AudioCircle
Audio/Video Gear and Systems => The Discless Circle => Topic started by: James Romeyn on 13 Jun 2013, 01:27 am
-
Parasound CEO Richard Schram, "Manufacturer's Comments," Stereophile June p.141 re. Parasound Halo CD-1 review by John Marks. No digital input; digital outputs only BNC and RCA, no AES/EBU balanced XLR digital.
...We hope (BNC is) the connection of choice...for anyone using...an outboard DAC. BNC connectors have a true 75 Ohms impedance, and will not create reflections that corrupt the integrity of digital signal transmission...
In the review Marks paraphrases Schram who "...doesn't think much of the AES3 digital-on-balanced-cables standard, which he believes is the result of an engineering compromise in the early days of digital intended to allow European broadcasters to save money by using their installed base of balanced analog microphone cables as digital-audio cables..."
Schram in the review also states a connector's mechanical dimensions must match the cable impedance, and XLR is "simply wrong for digital audio, in all mechanical respects"
XLR was designed for lower audio frequencies. ...The result is reflections-echoes of the signal travel back and forth, corrupting the real signal resulting in jitter and ringing near the signal edges which are the transitions between 0 and 1. The longer the cable and the higher the sampling rate, the worse the problem."
I have no idea how valid is the above. Schram's been at this for several decades. Has anyone compared BNC vs. XLR or BNC vs. RCA between the same digital source and DAC?
I suppose the most common IC in this forum is USB rather than any mentioned above.
-
"[...]BNC connectors have a true 75 Ohms impedance, and will not create reflections that corrupt the integrity of digital signal transmission..."
This is correct, up to some 3-4 GHz. With the mention that 50 Ohm BNC connectors are also available.
For RF application, BNC (and other connector types) came to replace the aging "UHF" connector which was distinguished by having "non-constant impedance" as described in the article here:
http://en.wikipedia.org/wiki/UHF_connector (http://en.wikipedia.org/wiki/UHF_connector)
I doubt that RCA and XLR are so tightly controlled in impedance over such a large spectrum as BNC is. They probably can't hold a candle to the poor UHF one.
Whether this has any effect on digital audio at common data rates - I don't know.
-
And here's a quick demo of what can happen to digital signals when encountering impedance mismatches.
Here's a 10MHz rectangular signal traveling through nearly ideal stuff - 50 Ohm source, connectors, cable and termination at oscilloscope end. Pretty easy to tell "0"s from "1"s at receiver's end:
(http://www.audiocircle.com/image.php?id=82094)
Now let's remove the 50Ohm terminator and use a 100Ohm resistor instead:
(http://www.audiocircle.com/image.php?id=82095)
And now let's go extreme and remove the resistor altogether:
(http://www.audiocircle.com/image.php?id=82096)
Which is "0" and which is "1" now and what's the clock of the signal?
Now instead of this rather controlled setup, let's imagine some generic microphone cable with RCA connectors at the ends.... each of a different impedance and all varying with frequency :jester:
-
:o I'd love to have an Oscilloscope. Don't know how to use one or what all I could do with it but still, I'd like to own one :)
There's a 2245a on CL in my area now for $150.00. Looks kinda rough from the front. Owner say's it's 4ch but he only has two leads.
-
Wow, those are amazing differences in the images. Thanks very much.
So do you agree with Schram 100%? That speaks badly of the industry in general, and especially manufacturers still including XLR and RCA digital connectors.
-
There are three reasons that RCA and XLR connectors are still used for digital audio. They are all spelled c-h-e-a-p! Regards
-
:o I'd love to have an Oscilloscope. Don't know how to use one or what all I could do with it but still, I'd like to own one :)
There's a 2245a on CL in my area now for $150.00. Looks kinda rough from the front. Owner say's it's 4ch but he only has two leads.
I would recommend a more contemporary model, as Tek has discontinued the analog ones like the 2245 cca 1996, they are all now at least 17 years old and will need some servicing. They're wonderful tools and if you can service them yourself sure, but otherwise you'd be soon left with an interesting looking doorstop :)
-
Wow, those are amazing differences in the images. Thanks very much.
So do you agree with Schram 100%? That speaks badly of the industry in general, and especially manufacturers still including XLR and RCA digital connectors.
Not really 100%. I agree that properly impedance matched transmission cables and connectors will degrade the signal much less than random cables/connectors. However we're talking digital signals - they can definitely take some degradation before the receiver would experience data errors (i.e. take an 1 as being 0 or the other way round). So as long as this degradation is kept under a certain threshold, all is well... Question is, how would the user know if he's doing ok or not? In the case of PCM, he'd probably hear "clicks" and other distinct effects when getting errors, as there's a good chance that the MSBs get erroneous too, and cause a large spike in the signal output by the DAC. The larger the BER (bit error rate) is, the more clicks he would hear. Thus he could decide that the cable is bad/mismatched and replace it. DSD would be more interesting, as theoretically you could get only some small white-noise-like degradation, at a level that would not yet attract your attention on the cables.
Here's a nice summary of cables/connectors for digital audio:
http://www.rane.com/note149.html (http://www.rane.com/note149.html)
you may easily decide from it that Schram is correct in saying that BNC+coaxial is introducing the least degradation of the signal.
BTW, BNC terminated RG59 or similar cable is not really expensive. If you have a ham radio guy nearby, he can make them for you for next to nothing. Definitely at a lesser price than XLR cables.
-
I still employ "lowly" one piece CDP, albeit a very well performing one (all SS TRL-modded Sony 900ES SACD multi-ch/DVD, multi-ch and DVD deleted). I'm daydreaming about future music server to outboard DAC.
My only current digital transmission is RCA co-ax, Blu-ray/Wi fi player to 2003 receiver for HT.
When I last experimented with outboard DACs for high end home audio (early-mid 00s) my RCA co-ax digital was (and still is above) a strange unidentified co ax cable (unidentified because the seller desired to keep it secret so he removed the jacket which apparently had stencil identifying it): dual layer silver plated copper braided shield over single center conductor comprising 16AWG single strand silver plated steel with white teflon dielectric. RCA plug is entry-level Neutrik then recommended by Stan Warren for analog IC.
Years ago in direct AB with $13k Krell transport/$10k Wadia 27x DAC the above described cable/RCA plugs compared well vs. commercial digital RCA up to $500...I think even one commercial cable was $1k silver type.
Do you recommend different RCA plugs?
What is potential audible improvement if one installed BNC jacks/plugs where now is located RCA jacks/plugs?
-
It is pretty expensive to characterize digital transmission cables, for an objective evaluation one would need some tools like those described here:
http://teledynelecroy.com/sistudio/?capid=131&mid=1084 (http://teledynelecroy.com/sistudio/?capid=131&mid=1084)
we're talking 5 to 6 figures in $...
The most important parameters for digital audio would be BER (bit error rate) and jitter.
BER would tell you what percentage of the transmitted bits have been received erroneously (1 mistaken for 0 or the other way round).
Jitter would tell you how stable in timing are the 0 to 1 and 1 to 0 transitions. (i.e. do they always occur exactly with the same delay as compared with the transmitter's transitions, or does this delay statistically vary around an average value http://www.tnt-audio.com/clinica/jitter1_e.html (http://www.tnt-audio.com/clinica/jitter1_e.html) )
A high BER would produce frequent "pops" and "cracks" or other forms of noise in the reproduced signal. You can test how well you're doing by creating a CD with two tracks, one with null signal ("silence") and one with a very low levle 1 kHz constant signal. Then just listen to it with the volume turned high. If it works as expected (i.e. there is no "crackling" or "popping" on either track) then your cable passes.
Jitter is more difficult to tell by just "listening", also some DACs are more immune to it than others as they employ some additional techniques to mitigate the effect of jitter. Normally the largest contribution toward jitter is the source (CD player) clock and electronic circuits, not the cable.
In the end I'd say that you can test for BER - it's easy and if your cable passes, then it would be very unlikely that you would get any benefit from upgrading the connectors. If, on the other hand, you can put in evidence some frequent, annoying noise, then you should probably revisit that cable.
As for reverse engineering your cable, you can check the table here:
http://en.wikipedia.org/wiki/Coaxial_cable (http://en.wikipedia.org/wiki/Coaxial_cable)
and look for cables having 75 in the impedance column, PF or PTFE in the dielectric one, double in the shields column, and an outer diameter close to yours. That assuming that the seller's description was correct.
-
Some more info and measurements on impedance matching and reflections HERE (http://ar-t.co/geekspeak.html). I do own a Legato but no affiliation otherwise.
I think one of the reasons studios can get away with using the AES/EBU interfaces is that everything is usually controlled by a master clock.
-
A terminator is one thing and a connector is another! Get out your Ohm-meter and measure the terminator, it will read it's listed value. Measure a connector and it will read near infinity. For a rather low frequency circuit like S/PDIF the connector is not a factor. For short interconnects like 1 or 2 meters the co-ax is not a factor either.
-
And here's a quick demo of what can happen to digital signals when encountering impedance mismatches.
....................
Now instead of this rather controlled setup, let's imagine some generic microphone cable with RCA connectors at the ends.... each of a different impedance and all varying with frequency .
Why don't you try it?
Then try it with a length of zip-cord or speaker wire & RCA connectors, but use a more reasonable S/PDIF like frequency.
Remember that S/PDIF is a bandwidth limited signal.
-
A terminator is one thing and a connector is another! Get out your Ohm-meter and measure the terminator, it will read it's listed value. Measure a connector and it will read near infinity. For a rather low frequency circuit like S/PDIF the connector is not a factor. For short interconnects like 1 or 2 meters the co-ax is not a factor either.
You can only measure resistance with an ohm meter (DC), you can't measure impedance (AC).
I have seen you post before that you don't think the connector matters, but I don't know what this is based on?
Experience? Measurements?
Was curious how you came to this conclusion?
RCA's and XLR's are not 75 ohm, so they will create reflections and degrade the signal. It is physics, and you can't get around the laws of physics. The question becomes whether or not the reflections degrade the sound.
Randy
-
RCA's and XLR's are not 75 ohm, so they will create reflections and degrade the signal.
And it's a good thing the XLR connector impedance isn't 75 ohm, as the specified impedance for an AES/EBU connection is 110 ohm. ;)
I have read that the XLR connector is actually closer to maintaining the target 110 ohm AES/EBU impedance than the RCA connector is to maintaining the target S/PDIF 75 ohm impedance, but I don't know if that's true and neither connector was designed for digital transmission impedances and frequencies.
Obviously the BNC terminated 75 ohm S/PDIF coax is the most consistent, but there are so many sources and DACs not offering BNC connections.
I guess I need a DAC with a built-in OLED oscilloscope display function!
Steve
-
Quite a few DAC's have BNC inputs, and there are a number of USB > S/PDIF convertors with BNC output.
-
And it's a good thing the XLR connector impedance isn't 75 ohm, as the specified impedance for an AES/EBU connection is 110 ohm. ;)
I guess I should have checked the XLR :oops:
I have seen many posts about RCA's, I saw some today that talk about impedance in the 30 ohm range.
Here, at diyaudio and at diyhifi.org there are many posts that talk about the impedance mismatches, sometimes with TDR plots which really show what happens to the signal from reflections.
There is also some information at diyhifi about adding a off the self terminator which might help fix some problems.
Randy
-
You can get 75 ohm rca connectors, but BNC is the way to go if you can.
-
You can get 75 ohm rca connectors, but BNC is the way to go if you can.
Examples?
Guys smarter than me say that the RCA's that claim to be 75 ohm are not really 75 ohm, and some of them have the instruments to really measure it.
Randy
-
Examples?
Guys smarter than me say that the RCA's that claim to be 75 ohm are not really 75 ohm, and some of them have the instruments to really measure it.
Randy
http://www.canare.com/ProductItemDisplay.aspx?productItemID=40
http://www.vhaudio.com/nextgenrcaspecsheet.pdf
-
Dave,
It doesn't matter what impedance a RCA plug has. What matters is what impedance the combined RCA plug and RCA socket has. Plug in those nominally 75 Ohm plugs into a 'normal' RCA socket and you've got a 30-50 Ohm impedance. To my knowledge (not verified by personal use or measurement) the only RCA plug/socket pairing that presents a complete 75 Ohm impedance is the WBT Nextgen RCA plug and socket. FWIW
-Paul
-
Dave,
It doesn't matter what impedance a RCA plug has. What matters is what impedance the combined RCA plug and RCA socket has. Plug in those nominally 75 Ohm plugs into a 'normal' RCA socket and you've got a 30-50 Ohm impedance. To my knowledge (not verified by personal use or measurement) the only RCA plug/socket pairing that presents a complete 75 Ohm impedance is the WBT Nextgen RCA plug and socket. FWIW
-Paul
Yes, that is a good point!
Canare offers a 75 ohm female connector as well...
http://www.canare.com/ProductItemDisplay.aspx?productItemID=101
-
...
Canare offers a 75 ohm female connector as well...
Good to know. Thanks. - Paul
-
It's interesting that using a 75 ohm cable with mismatched non-coaxially terminated RCA plugs (WBT 0144) that should likely result in unwanted signal reflections and square wave corruption, I still believe I hear a more solid and focused sound than with the Toslink optical connection that has no inherent mismatch.
Most of the articles and test reports I have read where they had the proper test equipment usually measured significantly higher jitter with the optical connection than with the coaxial. I guess this is to be expected when inserting another conversion layer in the signal path to convert electrical S/PDIF to optical and back again as well as the possible suspect quality of many inexpensive Toslink transceiver chipsets.
I'm not sure how much the impedance mismatch of RCA connectors affects the sound, but those who have sources and DACs with BNC, XLR and RCA connectors at least have the ability to do a rapid input switching A/B comparison to get some idea of real-world sound differences between BNC, XLR, regular RCA and "true" 75 ohm RCA, beyond electrical theory alone.
Steve
-
I use a scope to check every digital path I wire.
There is no substitute for checking signal integrity
at the receiver point.
One thing you guys don't really mention is the
affect on clock recovery as the edges degrade.
I'm not sure about the common DAC chips, but
most of the digital circuits I've worked on do
some sort of clock recovery from the incoming
signal (unless a separate clock is provided on
another pair) in order to correctly sample
incoming data edges into some sort of input flip-flop
for the rest of the chip.
AES is also differential and most modern
differential transmitters can drive a very wide
range of cable and connector conditions with
little impact to signal quality.
BNC started life as an RF connector and it's
used quite extensively in the test equipment
world where low noise is expected from DC
well into the 200MHz range. That makes it
a natural for digital signals but I don't think
it offers superior performance to an XLR
cable in the differential world.
My two cents.
-
A terminator is one thing and a connector is another! Get out your Ohm-meter and measure the terminator, it will read it's listed value. Measure a connector and it will read near infinity. For a rather low frequency circuit like S/PDIF the connector is not a factor. For short interconnects like 1 or 2 meters the co-ax is not a factor either.
I suggest you measure both the connector and terminator at something like 10-100Mhz and post the results here.
S/PDIF was designed to use rectangular ("digital") signals. By any means they are not "low frequency". An ideal rectangular signal has an infinite spectrum.
-
I use a scope to check every digital path I wire.
There is no substitute for checking signal integrity
at the receiver point.
One thing you guys don't really mention is the
affect on clock recovery as the edges degrade.
I'm not sure about the common DAC chips, but
most of the digital circuits I've worked on do
some sort of clock recovery from the incoming
signal (unless a separate clock is provided on
another pair) in order to correctly sample
incoming data edges into some sort of input flip-flop
for the rest of the chip.
AES is also differential and most modern
differential transmitters can drive a very wide
range of cable and connector conditions with
little impact to signal quality.
BNC started life as an RF connector and it's
used quite extensively in the test equipment
world where low noise is expected from DC
well into the 200MHz range. That makes it
a natural for digital signals but I don't think
it offers superior performance to an XLR
cable in the differential world.
My two cents.
I have tried many coax cables over the years, most recently the Lessloss coax for the past few years. After reading about potential problems caused by reflections, impedance matching, etc., I tried several cables of different lengths to see what difference (if any) would become manifest, and an inexpensive 10-foot Canare cable outperformed my expensive Lessloss cable, as well as the same cable in different lengths. I then decided to try Toslink and was amazed that it was the best yet. I can do instant switching comparisons and there is no doubt in my mind that the Toslink is much better. I now have a Wireworld Supernova glass cable that sounds better than any cable I have yet tried. I don't have the answer, but it just sounds better. I suppose it is useful sometimes to question the "known facts" that we have been fed.