Hi all
I have 2 Qs here that I know must seem very basic, but need help sorting it out in my head.....
Firstly:
I have a sound card - m-audio audiophile 192. It outputs SPDIF over a coax connection. I have a DMM and oscilloscope. Can someone tell me how can I measure the SPDIF output and what it should be... ie 0.5V square wave, sine wave, should it be 2V and so on..... yes I am a beginner....
Second:
Same SPDIF - and the 75R impedance question. I always think of impedance as resistance to flow, makes sense right? And I can imagine that a particular cable might have a certain impedance (kinda like the way a cable will have some capacitance too). But how is that measured, how is 75R arrived at and exactly what is meant by the 75R? I mean, do they take a standard length of this cable, and measure the attenuation of a signal at a certain frequency or something?
Why is a RCA connector not in compliance with this impedance but a BNC connector is?
I know they seem like stupid questions - but I see it often cropping up in arguments and I just don't get it. Seems like a basic thing I should get clear
Fran
BTW, I did google this but didn't find the answers, and really just found quite a bit fo sales talk as to why I should buy x, y or z cable.
I have 2 Qs here that I know must seem very basic, but need help sorting it out in my head.....
Firstly:
I have a sound card - m-audio audiophile 192. It outputs SPDIF over a coax connection. I have a DMM and oscilloscope. Can someone tell me how can I measure the SPDIF output and what it should be... ie 0.5V square wave, sine wave, should it be 2V and so on..... yes I am a beginner....
Second:
Same SPDIF - and the 75R impedance question. I always think of impedance as resistance to flow, makes sense right? And I can imagine that a particular cable might have a certain impedance (kinda like the way a cable will have some capacitance too). But how is that measured, how is 75R arrived at and exactly what is meant by the 75R? I mean, do they take a standard length of this cable, and measure the attenuation of a signal at a certain frequency or something?
Why is a RCA connector not in compliance with this impedance but a BNC connector is?
I know they seem like stupid questions - but I see it often cropping up in arguments and I just don't get it. Seems like a basic thing I should get clear
Fran
BTW, I did google this but didn't find the answers, and really just found quite a bit fo sales talk as to why I should buy x, y or z cable.
hi, this is the single most useful link:
http://www.rane.com/pdf/ranenotes/Interfacing AES3 and SPDIF.pdf
btw everyone is about dpll reclocking or asrc or direct i2s out hdd player nowdays, and i can understand this after looking at the miller analyzer measurements.
Also can someone point me to schematic of "Jocko's spdif receiver" ? I need it for aes ebu though.
http://www.rane.com/pdf/ranenotes/Interfacing AES3 and SPDIF.pdf
btw everyone is about dpll reclocking or asrc or direct i2s out hdd player nowdays, and i can understand this after looking at the miller analyzer measurements.
Also can someone point me to schematic of "Jocko's spdif receiver" ? I need it for aes ebu though.
At radio frequencies, a transmission line with a certain geometry will have a "characteristic impedance."
The way I think of it is for a cable with a 75R characteristic, placing a 75R resistor at one end and transmitting an RF signal at the other end, 75R will cause the least reflection.
The RCA connectors do not have the correct geometry to be 75 ohm, and will cause an impedance "lump", or reflection as the signal passes it, much like light waves from glass to air.
I believe that BNC connectors can be purchased in both 50R and 75R styles.
Much to the disappointment of my professor Electromagnetic field theory , I could not give you the theory off the top of my head.
HTH
Doug
FYI, Good paper above.
Noobie talk :
At high frequencies (people use to say RF for Radio Frequencies) all cables behave (or misbehave) as transmission lines.
Get a long pipe and talk into it, that's a transmission line. If it is long enough, you'll hear some echo coming back from the other end. That's a problem.
Whenever you send some stuff in a transmission line it will always travel at a certain velocity (the speed of sound in the case of you talking in a pipe, the speed of light modulo impedance, dielectric constant etc, in a cable).
So, if your signal is fast enough, that this propagation effects are not negligible anymore, you'll say it's a transmission line, and you have to treat it like this. It is not something difficult, just a part of your design.
For instance take a SATA cable. If you send some low frequency signals in it, it's a cable. If you send 3 billion bits per second in it, it's a transmission line, since there are several bits in the cable at any given time. Of course you'll want to minimize echoes if you wanna use that to transmit some useful information.
So if you put a resistor at the end of the cable, which has a certain value, you will get optimal damping and optimal tranfer of power : all the power that flows in the transmission line will end up in the resistor instead of bouncing back to the source and messing everything. The value of this resistor is your characteristic transmission line impedance. It can be calculated (depending on geometry etc) or it can be measured.
Of course an impedance discontinuity like someone stepping on your cable, a too-sharp kink, or a mis-specified RCA connector (those were never specified for anything anyway) will cause some amount of signal reflection. Depending on what you use the cable for, and the size of the impedance discontinuity, the effects might range from none to fatal.
Also what the SPDIF looks like on your scope depends on wether the scope input is the same impedance as your transmission line (it's written on your scope) ; if it isn't you'll get ringing which is nothing more than the signal bouncing back and forth in the cable.
Bottomline : to transport high frequencies with a good quality of signal and minimum leaking to close sensitive analog audio circuits, use parts specified for this use, as all RF designers have been doing since the dawn of time, ie. BNC or SMA connectors and coax of the proper impedance.
> Why is a RCA connector not in compliance with this impedance but a BNC connector is?
Because the BNC connector was specced and designed for this specific purpose, and the RCA was not. Specs on a RCA connector are nothing more than "make contact".
At high frequencies (people use to say RF for Radio Frequencies) all cables behave (or misbehave) as transmission lines.
Get a long pipe and talk into it, that's a transmission line. If it is long enough, you'll hear some echo coming back from the other end. That's a problem.
Whenever you send some stuff in a transmission line it will always travel at a certain velocity (the speed of sound in the case of you talking in a pipe, the speed of light modulo impedance, dielectric constant etc, in a cable).
So, if your signal is fast enough, that this propagation effects are not negligible anymore, you'll say it's a transmission line, and you have to treat it like this. It is not something difficult, just a part of your design.
For instance take a SATA cable. If you send some low frequency signals in it, it's a cable. If you send 3 billion bits per second in it, it's a transmission line, since there are several bits in the cable at any given time. Of course you'll want to minimize echoes if you wanna use that to transmit some useful information.
So if you put a resistor at the end of the cable, which has a certain value, you will get optimal damping and optimal tranfer of power : all the power that flows in the transmission line will end up in the resistor instead of bouncing back to the source and messing everything. The value of this resistor is your characteristic transmission line impedance. It can be calculated (depending on geometry etc) or it can be measured.
Of course an impedance discontinuity like someone stepping on your cable, a too-sharp kink, or a mis-specified RCA connector (those were never specified for anything anyway) will cause some amount of signal reflection. Depending on what you use the cable for, and the size of the impedance discontinuity, the effects might range from none to fatal.
Also what the SPDIF looks like on your scope depends on wether the scope input is the same impedance as your transmission line (it's written on your scope) ; if it isn't you'll get ringing which is nothing more than the signal bouncing back and forth in the cable.
Bottomline : to transport high frequencies with a good quality of signal and minimum leaking to close sensitive analog audio circuits, use parts specified for this use, as all RF designers have been doing since the dawn of time, ie. BNC or SMA connectors and coax of the proper impedance.
> Why is a RCA connector not in compliance with this impedance but a BNC connector is?
Because the BNC connector was specced and designed for this specific purpose, and the RCA was not. Specs on a RCA connector are nothing more than "make contact".
BNC connectors are available in both 50 and 75 ohm flavors.
A 75 ohm link will have a 75 ohm source, connectors all specified to maintain a 75 ohm impedance over the range of frequencies of interest, cables with a characteristic impedance of 75 ohms and a terminating impedance of 75 ohms.
Every discontinuity in the path results in some of the energy being reflected back to the transmitting end where it is once again reflected due to an imperfect termination. The same thing happens at each connection along the way and at the final termination as well. Well designed transmission lines will have very small percentages of reflected signal.
RCA jacks are not 75 ohm devices (caution there are some reputedly 75 ohm RCA jacks, very expensive and I have not tried them) for a number of reasons including the pin diameter, the dielectric used, and the distance/spacing to the outer conductor of the jack.
You want to minimize the number of discontinuities in the path and make sure all connections maintain the proper impedance, which incidentally with most devices is going to be impossible. The spdif connection point on the end of the shigaclone is not 75 ohms characteristic impedance, and arguably is far enough from the output port of the dsp for reflections to be a potential issue. (Reflections before the cable matching network.) Mine are mounted on a connector that plugs into the socket, and I found more than a couple of cm of wire prior to the network resulted in what I thought was audible degradation. Peter connected his directly at the digital out on the dsp chip, I did not do so because I wanted to avoid the possibility of mechanical damage to my boards.
Depending on the timing of the reflections they may contribute to aperture uncertainty at the receiver which cannot determine the exact moment the transition from one state to the other actually occurs. (The cause of some/most? of the jitter potentially present in a spdif signal.)
This is by no means the entire story, I would look for more information on wikipedia about transmission line theory and intersymbol interference in digital data transmission.
Dig far back, I've posted quite a lot on this subject in this thread.
The optimum resistor values are not necessarily the 300/100 cited often here because of the dsp digital out source voltage. I listed them long ago as just one option. They are a good choice. I use 392 and 93.1 ohms.. If you are fussy it is a good idea to measure the resistors and select ones closest to their nominal values. (1% standard E-96 series parts - mine are various Caddock mk-132 types)
A 75 ohm link will have a 75 ohm source, connectors all specified to maintain a 75 ohm impedance over the range of frequencies of interest, cables with a characteristic impedance of 75 ohms and a terminating impedance of 75 ohms.
Every discontinuity in the path results in some of the energy being reflected back to the transmitting end where it is once again reflected due to an imperfect termination. The same thing happens at each connection along the way and at the final termination as well. Well designed transmission lines will have very small percentages of reflected signal.
RCA jacks are not 75 ohm devices (caution there are some reputedly 75 ohm RCA jacks, very expensive and I have not tried them) for a number of reasons including the pin diameter, the dielectric used, and the distance/spacing to the outer conductor of the jack.
You want to minimize the number of discontinuities in the path and make sure all connections maintain the proper impedance, which incidentally with most devices is going to be impossible. The spdif connection point on the end of the shigaclone is not 75 ohms characteristic impedance, and arguably is far enough from the output port of the dsp for reflections to be a potential issue. (Reflections before the cable matching network.) Mine are mounted on a connector that plugs into the socket, and I found more than a couple of cm of wire prior to the network resulted in what I thought was audible degradation. Peter connected his directly at the digital out on the dsp chip, I did not do so because I wanted to avoid the possibility of mechanical damage to my boards.
Depending on the timing of the reflections they may contribute to aperture uncertainty at the receiver which cannot determine the exact moment the transition from one state to the other actually occurs. (The cause of some/most? of the jitter potentially present in a spdif signal.)
This is by no means the entire story, I would look for more information on wikipedia about transmission line theory and intersymbol interference in digital data transmission.
Dig far back, I've posted quite a lot on this subject in this thread.
The optimum resistor values are not necessarily the 300/100 cited often here because of the dsp digital out source voltage. I listed them long ago as just one option. They are a good choice. I use 392 and 93.1 ohms.. If you are fussy it is a good idea to measure the resistors and select ones closest to their nominal values. (1% standard E-96 series parts - mine are various Caddock mk-132 types)
Frankly it is interesting that anyone chooses 75 ohms as a standard source or input impedance for modern products. There seems to be relatively little standard test equipment with 75 ohm input impedance even as an option, most scopes for example are 50 ohms and/or 1M.. Most network, tdrs, and spectrum analyzers are also 50 ohm..
I work for a relatively large electronics firm with large labs, and almost nothing in our lab pool is 75 ohms.. hmmm...
You can get 75 ohm terminators for your scope or even make them if necessary, but a scope only provides good information on observable characteristics like rise time, amplitude, and waveform fidelity (eye diagram) beyond that you need something like one of the devices Wenzel makes to measure jitter. Most tdrs don't work well over very short distances making it hard to check the integrity of the path...
This might all be moot with a really good receiver IC like the WM8804 or in a case where asynchronous resampling is being used with a very stable local clock.
Thread I was talking about in the previous post was Peter's Shigaclone thread.. FYI Too late to edit..
I work for a relatively large electronics firm with large labs, and almost nothing in our lab pool is 75 ohms.. hmmm...
You can get 75 ohm terminators for your scope or even make them if necessary, but a scope only provides good information on observable characteristics like rise time, amplitude, and waveform fidelity (eye diagram) beyond that you need something like one of the devices Wenzel makes to measure jitter. Most tdrs don't work well over very short distances making it hard to check the integrity of the path...
This might all be moot with a really good receiver IC like the WM8804 or in a case where asynchronous resampling is being used with a very stable local clock.
Thread I was talking about in the previous post was Peter's Shigaclone thread.. FYI Too late to edit..
It is quite hard to lose bits with SPDIF, it can still work even with rather horrible signal quality... (the quality of the recovered clock is another matter of course).
I wonder about the relevance of worrying about cables when your precious signal comes out of a big fat noisy VLSI chip like that, those things can create really huge amounts of jitter. Unless you slave the transport, then it doesn't matter, of course.
kevinkr said:
I wonder about the relevance of worrying about cables when your precious signal comes out of a big fat noisy VLSI chip like that, those things can create really huge amounts of jitter. Unless you slave the transport, then it doesn't matter, of course.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.