|
Audio Asylum Thread Printer Get a view of an entire thread on one page |
For Sale Ads |
66.220.121.48
In Reply to: RE: Elizabeth is right posted by jllaudio on August 13, 2012 at 05:55:37
1.5m should be sufficient in most cases. Risetimes are usually 25nsec or faster. The first refection is the important one.
See:
http://www.positive-feedback.com/Issue14/spdif.htm
Follow Ups:
I agree that the first reflection is the most important one, as it is the largest one.
I disagree that the "average transport" (whatever the heck that is) has an average rise time of 25 nsec. In fact I remember about 10 years ago when Hi-Fi News measured a Theta DVD player and measured a risetime on the the S/PDIF output of around 3 nsec. This is about what I measure on our digital audio products also. So we are already off by around an order of magnitude.
Light in a vacuum travels about 1 foot per nanosecond. In a copper cable it is about 75% to 80% that speed (typically). So if a cable is 3 feet long long, each trip will take roughly 4 nanoseconds. Then three trips means the first reflection will arrive around 12 nanoseconds after the original signal. Depending on the circuitry in the transport and the DAC, this could be in exactly the right time range to screw up the original signal. It will all depend on the circuitry of the transport (mainly) and the DAC (partly).
On the other hand, using a longer cable of say 8 meters (25 feet) will change the time of the first reflection to around 70 nanoseconds and there is no possibility of a problem with CD. Even cutting it in half to 4 meters would give a much better chance of good sound with a broader range of equipment. But going to 1 or 1.5 meters is almost like asking for trouble.
I wrote that the first reflection will arrive 12 nsec after the original signal. This is not quite right. While it is true that the first reflection will arrive 12 nsec after the original signal is *transmitted*, it will arrive only 8 nsec after the original signal *arrives* at the receiver.
This secondary signal will then happen at almost the perfect time to disrupt the signal from most digital audio sources. One is much better off to use at least a 5 meter or longer cable to avoid these sorts of problems altogether.
Nice sq pop. Thanks Charles. I put in a 4 meter extension to my 1 meter spdif cable (I know adapters are not good) and it sounds very very good.
means that most excess length will be coiled: won't this hinder the transmission?
Also, is this visible in reflectometers?
Stu
I am an RF engineer - and to really understand what's going on you would need to use a Time domain reflectometer - you put a pulse down the cable and you will see every reflection as the signal goes down the line. In my experience the largest reflections will come from the connectors and adapters, less so cable bends unless they are excessively tight. And based upon empirical experience of sound, this makes sense.I have not seen normal cable bends creating more reflections - and with an eye diagram (a way of demodulating the digital on a scope and measure timing errors, clock errors, etc.) you see any small effect getting swamped out entirely by connector, clock and other effects. However if you violate the manufacturer's "minimum bend radius" you will see problems form that bend. And since the cable may be damaged, it may never come back once it has gone past that radius. A cable won't generally naturally lay in that sort of radius - you usually have to force it - and if you do the cable pinches, and the geometry deforms - sometimes permanently. If you don't know what it is, normal use will be fine - just don't force the cable into a tight bend or routing and you shoudl be 100% fine 100% of the time.
Also ...
A good impedance match is not always reflection free(*) - the design of the connectors and the method of attachment will influence this. If you could have a reflection free 75 Ohm cable - then the musings on the right length would be a non issue as it would not matter. BNC connections are generally pretty good if specified into the upper UHF to low microwave ranges. RCA connectors won't be good - even "true 75 Ohms" connectors if hooked to a typical SPDIF input won't be good.
SO my rules for coaxial RF cables for SPDIF:
1. Connectors - BNC first and specified to 1-6GHz somewhere.
2. Cable type - Specified to Microwave range (3-10GHz) with high coverage shielding. Skinny with a very narrow minimum bend radius.
3. CRIMP connections before solder. Passive Intermodulation is the only working theory for this - since multiple solder connections can be poor for this - and a good solid crimp may be a better interface.
5. Adapt from BNC to RCA - and if you have "True 75 Ohm" adapters it should work good. But don't sweat it too much - since the RCA connector itself may be a source of issues.
I use a Squeezebox into a Berkeley audio DAC right now - I have a homebrew cable at about 1.7m, and I also have a "Black Cat Silver Star" which is 1.23m - both sound great. I had an Audioquest cable that was "ok" but overall somewhat crummy and expensive. AQ I have egenrally liked for analog - but not SPDIF!However the RCA on the SB end of it doesn't make an intimately tight connection. I haven't noticed a large effect in poor sound with one that is tighter (but not 75 Ohms) and looser (and is true 75 Ohms)- but I suspect that if I were to correct this with a hardware mod to the SB, I am sure there will be a marginal improvement.
The SB is a great source - but the source itself isn't perfect - and there will be power supply noise, etc, that once you have a cable you are happy with that you may want to address before fussing with more cables. (I also have a Sonos - and use it for a casual whole house audio. In addition to not being bit perfect, it is very jittery - the SB is head and shoulders better than that)
(*) It is possible to have a perfect match hand still have large reflections - in many connectors they use various size and phase of reflections from various parts to create an overall match when they all ring back and forth.
"Knowledge is knowing that a tomato is a fruit. Wisdom is knowing not to put it in a fruit salad"
Edits: 08/26/12 08/26/12
.
.
It's rewarding (and very rare, unfortunately) when theory leads to an actual improvement in sound quality. I'm glad that worked for you. Try taking out the first meter and adapter and just going with the 4 meter cable. There is nothing magic about 5 meters. I think 4 meters will give you pretty much all of the advantages without the disadvantages of using two different cables plus an adapter.
As always, please let us know what you find out. Thanks!
Unfortunately the other 1.5 meter cable is hardwired into my Buf32s dac. But no matter adding the 4 meters made a substantial difference for the better.
Thanks for the update.
Won't a very long cable have a reduced bandwidth which will round off the square wave edges making a more difficult job for the receiver to retrieve the clock and determine the bit values?
At the lengths we are talking about dispersion shouldn't be an issue.
"Knowledge is knowing that a tomato is a fruit. Wisdom is knowing not to put it in a fruit salad"
No, you would have to get to hundreds of feet of cable for these effects to become problematic.
Any pictorial depiction?
And does this "reflection" problem also plagues spdif/toslink and st (with bnc terminals) optical cables?
Its like a "bounce" of the signal edge from the destination back to the source and then back to the destination, like a ball ricocheting from one wall to another and back again.
You will probably be familiar with these reflections in the form of "ghosts" on analog TV signals. The link below has an animation of this effect.
Tony Lauck
"Diversity is the law of nature; no two entities in this universe are uniform." - P.R. Sarkar
Thank you very much.
Post a Followup:
FAQ |
Post a Message! |
Forgot Password? |
|
||||||||||||||
|
This post is made possible by the generous support of people like you and our sponsors: