[rajacat]

Pat,

I have a 8' RCA/RCA so-called digital cable (Blue Jeans LC-1) which I inserted in place of a much shorter cable and did notice, I think, some improvement. Certainly My DAC does have a BNC input but my SB3 doesn't have a BNC output. Would I have better SQ if I changed one RCA for a BNC and used a RCA to BNC adapter on the other? Also is there an optimum length or is longer always better.

Thanks,

Roy

[wakibaki]

He said : "long pulses" - 1.4112 MHz
Two posts later he said:
"there are also short pulses, at 2.822 MHz."

So, what is your point here?

The question he was asked is 'what is the fundamental pulse frequency of SPDIF'

If you read posts #4 thru #9 it's quite clear that he believes the fundamental bitrate is 44100*32 and that what he is referring to when he talks about the 'frequency' of the long pulses is the pulse repetition rate.

w

[WGH]

Pat,

I have a 8' RCA/RCA so-called digital cable (Blue Jeans LC-1) which I inserted in place of a much shorter cable and did notice, I think, some improvement. Certainly My DAC does have a BNC input but my SB3 doesn't have a BNC output. Would I have better SQ if I changed one RCA for a BNC and used a RCA to BNC adapter on the other? Also is there an optimum length or is longer always better.

Thanks,

Roy

Here is some info from one of Pat's previous posts:

All 3 components of the chain contribute reflections: TX end, RX end, and the cable.

On a typical system, with RCAs and poor control of output Z (especially over a wide frequency range), you can easily get 20% reflection from either end. If 20% bounces back from the RX end, and 20% of that bounces back from the TX end as it is re-reflected forward, you are down to 4% total energy ending up at the RX end as an unwanted signal. If you can reduce the TX end's contribution to only 5%, then you will only have 1% unwanted energy arriving at the RX end.

If you can get both ends to reflect only about 4-5% of the signal, you can see how you can get far below 1% total reflection.

(All this ignores the cable; we shall assume it is matched fairly closely.)

Since I have done RF design for a living, I can get below 1% on both ends. You need good test equipment to be able to do this.

Some of you may recall from the data that I posted a few years back that some of my stuff has 4% or so of reflections. That is true, but I am measuring out to over 1 GHz. Over 10 times the BW of the SPDIF signal that you will ever encounter. When you reduce the BW, it drops to 1%.

Hope this helps to clarify things.

Pat

[rajacat]

So, I guess that means that RCA/BNC should reduce reflections more than RCA/RCA therefore that is good.

Cable length, hmmm...? Maybe the little digital packets are like a steam that likes a clear, unobstructed channel.
When you implement a connector, it's like a small dam that introduces turbulence(reflections) but eventually they smooth over a distance. So a longer cable smooths out the reflections more than a short one. Therefore, the more compromised your connectors; the longer your cable should be.

-Roy

[stormsonic]

Longer cable is needed to delay reflections, to arrive later in time. Reflections don't smooth over few meters distance, adleast not enough to dissapear.

Theoretical max. speed is 300.000 km/S or 0.3 m/nS (nano-Second). Inside cable signal is slowed down. Velocity factor for 75R coax is 0.67 for RG-59. Speed of signal inside coax cable is 0.3 m/S * 0.67 = 0.201 m/nS or 4.975 nS for 1 meter.

If cable used is 1 m, then reflection must travel this distance twice and will appear 9.95 nS later (2 * 4.975 nS = 9.95 nS).

[WGH]

Pat's long cable post from 3 years ago:

http://www.audiocircle.com/index.php?topic=49432.msg443440#msg443440

[wakibaki]

Here is what attenuators do:

1.) They attenuate the incident signal, by the amount of the pad.

Yeah, what is the effect of that?

It increases the jitter.

Why does it increase the jitter?

Because it degrades the SNR (signal to noise ratio).

And?

And reduces the slew rate...

w

[stormsonic]

Link removed

[art]

So, I guess that means that RCA/BNC should reduce reflections more than RCA/RCA therefore that is good.

NO! It will not reduce reflections. It will not create new ones, by sticking something that is closer to 30 ohms, in the system.

Cable length, hmmm...? Maybe the little digital packets are like a steam that likes a clear, unobstructed channel.
When you implement a connector, it's like a small dam that introduces turbulence(reflections) but eventually they smooth over a distance. So a longer cable smooths out the reflections more than a short one. Therefore, the more compromised your connectors; the longer your cable should be.

-Roy

NO! It does not "smooth out" any "turbulence". It merely acts as a delay line.

How long the cable should (or shouldn't be) is a function of the rise time, of the transmit side of the chain. It has nothing to do with what sort of connector you use, or whether you believe in fairy tales. (IOW, the mythical "75 ohm RCA".)

Your Duet has a HC logic output. Its rise time is around 3 nSec. Maybe a bit more, as they stick a ferrite bead, in the output. You would probably prefer the sound of a longer cable, as opposed to a shorter one, of identical construction. It is as simple as that.

Part of the reason I post this stuff is to have a place to point folks to, for all the questions we get, via e-mail. And, to correct some of our stuff, that has been misquoted, taken totally out of context, and splattered across any number of hobbyist and/or other manufacturer's websites. (That last part really galls the PTB, and I eventually hear about it.)

Hopefully, this time, some of it will make sense.

If not, ask here. Not somewhere else. Please. We don't read those other places (although we eventually hear about them), so we would appreciate confining any questions to this thread.

Besides, look at all the neat pictures I made! Too bad I am lousy with a camera...................

Pat

[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.

Anyone interested in the relationship between SNR, slew rate and jitter should look here:-http://pdfserv.maxim-ic.com/en/an/AN3631.pdf



The bottom image shows jitter vs. noise for a sinewave (zero crossing). In this case the signal is constant amplitude and the noise is increasing, but the effect is exactly the same if you attenuate the signal and the noise is constant. The RMS noise in the system is related to the bandwidth.



w

[art]

Decrease the slew rate? Right...............dv/dt is lowered. The whole RF world has just imploded.

Pat

[art]

Anyone interested in the relationship between SNR, slew rate and jitter should look here:-http://pdfserv.maxim-ic.com/en/an/AN3631.pdf

You will notice they use a sine wave. Too bad you failed to point that out. Perhaps you should read it, more closely, before you trumpet it.

Pat

[wakibaki]

Perhaps you should read what I write...

The bottom image shows jitter vs. noise for a sinewave (zero crossing).

You think it doesn't count because it's a sinewave? Even square waves have a sampling threshold.

Quote:- 'Just as depicted in Figure 3, at time ∆t, there is a probability that the noisy signal ∆y will reach the threshold; thus the jitter is added to the histogram at ∆t sooner or later than the anticipated sampling point.'

Quote:- With slight modification, Equation 6 can also accommodate the jitter translation of other waveforms. (page 5)

...and read the articles I quote.

You're just wriggling, just as you did with the answer about the network analyser. In your own words you laid out that HCMOS rise times (6nS typ) will not reveal a significant discontinuity in an attenuator pad made from discrete resistors. Why then would they reveal a discontinuity such as an RCA connector or a few centimetres of PCB track? No. The fact is that a simple DMM measurement of resistance is perfectly adequate to assess the quality of the match.

I note also your suggestions that reflections are:

1. Caused by inaccuracies in the cable impedance.

2 Pass through the attenuator twice.

If a reflection results due to a mismatch between the cable and 75 ohms, then a reflection will occur at the junction of the attenuator and the cable. If the attenuator is at the receive end then the reflection will pass through the attenuator only once. If it is at the transmit end then it will never pass through the attenuator at all.

Let us now look at the probability that a reflection will actually cause jitter.

In order to arrive at the receive end coincident with a signal edge, the cable must be of the correct length. The reflection must have travelled from the receive end, back to the transmit end and back to the receive end again. That is to say, depending on the length of the cable, the reflection can arrive anywhere along the signal.

The duration of the shortest pulse at 96kHz is 1/96000*128 seconds or ~81nS. The rise time being 6nS, the flat-topped portion of the signal therefore has a duration of 75nS. Since there are two edges, the likelihood of interference occurring for any given cable length is 75/12 or approximately 1 in 6 absolutely worst case, i.e. there are 75nS during which 1 of 2 reflections of 6 nS have no effect on the jitter whatsoever. For a sample rate of 44100 the likelihood falls to ~1 in 15.

Balancing the certainty of worsening the jitter due to slew rate and SNR considerations against a 1 in 15 chance of degradation due to a reflection which may not exist anyway, isn't it simply the case that your suggestions regarding attenuators are ill-thought-through, and that the information you are providing to your readers is less than complete?

w

[jkeny]

Waki,
You have been trying to prove something (I don't know what) both here & on DIYA for months now - denying the effectiveness of attenuators using convoluted arguments (not least the idea that characteristic impedance can be measured using a DMM). Have you ever tried attenuators yourself or is your mission to try to disprove reality? You have seen the scope shots that JosephK & now Jocko has produced. What exactly is your point - RF attenuators don't work & cause more harm than good ?

[art]

You call it wriggling.

I call it feeding a troll.

Pat

[wakibaki]

(not least the idea that characteristic impedance can be measured using a DMM).

I have never suggested this. When are you going to understand the difference between characteristic impedance and terminating impedance?

RF attenuators ... cause more harm than good

w

[wakibaki]

You call it wriggling.

I call it feeding a troll.

Pat

So, in fact, you have no answers to these points.

w

[Joseph K]

Ok, let's proceed "scientifically". This is your favorite procedure.
You say:

If you read posts #4 thru #9 it's quite clear that he believes the fundamental bitrate is 44100*32 and that what he is referring to when he talks about the 'frequency' of the long pulses is the pulse repetition rate.

We were talking about ONE SPECIFIC oscilloscope shot, which had shown a LONG PULSE.
(You know, people with a bit of experience, could guess that - do you have any of this?)
Speedskater wanted to make a guess about the time scale. So he came up with his question, about the bitrate.
Pat had responded on the flight, by memory - as opposite to You, who had spent some hours to dig up your "documentation". Pat was concentrated on a real problem, and not searching fight and trouble, as you are doing here. He had made a guess referring to that pulse, in front of us. In the background also I was guessing in a similar way, also Storm, and EVERYBODY was content with the answer, which was totally satisfying for that problem.

At this moment here comes Mr. BAKI, the unique, desperately searching for anything which could give him the occasion to put on stage his bloated ego and "scientific" superiority.
It's a bit problematic, that Pat had corrected himself ONLY two posts later, and had concluded exactly in the same way what Mr. BAKI is pushing..
So Mr. BAKI neglects this small particular, and sticks and quotes ONLY THAT PART which could seem to be a mistake, for the less attent..

I can not arrive to any other conclusion - this is pure trolling.

George

[art]

I have never suggested this. When are you going to understand the difference between characteristic impedance and terminating impedance?

w

http://www.diyaudio.com/forums/digital-source/168901-rf-attenuators-jitter-reducers.html#post2221444

Please take your food fight, back to where it came from.

Pat

[art]

We had to go to considerable lengths with very fast rise times, a very expensive oscilloscope and long, long cables for a university lab demonstration of reflection in digital systems, so I don't believe you've got plots. Not in an SPDIF system.

I have demonstrated the fallacy of that. I have also shown, with a "broken" 'scope, how much one can resolve:



If one looks closely, you can see a small downward step, where the impedance of the cable changes, very slightly. Sorry, didn't need the fancy TDR, we have, back at the lab, or your fancy university, to demonstrate it.

But, please, continue on how a resistive element, an attenuator, can lower the slew rate. We all need a good laugh.

Almost as funny as your claim on how a DMM can measure impedance. Terminating, or otherwise.

I come here, as a manufacturer, to interact with potential customers. Not educate manufacturer wanna-be kids, or participate in a flame war, with a self-appointed expert. I chose Audio Circle, because it is not like those other forums, some of you like to frequent. Kindly take your personal feud, back to where it came from.

I can think of at least 4 examples, off of the top of my head, where your DMM model falls apart. You can read one of them, here:

http://www.audiocircle.com/index.php?topic=91493.0

Of course, if you had actually looked at the photos, at the beginning of this thread, you would know why one needs a vector network analyzer (or TDR), as opposed to a DMM.

I see no need to further amuse you. Good day, sir.

Pat