multiple of the symbol time of the waveform.

Cheers, Joe

Do you think the return loss would be so significant from 0.5m to 1.5m? That would have to be a very lossy cable I would think. Going from 1m to 100m I think you have a good point but such a short distance? My experiments with this on scientific instrumentation showed no real losses going from 5 to 10m (with standard RG58 coax) but of course the period of the reflections changed as the propagation time changed.

 

[microstrip]

Have you designed a mass spectrometer with a dual polarity floating source? I doubt many have so take care in what you describe as "trivial". Matching the impedance of a flat plate and a scope input through a floating power supply is anything but trivial...and took months of work to optimize...in the end elimination of all ringing was impossible but was minimized.

What is the point of bringing a mass spectrometer to this debate? To dazzle people? I brought a comparable situation - nuclear electronics - where source and load impedance are similar to audio and just said it is trivial. Surely I could write about matching impedance in LHC detectors or tokamak for fusion studies, but IMHO it is meaningless for this thread.

Assuming the cables are constructed identically, the losses in 0.5 and 1.5meter will be trivial and the impedance should be essentially the same. There will always be some reflection, which is why the length probably matters...the question is what is it about that particular length that is special? Or is it only approximate? Maybe 1.45 m would be even better or 1.6? Has anyone experimented outside standard available lengths?

Although we do not have measurements, the question is that some people report that some digital cables are designed to have high losses and sometimes impedance varies significantly with frequency . I remember that some years ago someone in WBF was measuring impedance of digital cables and found that some cables reported for sounding good had a poor behavior in measurements.

But yes, you are right, all this debate is just audio gossip with a few data points and individual opinions on sound quality.

 

[morricab]

What is the point of bringing a mass spectrometer to this debate? To dazzle people? I brought a comparable situation - nuclear electronics - where source and load impedance are similar to audio and just said it is trivial. Surely I could write about matching impedance in LHC detectors or tokamak for fusion studies, but IMHO it is meaningless for this thread.



Although we do not have measurements, the question is that some people report that some digital cables are designed to have high losses and sometimes impedance varies significantly with frequency . I remember that some years ago someone in WBF was measuring impedance of digital cables and found that some cables reported for sounding good had a poor behavior in measurements.

But yes, you are right, all this debate is just audio gossip with a few data points and individual opinions on sound quality.

I never bring in items to a debate to dazzle people, perhaps this is something you do but you won't get that from me. Matching of impedance is a problem in a lot of fields, including audio and I don't think experiences from outside fields is either irrelevant or trivial, which is why I introduced my experience in combating the problem. I doubt many in audio have experience in dealing with and finding solutions to this kind of problem and therefore think sharing said experience could be helpful at least in understanding the problems if not the exact solutions. Even fewer have likely seen what data of this sort actually looks like or played with a network analyzer.

I guess a cable with high losses would also probably suffer less from reflections as they would get "lost" so to speak and as long as the signal was large enough you still have something the other side can lock onto. I wonder if also a device that boosts the output signal of the digital source would have a similar effect by giving the DAC something strong and sharp to lock onto? I have a Monarchy Audio DIP, that has almost never failed to improve a source that galvanically isolates the digital signal, reclocks it and boosts it to a full 5V before sending it on to the DAC. Does it improve the sound because of the (measurably) lower jitter (confirmed by independent test) or because it boosts the signal to give the receiver something strong to lock onto or both? The galvanic isolation was demonstrated (in the same independen test) to prevent jitter being introduced through the power supply of the Jitter box itself, something the other boxes on test did not prevent and indicate a less than thorough design.

 

[mountainjoe]

Do you think the return loss would be so significant from 0.5m to 1.5m? That would have to be a very lossy cable I would think. Going from 1m to 100m I think you have a good point but such a short distance? My experiments with this on scientific instrumentation showed no real losses going from 5 to 10m (with standard RG58 coax) but of course the period of the reflections changed as the propagation time changed.

I was just making a general point about the mechanisms at play in regards to return loss and reflection timing through cables - one is a direct function of cable length (loss) while the other is a periodic function of length and symbol timing. The original comment implied the latter was a simple function of length and that is not correct.

I did not make any comments in regards to the magnitudes of these effects but you are correct in that I would not expect a big difference between a 0.5m cable and a 1.5m cable.

Also this understanding provides a means to identify which of these mechanisms may be at play - for example if you add a 3dB attenuator on the cable, you will improve return loss by 6dB whereas there shouldn’t be a significant change in the timing of reflections (assuming the electrical length of the attenuator is small relative to the symbol timing of the waveform).

Cheers, Joe

 

[microstrip]

I never bring in items to a debate to dazzle people, perhaps this is something you do but you won't get that from me. Matching of impedance is a problem in a lot of fields, including audio and I don't think experiences from outside fields is either irrelevant or trivial, which is why I introduced my experience in combating the problem. I doubt many in audio have experience in dealing with and finding solutions to this kind of problem and therefore think sharing said experience could be helpful at least in understanding the problems if not the exact solutions. Even fewer have likely seen what data of this sort actually looks like or played with a network analyzer.

IMHO the experience you are addressing is irrelevant to this debate. I systematically bring electronic and instrumentation facts that are objectively relevant, addressing they are many times weak and too superficial to be correlated with the subjective aspects.

I guess a cable with high losses would also probably suffer less from reflections as they would get "lost" so to speak and as long as the signal was large enough you still have something the other side can lock onto. I wonder if also a device that boosts the output signal of the digital source would have a similar effect by giving the DAC something strong and sharp to lock onto? I have a Monarchy Audio DIP, that has almost never failed to improve a source that galvanically isolates the digital signal, reclocks it and boosts it to a full 5V before sending it on to the DAC. Does it improve the sound because of the (measurably) lower jitter (confirmed by independent test) or because it boosts the signal to give the receiver something strong to lock onto or both? The galvanic isolation was demonstrated (in the same independen test) to prevent jitter being introduced through the power supply of the Jitter box itself, something the other boxes on test did not prevent and indicate a less than thorough design.

Yes, but galvanic isolation has its intrinsic problems, mainly that the impedance of this devices is variable versus frequency and have resonances. It is a classical compromise in instrumentation. Can you provide a link to this independent test?

 

[Empirical Audio]

Have you designed a mass spectrometer with a dual polarity floating source? I doubt many have so take care in what you describe as "trivial". Matching the impedance of a flat plate and a scope input through a floating power supply is anything but trivial...and took months of work to optimize...in the end elimination of all ringing was impossible but was minimized.

Assuming the cables are constructed identically, the losses in 0.5 and 1.5meter will be trivial and the impedance should be essentially the same. There will always be some reflection, which is why the length probably matters...the question is what is it about that particular length that is special? Or is it only approximate? Maybe 1.45 m would be even better or 1.6? Has anyone experimented outside standard available lengths?

If you read my white paper, you will see that it is the risetime of the source that determines the minimum length. Very short, like 6 inches is good and longer than 1.25m is good. Between those lengths, most cables will add significant jitter. This has been independently ABX tested BTW.

Steve N.
Empirical Audio

 

[Empirical Audio]

Have you looked at this timing on a scope with discrete signal inputs to clearly see the ringing? With a continuous signal I guess this could be extracted with an FFT but it would be more interesting to see it discretely in time. Perhaps it is not possible to create a short enough pulse with such a short cable to see the effects directly. We were using several meters of cable and the pulses were ion packets that were a few ns wide.

I have looked at it with my 7GHz B/W programmable scope. The effect is really dependent on the risetime. If you drive the cable with a really short pulse or step with 20psec risetime, this will not tell you how it will behave with the actual driver, which probably has a >3nsec risetime.

Steve N.
Empirical Audio

 

[Empirical Audio]

This doesn't quite make sense to me...the cable itself is often the problem with impedance matching unless it is a true 75 ohms. Therefore, the cable and it's length will always both matter.

Certainly very long lengths will put the receiver out of the window for a large reflection, however they will also add a lot of loss and dielectric absorption, which causes intersymbol interference and slower risetime.

BTW, even a Belden 1694A impedance is not very close to 75 ohms. I used to use this as a reference until I realized how far off it is.

The bigger problem in audio is not so much the cable impedance, but the driver impedance not being 75 ohms. This is one aspect that I have mastered in my 22 years in this business.

Steve N.
Empirical Audio

 

[Empirical Audio]

I wonder if also a device that boosts the output signal of the digital source would have a similar effect by giving the DAC something strong and sharp to lock onto?

This would likely be a new driver with a new output impedance and a different risetime. If all of these are optimized, the DAC will generate lower distortion due to lower jitter.

I have a Monarchy Audio DIP, that has almost never failed to improve a source that galvanically isolates the digital signal, reclocks it and boosts it to a full 5V before sending it on to the DAC. Does it improve the sound because of the (measurably) lower jitter (confirmed by independent test) or because it boosts the signal to give the receiver something strong to lock onto or both?

There is no "signal boost" going on. The DIP reclocks the data and a new driver outputs this to the DAC. It is all about jitter. I have modded the DIP in past years to improve the output section and the clock. There is only so much that can be done with mods though. If you want to hear REALLY LOW jitter, try out one of my Synchro-Mesh reclockers with 7psec of directly measured output jitter at the end of my 4 foot BNC cable. You will not find anything else on the market that has jitter this low. Here is are the measurements:

https://www.audiocircle.com/index.php?topic=157348.0

The galvanic isolation was demonstrated (in the same independen test) to prevent jitter being introduced through the power supply of the Jitter box itself, something the other boxes on test did not prevent and indicate a less than thorough design.

Galvanic isolation is a double-edged sword. Yes, you will eliminate common-mode noise from ground-loops that can impact jitter. On the other hand, if you are using a pulse transformer, this will add some jitter, depending on the quality of the transformer. I like to do it without transformer on the output if possible, but I offer both versions on all of my products.

Steve N.
Empirical Audio

 

[Empirical Audio]

I was just making a general point about the mechanisms at play in regards to return loss and reflection timing through cables - one is a direct function of cable length (loss) while the other is a periodic function of length and symbol timing. The original comment implied the latter was a simple function of length and that is not correct.

I did not make any comments in regards to the magnitudes of these effects but you are correct in that I would not expect a big difference between a 0.5m cable and a 1.5m cable.

Also this understanding provides a means to identify which of these mechanisms may be at play - for example if you add a 3dB attenuator on the cable, you will improve return loss by 6dB whereas there shouldn’t be a significant change in the timing of reflections (assuming the electrical length of the attenuator is small relative to the symbol timing of the waveform).

Cheers, Joe

Adding attenuation to the cable will do a lot of harm, as will adding a ferrite. Both of these will add jitter.

Steve N.
Empirical Audio

 

[microstrip]

Certainly very long lengths will put the receiver out of the window for a large reflection, however they will also add a lot of loss and dielectric absorption, which causes intersymbol interference and slower risetime.

BTW, even a Belden 1694A impedance is not very close to 75 ohms. I used to use this as a reference until I realized how far off it is.

The bigger problem in audio is not so much the cable impedance, but the driver impedance not being 75 ohms. This is one aspect that I have mastered in my 22 years in this business.

Steve N.
Empirical Audio

Nice to see we agree on the reality of life, although my experience in these audio matters is very limited - in audio I am mainly a consumer with a pair of ears and some preferences!

 

[NorthStar]

Francisco, I don't see any Schiit in your signature.

 

[microstrip]

Francisco, I don't see any Schiit in your signature.

Thanks for highlighting it, Bob!

 

[mountainjoe]

Adding attenuation to the cable will do a lot of harm, as will adding a ferrite. Both of these will add jitter.

Steve N.
Empirical Audio

You can't argue that these cables lengths are so short that the losses are irrelevant to return loss and then argue that 3dB of attenuation will add jitter. On such short lengths, 3dB of attenuation should not materially affect the eye pattern unless there is a lot noise being introduced into the system. Resistive attenuation by itself will not increase jitter - this will only happen if the SNR at the receiver is impacted.

And if jitter is dominated by the reflections in the system, then adding 3dB of attenuation could actually improve jitter performance.

Lastly, if any such implementation doesn't have at least 3dB of margin, then it is grossly under-designed imo..

Cheers, Joe

 

[NorthStar]

Thanks for highlighting it, Bob!

Like I said before; with a name like that what can you do. :b