[Joseph K]

Wakibaki,

As I said already, if you put an attenuator in the SPDIF connection, you degrade the SNR and decrease the slew rate. So if you don't have significant reflections in your system, you're definitely making things worse.

Also here you had picked up only a part of  the original message. All this debate started by saying that attenuators could do good in a system with one special unit, the Hiface.
Where the signal level was already way above the standard. Placing an attenuator in the line just made all things turn back to the normal situation.
So the extra jitter "introduced" this way - is exactly the same jitter level that we are supposed to have with standard signals!
That is the jitter  "designed in" into the receiver, when it works with 500mVpp levels.

So, when we insert an attenuator - and so we turn back to normal - as a side-effect, we also gain a reduction in the level of reflections.
Reflections are causing data-correlated jitter.
So - in general, through application of attenuators, one is 'exchanging' possible data-dependent jitter for a slightly raised level of random noise at the input. Also, while the relative noise is raised proportionally with the attenuation, the reflections are reduced by twice as much.
But, in the original debate, with that device, your point is totally moot: the levels only got attenuated back to normal.

George

[Joseph K]

Also:

I note also your suggestions that reflections are:

1. Caused by inaccuracies in the cable impedance.

No. It was not stated here. Primarily - as it had been clearly shown up above - reflections are caused by wrong, reactive termination impedance. And as such, they travel across the attenuators.

[art]

Perhaps you should point out to the readers that placing an attenuator at the receiver attenuates both the signal and the noise. Ignoring for a moment any noise caused by the input stage, the SNR essentially is the same.

Slew rate is not changed.

Part of the rationale for a fast rise time is the amount of time spent in the "dead zone", around the zero crossing, is lessened. Clearly, this is only the case in a digital signal. As this time is lowered, the ability of noise or reflections to corrupt the signal is minimized.

When we have to send a pure sine wave, across a long line, it is customary to pass it through a bandpass filter, to remove any of the noise. Yes, the dv/dt with a sine wave is going to be much lower, just by its nature. Trying to compare digital signals to sine waves is just plain silly.

Pat

[Mike B.]

I saw the thread at DIYAudio and ordered a 6 dB attenuator. It did make a sonic improvement IMO. I also have a coax cable I got with a modified Magnavox CD player many years ago. Maybe Audio Alchemy? It is a coax cable with a RF connector on one end and a RCA male on the other. The RF end screws into a large metal cased adaptor with a RF female on one end and a RCA male on the other. I have no idea what is inside the metal case. The reason I bring this up is because this old cord has bested some well respected digital cables. I am wondering if a similar thing is happening with it?

[Joseph K]

Mike,

Hm, sounds suspicious.. it could even be..

Pat,

I think I understand his point of view: he says  that the noise inherent to the receiver side remains the same, and so the ratio deteriorates. This is why I had shown that even taking account of this, we are still better with the attenuators. Not talking about the fact that he is cheating, and likes to forget that we were only attenuating back to normal.

[HAL]

Mike B.
Was it a cylindrical metal grey with the RF connector?

IIRC it was called the AA Clearstream cable for S/PDIF.  There is a pulse transformer in the metal cylinder.  It was there for galvanic isolation. 

[Mike B.]

Hal, that sure sounds like it.

[wakibaki]

All this debate started by saying that attenuators could do good in a system with one special unit, the Hiface.

I searched the whole thread. Hiface is not mentioned until this post of yours. It's obvious that the application is meant to be general, and no-one coming on the thread without reference to anything else would ever imagine it was only intended to apply to the Hiface.

So if nothing else, I've clarified this point.

w

[wakibaki]

Perhaps you should point out to the readers that placing an attenuator at the receiver attenuates both the signal and the noise. Ignoring for a moment any noise caused by the input stage, the SNR essentially is the same.

Oh, right, that'll be why so many radio receivers have attenuators at the input, to improve the sensitivity. Not.

You've obviously never designed a radio receiver.

w

The (thermal noise) power spectral density per hz of bandwidth vn(bar)^2=4kBTR Where kB is Boltzmanns constant, i.e. the thermal noise in a system is related to temperature, resistance and bandwidth. Attenuators do not attenuate noise, just the signal, resulting in a degraded SNR. Basic.

[wakibaki]

Slew rate is not changed.

OK, a 6dB attenuator halves the voltage.

Slew rate =dv/dt or rate of change of voltage versus time interval (rise time).

The rise time is unchanged by the attenuator.

Therefore a 6dB attenuator halves the slew rate.

w

[wakibaki]

possible data-dependent jitter

Yes, iff there is a mismatch, iff the reflection coincides with the edge, with a worst-case possibility of 1 in 15 for 44.1. Even if the reflection does coincide with the data edge, the placement is critical and may not result in jitter.

Most DACs have a 75R resistive termination which at these rise times can be checked with a DMM. If you pointed this out before getting out your oscilloscopes you might be doing something useful.

w

[sts9fan]

Let's keep this topic civil

[wakibaki]

Let's keep this topic civil

I'm trying.

w

[sts9fan]

try harder

[art]

Oh, right, that'll be why so many radio receivers have attenuators at the input, to improve the sensitivity. Not.

You've obviously never designed a radio receiver.

Ok, you have now just crossed over the line.

I bet every person here has made a long-distance call, over more than one brand of microwave radio. Almost every brand had a pad, between the mixer, and the preamp.* Yes, it added to the noise figure. Despite that, they all still had one!

And the reason: impedance mismatch.

The pad made for a more resistive termination, for the mixer. Those impedance perturbations added to the group delay, which lead to increased AM-PM. Yet, we sacrificed noise figure, and also fade margin, by sticking in those pads.

No, I didn't design the "rat race" mixer. I just worked for the guy who did. May he rest in peace. He was a gentleman, and an honor to work for. (My job was to get the group delay down.)

So, don't make asinine assumptions about someone you know nothing about.

Do you have any further nonsense to enlighten us with? 

Please, tell us more about slew rate. We are fascinated. Next thing you know, you will be lecturing us on rise time.

(Maybe I need to add "rolling eyes" smiley thingies, so you can spot obvious sarcasm and derision.)

Every text I have mentions C/I vs BER. None of them talk about slew rate. Yes, as "C" goes down, when you lower the signal, the BER deteriorates. I think most here know that, without a college-level communications theory course. Most receivers lose phase lock, when it is too low. Only an idiot would insist on running it at the ragged edge. (I bet we both know some. It would be erroneous for either of us to slag the other, based on idiots we had the misfortune of meeting.)

I thought slew rate was something only audio designers use. You know, those people you accuse of being down to their last brain cell. I am surprised that you would lower yourself to our level.

I am trying to be civil, but your insults can not stand.

Pat

* = Never worked on Western Electric/Bell Labs/ATT stuff. They may be an exception. Collins, Raytheon, even GEC; they all used those tricks. (Ever hear of the last one? You should, if you have been in telecom, on your side of the pond.)

[art]

Most DACs have a 75R resistive termination which at these rise times can be checked with a DMM. If you pointed this out before getting out your oscilloscopes you might be doing something useful.

w

No, they don't!!! I have measured more brands of DACs than you have probably seen. Most have horrible mismatches. If you had actually measured some, you would know I am right.

(The consulting side of the business is regularly paid to evaluate different brands, for the companies we consult for. Measured more DACs, in the 90s, than you have seen. Again, bogus assumption.)

They were all measured with a VNA and TDR. If we had charged our clients, based on an ohmmeter measurement, we would have been laughed out of business. We are still here. I wonder how? We are obviously clueless.

Pat

[JoshK]

Pat...don't feed the troll.  I know only enough about this subject to fill a thimble full but I can spot who knows what he is talking about and who doesn't.  Let's leave it at that.

[art]

Ah, but he hurled an insult at me. Them's fightin' words, down here in Texas.

I am trying hard, to keep this thread on track. Some folks seem intent on derailing it. They have lots of practice. Since last July, or so.

(I could suggest they both get a room, but some may take that as an insult!)

Pat

[JoshK]

I get it.  When you answer him on a technical only level it levels him more IMO.

[Joseph K]

No, I would not take that as an insult, and already was thinking of it.
One last comment before I'm off:
This plot is from NIST, it is a suggested test setup measuring phase noise of oscillators.
They use the attenuators (called PAD) exactly the same way that is suggested here.
The question is: if they only raise the jitter at the mixer, why did they put them there, in a phase noise setup aiming for the lowest jitter possible?


Ciao, George