View Full Version : Digital Cable Lengths and Other Stuff.
Hi Gentlefolk,
After being directed to a few posts HERE (http://theartofsound.net/forum/showthread.php?27847-Bushmaster-Mk-II-(TC-7533)&p=482312#post482312) I thought I'd start a thread about cable length for digital I/C's, and here it is!
There's been much chatter about this over the years and wondered what the general consensus was... I have a hypothesis or two but would appreciate your thoughts first. :)
ATB,
Mike.
I don't have any thoughts of my own but am following the general depth of technical opinion out there that suggests 1.5m or multiples of are a good length to minimise internal reflections.
I only use one digital co-ax run: from Logitech Touch to Bushmaster Mk.II, a 1.5m length of Blue Jeans co-ax (Belden 1694A).
Thanks, Martin, good stuff that Belden cable, I have quite a few 'flavours' of it. But what made you go for 1.5m rather than 'multiples of', if I may ask?
Cheers...
Stratmangler
17-09-2013, 20:53
Shorter certainly isn't the way to go, but what length is optimal?
I remember Siemens having a cable of 10m length with an ISDN interface box, the cable fitting several designed in parameters.
I suppose that the optimal length for SPDIF is going to be something related to the switching frequency of the analog representation of the digital stream (yes folks, digital is analogue - get yer bonces round that one ;)).
I don't have the maths knowledge (or interest - life's too short, and in all likelihood I have more days behind me than I have ahead of me) to work it out.
(yes folks, digital is analogue - get yer bonces round that one ;)).
Yeah, it took a while, but I eventually managed it!
Stratmangler
17-09-2013, 21:13
I'm going to guess that 200m is the optimal length for a SPDIF cable.
That's roughly the distance between me and my local Virgin Media roadside cabinet.
It's just a guess mind - there's no claim for my guess being anywhere in the ballpark :eyebrows:
Stratmangler
17-09-2013, 21:20
Silly boy! :lol: ;)
So the clues mount up (says he, noting the little wink).
What about the 10m that the Siemens cable was - that's got to be a lot closer to the mark?
AFAIU its the ratio of cable length vs signal rise time that the recommendation for 1.5m (SPDIF) cables arises from. The return loss due to cable length is negligible in itself but because the end points are terminated with non optimal RCA connections that don't match the characteristic impedance of 75Ohms the return loss due to the mismatch is poor so a lot of signal gets reflected. If we could terminate with matched BNC connectors and the manufactures ensured all internal layouts where also 75Ohm then we could use any length of cable we want.
Another thought: I believe the typical RCA phono is somewhere in the region of 35 to 45 Ohm impedance so maybe a better interconnect for SPDIF would be a 50Ohm cable, yes, there would still be a mismatch but it wouldn't be as great and maybe it would perform better. It's not something I've tried though.
No idea, mate (Chris)... I just think there's a better way! (keeping my cards close to my chest for the moment ;))
AFAIU its the ratio of cable length vs signal rise time that the recommendation for 1.5m (SPDIF) cables arises from. The return loss due to cable length is negligible in itself but because the end points are terminated with non optimal RCA connections that don't match the characteristic impedance of 75Ohms the return loss due to the mismatch is poor so a lot of signal gets reflected. If we could terminate with matched BNC connectors and the manufactures ensured all internal layouts where also 75Ohm then we could use any length of cable we want.
Another thought: I believe the typical RCA phono is somewhere in the region of 35 to 45 Ohm impedance so maybe a better interconnect for SPDIF would be a 50Ohm cable, yes, there would still be a mismatch but it wouldn't be as great and maybe it would perform better. It's not something I've tried though.
Ah-ha.... now were getting somewhere..... :)
Stratmangler
17-09-2013, 21:43
I'm gonna shut up and (hopefully) learn summat! :)
Has anyone actually noticed a difference with shorter cables?
I had some short cables made up for me by Mark Grant - 0.5m Coax - that is what i uses now.
When i ordered them, although i had read the 1.5m guideline in the past - i did not remember it at the time
Does anyone think the 0.5m is a real impact?
I don't have any thoughts of my own but am following the general depth of technical opinion out there that suggests 1.5m or multiples of are a good length to minimise internal reflections.
I only use one digital co-ax run: from Logitech Touch to Bushmaster Mk.II, a 1.5m length of Blue Jeans co-ax (Belden 1694A).
Has anyone actually noticed a difference with shorter cables?
I had some short cables made up for me by Mark Grant - 0.5m Coax - that is what i uses now.
When i ordered them, although i had read the 1.5m guideline in the past - i did not remember it at the time
Does anyone think the 0.5m is a real impact?
Perfect!.... this is exactly what I'm wanting to know, thanks for asking the question. :)
(Stay with me here, folks... I'll eventually get to where I'm trying to go. But it might take a while :eyebrows:)
Ah-ha.... now were getting somewhere..... :)
OK, Belden 9395, 35 Ohm.....or 9778, 40 Ohm...What where you thinking of Mike?
Edit: Yes I noticed short RCA terminated 75Ohm sounded worse than a 1.5m version...
OK, Belden 9395, 35 Ohm.....or 9778, 40 Ohm...What where you thinking of Mike?
I did once think of exactly the same thing, as in 50ohm(ish) might be better.... but I think there's a better way! :eyebrows:
<caveat: BNC would have been *SO* much more simple. Which idiot thought the RCA connector would be "best"? :doh:>
Audioman
17-09-2013, 22:04
SPDIF (COAX or Optical) is suitable for short distances of max 10m. Toslink (optical) has a practical limit of 6.1M due to high light attenuation. BTW Marc grant HD1000 claims the ideal 75 ohm RCA connectors and cable. Can't see why lengths shorter than 1.5 m would affect sound though in practical terms less than 0.5m is not generally usable.
.... but I think there's a better way! :eyebrows:
Toslink!! :D
Stratmangler
17-09-2013, 22:07
Toslink!! :D
Probably closer to the truth than you realise :)
SPDIF (COAX or Optical) is suitable for short distances of max 10m.
Maybe, but what about 'very' short lengths? Say, 50cm or less?
Toslink!! http://theartofsound.net/forum/images/smilies/biggrin.gif
Not much wrong with that when done properly! :confused:
Stratmangler
17-09-2013, 22:13
Maybe, but what about 'very' short lengths? Say, 50cm or less?
The amplitude of the signal is likely to create reflections at non optimal junction points in the cable.
The amplitude of the signal is likely to create reflections at non optimal junction points in the cable.
Indeed!... so what should we do?
IMHO, silly lengths of cable that only need to connect equipment a few centimetres apart also have problems. what if the cable length didn't matter? How about a 20cm cable where reflections weren't an issue?
Would that be of interest to anyone? Is it possible? :popcorn:
Stratmangler
17-09-2013, 22:25
OK.
We're tied to having a single pair of conductors - to parallel anything up is going to add indefinable clocking issues into the equation.
To answer your last question, the answer is yes, it is of interest, but is it attainable?
Stratmangler
17-09-2013, 22:29
To push a bit more on the "silly" cable lengths thing - I've got a cable that's (IIRC) 3.6m long coiled up neatly behind my Squeezebox Touch and M2Tech Evo DAC, two devices that are physically very close, yet the cable length is of little consequence, as it's coiled up neatly out of the way.
To push a bit more on the "silly" cable lengths thing - I've got a cable that's (IIRC) 3.6m long coiled up neatly behind my Squeezebox Touch and M2Tech Evo DAC, two devices that are physically very close, yet the cable length is of little consequence, as it's coiled up neatly out of the way.
That's exactly what I thought for quite a long time, but 'apparently' there are issues with coiled cables too.... but that's another story. Probably.
Stratmangler
17-09-2013, 22:38
That's exactly what I thought for quite a long time, but 'apparently' there are issues with coiled cables too.... but that's another story. Probably.
Additional capacitance I'd have thought negligible.
The additional inductance likewise.
There is the potential for increased RF reception, as the coil could act as an aerial.
To answer your last question, the answer is yes, it is of interest, but is it attainable?
Indeed... that's what I think I might have found an answer to but I need some help proving the hypothesis! :)
It's a bit of a long story which I'll tell later, but it's getting late and I'm off to bed. :goodnight:
Chat amongst yourselves for a while. :wave:
Additional capacitance I'd have thought negligible.
The additional inductance likewise.
There is the potential for increased RF reception, as the coil could act as an aerial.
Yeah, it's all a bit weak, I agree.... but I'm trying to address this 'issue' of short cables and the 'problems' caused by reflections etc. Apparently it's a serious problem. :confused:
Stratmangler
17-09-2013, 23:17
I no longer have a coil of SPDIF coax cable between my Squeezebox Touch and M2Tech Evo DAC.
The cable connecting the two is the same cable, but it's no longer a coil.
I wouldn't care to state that it's an improvement, but the subjectively the L/R spread of the music I'm playing seems to have widened slightly.
Maybe a can of worms has been opened...
Maybe the SPDIF is far more susceptible to analogue states than previously thought ...
StanleyB
18-09-2013, 07:18
Not much wrong with that when done properly! :confused:
Which it is in the Bushmaster MKII. For high bitrate and sampling rate use the TOSLINK sockets need a dedicated low noise supply. Otherwise the nose will cause intermodulation with the signal.
As for coax: it is a digital signal, not an analogue one. A crystal controlled phase locked loop as used on a decent FM tuner and sliced by a schmitt trigger is one solution to take care of reflections. That's what I use, which negates the effect of cable lengths mismatch.
But what made you go for 1.5m rather than 'multiples of', if I may ask?
The simple fact that the actual run is 0.5m and I didn't want a huge run of cable behind the scenes!
sq225917
18-09-2013, 07:41
I see lots of speculation but not one thing showing any chance in the analogue output from the DAC by swapping cables, below a practical length. Any decent DAC, and that includes budget stuff will have no issue separating signal from reflections that are -80db down.
Its a none issue created in the fertile minds of cable company's and paranoid HiFi owners. Try swapping a few cables blind, and you'll reach the same logical conclusion.
tried that, kept putting the cable in the wrong socket...
The Black Adder
18-09-2013, 08:30
I see lots of speculation but not one thing showing any chance in the analogue output from the DAC by swapping cables, below a practical length. Any decent DAC, and that includes budget stuff will have no issue separating signal from reflections that are -80db down.
Its a none issue created in the fertile minds of cable company's and paranoid HiFi owners. Try swapping a few cables blind, and you'll reach the same logical conclusion.
I must admit I can't see how this hasn't come to light a lot earlier. All dacs should be able to cope with any length of cables, or if they can't then maybe each dac has an optimal length? If that was the case then all cdp's/dacs may have a different optimal length?
SPDIF (COAX or Optical) is suitable for short distances of max 10m. Toslink (optical) has a practical limit of 6.1M due to high light attenuation. BTW Marc grant HD1000 claims the ideal 75 ohm RCA connectors and cable. Can't see why lengths shorter than 1.5 m would affect sound though in practical terms less than 0.5m is not generally usable.
It's physically impossible for the RCA connection to be 75 Ohm due to the pin size. What is possible is to minimise the length at which impedance changes occur, which is what the Canare RCAs do. My DAC doesn't even come with RCAs, BNC only, so stanch is the designer's view on impedance matching.
StanleyB
18-09-2013, 08:54
I must admit I can't see how this hasn't come to light a lot earlier. All dacs should be able to cope with any length of cables, or if they can't then maybe each dac has an optimal length? If that was the case then all cdp's/dacs may have a different optimal length?
As long as we separate ticket price from technical construction we can then have something to go from. The last thing I want to do is to have to increase my own prices by say 5 or ten times before the techniques I use are considered as technically advanced. It's not rocket science to tackle reflections. They already do that in the bog standard Y-splitter.
I must admit I can't see how this hasn't come to light a lot earlier. All dacs should be able to cope with any length of cables, or if they can't then maybe each dac has an optimal length? If that was the case then all cdp's/dacs may have a different optimal length?
yes... agree...
I see lots of speculation but not one thing showing any chance in the analogue output from the DAC by swapping cables, below a practical length. Any decent DAC, and that includes budget stuff will have no issue separating signal from reflections that are -80db down.
Its a none issue created in the fertile minds of cable company's and paranoid HiFi owners. Try swapping a few cables blind, and you'll reach the same logical conclusion.
Yes, agree again, it should be this way, but....
Edit: Yes I noticed short RCA terminated 75Ohm sounded worse than a 1.5m version...
If neal or others hear a difference then maybe we should stick with 1.5m cable to be safe?
Has anyone else heard a difference??
Or could be Neal's dac that was more sensitive to feedback??
Please see my thread here: http://theartofsound.net/forum/showthread.php?21670-Esoteric-SPDIF-cable
The offer still stands whoever wants to try. Please also see: http://ar-t.co/ubyte.html
wee tee cee
18-09-2013, 09:06
tried that, kept putting the cable in the wrong socket...Aye, thats a lame excuse ya mucky bugger!!!!!
Feedback? :scratch:
If the SRL was truly -80db down then there wouldn't be a problem. I've still got my blind test T-Shirt, although the kit I used at the time is no longer with me.
Good thread this one, worth a read:
http://www.diyhifi.org/forums/viewtopic.php?f=2&t=1943
Maybe skip right to the very last post to cut out all the normal forum sh*t slinging...
Stan has reminded me, I must try the Toslink on the BM MKII....
It's physically impossible for the RCA connection to be 75 Ohm due to the pin size. What is possible is to minimise the length at which impedance changes occur, which is what the Canare RCAs do. My DAC doesn't even come with RCAs, BNC only, so stanch is the designer's view on impedance matching.
But then the Canare plug into a standard RCA socket...AFAIK somebody measured the Canare plugs on DIYAudio a long time back and questioned their claim to have a characteristic impedance of 75R
Feedback? :scratch:
If the SRL was truly -80db down then there wouldn't be a problem.
Good thread this one, worth a read:
http://www.diyhifi.org/forums/viewtopic.php?f=2&t=1943
Maybe skip right to the very last post to cut out all the normal forum sh*t slinging...
Stan has reminded me, I must try the Toslink on the BM MKII....
Ah yes DIYHiFi.org. :) "Jocko Homo" (Pat DiGiacomo) is an RF engineer turned digital engineer, hugely respected (people like Gordon Rankin from Wavelength turn to him for advice), and responsible for planting the seed of longer SPDIF cables into my head...
But then the Canare plug into a standard RCA socket...AFAIK somebody measured the Canare plugs on DIYAudio a long time back and questioned their claim to have a characteristic impedance of 75R
Yes I agree the Canare physically can't be 75R, but the plug is designed to minimise the lengths where it's not 75R. So it's making the best of an impossible situation...
As someone who makes digital cables, I still point out that this old chestnut about cable lengths is just that in terms of domestic audio applications, an old chestnut. Forget about length paranoia and fit the cable you have. If you get a signal, you get a signal. It's also true that there is NO true 75 Ohm RCA and that includes Canare, since the characteristic impedance is to do with the spacing between signal and signal return so by definition the standard RCA phono socket spacing determines this and not whether you use a crimped RCA plug or not.
I use short barrelled RCAs (Furutech FP126) for digital cables to minimise any mis-match purely as good practice but to be honest, it all probably makes very little difference. The quality of the cable is more important to me and there are some cables (particularly some of Chinese origin) purporting to be true 75 Ohm, but so badly manufactured that there is up to 0.5mm difference between signal and shield spacing along the cable length. Doesn't sound much but it's enough to throw constant 75 Ohm impedance out the window and more of an issue over longer lengths than the plugs at each end sometimes.
My advice is just use a decent quality 75 Ohm cable of any length you wish, make sure it's well terminated to minimise impedance mis-matches and then forget about it. There's enough signal redundancy in the data and sufficient competence in most DACs to ensure that the resultant signal reaches your amp without too much "missing" if anything.
sq225917
18-09-2013, 10:30
Thing is Jocko doesn't say use longer cables to attenuate reflections, he says it doesn't matter what length they are as long as the terminations are adequate. Signal reflection is likely down at 0.01% by the time it hits the dac for the 2nd time and that's regardless of cable length. The greater the impedance mismatch the greater the reflection coefficient. In any case it's a trivial none issue designed to get you to buy longer and more expensive fancy cables.
Thing is Jocko doesn't say use longer cables to attenuate reflections, he says it doesn't matter what length they are as long as the terminations are adequate. Signal reflection is likely down at 0.01% by the time it hits the dac for the 2nd time and that's regardless of cable length. The greater the impedance mismatch the greater the reflection coefficient. In any case it's a trivial none issue designed to get you to buy longer and more expensive fancy cables.
The point is that the terminations aren't adequate in general e.g. RCA and poorly designed SPDIF inputs / outputs, so longer cables help to address this. If anyone actually fancies measuring this stuff (I don't):
http://www.diyhifi.org/forums/viewtopic.php?f=2&t=1943&start=15#p41360
The Black Adder
18-09-2013, 10:51
It's strange that too because generally nobody really wants a load of flappy cabling hanging about. It's a pain to route especially if it has to drag on the floor past/near mains cabling.
If there is something audible in this then I guess I will just have to find out and buy a cheapo 1.5m and trust my ears. Or make one...
Right now I'm using optical.
What is possible is to minimise the length at which impedance changes occur, which is what the Canare RCAs do.
The Canare RCAs, as used in my Blue Jeans cable, is widely acknowledged to be good for digital cables.
The Canare RCAs, as used in my Blue Jeans cable, is widely acknowledged to be good for digital cables.
Widely acknowledged by whom? It's no better in reality than any short barrelled well designed RCA plug which has been carefully terminated. The main thing going for it is that it is relatively cheap and that the screen is crimped in a way that maintains (or is meant to) a consistent distance between signal and screen. In reality, it's not truly consistent as there will always be an impedance mis-match with the RCA standard phono geometry so many solder-type RCAs can be carefully terminated to be just as effective.
People worry too much about this. It really is a non-issue. If that worried, use 75 Ohm BNC kit or buy DACs that allow 75 Ohm BNC inputs. I've yet to hear any difference between BNC and a decent SPDIF connection but that's a whole other topic.
As someone who makes digital cables, I still point out that this old chestnut about cable lengths is just that in terms of domestic audio applications, an old chestnut. Forget about length paranoia and fit the cable you have. If you get a signal, you get a signal. It's also true that there is NO true 75 Ohm RCA and that includes Canare, since the characteristic impedance is to do with the spacing between signal and signal return so by definition the standard RCA phono socket spacing determines this and not whether you use a crimped RCA plug or not.
I use short barrelled RCAs (Furutech FP126) for digital cables to minimise any mis-match purely as good practice but to be honest, it all probably makes very little difference. The quality of the cable is more important to me and there are some cables (particularly some of Chinese origin) purporting to be true 75 Ohm, but so badly manufactured that there is up to 0.5mm difference between signal and shield spacing along the cable length. Doesn't sound much but it's enough to throw constant 75 Ohm impedance out the window and more of an issue over longer lengths than the plugs at each end sometimes.
My advice is just use a decent quality 75 Ohm cable of any length you wish, make sure it's well terminated to minimise impedance mis-matches and then forget about it. There's enough signal redundancy in the data and sufficient competence in most DACs to ensure that the resultant signal reaches your amp without too much "missing" if anything.
Well I disagree somewhat there Paul, you may as well use 50Ohm cable or one of the Belden cables I mentioned earlier the mismatch will be less and you can sleep well at night. Also, what is well terminated? Its a mismatch with the use of RCA so can never be well terminated.
On another point, Joko never inferred cable length would improve the return loss he was advocating the use of a sized length of cable to change the timing or arrival of the reflected signal so it had less impact durning the sample window. As I mentioned before if the termination where adequate then it really becomes an non issue.....it may already be as Stan suggested.
The Black Adder
18-09-2013, 13:25
Moving away slightly from the discussion regarding RCA connectors, would it be beneficial then to install BNC connectors instead of RCA's on the transport and dac? Surely then the connectors and cable will be more accurate?
I have only ever had one dac with BNC's and that was a Trichord.
Well I disagree somewhat there Paul, you may as well use 50Ohm cable or one of the Belden cables I mentioned earlier the mismatch will be less and you can sleep well at night. Also, what is well terminated? Its a mismatch with the use of RCA so can never be well terminated.
On another point, Joko never inferred cable length would improve the return loss he was advocating the use of a sized length of cable to change the timing or arrival of the reflected signal so it had less impact durning the sample window. As I mentioned before if the termination where adequate then it really becomes an non issue.....it may already be as Stan suggested.
I would consider "well terminated" to mean that it was terminated with care so that the deviation between the screen position and that small portion of it that needs soldering to the RCA barrel is minimised in length and that the central signal conductor is soldered very neatly and central in the signal pin solder bucket. Not all are. You'd be surprised. The deviation between 75 Ohms for a well soldered RCA of this type and a Canare type plug are really minimal, miniscule and not worth worrying about. The RCA socket though is not 75 Ohm so there will always be a slight mismatch here but the question is, how much difference does this really make to sound? Not a lot with a half decent DAC which almost all are nowadays.
Joe makes a valid point and that is why are not all DAC inputs of this type 75Ohm BNC? Well perhaps they should be but the fact that they're not and that the difference to sound being minimal is good enough for most people. There's more differences in the sound of DACs than there is between the cables connecting them IMHO, but as mentioned, if it matters that much, then there are DACs with 75 Ohm BNC inputs.
As the chap who (inadvertently) started this all off http://theartofsound.net/forum/showthread.php?27847-Bushmaster-Mk-II-(TC-7533)/page6 (picked up and run with by Mike) I hope soon to be able to report on an A/B demo.
Mike is currently making me a 1.5m length of my current 0.5m cable - same cable, same WBT plugs, same chap making them.
Clearly, won't be a double blind placebo controlled trial but may shed some light on whether the theoretical "issues" are audible or not. I'll probably stick to CD into the BM II and then headphones only to keep things as simple as possible.
Steve
Given how Stan engineers his DACs I think I can predict the outcome. ;)
sq225917
18-09-2013, 18:59
There's plenty of dacs with bnc sockets, it doesn't mean they are 75 ohms or that the termination to the pcb and the traces heading to the receiver chip continue at 75 ohms. Most designs don't...
Some interesting opinions here, chaps. But even more fascinating (to me) is how most people seem to be focused on what happens at the DAC end! :hmm:
Consider, for a moment, what these 'reflections' do on their way back UP the cable, and what happens when they get back to the transmitter! :)
I just ordered a 0.5m version of my current 5m cable that I use with the Bushmaster (Belden 1694A / Canare RCAs from Blue Jeans), so will have a play!
I’m finding this to be a rather interesting thread from an engineering point of view. In some ways I can only comment on half of the problem. Having spent thirty years in electronics research, predominately in microwave communications and radar, I am quite familiar with transmission lines, matching conditions and of mismatch reflections. What I am almost completely ignorant of is digital technology; so I have had to do a bit of research on the web – and I might have misunderstood it all. If I have, perhaps someone (StanleyB?) can put me right.
The CD transport (CDT) reads the disk and produces a stream of “1”s and “0”s. The “1”s are represented by a +ve voltage, and the “0”s by a –ve voltage (but not 0 V as might be assumed). The NRZ (“non return to zero”) frequency is 2.8224MHz for a sampling frequency of 44.1kHz (where does the factor of 64 arise?). This means a 0 → 1 or 1 → 0 transition might occur every 1/ 5.6448 .106 = 177.154 ns. However the transition is not instantaneous, it takes a small amount of time: the rise time (or “decision window”) for the change to occur. Usually the rise time quoted is the time taken to change from 10% to 90% or vice versa.
I cannot find the size of the rise time (or length of the decision window), but I have seen various figures quoted ranging from 1ns to 25ns. If we take the longer and worse case figure of 25ns, this suggests we are dealing with frequencies of ~ 14MHz (1ns ≡ 350MHz), so it is possibly valid and correct to consider the cable connecting the CDT to the DAC as a transmission line. To do this the cable length needs to be greater than the wavelength/2π, which for a typical cable with a 70% velocity factor corresponds to 2.4m.
Now the argument goes that if during the rise time (or decision window), a portion of the outgoing signal is reflected off the DAC and returns to the CDT, the timing of the final part of the change is corrupted in time, leading to a phenomenon called “jitter”.
Thus the ‘out and back’ journey must be greater than the rise time. Again taking a 70% velocity factor, the required cable length must be not be less than 0.7 (3.108). (12.5 10-9) = 2.62m. This length can be reduced pro rata for a corresponding shorter rise time.
All this argument is based on there being an impedance mismatch between the CDT, the cable and the DAC. If there was perfect matching, then any length of cable will do. Of course perfect matching will not occur, so how much of a mismatch is acceptable? Again the figures vary from an impossible –80dB (0.01% of the signal is reflected), to a much more reasonable –30dB (3% of the signal reflected).
And here lies the rub. For a typical coaxial cable having a characteristic impedance of 75Ω (such as RG59), to account for manufacturing tolerance there can be a ± 3Ω variation. Now assuming the CDT and DAT both have perfect output and input impedance respectively of 75Ω, the 3Ω tolerance of the cable results in a return loss of –34dB. This is all well and good, but the CDT and DAC have their impedances set by a simple resistor and are unlikely to be sufficiently close to 75Ω.
So if the return loss is poor (say –8dB to –20dB), the cable length is important.
If the return loss is good at one end and poor at the other, it is probably still better to use a long cable length.
It is now clear that the argument over RCA phono plugs and (75Ω) BNC plugs is a bit of a red herring. From the geometry of the RCA phono plug the characteristic impedance cannot be greater than 57Ω, and will be less, depending on the material used to support and insulate the centre pin from the coaxial outer conductor. So in practice the impedance could be anywhere between 30 and 40Ω (-7.36db to –10.3dB)*. And even if a true 75Ω BNC connector is used, it is unlikely the rest of the circuitry will maintain this impedance.
An alternative solution might be to improve the return loss through the use of an attenuator. Again SPDIF specifies an output level of 0.5 - 0.6V peak to peak, and a minimum input level of 0.2V. So if a 6dB attenuator is used the overall return loss will be reduced by 12dB.
* There are some manufacturers claiming to make 75Ω RCA phono plugs. They have done this by deviating from a coaxial geometry towards quasi-two conductor geometry. Whilst analysis might show that such geometry does have this impedance, it will be no different to that of a conventional connector once it is mated with the corresponding socket.
I’m finding this to be a rather interesting thread from an engineering point of view.
Glad you think so... :)
An alternative solution might be to improve the return loss through the use of an attenuator. Again SPDIF specifies an output level of 0.5 - 0.6V peak to peak, and a minimum input level of 0.2V. So if a 6dB attenuator is used the overall return loss will be reduced by 12dB.
Excellent... I'm getting to that. With a 'twist'!!! :) :)
* There are some manufacturers claiming to make 75Ω RCA phono plugs. They have done this by deviating from a coaxial geometry towards quasi-two conductor geometry. Whilst analysis might show that such geometry does have this impedance, it will be no different to that of a conventional connect once it is mated with the corresponding socket.
And that too! :D
BTW, My apologies for taking ages to get to the bloody point, I was on-call last night and tonight I've been making cables.... I've not had me tea yet! :rolleyes:
Thank you Barry that was a fascinating post and really helped break it down for me. I am aware of such things, but really understanding why is a different matter.
Well a nice summary of the DIYHiHi.org discussion I linked to earlier...
sq225917
20-09-2013, 06:54
The 64x comes from the master clock rate on the receiver, typically 64fs. Check out the Wolfson Wm8804 data sheet it is all explained in there.
Your numbers for attenuation are way off. -30db reflection loss at each end is very achievable, even in a $10 eBay DAC. So a figure of 0.01% for the round trip is spot on. Cable reflections are a none issue for any reasonable DAC, those expecting to do 24/192 over spdif need to try a 'little' harder but its still relatively trivial.
The 64x comes from the master clock rate on the receiver, typically 64fs. Check out the Wolfson Wm8804 data sheet it is all explained in there. Thanks - I'll do that. Does the data sheet give the rise time?
Your numbers for attenuation are way off. -30db reflection loss at each end is very achievable, even in a $10 eBay DAC. So a figure of 0.01% for the round trip is spot on. Cable reflections are a none issue for any reasonable DAC, those expecting to do 24/192 over spdif need to try a 'little' harder but its still relatively trivial.
The numbers quoted were taken from other web sites. -40dB return loss (required for the figure of 0.01%) implies a source output impedance and a DAC input impedance of 75 +/- 1.5 Ohm. I find it hard to believe that the impedances can be held to within 2% of the system impedance.
Cable reflections are important if the cable is to be regarded as a transmisssion line, which it ought to be under these conditions. Cable attenuation is unimportant though.
sq225917
20-09-2013, 15:40
But it's not a transmission line, and the reflections are unimportant as they are way below the signal level required to latch, even if they were a tenth of the signal they still wouldn't latch. Add onto this the fact that most dacs either fifo or ASRC and it's a total and utter none issue for anything but the worst implemented designs. If length of cables made an audible difference to what comes out of the other end of the dac then why isn't a clever cable brand cornering the market by marketing their cables as the solution to this problem?
That's a rhetorical question.
But it's not a transmission line, and the reflections are unimportant as they are way below the signal level required to latch, even if they were a tenth of the signal they still wouldn't latch. Add onto this the fact that most dacs either fifo or ASRC and it's a total and utter none issue for anything but the worst implemented designs. If length of cables made an audible difference to what comes out of the other end of the dac then why isn't a clever cable brand cornering the market by marketing their cables as the solution to this problem?
That's a rhetorical question.
Spot on. Cable lengths are the real red herring here. We all know that RCA isn' 75 Ohms and there's no such thing as the perfect RCA for the job (nor BNC for different reasons) but that slight mis-match we also know to be a non-issue. Even if Jitter were a real issue, almost all DACs these days have very good jitter correction/compensation clocks so it really is a non issue. What is surprising is that people continue to convince themselves that it is an issue when I'd like to bet that most use a 1m to 1.5m digital cable length with no problems at all. I'm more than happy to demonstrate to anyone that even 300mm long cable can be used with no discernable data drop out (ie audible). I've tried it.
Oldpinkman
20-09-2013, 16:32
Spot on. Cable lengths are the real red herring here. We all know that RCA isn' 75 Ohms and there's no such thing as the perfect RCA for the job (nor BNC for different reasons) but that slight mis-match we also know to be a non-issue. Even if Jitter were a real issue, almost all DACs these days have very good jitter correction/compensation clocks so it really is a non issue. What is surprising is that people continue to convince themselves that it is an issue when I'd like to bet that most use a 1m to 1.5m digital cable length with no problems at all. I'm more than happy to demonstrate to anyone that even 300mm long cable can be used with no discernable data drop out (ie audible). I've tried it.
Erm, is this a good moment to say I have been watching this thread with interest whilst listening to the awesome DaCapo -connected using half a £1 phono rca pair 60cm long with BNC adaptors at both ends? Sounds ok to me. ;)
I use the red wire - much better sounding than the white one :eyebrows:
Erm, is this a good moment to say I have been watching this thread with interest whilst listening to the awesome DaCapo -connected using half a £1 phono rca pair 60cm long with BNC adaptors at both ends? Sounds ok to me. ;)
I use the red wire - much better sounding than the white one :eyebrows:
Red wire = more full blooded sound, white wire = lighter sound :lol: :ner:
But it's not a transmission line, and the reflections are unimportant as they are way below the signal level required to latch, even if they were a tenth of the signal they still wouldn't latch. Add onto this the fact that most dacs either fifo or ASRC and it's a total and utter none issue for anything but the worst implemented designs.
I was probably being a bit too dogmatic in saying digital cables should be treated as transmission lines. To follow the fast rise time, the cable has to handle high frequencies (1ns rise time (10% to 90%) implies frequencies up to 350MHz). If the length of the cable is much greater than wavelength/(2*pi), the cable ought to be treated as a transmission line. The two-way delay time required so that the reflection from the DAC does not interfere with pulse edge, requires the cable to have a certain length, and this length is ~ wavelength/(2*pi); so treating the cable as a transmission line in this instance is probably borderline.
But if it's not a transmission line, then why is there the fuss over the system impedance of 75Ohm?
The interference caused by the reflection has nothing to do with the amplitude; rather it is, if I have understood it correctly, all to do with the uncertainty in the timing of the pulse edge, which causes “jitter”.
If length of cables made an audible difference to what comes out of the other end of the dac then why isn't a clever cable brand cornering the market by marketing their cables as the solution to this problem?
It would seem that there are companies specialising in cables that purportedly overcome this problem. One such manufacturer, mentioned in the DIYHiFi.org link, markets a cable that is 16’ long (nearly 5m)!
I can’t provide the definitive answer to how long the cable should be. It would seem to depend on the pulse rise time and whether the source and DAC are well matched or not. I don’t know the answers, as I’m not an expert or designer of digital electronics. All I was trying to do in my post was to independently support, or otherwise, the opinion of posters in the DIYHiFi.org thread, with calculations of my own.
My own personal experience with using separate external DACs, was that the cables used made no difference; the characteristic sound of the various DACs was preserved, and did not change, regardless of the cable used. I tried various cables, all of them having 75 Ohm impedance, with lengths varying from less than 0.5m to ~2m. The infererence being that my CD transport must have a fast enough rise time to the pulse edges so as to not need long cables
Out of curiosity, I did try a 1m long 'analogue' audio interconnect that did not have an impedance of 75 Ohm. Using that cable made a big difference to the sound - the upper registers sounded ‘tinny’, and there seemed to be a ‘smearing’ of the sound: the notes tended to run one into another with little space between them.
I was probably being a bit too dogmatic in saying digital cables should be treated as transmission lines. To follow the fast rise time, the cable has to handle high frequencies (1ns rise time (10% to 90%) implies frequencies up to 350MHz). If the length of the cable is much greater than wavelength/(2*pi), the cable ought to be treated as a transmission line. The two-way delay time required so that the reflection from the DAC does not interfere with pulse edge, requires the cable to have a certain length, and this length is ~ wavelength/(2*pi); so treating the cable as a transmission line in this instance is probably borderline.
But if it's not a transmission line, then why is there the fuss over the system impedance of 75Ohm?
The interference caused by the reflection has nothing to do with the amplitude; rather it is, if I have understood it correctly, all to do with the uncertainty in the timing of the pulse edge, which causes “jitter”.
It would seem that there are companies specialising in cables that purportedly overcome this problem. One such manufacturer, mentioned in the DIYHiFi.org link, markets a cable that is 16’ long (nearly 5m)!
I can’t provide the definitive answer to how long the cable should be. It would seem to depend on the pulse rise time and whether the source and DAC are well matched or not. I don’t know the answers, as I’m not an expert or designer of digital electronics. All I was trying to do in my post was to independently support, or otherwise, the opinion of posters in the DIYHiFi.org thread, with calculations of my own.
My own personal experience with using separate external DACs, was that the cables used made no difference; the characteristic sound of the various DACs was preserved, and did no change, regardless of the cable used. I tried various cables, all of them having 75 Ohm impedance, with lengths varying from less than 0.5m to ~2m. The infererence being that my CD transport must have a fast enough rise time to the pulse edges so as to not need long cables
Out of curiosity, I did try a 1m long 'analogue' audio interconnect that did not have an impedance of 75 Ohm. Using that cable made a big difference to the sound - the upper registers sounded ‘tinny’, and there seemed to be a ‘smearing’ of the sound: the notes tended to run one into another with little space between them.
Highlighted: about sums it up Barry :exactly:
Paul, what are your thoughts concerning my last paragraph?
Hmm... this is all getting even more interesting! :)
Before I start prattling on about what happened when someone contacted me about fitting RF attenuators to cables, are we saying that any cable 'designed' for digital transmission and having 75Ohm (or thereabouts) characteristic impedance will 'sound' the same regardless of it's length? Within practical reason of course! ;)
Hmm... this is all getting even more interesting! :)
Before I start prattling on about what happened when someone contacted me about fitting RF attenuators to cables, are we saying that any cable 'designed' for digital transmission and having 75Ohm (or thereabouts) characteristic impedance will 'sound' the same regardless of it's length? Within practical reason of course! ;)
Well that is my practical experience Mike.
I have also found that a coaxial cable normally used for analogue duties, and most certainly not having a characteristic impedance of 75 Ohm, sounded awful when used as digital interconnect. But there are some saying digital cables are not transmission lines and their characteristic impedance is irrelevant.
Paul, what are your thoughts concerning my last paragraph?
I agree with you Barry having experienced similar things myself. When I started experimenting (because the output from the DAC is what should be measured), I used a variety of cables, all set initially at 1.5m and all terminated with the same plugs (Canare and Neutrik originally). The cables ranged from 50 Ohms through to 75 (about 3 or 4 variations). The only cables that gave consistent results most noticeably in the HF region were the 75 Ohm cables and that seemed to be irrespective of whether BNC, Neutrik or Canare was used. I used some Van Damme 50 Ohm which sounded particularly horrible. I cannot say for sure whether this is due to cable impedance nor can I say that the 75 ohm cable sounded better because it WAS acting as a transmission line but my understanding is that for the lengths connected and the data rates, digital SPDIF doesn't act like a true transmission line. I am fully open to correction here as my understanding may be lacking.
This led me to consider other things such as cable shielding and sure enough, most of the cheaper cables used spiral shields, some were better than others, whilst the 75 ohm cables I used used braided shields. of those, the least dense shielding from cheap and cheerful RG cable didn't sound as good as the cables employing dense shielding, so I surmised from those experiments that GOOD SHIELDING is as vital to a decent digital cable, almost as much as the impedance.
The next thing tried was the use of a variety of 75 Ohm cables this time at 1m length. I tried cheaper Chinese sourced cables and carefully cut one up to measure accuracy of placement of signal conductor to shield. The worst used pretty horrible PVC dielectric and had significant inconsistencies in signal to shield, and by significant, I mean bordering on 0.5 to 1mm! Those were binned. I settled on the cable I now use because it employed really good shielding, and is a high quality cable because I do think that matters. The end result may be all but inaudible between many 75 ohm cables, but the proof for me was in listening tests between quality cables, awful 75 ohm cables and non 75 Ohm cables. The SQ difference between those and a decent cable was audible almost every time. I just wished that I had the means to measure and test the output from the DAC so that the reasons behind this and the effect on the signal could be more fully established.
The long and short of it Barry is that I think cable length doesn't matter as much as cable quality. I also believe in using shielded and high quality (in respect of electrical connectivity particularly) connectors. The end result is most definitely the sum of the parts when it comes to these things IME.
Well that is my practical experience Mike.
Yeah, that's what I initially thought too. But then I (foolishly?) made some cables for other folk and started to get some very good feedback, so tried it for myself. Now I'm not so sure! :scratch:
Cool, we're getting somewhere, impedance differences do make a difference.
The best "75 Ohm" cable will have a manufacturing tolerance of +/- 1.5 Ohm. Agree the shielding is important; braided shields are always better than lapped. I don't know of any RF cable using lapped shields.
All the cables I used were professional-grade RF cables. Some were triaxial designs, where the inner coaxial cable had a 75 Ohm impedance. One was a triaxial version of RG59. In my experience all RG cables from established cable manufacturers: Amphenol, Times Wire, Huber & Suhner, Trompeter, are all of high quality having a shielding braid with high optical coverage (at least > 92%) and can be trusted. RG223 for example, uses two outer braids in contact with one another; with very good shielding performance (that is, a low surface transfer impedance). Another professional grade RF cable is Belden 9907, this having a tight braided screen covered with a tinned-copper tape. The optical coverage is thus 100%.
The last two cables mentioned have a 50 Ohm impedance. Perhaps I ought to try them, to see (or hear!) if digital cables have to have a 75 Ohm impedance?
If the signal locks Barry you may well find that they work. How much data redundancy is lost (and therefore how much interpolation that the DAC is left with) is another matter. My own conclusions are that it will affect sound quality and not necessarily affect whether you signal lock or not. Interestingly, the only RG59 cable I had to hand was a cheap and cheerful unbranded version I had as an offcut from re-routing an aerial connection and the sheild coverage, braided as it was, was probably only 50% if that. No doubt at all that a decent quality RG59 is fine but I prefer to use something a little more flexible.
Another quick point: I don't use RF cable. It's capable of doing so (obviously), but that's not it's intended function. ;)
sq225917
20-09-2013, 18:51
Paul, maybe you could show measurements for some of these differences, i've never been able to get any differences out the back of any dac I've owned by swapping input cables- and boy have i tried.
Paul, maybe you could show measurements for some of these differences, i've never been able to get any differences out the back of any dac I've owned by swapping input cables- and boy have i tried.
If I had the means, trust me I would Simon. I'm as keen as you or anyone to understand why. I fully admit that I do not have the evidence to explain it but I have heard differences and I've heard these using more than one DAC. Then again, I've gone through many dozens of cables and connectors when trying to pick one that sounded best to me for the job. Having done that, I'm less interested now in the reasons and content that I've picked a decent quality match that does the job admirably. Perhaps though it would be a great and very telling experiment of someone has the requisite software to try some signal analysis on the back end of the DAC (output) for various cable and connector combos at the input.
I've got to admit that I always thought that if the transport (or whatever) sent; "011010" and the DAC read; "011010" then that was it, job done.
TBH, I'm still trying to hang on to that thought, but 'recent' (and still going on after several years!) experiences have caused me to question that belief.... I'm at a complete loss to explain what's going on though! :scratch:
I've got to admit that I always thought that if the transport (or whatever) sent; "011010" and the DAC read; "011010" then that was it, job done.
TBH, I'm still trying to hang on to that thought, but 'recent' (and still going on after several years!) experiences have caused me to question that belief.... I'm at a complete loss to explain what's going on though! :scratch:
That is more or less it, but as with all these things its not quite so simple. As I understand it, there's up to 8 times redundancy to account for things like jitter (so you have 8 "0's" and 8 "1's") and for those zeros and ones which dont make it, the DAC has the ability to interpolate bvetween missing bits...up to a point as when so much of the data stream becomes corrupted, the DAC cannot do its job and the signal drops out. Its the in-between state that interests me and to say that cables make no difference at all is something I'd contest as clearly there is a state between a "too corrupted" data stream and a perfectly intact data stream, so reason would suggest that things like cables matter. However, that's still conjecture and Simon has a point, it does need measuring.
At this point I wouldn't be surprised if 'someone' chimed in about the 'so called importance' of measurements! :D
Oldpinkman
20-09-2013, 19:39
At this point I wouldn't be surprised if 'someone' chimed in about the 'so called importance' of measurements! :D
Not me mate. I'd get banned. I happen to know a particularly good digital engineer i could ask but I see no point. I trust my ears :)
wee tee cee
21-09-2013, 09:10
I tried several co-ax cables with my Dac last night.......I know its ones and zeros but everyone of them sounded different. I a cant and have no intention of trying to explain why, But if you enjoy your music more with one than another that's what counts in my book.
Oldpinkman
21-09-2013, 11:32
I apologise. I mislead the forum. I've just been to have a look and check
It was a video cable
I stripped off the red and the white to be tidier
It was the yellow. This may account for the sound. :)
I know you won't believe me - but I had a listen too, and it sounded pretty damn good to me . The choice of cable was due to impatience and expediency (just picked it up from Norfolk at the time) rather than machismo, and I haven't tried any other cable which therefore might be better, but the sound is amazing, and EXACTLY how I remembered it 20 years ago.
http://i1315.photobucket.com/albums/t582/oldpinkman/20130921_1215021_zps776c1642.jpg (http://s1315.photobucket.com/user/oldpinkman/media/20130921_1215021_zps776c1642.jpg.html)
The 1.5m cable arrived from Mike this morning and I've been listening for the last couple of hours and I actually think that there is a difference between the 0.5m one! It's not night and day but definitely there.
I can detect no difference in tone, pace or rhythm but I do believe the timing is sharper. Vocals and the majority of instrumental sound the same to me but there is a subtle difference in percussion especially fast hard drum kit which has more attack and definition particularly noticeable when the sticks are struck together or on a wood block or the rim. Cymbals/high hat have a bit more "zing and ting". Stereo imaging between the speakers is tighter focused.
As I say these changes are subtle - nowhere near the differences I noted between the cables Mike supplied and others I've tried (the QED quenex sr 75 by comparison is flat and dull - how did this get a What HiFi 5 star product of the year award?).
I know nothing about electronics and assumed that a digital signal of 0's and 1's ought to sound the same regardless of cable type/length but my ears tell me differently. I'll keep this cable in for a few days and put the 0.5m back in and see if I've been deluding myself but I don't think so.
Steve
Oldpinkman
21-09-2013, 14:53
I have ordered a 1m bnc bnc 75ohm van damme since all my mates are incommunicado for now. Be interesting to see if it makes a difference. I doubt it, not so much because of my issues with other cables, but because it sounds so good now, and so much like I remember it from 20 years ago, there really doesnt seem scope for improvement. But quicker to "suck it and see" on this occasion than seek advice.
I have ordered a 1m bnc bnc 75ohm van damme since all my mates are incommunicado for now. Be interesting to see if it makes a difference. I doubt it, not so much because of my issues with other cables, but because it sounds so good now, and so much like I remember it from 20 years ago, there really doesnt seem scope for improvement. But quicker to "suck it and see" on this occasion than seek advice.
If you'd looked through the thread it's 1.5m you should have ordered. :P :D
My 0.5m cable arrived, will have a play around between this and the 5m cable over the next few days.
Oldpinkman
24-09-2013, 11:52
If you'd looked through the thread it's 1.5m you should have ordered. :P :D
My 0.5m cable arrived, will have a play around between this and the 5m cable over the next few days.
Hmm - I thought the thread conclusion was size didn't matter - or am I confusing another recent discussion? I wasn't worried about lengths (surely if that had mattered one or two of the nobs in the know who manufacture dacs would have bought the subject up). I was more worried that I had gone a bit too philistine and connected up something sufficiently wrong to create a real problem. Should have been here today - but not in the post. However, a quick burst of rationality tells me that the old "trust your ears" means I am likely to have been fretting needlessly about a wholly unsuitable cable, cos it sounds fine.
In support of this I have now established the connectors are 75ohm, and any shortcomings the RCA's may have theoretically are unlikely to matter in practice. That and the fact that when we rummaging round in Norfolk for a cable just to "have a listen" nobody suggested to me a standard RCA interconnect wouldn't do the job (we were just thwarted by the fact the dac only had a BNC and not rca socket)
So - I'll watch the post, but not expecting to get as excited as I had originally hoped. :(
Well it is a very good sign that your DAC has a BNC input as it suggests that the PCB traces etc. in the DAC are also 75 ohm. It should matter a lot less than RCA. In any case I will try to find an area where I can reliably tell the difference between my 0.5 and 5m cable and, in my system, would be happy to arrange an ABX test. I don't believe in such tests in unfamiliar systems and environments as there are too many disruptive variables and auditory memory limitations.
Stratmangler
24-09-2013, 16:43
Back to the OP - what thoughts did you have when you started the thread Mike?
We're on the OP? Talking about digital cable lengths and stuff, but yes, interested to hear some more thoughts from Mike too.
Edit: ah I get it, back to the original poster, not the topic... Sorry!
Oldpinkman
26-09-2013, 12:41
New van Damme 75ohm proper bnc to bnc arrived today, and very pretty it is too.
And it fits
And it works - I have music coming out of the DaCapo. Just like I did before.
(For all cable fans I should note I have plugged it in and listened to 10 minutes of Katie M which was in the tray at the time. This is not a "no difference between cables" soak test.)
But - my flaky old RCA cable with a couple of adaptors clearly wasn't that flaky. The van damme is much prettier though, so I'll leave it in :)
So what is the conclusion? Does length matter? Does it matter if the cable is not of 50 Ohm characteristic impedance? Does length matter depending on the rise and fall time of the sampling waveform?
Question, questions, questions ......
Oldpinkman
07-10-2013, 18:56
So what is the conclusion? Does length matter? Does it matter if the cable is not of 50 Ohm characteristic impedance? Does length matter depending on the rise and fall time of the sampling waveform?
Question, questions, questions ......
Shouldn't that be not of 75 ohm?
In my humble experience size doesn't matter. My little one seems just fine :)
Shouldn't that be not of 75 ohm?
In my humble experience size doesn't matter. My little one seems just fine :)
Yes, sorry - I should have written 75 Ohm characteristic impedance. It's been a long day....
But as you say, size doesn't matter; it all depends on how it is used! :eyebrows:
Well unlike Richard, who just changed the cables and thought ho-hum, I have been listening to a single track for the past 20mins with my 5m cable after some quicker comparisons with my (otherwise identical) 0.5m cable. Luckily it is track 1 from "Wu Tang Forever". Will report back.
Things are a tad more natural with the 5m cable, the 0.5m is a bit hotter, some digital sheen. Piano decays, Joni Mitchell's vocal sibilance etc. are a bit more realistic and less digital-sounding with the longer cable. Having said this it's not the sort of thing readily identifiable when A/B testing, more like it sinks in as I spend time with the different cables. Again to reiterate I think the differences are always subtle unless there's a deliberate flavouring added by a manufacturer. The problem is that long-term listening is really the true judge of what an individual wants from their system, so these comparisons in a way are a bit moot.
Oldpinkman
08-10-2013, 05:57
Things are a tad more natural with the 5m cable, the 0.5m is a bit hotter, some digital sheen. Piano decays, Joni Mitchell's vocal sibilance etc. are a bit more realistic and less digital-sounding with the longer cable. Having said this it's not the sort of thing readily identifiable when A/B testing, more like it sinks in as I spend time with the different cables. Again to reiterate I think the differences are always subtle unless there's a deliberate flavouring added by a manufacturer. The problem is that long-term listening is really the true judge of what an individual wants from their system, so these comparisons in a way are a bit moot.
Surely this is subjectivity taken to the extremes of fantasy. A digital cable catches the 0s and 1s or drops them. No manufacturer can add a flavour. If digital cables make a difference- then its binary. They fuck up, or they don't. *You don't get to custom build flavour.
*ok not strictly true. They can fuck up a tiny bit or a bigger bit. You still can't custom build "flavours"
Surely this is subjectivity taken to the extremes of fantasy. A digital cable catches the 0s and 1s or drops them. No manufacturer can add a flavour. If digital cables make a difference- then its binary. They fuck up, or they don't. *You don't get to custom build flavour.
*ok not strictly true. They can fuck up a tiny bit or a bigger bit. You still can't custom build "flavours"
I was just reiterating my general stance on cables, sorry I wasn't more specific. Having said this you are missing a vital part of those 1s and 0s, which is the analogue component of the signal (clock). Sending the data is unquestionably the easy bit, but there is such a thing as cable jitter, which can arise due to signal reflections or, in the case of Toslink in particular a lack of bandwidth that 'rounds off' / smears the signal... there's a lot of stuff out there that shows it's measurable, but people never seem to talk about RF transmission & clock signals, just those 1s & 0s...
I was just reiterating my general stance on cables, sorry I wasn't more specific. Having said this you are missing a vital part of those 1s and 0s, which is the analogue component of the signal (clock). Sending the data is unquestionably the easy bit, but there is such a thing as cable jitter, which can arise due to signal reflections or, in the case of Toslink in particular a lack of bandwidth that 'rounds off' / smears the signal... there's a lot of stuff out there that shows it's measurable, but people never seem to talk about RF transmission & clock signals, just those 1s & 0s...
Any (most) decent DACs will sort out jitter. It's a bit of a hairy old chestnut I'm afraid. It does exist but part of the DACs job, long with the clock signal is to re-build the signal exactly as it left the transport, and to most intents and purposes, they do exactly that. Whilst it's purportedly noted that there are some data drop-outs with impedance mis-matches, many people forget that there is up to 8 times redundancy of the signal information (the 0's and 1's). In the event that most or all of that redundancy is taken up through data loss for a tiny part of the signal, signal interpolation plugs the gap, and it'd be a braver person than I to say for sure they're hearing differences due to data drop out. Richard has a point...it is largely subjective and as you admit, subtle. If it's that subtle, one has to ask, is it relevant and does it matter? Especially so when I personally find there to be much larger deviations in DAC implementation and sound quality....if people expermented with various DACs they'd soon see that DACs matter far more than the connecting cables.
Oldpinkman
08-10-2013, 11:19
Any (most) decent DACs will sort out jitter. It's a bit of a hairy old chestnut I'm afraid. It does exist but part of the DACs job, long with the clock signal is to re-build the signal exactly as it left the transport, and to most intents and purposes, they do exactly that. Whilst it's purportedly noted that there are some data drop-outs with impedance mis-matches, many people forget that there is up to 8 times redundancy of the signal information (the 0's and 1's). In the event that most or all of that redundancy is taken up through data loss for a tiny part of the signal, signal interpolation plugs the gap, and it'd be a braver person than I to say for sure they're hearing differences due to data drop out. Richard has a point...it is largely subjective and as you admit, subtle. If it's that subtle, one has to ask, is it relevant and does it matter? Especially so when I personally find there to be much larger deviations in DAC implementation and sound quality....if people expermented with various DACs they'd soon see that DACs matter far more than the connecting cables.
:exactly:
:exactly:
For the record, I would have to agree. My nights spent comparing 3x DACs have been VERY clearly audible compared to my SPDIF cable testing this week, which I concede is something that's really not worth worrying about in the grander scheme of things. Still some more listening to do, but in any case, moving my speaker a few inches, a DAC change etc. yields far more important and defined results.
Last time I looked there was a single Even Parity bit for every 32 bit sub-frame in the AES/SPDIF spec. Not sure where the 8 times redundancy comes from Paul, unless you are referring to CD CIRC which applies to the reading of the disk not SPDIF. Personally I've found cable length and correct termination of the end points beneficial in more than a subtle way with SPDIF, YMMV.
Oldpinkman
09-10-2013, 04:10
I was brung up to use coax. Made easier for me this time round since the only digital out jw saw fit to put on the cd6 I am using as transport is a bnc :)
Powered by vBulletin® Version 4.2.3 Copyright © 2026 vBulletin Solutions, Inc. All rights reserved.