I don't think I argued it never matters (nobody here did) but that there are cases where it is far less important because the transmission system itself is very robust, S/PDIF being one example.
Real high speed databusses like HD-SDI, DVI, HDMI, SATA need a well matched transmission line, no arguing about that, altough these standards employ tricks like EQ on the (differential) datalines that really help with keeping the signal integrity, this is all 100% analog transmission, carrying digital information...
More to the point, spdif if done right has a really slow edge rate compared to something like HD-SDI (1.55Gb/s) or sata, and the edge rate determines the smallest feature that needs to be considered as a transmission line. For spdif this is always larger then any of the likely connectors, so even a connector that is very wrong (Phono plug) will do just fine.
If you are genuinely hearing differences between cables, then I would be giving the clock recovery circuits, pll loop bandwidth and lock range the hairy eyeball, doing it really right means locking a low phase noise VCXO to the recovered sclk with a slow loop filter (Above loop filter cutoff you get the PN from the VCXO, below you get the PN coming down the wire). Note that low PN implies a narrow tuning range on the VCXO.
The studio gear design cats have a simple way to design kit to make the gear insensitive to jitter: stick an asrc between the line rx and the DAC! Modern asrc chips have better performance then the eventual DAC, and use a delay locked loop with a small buffer in the rate estimator so that a modest amount of jitter gets efficiently filtered without interfering with rate estimate.
We are increasingly seeing this in boat ADCs and DACs used in studios as it pretty much just works.
It has its limits (Which are easily met) but is very effective as a local rock on the converter board will always have lower phase noise then a pll recovered clock.
Regards, Dan.
If you are genuinely hearing differences between cables, then I would be giving the clock recovery circuits, pll loop bandwidth and lock range the hairy eyeball, doing it really right means locking a low phase noise VCXO to the recovered sclk with a slow loop filter (Above loop filter cutoff you get the PN from the VCXO, below you get the PN coming down the wire). Note that low PN implies a narrow tuning range on the VCXO.
The studio gear design cats have a simple way to design kit to make the gear insensitive to jitter: stick an asrc between the line rx and the DAC! Modern asrc chips have better performance then the eventual DAC, and use a delay locked loop with a small buffer in the rate estimator so that a modest amount of jitter gets efficiently filtered without interfering with rate estimate.
We are increasingly seeing this in boat ADCs and DACs used in studios as it pretty much just works.
It has its limits (Which are easily met) but is very effective as a local rock on the converter board will always have lower phase noise then a pll recovered clock.
Regards, Dan.
hmm, I used cheap RGB6 cable with 70% aluminum shield and compression rca jacks for my 25 foot digital cable run from my TV to my ancient Yamaha RX-V992 amp.
No noise, works great.
That cable is laying right with L/R & Center speaker cable runs and the last 4 feet it passes a wall outlet, with a power strip where the computer, monitor, 12v light and stereo gear all meet.
Quite often people tend to over think stuff. If it was complicated it would have never became a standard.
No noise, works great.
That cable is laying right with L/R & Center speaker cable runs and the last 4 feet it passes a wall outlet, with a power strip where the computer, monitor, 12v light and stereo gear all meet.
Quite often people tend to over think stuff. If it was complicated it would have never became a standard.
The rise time of the spdif signal will be constant for all clock rates as it is not dependent on the clock rate but on the driver used, unless a circuit switches drivers depending on the bit rate which is very unlikely.
NOTE: The edge rate is faster than the specification as the specification is the worst case allowable rise time.
A 74AHC04 has a specified worst case rise/fall time of less than or equal to 3ns (NXP Datasheet 15, 50pf load full temp range).
A 74VHC14 (ST Micro) shows the same specifications.
In reality the rise/fall time is closer to 1ns and probably even less since our applications will not operate any where near the allowed temperature extremes.
A good rule of thumb would be to use 500ps as the rise/fall time. Most circuits I've seen posted used a 74AHC04, 74VHC14, or similar driver.
This is at the driver pin, not at the output connector.
NOTE: The edge rate is faster than the specification as the specification is the worst case allowable rise time.
A 74AHC04 has a specified worst case rise/fall time of less than or equal to 3ns (NXP Datasheet 15, 50pf load full temp range).
A 74VHC14 (ST Micro) shows the same specifications.
In reality the rise/fall time is closer to 1ns and probably even less since our applications will not operate any where near the allowed temperature extremes.
A good rule of thumb would be to use 500ps as the rise/fall time. Most circuits I've seen posted used a 74AHC04, 74VHC14, or similar driver.
This is at the driver pin, not at the output connector.
Dan, do you see this done on HD video lines, too? Or is it not cost effective. I don't know what's in the HD-SDI boxes I use, I know they just work!
Classical ASRC at a HD-SDI bitrate would be hard, for video we usually work in terms of entire fields (or frames, depends) rather then lines or pixels.....
Besides the only places small amounts of jitter matter are when the device on the end of the link is a badly designed AD or DA converter and you are using a self clocked link, otherwise as long as the eye pattern is ok there is no need to worry.
SDI line extenders usually just use a pll to regenerate the clocks and then a transmitter to put reclocked data onto the output, they are usually fairly dumb.
SDI is kind of fun to play with if you have a reasonably serious fpga dev kit.
Regards, Dan.
Besides the only places small amounts of jitter matter are when the device on the end of the link is a badly designed AD or DA converter and you are using a self clocked link, otherwise as long as the eye pattern is ok there is no need to worry.
SDI line extenders usually just use a pll to regenerate the clocks and then a transmitter to put reclocked data onto the output, they are usually fairly dumb.
SDI is kind of fun to play with if you have a reasonably serious fpga dev kit.
Regards, Dan.
Hmmm ... a bit vicious,
... I note a category in Wikipedia for such: http://en.wikipedia.org/wiki/Category:Quantum_electronics, which has 'real' things in it ... Well!!!!!!!!
Some interesting posts, considering one of the first things put up were links to the likes of Howard Johnson and Eric Bogatin, but obviously you can point an audiophile to knowledge, etc etc.
Digital signal transfer, signal integrity and EMC are all well studied phenomena, we don't even have to make stuff up there is so much info out there regarding it, why do more silly myths and BS have to be created...
I second DF96, when quantum is mentioned in relation to audio sound reproduction you know a big barrel of BS is heading towards you at a very fast rate....
For the silver supporters, lots of hot air but NO CONTENT, how does silver improve digital signal transmission...
Some interesting posts, considering one of the first things put up were links to the likes of Howard Johnson and Eric Bogatin, but obviously you can point an audiophile to knowledge, etc etc.
Digital signal transfer, signal integrity and EMC are all well studied phenomena, we don't even have to make stuff up there is so much info out there regarding it, why do more silly myths and BS have to be created...
I second DF96, when quantum is mentioned in relation to audio sound reproduction you know a big barrel of BS is heading towards you at a very fast rate....
For the silver supporters, lots of hot air but NO CONTENT, how does silver improve digital signal transmission...
Hmmm ... a bit vicious,
umm, I dont think that page says what you think it does
Sorry, obviously a poor attempt to point out that that some others are not so averse to associating the word "quantum" with the word "electronics", .
In other words, there is a whole variety of electrical phenomena, linked to on that page, where the behaviour is considered to have quantum characteristcs ...
.In other words, there is a whole variety of electrical phenomena, linked to on that page, where the behaviour is considered to have quantum characteristcs ...
Some confusion there.
RCA is not 50R, or anything in particular.
Impedance controlled connectors (50R for instruments and wireless transmission, 75R for video) are built to the correct mechanical dimensions and dielectric parameters to meet the impedance specs. Ditto with coaxial cables.
RCAs were "designed" as a cheap and cheerful audio connector, not needing impedance matching. They happened to be close to 75R (but that would depend on who built them) so got used for video connections in consumer gear as well.
RCA is not 50R, or anything in particular.
Impedance controlled connectors (50R for instruments and wireless transmission, 75R for video) are built to the correct mechanical dimensions and dielectric parameters to meet the impedance specs. Ditto with coaxial cables.
RCAs were "designed" as a cheap and cheerful audio connector, not needing impedance matching. They happened to be close to 75R (but that would depend on who built them) so got used for video connections in consumer gear as well.
The Only Thing Worse Than Being Talked About... Is Not Being Talked About....
Dan.
Yes, IIRC Jocko had words to say about the importance of drive and termination impedances, and stronger words about particular cable length in relation to cable propagation velocity...think arrival times of multiple reflections at the receiving end.
Dan.
Last edited:
That seems to be mainly physics or applications of physics, so genuine quantum stuff. Note that real physics is amazing enough, so why do people have to invent stories? I know, it's because they don't understand the real stuff.fas42 said:
RCA is not 50R. It is not constant impedance at all. However it may, at sufficiently low frequencies, be a rough approximation to 50R. To make it look more like a rough (even rougher!) approximation to 75R you would need to add a little inductance in series with the centre pin. Some people may do this accidentally with their wiring.Triodethom said:
75R, as has already been said, is the standard for video and the same drivers can be used for digital audio.
120R can be fairly easily obtained in twisted pair - I think. Generally, twin cables have higher impedance than coax because there is less capacitance between the conductors.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.